[{"id":"sync-wsd-h20--0.008--0.9--0.9--s42","configuration":"sync-wsd-h20--0.008--0.9--0.9","method":"sync","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.008,"seed":42,"loss":3.577678713712776,"jobId":"2450844","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"fc01b936f4e98df37aff5e4aeab1508053ac81ec703b37e41af6da964b9ae854","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.9--0.9--s43","configuration":"sync-wsd-h20--0.008--0.9--0.9","method":"sync","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.008,"seed":43,"loss":3.5845053023134006,"jobId":"2450845","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"ed25802775448a9182f4c27018093fc501d9a8eca23b962407143dfd5eda81de","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.9--0.9--s44","configuration":"sync-wsd-h20--0.008--0.9--0.9","method":"sync","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.008,"seed":44,"loss":3.5951866755998427,"jobId":"2450846","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"9d0cfb812e923b3af62f9d34ba3ba3ed96c81c26343cddd42b3869bc5f889cca","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.9--0.9--s42","configuration":"sync-wsd-h20--0.006--0.9--0.9","method":"sync","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.006,"seed":42,"loss":3.5710420365123547,"jobId":"2450847","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"b22e5f7bdba2f6a49db4e0e688c9ecf1746f36a7fe25a7ea6fe596776e902e58","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.9--0.9--s43","configuration":"sync-wsd-h20--0.006--0.9--0.9","method":"sync","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.006,"seed":43,"loss":3.580096477316119,"jobId":"2450848","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"bc8307be1698222ea388d9ac28c301d18f0194ac1f1f13c14d20057bd392d328","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.9--0.9--s44","configuration":"sync-wsd-h20--0.006--0.9--0.9","method":"sync","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.006,"seed":44,"loss":3.5732184681253227,"jobId":"2450849","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"0904b7b84d0382b171cbc6f3d7b9f6bf50775b2531a830606c24ba4611720f06","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.004--0.9--0.9--s42","configuration":"sync-wsd-h20--0.004--0.9--0.9","method":"sync","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":42,"loss":3.584592103854237,"jobId":"2450850","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"54d6dc8e326ba85d05f0aaae6980ce6c64c251fe6a417dce217ef8375564edd9","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.004--0.9--0.9--s43","configuration":"sync-wsd-h20--0.004--0.9--0.9","method":"sync","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":43,"loss":3.5795696503595553,"jobId":"2450851","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"535472da4b118a09df64401538c9c3bd0d6d6feca8c65148300a2a95ad61f525","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.004--0.9--0.9--s44","configuration":"sync-wsd-h20--0.004--0.9--0.9","method":"sync","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":44,"loss":3.5820670418408516,"jobId":"2450852","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"5aa9ba7ee236df30fa8bec4ddd8c4262025c68c21fe5e626ebad2be6aad5fef7","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.9--0.9--s42","configuration":"sync-wsd-h20--0.003--0.9--0.9","method":"sync","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.590969326333108,"jobId":"2450853","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"800d011d12404717311dbeb5c3a4558cbf0faac0f40e68a3f656ddb1856bc9e5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.9--0.9--s43","configuration":"sync-wsd-h20--0.003--0.9--0.9","method":"sync","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.581787281399286,"jobId":"2450854","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"ec856bd6402e2c28c29059d28d70c3c70786159194908429a57ea1dc199a4f39","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.9--0.9--s44","configuration":"sync-wsd-h20--0.003--0.9--0.9","method":"sync","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.5916640382110243,"jobId":"2450855","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"3b48d7bcf540e9099ad1d9331636ad6d2f1ff870131ed75d9a45ad9ddc7ce6cb","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.9--0.9--s42","configuration":"sync-wsd-h20--0.002--0.9--0.9","method":"sync","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.612617613381496,"jobId":"2450856","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"9b5b93607e0d54dcd5fa03be06d5e503477b4d69e007b387d9286519a324d77a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.9--0.9--s43","configuration":"sync-wsd-h20--0.002--0.9--0.9","method":"sync","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.595483762233382,"jobId":"2450857","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"fb58acd1e3e8a31c52ae65d7659745e904a95f38da1c39b562959add19f17e9b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.9--0.9--s44","configuration":"sync-wsd-h20--0.002--0.9--0.9","method":"sync","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.599385396911696,"jobId":"2450858","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"7e33810613fa5bcbf16e5236d958221dc951db67b3b27e536a9ba77ef1c41851","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.9--0.98--s42","configuration":"sync-wsd-h20--0.008--0.9--0.98","method":"sync","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.008,"seed":42,"loss":3.571040624847379,"jobId":"2450859","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"63edcc2ae056f61987339bcdb23b26553fd49d83d10ffb9b45321245defb7335","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.9--0.98--s43","configuration":"sync-wsd-h20--0.008--0.9--0.98","method":"sync","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.008,"seed":43,"loss":3.5802491183136085,"jobId":"2450860","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"f1b144ba701426367f06e44590133b145874c8f4bb952a46ae0218b94454b139","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.9--0.98--s44","configuration":"sync-wsd-h20--0.008--0.9--0.98","method":"sync","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.008,"seed":44,"loss":3.569550507584654,"jobId":"2450861","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"8a60ea464f96f48217d7ca97e639d3c14c586d4af069b844be63a236763bd42e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.9--0.98--s42","configuration":"sync-wsd-h20--0.006--0.9--0.98","method":"sync","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.006,"seed":42,"loss":3.568652845270944,"jobId":"2450862","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"df1cda8f98b14f62dbb428fd1201a7fd78ca03e55698014a1340ee87a98b417a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.9--0.98--s43","configuration":"sync-wsd-h20--0.006--0.9--0.98","method":"sync","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.006,"seed":43,"loss":3.5643524773799378,"jobId":"2450863","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"c55f4a894e3ed8389e4f252e078bb5e0c8219d38f17e95c01fb689fe77d5903c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.9--0.98--s44","configuration":"sync-wsd-h20--0.006--0.9--0.98","method":"sync","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.006,"seed":44,"loss":3.573471836063833,"jobId":"2450864","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"51658740bb7cb3d9499bf5686151fb5beb1cfe3828871f7628594c1bbbad1231","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.004--0.9--0.98--s42","configuration":"sync-wsd-h20--0.004--0.9--0.98","method":"sync","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":42,"loss":3.5599536061188073,"jobId":"2449510","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"fbfac5d6f7fde07f4d3a17d75cec051ff6bff880a43c9bad8b2b816e955171b4","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s42/attempt-003","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s42/attempt-003/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.004--0.9--0.98--s43","configuration":"sync-wsd-h20--0.004--0.9--0.98","method":"sync","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":43,"loss":3.5779362573653644,"jobId":"2450865","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"244a0a4f330887756c78df3e67dd685586c23e8576d01dfcf90e908a03d10f1c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.004--0.9--0.98--s44","configuration":"sync-wsd-h20--0.004--0.9--0.98","method":"sync","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":44,"loss":3.576087391373635,"jobId":"2450866","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"460b5b352c7281894b7d5e954b964259a6982885dc908b1291237dd3f3a33ccd","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.9--0.98--s42","configuration":"sync-wsd-h20--0.003--0.9--0.98","method":"sync","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.584645328549556,"jobId":"2450867","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"dbec4bad17a92472c0d1f38ce46ee688df78d35aeb09c20233b9b3365b2163d8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.9--0.98--s43","configuration":"sync-wsd-h20--0.003--0.9--0.98","method":"sync","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.5786623300670466,"jobId":"2450868","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"e9afde9cbafe9533ae8a758d741bac74fc0a95201f617946ec567ebd5b076a11","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.9--0.98--s44","configuration":"sync-wsd-h20--0.003--0.9--0.98","method":"sync","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.5863996854865703,"jobId":"2450869","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"f64cefe8c9e4d238872951fa098619e49fd72c6e37f2a12ebf119beffd5031af","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.9--0.98--s42","configuration":"sync-wsd-h20--0.002--0.9--0.98","method":"sync","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.594647453114633,"jobId":"2450870","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"bfe6f702a4e125e905088e5864c21d042176fde600002e5632562c7904149cae","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.9--0.98--s43","configuration":"sync-wsd-h20--0.002--0.9--0.98","method":"sync","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.5819512454868043,"jobId":"2450871","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"59fc08d263067e76cfda5d21298061c03a966803c1a315d06147b737d2a6ccc7","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.9--0.98--s44","configuration":"sync-wsd-h20--0.002--0.9--0.98","method":"sync","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.5940430049418595,"jobId":"2450872","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"30d49c03e5e0ecc9c431f49c252bbf98896d26e41f136d22d61b391045880886","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.9--0.999--s42","configuration":"sync-wsd-h20--0.008--0.9--0.999","method":"sync","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.008,"seed":42,"loss":4.278052927002443,"jobId":"2450873","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"8e83a979a769f054a02a9de0d7970b8f06c9f1d22b6c384c69a83b29c5f60f03","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.9--0.999--s43","configuration":"sync-wsd-h20--0.008--0.9--0.999","method":"sync","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.008,"seed":43,"loss":3.6021826013279674,"jobId":"2450874","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"f8ad393caed352eef89c3fba33e927c768bcf321412fb82c50dad78fefeabb6d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.9--0.999--s44","configuration":"sync-wsd-h20--0.008--0.9--0.999","method":"sync","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.008,"seed":44,"loss":3.5910854813113016,"jobId":"2450875","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"dd5d5db2cb610c1ad787eb008c10841dd14de51d2520ced26cacef7e15e2393c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.9--0.999--s42","configuration":"sync-wsd-h20--0.006--0.9--0.999","method":"sync","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.006,"seed":42,"loss":3.576374475218168,"jobId":"2450876","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"55fcc489cc29e17130ad11cba95393adc475b8ad6524dd9effb94485bde52e76","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.9--0.999--s43","configuration":"sync-wsd-h20--0.006--0.9--0.999","method":"sync","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.006,"seed":43,"loss":3.5731031750731845,"jobId":"2450877","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"1d55fff59934098db9320882537ea06b51eb205545e79c44d4cf3328a94852c1","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.9--0.999--s44","configuration":"sync-wsd-h20--0.006--0.9--0.999","method":"sync","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.006,"seed":44,"loss":3.644299475739938,"jobId":"2450878","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"0599194c104e352d6d125f58770830c04fc142c0d7f55bf36d2d91e725551229","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.004--0.9--0.999--s42","configuration":"sync-wsd-h20--0.004--0.9--0.999","method":"sync","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":42,"loss":3.5796302724512543,"jobId":"2450879","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"9b28a46ecb73b427269826507c2f3e3f167db0f5ce2b5bfa5d58cd857f4732e0","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.004--0.9--0.999--s43","configuration":"sync-wsd-h20--0.004--0.9--0.999","method":"sync","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":43,"loss":3.5838644489429505,"jobId":"2450880","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"1a0820bc550260835a8d2ff245ed6f7262ccc130f6a3293a712783e94501dffd","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.004--0.9--0.999--s44","configuration":"sync-wsd-h20--0.004--0.9--0.999","method":"sync","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":44,"loss":3.575918023174999,"jobId":"2450881","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"9ed724fd6d66dc1d270cd7502700ffa9b3c6ce23592cbbd592f1cd0f492995fe","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.9--0.999--s42","configuration":"sync-wsd-h20--0.003--0.9--0.999","method":"sync","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.574305882349551,"jobId":"2450882","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"da53fca47fea2be4fa03b06a5ab1d2a3c1be3fc854325cf0ef670499bc7dc593","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.9--0.999--s43","configuration":"sync-wsd-h20--0.003--0.9--0.999","method":"sync","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.6011489677024087,"jobId":"2450883","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"840287bd4ea4758f204c763cae7c0075e49ca3944c801fa18c6f60d883d4104b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.9--0.999--s44","configuration":"sync-wsd-h20--0.003--0.9--0.999","method":"sync","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.6022701932327457,"jobId":"2450884","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"5aaaf035810427de61588c10be5a435509199a408ba6f031ca30b99aef4b335f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.9--0.999--s42","configuration":"sync-wsd-h20--0.002--0.9--0.999","method":"sync","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.6048555121924224,"jobId":"2450885","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"5ef8ad14a4a8cb6eb89a7f625c7e3b71a5a0bb4e5c40a32ca0881234d8fd3709","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.9--0.999--s43","configuration":"sync-wsd-h20--0.002--0.9--0.999","method":"sync","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.5988897075367965,"jobId":"2450886","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"0f8cfc9dc229ba0cea536e36ba09d8c5cba7bbb5a6bb2b340c1b571ec1316454","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.9--0.999--s44","configuration":"sync-wsd-h20--0.002--0.9--0.999","method":"sync","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.5969445118891783,"jobId":"2450887","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"f92dc85f08c6a7a690497a280d6a347088489e34cc7a454e352a22731b12005e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.95--0.9--s42","configuration":"sync-wsd-h20--0.008--0.95--0.9","method":"sync","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.008,"seed":42,"loss":3.5809341788065727,"jobId":"2450888","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"b2d2b18b3ffcffe75d9619dfd196ef24b83fb3dbb46a00c4033e0d6b698b38f6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.95--0.9--s43","configuration":"sync-wsd-h20--0.008--0.95--0.9","method":"sync","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.008,"seed":43,"loss":3.5790427700718372,"jobId":"2450889","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"78b041759ea45deedd561a1ba47783100649b4265a455c0e06a3d59a8dc47d18","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.95--0.9--s44","configuration":"sync-wsd-h20--0.008--0.95--0.9","method":"sync","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.008,"seed":44,"loss":3.583277486065434,"jobId":"2450890","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"abef7acb18acfdefef548e8ab6f9845c41a931401dda81624122727f07cbffe6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.95--0.9--s42","configuration":"sync-wsd-h20--0.006--0.95--0.9","method":"sync","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.006,"seed":42,"loss":3.569095058628378,"jobId":"2450891","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"23caf036a37e08187fd15cc37a6ae74b8654817763d6bf4f92c6e6194fdf85bc","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.95--0.9--s43","configuration":"sync-wsd-h20--0.006--0.95--0.9","method":"sync","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.006,"seed":43,"loss":3.577184415113812,"jobId":"2452072","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"1f44e2d5a3a3a75cf74db9e8248bc6d0eaeb94c506f7f67214683896c8210e9c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.95--0.9--s44","configuration":"sync-wsd-h20--0.006--0.95--0.9","method":"sync","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.006,"seed":44,"loss":3.571636840380292,"jobId":"2452420","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"1d4fbaee29ccc1cb6b7e860619c5e8d1ae2082224c2e7e84ac489ca86eeae96e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.004--0.95--0.9--s42","configuration":"sync-wsd-h20--0.004--0.95--0.9","method":"sync","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":42,"loss":3.56448257767416,"jobId":"2452421","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"f60d722685c2a4866dff4e16586845a0b55cb1a00e1ea8e3a4f5fe2826515360","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.004--0.95--0.9--s43","configuration":"sync-wsd-h20--0.004--0.95--0.9","method":"sync","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":43,"loss":3.5674018363182567,"jobId":"2452422","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"8c1e550e2f53d514b1088f13b06292b79eb7b5384915a73b24eef0db2d2abb40","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.004--0.95--0.9--s44","configuration":"sync-wsd-h20--0.004--0.95--0.9","method":"sync","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":44,"loss":3.570829121158611,"jobId":"2452423","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"eeab317f65d1d74056a9937d793fb495a528032fad1ec3126d2a18e40a7059f2","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.95--0.9--s42","configuration":"sync-wsd-h20--0.003--0.95--0.9","method":"sync","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.565451243638595,"jobId":"2452424","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"5dd5fc2cc428fdf8761089f84a14dee1b29e32069a6700a9b190b58bd610b72f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.95--0.9--s43","configuration":"sync-wsd-h20--0.003--0.95--0.9","method":"sync","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.5748634036493634,"jobId":"2452426","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"584e34cc76aa15b70badff5da687df642cfc1762cf9ee0253038378e5df887ba","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.95--0.9--s44","configuration":"sync-wsd-h20--0.003--0.95--0.9","method":"sync","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.5736172870777962,"jobId":"2452427","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"b516738123645649eb78cc3175378ac7fbabd40c5a9bb118a3799d50946a217f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.95--0.9--s42","configuration":"sync-wsd-h20--0.002--0.95--0.9","method":"sync","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.5702033761528265,"jobId":"2452428","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"bbc5110848251cc53f8993ea9be4ce46edfd927e7ccdee4cc9e1fc84aefc11c1","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.95--0.9--s43","configuration":"sync-wsd-h20--0.002--0.95--0.9","method":"sync","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.577544288226398,"jobId":"2452429","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"d0cccdc814580629bc2ec3b69fa2fe573ec3c07cbd04d49a5b0cbd56917c4d22","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.95--0.9--s44","configuration":"sync-wsd-h20--0.002--0.95--0.9","method":"sync","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.580599891553157,"jobId":"2452430","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"c7a99c746d212c53e6e2dc86c00f3a44d116b13047529f88d2671db41da0adc0","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.95--0.98--s42","configuration":"sync-wsd-h20--0.008--0.95--0.98","method":"sync","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.008,"seed":42,"loss":3.567332756965113,"jobId":"2452431","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"a4cd01732ded3495fc208c7edb0aae9386e7699d42e181edc6d60f4b55d01841","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.95--0.98--s43","configuration":"sync-wsd-h20--0.008--0.95--0.98","method":"sync","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.008,"seed":43,"loss":3.5687355655742645,"jobId":"2452432","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"dd1222ca36bb2aec2958e14601bdec6fe636f23385c464c49e655f64702319f7","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.95--0.98--s44","configuration":"sync-wsd-h20--0.008--0.95--0.98","method":"sync","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.008,"seed":44,"loss":3.5693477999073466,"jobId":"2452433","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"639452fd324355f5e402d15f0d912abe67c58a4966afd642ade77301c70216cc","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.95--0.98--s42","configuration":"sync-wsd-h20--0.006--0.95--0.98","method":"sync","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.006,"seed":42,"loss":3.559489830036192,"jobId":"2452434","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"eb8cf61827cce2d9a57acee0a4ae16026112d8bb321d19efa7e1681c91c243ba","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.95--0.98--s43","configuration":"sync-wsd-h20--0.006--0.95--0.98","method":"sync","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.006,"seed":43,"loss":3.564557421940719,"jobId":"2452439","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"599f48ff3fffe43f8e1f0c560f21db5fae0a785d24e31cb0328d6de34983b376","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.95--0.98--s44","configuration":"sync-wsd-h20--0.006--0.95--0.98","method":"sync","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.006,"seed":44,"loss":3.565673217273487,"jobId":"2454799","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"6d450712e6ab215c4d8e711df72b90edba1d801e2f668aadeb9da64a15634512","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s44/attempt-002","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s44/attempt-002/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.004--0.95--0.98--s42","configuration":"sync-wsd-h20--0.004--0.95--0.98","method":"sync","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":42,"loss":3.5544962947525507,"jobId":"2452443","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"b9917fd47aa835a95becb81e8b319d1714035d318f9ce084d668252c8390a7e7","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/final.pt"},{"id":"sync-wsd-h20--0.004--0.95--0.98--s43","configuration":"sync-wsd-h20--0.004--0.95--0.98","method":"sync","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":43,"loss":3.557062698977349,"jobId":"2452447","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"d7d0f17990cee0856327fd6bf5ab3e82806bfbaedcb5ae4040c33a1a701dd38e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/final.pt"},{"id":"sync-wsd-h20--0.004--0.95--0.98--s44","configuration":"sync-wsd-h20--0.004--0.95--0.98","method":"sync","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":44,"loss":3.5585373031785137,"jobId":"2452449","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"f47d72255d99d1c063cd9af85acc3fdd7f7b5ef6c4f5d626bc5928d0463506f6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/final.pt"},{"id":"sync-wsd-h20--0.003--0.95--0.98--s42","configuration":"sync-wsd-h20--0.003--0.95--0.98","method":"sync","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.558380087390087,"jobId":"2452450","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"3806f8012353f95b959acf9f17d705bbb4d0d7d80389974fc4fbffb9377f2514","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.95--0.98--s43","configuration":"sync-wsd-h20--0.003--0.95--0.98","method":"sync","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.560115019637252,"jobId":"2452451","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"bc727f09b952ac9ff11ecda9173bde1fb5d21cde47597b56424071a8e05ea15f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.95--0.98--s44","configuration":"sync-wsd-h20--0.003--0.95--0.98","method":"sync","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.563765453203556,"jobId":"2452452","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"5684cc969d18e7a8638015b7bc30cf72355ba58c28cc373c0ab30844c5bb298a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.95--0.98--s42","configuration":"sync-wsd-h20--0.002--0.95--0.98","method":"sync","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.567208347256614,"jobId":"2452453","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"00f7c46b70aab600be222a583e855cbf1235ab9f33b0d3e7a1ca4ab2957723b2","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.95--0.98--s43","configuration":"sync-wsd-h20--0.002--0.95--0.98","method":"sync","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.576001709655114,"jobId":"2452454","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"1ffa38904f628ed9d48d0eecc32086871441d0c2609b959096f5cd2bfd206838","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.95--0.98--s44","configuration":"sync-wsd-h20--0.002--0.95--0.98","method":"sync","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.5661558365932238,"jobId":"2452455","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"916bf5b877d6d81347b1922c7a4c7180398da5f1bc34b4a7acd01dfa50a4f5ee","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.95--0.999--s42","configuration":"sync-wsd-h20--0.008--0.95--0.999","method":"sync","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.008,"seed":42,"loss":3.6366870581997817,"jobId":"2452456","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"02b026372048ff204e10fc309c8c53e3b64c8f47d4dc828b6f93bb238d6bdfd6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.95--0.999--s43","configuration":"sync-wsd-h20--0.008--0.95--0.999","method":"sync","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.008,"seed":43,"loss":3.592016682530186,"jobId":"2452457","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"1d95394fa5d954dce4cbd6e7dadd6e7af12122c7a38fe96273859c92071b4077","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.008--0.95--0.999--s44","configuration":"sync-wsd-h20--0.008--0.95--0.999","method":"sync","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.008,"seed":44,"loss":3.583236902938327,"jobId":"2452458","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"015e5851ae29a63d5e14b64ba722045f84297d4f82735adb9a52ee280e68395f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.95--0.999--s42","configuration":"sync-wsd-h20--0.006--0.95--0.999","method":"sync","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.006,"seed":42,"loss":3.5781384521506574,"jobId":"2452459","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"878e08fa30a42fae12d1d483c88093b533381fbd5cc38c52f68681f0e9d2b5b8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.95--0.999--s43","configuration":"sync-wsd-h20--0.006--0.95--0.999","method":"sync","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.006,"seed":43,"loss":3.579306858623379,"jobId":"2452460","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"46a3f8b4ddeb3481a2c7b935695194ffe84798d081e948e2916aad37226d0a79","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.006--0.95--0.999--s44","configuration":"sync-wsd-h20--0.006--0.95--0.999","method":"sync","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.006,"seed":44,"loss":3.5734827982720696,"jobId":"2452461","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"185dc45642683098ce2e7cf393a232b78825a31abcea2fcb5829b43b1605e158","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.004--0.95--0.999--s42","configuration":"sync-wsd-h20--0.004--0.95--0.999","method":"sync","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":42,"loss":3.5627517779599316,"jobId":"2452462","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"7a9def965aa676ddaebc4af68bfc07c3d891886cc1a5220273a310e1ee41a002","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.004--0.95--0.999--s43","configuration":"sync-wsd-h20--0.004--0.95--0.999","method":"sync","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":43,"loss":3.568954754386764,"jobId":"2452463","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"1839067f3a053254a27d3fe93aebaaa467b0c97b8ba7c19bc95d00f72311489b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.004--0.95--0.999--s44","configuration":"sync-wsd-h20--0.004--0.95--0.999","method":"sync","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":44,"loss":3.5696288155343874,"jobId":"2452465","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"012b311a4fbb439c0aa5f9383dc99ceed882ddc80e03145aa2aac5b23f4ce6ce","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.95--0.999--s42","configuration":"sync-wsd-h20--0.003--0.95--0.999","method":"sync","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.56926195266135,"jobId":"2452466","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"32b8adddfc7344bb26b90362c7c9a892c4e4270380db2e0f83469989e35e9d76","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.95--0.999--s43","configuration":"sync-wsd-h20--0.003--0.95--0.999","method":"sync","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.5674922862088256,"jobId":"2452467","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"6472bcff2340dab0523131708cf3d0caed47268831ab992701149e32eb39dad4","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.003--0.95--0.999--s44","configuration":"sync-wsd-h20--0.003--0.95--0.999","method":"sync","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.570816793700776,"jobId":"2452468","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"dee49cf1f9785a4bf1d721580686ac9814141590e8282877ca95703b3dd8842b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.95--0.999--s42","configuration":"sync-wsd-h20--0.002--0.95--0.999","method":"sync","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.5726546033636044,"jobId":"2452469","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"9faa0f5892a72d47793cfc6cdf09077681c9d681ef1b4f0717930aa2a1518b49","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.95--0.999--s43","configuration":"sync-wsd-h20--0.002--0.95--0.999","method":"sync","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.578716270790343,"jobId":"2454800","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"959f20efd4d07b81c4ff1d15147b0466e5d7457c110e7d9cdb764d88703367d3","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s43/attempt-002","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s43/attempt-002/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h20--0.002--0.95--0.999--s44","configuration":"sync-wsd-h20--0.002--0.95--0.999","method":"sync","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.5747387670841677,"jobId":"2452471","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"07922fc5b47e3f7e2964e4c181fccec5ce2e575fabba8a8497e1657dee05e8bb","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.004--0.9--0.9--s42","configuration":"sync-wsd-h40--0.004--0.9--0.9","method":"sync","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":42,"loss":3.4755529848956312,"jobId":"2456190","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"302c51f95f86bf4f3afae463701ed00417f7ed6082456399f91ddd3d50266493","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.004--0.9--0.9--s43","configuration":"sync-wsd-h40--0.004--0.9--0.9","method":"sync","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":43,"loss":3.4788332237586945,"jobId":"2456193","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"d639aaadf01a6c3fdf9501925d52e937a7f30f2de708c7404ece5a0c6340bf72","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.004--0.9--0.9--s44","configuration":"sync-wsd-h40--0.004--0.9--0.9","method":"sync","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":44,"loss":3.4777120132638637,"jobId":"2456200","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"f936f71bc4dc4210dc4d132b253b4a2ef3ae12986fb1d05de8133806e4d9e5eb","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.9--0.9--s42","configuration":"sync-wsd-h40--0.003--0.9--0.9","method":"sync","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.4843115747706386,"jobId":"2456205","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"4b12e59aa6a342f251b89ef63ead8a8e1ba5e92cf88d70b6db8346d68137769c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.9--0.9--s43","configuration":"sync-wsd-h40--0.003--0.9--0.9","method":"sync","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.476859046660619,"jobId":"2456206","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"c8132a16f74661dd98af3f377f51f5a9bb4d1a7f343d95ded98a5d8c515505f8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.9--0.9--s44","configuration":"sync-wsd-h40--0.003--0.9--0.9","method":"sync","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.4856309002882964,"jobId":"2456207","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"8b583b6c0fe88384b694511e6bef89c5ce1fdc7815e8e22fe4386a6233792596","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.9--0.9--s42","configuration":"sync-wsd-h40--0.002--0.9--0.9","method":"sync","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.496402676426997,"jobId":"2456208","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"60b6f06523e461bddefbf4847990b2724cc01024a051634648a79d558a74dbd3","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.9--0.9--s43","configuration":"sync-wsd-h40--0.002--0.9--0.9","method":"sync","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.484004256418404,"jobId":"2456209","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"244052dab9201ece5211209c2bbe18700cae4838a91b091e4a1b313c2d51f79b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.9--0.9--s44","configuration":"sync-wsd-h40--0.002--0.9--0.9","method":"sync","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.4881544762436234,"jobId":"2456210","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"1c052013a580b35d21b6f4c74b6206d9b04e43be7c80f95448476aca1eb5ff0a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.004--0.9--0.98--s42","configuration":"sync-wsd-h40--0.004--0.9--0.98","method":"sync","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":42,"loss":3.4548085307638963,"jobId":"2456211","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"2f6405e8f779bfad9dc3b2579c9bc7231f7546e066268c8cfe50e51c41a41344","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.004--0.9--0.98--s43","configuration":"sync-wsd-h40--0.004--0.9--0.98","method":"sync","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":43,"loss":3.4679864698303144,"jobId":"2456213","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"8b8e9fcc0dcf358804bcfc043a21c1f43474cdb9c368077503826d23008ad177","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.004--0.9--0.98--s44","configuration":"sync-wsd-h40--0.004--0.9--0.98","method":"sync","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":44,"loss":3.4673522786949325,"jobId":"2456214","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"97c6639848cb4eb4b38489a677a87b5535bfdcf6f4ebd56f824f3d6d39efeed3","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.9--0.98--s42","configuration":"sync-wsd-h40--0.003--0.9--0.98","method":"sync","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.473158218724192,"jobId":"2456215","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"c7b387838d2b2d324bf3efb81059fde2d66709c61b4e7b495a211eb58866f6a6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.9--0.98--s43","configuration":"sync-wsd-h40--0.003--0.9--0.98","method":"sync","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.4681682661046906,"jobId":"2456216","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"8d5ebf9ff14db86a803c1b7294672584ea1b8d41cdd62811b5125b477f1dbc4d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.9--0.98--s44","configuration":"sync-wsd-h40--0.003--0.9--0.98","method":"sync","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.475472607358183,"jobId":"2456217","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"3888794ae6a812a624d32d79cbe9b18a96408624bc711130e4ae7e4a7ec13370","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.9--0.98--s42","configuration":"sync-wsd-h40--0.002--0.9--0.98","method":"sync","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.479317315960111,"jobId":"2456218","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"418c4ef9f43f2bde0daa6ee28eee9cb6da3a015a126564844aea51931abf203b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.9--0.98--s43","configuration":"sync-wsd-h40--0.002--0.9--0.98","method":"sync","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.4667723287160865,"jobId":"2456219","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"2c1360c99b9dc7b32dcbe5a4e8ad4b703ff086f008fb45a44ce9425071c2d45f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.9--0.98--s44","configuration":"sync-wsd-h40--0.002--0.9--0.98","method":"sync","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.478118580700111,"jobId":"2456220","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"6c46b6f57ecc24b4c5ddf4565637d0891e5948119b0b330576f011545ccf9da3","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.004--0.9--0.999--s42","configuration":"sync-wsd-h40--0.004--0.9--0.999","method":"sync","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":42,"loss":3.4735047162075072,"jobId":"2456221","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"6b3197b0c883a8e9a85ecf7bf586da7fe161c9b1438367331756abf7789e9f99","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.004--0.9--0.999--s43","configuration":"sync-wsd-h40--0.004--0.9--0.999","method":"sync","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":43,"loss":3.492706443056062,"jobId":"2456269","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"0513547ebe96c2504ecf688bebccbcbae61fb1453e075f6d64b185ed0742e177","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.004--0.9--0.999--s44","configuration":"sync-wsd-h40--0.004--0.9--0.999","method":"sync","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":44,"loss":3.470226172025489,"jobId":"2456270","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"defd4aa99cf6d2ac9c12da1efe0f74b5d85898ed15457b9c99c7aefc69273870","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.9--0.999--s42","configuration":"sync-wsd-h40--0.003--0.9--0.999","method":"sync","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.4631896312463604,"jobId":"2456271","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"d2c8414f2fb82e9a7d1cfc92746d4582a5e8776ba107a6a7a5757db70c9fcf79","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.9--0.999--s43","configuration":"sync-wsd-h40--0.003--0.9--0.999","method":"sync","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.476197257843589,"jobId":"2456272","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"f0e5394c389af0a7893f5826b180ff9a4ac9451394c4337712941a271008aa23","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.9--0.999--s44","configuration":"sync-wsd-h40--0.003--0.9--0.999","method":"sync","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.4755693040708757,"jobId":"2456273","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"e1ae2184f4151d958ddc58d860b413f994203b60c09e8464f7ea7ae7f26372ff","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.9--0.999--s42","configuration":"sync-wsd-h40--0.002--0.9--0.999","method":"sync","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.4824460882651778,"jobId":"2456274","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"0c57936e6f97fb101036898590254a47bd259dad44b11ad31d242cf199dcaffa","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.9--0.999--s43","configuration":"sync-wsd-h40--0.002--0.9--0.999","method":"sync","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.4771503695215995,"jobId":"2456275","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"3adf90d7b05ed2015a29f06942899da4abbb5b2e574414a900e2deb0cd0c6b41","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.9--0.999--s44","configuration":"sync-wsd-h40--0.002--0.9--0.999","method":"sync","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.4767636819031935,"jobId":"2456276","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"5b08ed312d173a3a1101c0daf4c42b52217dbf1251ffee0dab557d4810cc09f9","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.004--0.95--0.9--s42","configuration":"sync-wsd-h40--0.004--0.95--0.9","method":"sync","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":42,"loss":3.457830407098489,"jobId":"2456277","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"adf17e69c1a1c29050d4096fe1baa27941863bfecc98bd511bd758f66f339c3f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.004--0.95--0.9--s43","configuration":"sync-wsd-h40--0.004--0.95--0.9","method":"sync","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":43,"loss":3.4625445275647917,"jobId":"2456278","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"645937ef2b90ed2f75dfb2df714cfac0b9740dbdcc58d708f40d1a1aa03f26a7","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.004--0.95--0.9--s44","configuration":"sync-wsd-h40--0.004--0.95--0.9","method":"sync","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":44,"loss":3.464490102473478,"jobId":"2456279","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"5ca8d1ca93ed3e3da00593aeff0086482f03c5516eeb05d6ee8640cf06b07b02","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.95--0.9--s42","configuration":"sync-wsd-h40--0.003--0.95--0.9","method":"sync","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.4577991555613536,"jobId":"2456280","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"1e3b203729b323ce163b556cc93b9bf71859766b6d473f576585d51757e52de0","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.95--0.9--s43","configuration":"sync-wsd-h40--0.003--0.95--0.9","method":"sync","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.4647956002771982,"jobId":"2456281","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"1efd5fd8b5b8265ed3d73eddbe407e93e07b68afd944e9742c1faf29274944b9","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.95--0.9--s44","configuration":"sync-wsd-h40--0.003--0.95--0.9","method":"sync","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.4664831987501215,"jobId":"2456282","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"6aa7d90527dd17afdb748b62d248cac692496cc0c06172f4b9eac09cf5a03d20","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.95--0.9--s42","configuration":"sync-wsd-h40--0.002--0.95--0.9","method":"sync","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.4613947454175875,"jobId":"2456283","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"24ddf086f12daaa72f6c8f5c97ea6dfe84a66a4da3c93b78b6dcc330d0097bde","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.95--0.9--s43","configuration":"sync-wsd-h40--0.002--0.95--0.9","method":"sync","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.4676512611340207,"jobId":"2456284","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"adc23815adf8e2b13a725e575b8a7056b28165e2d61b4cdd7254f364fcc4a55d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.95--0.9--s44","configuration":"sync-wsd-h40--0.002--0.95--0.9","method":"sync","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.4697875636146684,"jobId":"2456285","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"1d033d95cb950b45c886b6d270b7a529af33b21475ca791ad575dca7418be2a0","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.004--0.95--0.98--s42","configuration":"sync-wsd-h40--0.004--0.95--0.98","method":"sync","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":42,"loss":3.4492329075080965,"jobId":"2456286","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"9167733dbe45d80547253cc54cd9a7b1ba4f0facdf3f3727484d33a326c40d40","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/final.pt"},{"id":"sync-wsd-h40--0.004--0.95--0.98--s43","configuration":"sync-wsd-h40--0.004--0.95--0.98","method":"sync","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":43,"loss":3.4551217690214986,"jobId":"2456287","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"c63c8ab822ca0bb7bec296eec0626fe054965c5940b6d9c3e2608c18b3431017","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/final.pt"},{"id":"sync-wsd-h40--0.004--0.95--0.98--s44","configuration":"sync-wsd-h40--0.004--0.95--0.98","method":"sync","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":44,"loss":3.4540458228008246,"jobId":"2456288","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"8890d7ce98f1d566449596664442e1ff8e67813bb63dc1c358c6295fe180d24d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/final.pt"},{"id":"sync-wsd-h40--0.003--0.95--0.98--s42","configuration":"sync-wsd-h40--0.003--0.95--0.98","method":"sync","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.4501401313581113,"jobId":"2456289","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"0d7722d5675bffbbbbbdf58303ce9a27b4c8e729eba2569ce8520d20a5119ecc","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.95--0.98--s43","configuration":"sync-wsd-h40--0.003--0.95--0.98","method":"sync","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.454050054316837,"jobId":"2456290","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"b91d431e5dba0df6f9c35fcb7ce1bcff74510f55ea7379068ff3da70acc940f8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.95--0.98--s44","configuration":"sync-wsd-h40--0.003--0.95--0.98","method":"sync","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.4556786315073427,"jobId":"2456291","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"51f143dfb4573d23b4e7809e94e281c98bd55d8f3083d4e2c4bc77d7af0755e6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.95--0.98--s42","configuration":"sync-wsd-h40--0.002--0.95--0.98","method":"sync","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.453656531049864,"jobId":"2456292","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"e49d6a19595dad1d4fc245b8fdbb0779d4e8718ad0bc3745e29cba5ed0c711c5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.95--0.98--s43","configuration":"sync-wsd-h40--0.002--0.95--0.98","method":"sync","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.464871085681808,"jobId":"2456293","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"98ff93c4852447fae32a4bea110beddaabb15c63e0da884fbf217d8cf0547a9d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.95--0.98--s44","configuration":"sync-wsd-h40--0.002--0.95--0.98","method":"sync","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.45512636877735,"jobId":"2456294","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"fc2e9d5bfe6c344df81d884bcdc8e7ebe0556cddcac7c18b2408ca1b0630167c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.004--0.95--0.999--s42","configuration":"sync-wsd-h40--0.004--0.95--0.999","method":"sync","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":42,"loss":3.4520953789037487,"jobId":"2456295","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"e6a22cfe38823c9e46d2095a18739e777ff36b0ec1ca27289197e07c069c9007","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.004--0.95--0.999--s43","configuration":"sync-wsd-h40--0.004--0.95--0.999","method":"sync","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":43,"loss":3.4580368830143953,"jobId":"2456296","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"29f9dd585bca19eb140978c692c59850b239137ed186a41202361921b4532c25","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.004--0.95--0.999--s44","configuration":"sync-wsd-h40--0.004--0.95--0.999","method":"sync","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":44,"loss":3.4554055779772432,"jobId":"2456297","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"040a82ca6726124539b62558deb1a4303b18c050a7f42c4c96ea09ac768aed16","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.95--0.999--s42","configuration":"sync-wsd-h40--0.003--0.95--0.999","method":"sync","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.452309710920177,"jobId":"2457298","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"24cf9859fa78e12ea697974d6c43abd0570001d5926536c2e4da988c774c341a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.95--0.999--s43","configuration":"sync-wsd-h40--0.003--0.95--0.999","method":"sync","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.455872899370765,"jobId":"2457299","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"0584d7e3d6e23d96e4089284e7b144ed42e49ca3c66cf4b4ac21e73f79008ee0","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.003--0.95--0.999--s44","configuration":"sync-wsd-h40--0.003--0.95--0.999","method":"sync","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.455832290668222,"jobId":"2457300","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"23c73841b0d9e7d9869faf5dc75f6557a62959ff4161478fb981742d2573f48b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.95--0.999--s42","configuration":"sync-wsd-h40--0.002--0.95--0.999","method":"sync","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.4549467369817295,"jobId":"2457301","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"663afa44c4a430a0efaf6c95b14587475e07dd6d71cf66248e917090cdc2db12","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.95--0.999--s43","configuration":"sync-wsd-h40--0.002--0.95--0.999","method":"sync","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.4634011282309762,"jobId":"2457302","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"ceecd345386516d1361ae215c344b1c839b14a0c465dea3931948ff59b0676dc","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h40--0.002--0.95--0.999--s44","configuration":"sync-wsd-h40--0.002--0.95--0.999","method":"sync","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.4576668361233858,"jobId":"2457303","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"8fc3a65934a1ff2f23e2f2a3f23b262daa7d3faa5d8da474b746544dd2e186e8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.9--0.9--s42","configuration":"sync-wsd-h80--0.004--0.9--0.9","method":"sync","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":42,"loss":3.400883827208692,"jobId":"2458992","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"f740e816691f99767dc6700ec3ccbcfdc10205a4de19993549875899ef302535","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.9--0.9--s43","configuration":"sync-wsd-h80--0.004--0.9--0.9","method":"sync","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":43,"loss":3.404250131254967,"jobId":"2458993","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"367af306003c82095c639f04a380ed4f7cefb1348959e5eb3248e843e9df3cfe","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.9--0.9--s44","configuration":"sync-wsd-h80--0.004--0.9--0.9","method":"sync","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":44,"loss":3.4025661407826067,"jobId":"2458994","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"15b8c78ab97a784a7876c4c09c314ee7ed3707cd92a7291633d75ef50eb826f9","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.003--0.9--0.9--s42","configuration":"sync-wsd-h80--0.003--0.9--0.9","method":"sync","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.406432859902007,"jobId":"2458995","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"1cd13860e821c339a3e07df7e607bc3ef38a7afdb1f67850da89dd516d72b81e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.003--0.9--0.9--s43","configuration":"sync-wsd-h80--0.003--0.9--0.9","method":"sync","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.4009907702655466,"jobId":"2458996","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"43ce31887392f962466387fb450cfbde347d49c252662cfe1e74400ef3a115f5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.003--0.9--0.9--s44","configuration":"sync-wsd-h80--0.003--0.9--0.9","method":"sync","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.406536946355269,"jobId":"2458997","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"25f691cbff5df50ce00e22d786622d75a00855879a66c803c0e580ad37d3daa4","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.9--0.9--s42","configuration":"sync-wsd-h80--0.002--0.9--0.9","method":"sync","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.4138426364204353,"jobId":"2458998","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"c19be9d22819877d62c65ca64121218cd08372ce460640572eae2f527caa894b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.9--0.9--s43","configuration":"sync-wsd-h80--0.002--0.9--0.9","method":"sync","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.404958759804277,"jobId":"2458999","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"3b318251c23eac7a41e111b12a20e722cb594442edaeeaa3d62659832b66a6b0","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.9--0.9--s44","configuration":"sync-wsd-h80--0.002--0.9--0.9","method":"sync","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.408059286757931,"jobId":"2459000","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"78c12c9ad6509d3e0c3222bdb4ffde2f9b95c911ce8a671a5d7bdc4614cc3d4d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.9--0.98--s42","configuration":"sync-wsd-h80--0.004--0.9--0.98","method":"sync","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":42,"loss":3.3827230932190484,"jobId":"2459001","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"2509b5ea32ef0ffb0ebf005ceecc156f22cbe6d8f76d25c10686b3a0b244096f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.9--0.98--s43","configuration":"sync-wsd-h80--0.004--0.9--0.98","method":"sync","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":43,"loss":3.3892127433880943,"jobId":"2459002","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"6f5a699d87fd16d4dd161ae06dc8e50b1f139a4076e30be07feb3f23d548ee10","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.9--0.98--s44","configuration":"sync-wsd-h80--0.004--0.9--0.98","method":"sync","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":44,"loss":3.3905290388975198,"jobId":"2459003","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"2fa4ab3b5a62966b8d21fa0e18a544e3eabdb3e2b29ebdbc2c7c69964acd2fc3","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.003--0.9--0.98--s42","configuration":"sync-wsd-h80--0.003--0.9--0.98","method":"sync","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.3958272631829294,"jobId":"2459004","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"b90f9eedd876f34e9961dcdbbf1a06be1214867e3222bf82039facaec7d68cb7","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.003--0.9--0.98--s43","configuration":"sync-wsd-h80--0.003--0.9--0.98","method":"sync","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.390542539821416,"jobId":"2459005","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"b72ab8427c791c09952459ad6d404041f966d43c49a0976833474c8b17e6864d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.003--0.9--0.98--s44","configuration":"sync-wsd-h80--0.003--0.9--0.98","method":"sync","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.3953619767770142,"jobId":"2459006","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"ef9986b011c4f0c1564defb1e51b93652038550b3cd8cc3f15ff5dd6003dc94a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.9--0.98--s42","configuration":"sync-wsd-h80--0.002--0.9--0.98","method":"sync","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.3957562129668304,"jobId":"2459007","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"550dac5eeb931c2fe5f59da55409dcbead9999b24264ed86885ce5aaf34347a5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.9--0.98--s43","configuration":"sync-wsd-h80--0.002--0.9--0.98","method":"sync","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.3862477731450116,"jobId":"2459008","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"001422b547c59960507e6eec2039a76ca445fbda0f74f6dca72d8fceb17e6970","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.9--0.98--s44","configuration":"sync-wsd-h80--0.002--0.9--0.98","method":"sync","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.394172872671825,"jobId":"2459009","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"34bf5bbe8b405ad509a268bd545cb8d310c46b8919cc0e74ac24fe98791c57cc","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.9--0.999--s42","configuration":"sync-wsd-h80--0.004--0.9--0.999","method":"sync","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":42,"loss":3.386918653844721,"jobId":"2459010","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"e2908dd8e95cc9990312290396d32a4fe53eb94bad4a66b88e5be32e165c50e3","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.9--0.999--s43","configuration":"sync-wsd-h80--0.004--0.9--0.999","method":"sync","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":43,"loss":3.399715728285496,"jobId":"2459011","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"93f692d94b5ffa374c408994a92da0127dafa06e13ead94f3421b1725b5038a5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.9--0.999--s44","configuration":"sync-wsd-h80--0.004--0.9--0.999","method":"sync","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":44,"loss":3.3833897058015197,"jobId":"2459012","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"cba9603ed965bf533deacd8f349ad1a6129c48bb12d53bcd88bbd15d14c1e7bb","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.003--0.9--0.999--s42","configuration":"sync-wsd-h80--0.003--0.9--0.999","method":"sync","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.3853547655989265,"jobId":"2459013","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"8bc1f7d21bdc5ec83dd4becb872766371768dd18aba2ad4c577e5b51f633f3ae","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.003--0.9--0.999--s43","configuration":"sync-wsd-h80--0.003--0.9--0.999","method":"sync","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.3898292523767624,"jobId":"2459017","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"a479553590175c90429bd0455b4e23ab00c7e9f2a9f52469930c29fcf058fad2","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.003--0.9--0.999--s44","configuration":"sync-wsd-h80--0.003--0.9--0.999","method":"sync","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.3894190289794257,"jobId":"2459020","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"27980fc7e980591d2d83efb3e531a8f0c2adf69841bda00affa56d30d16cc488","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.9--0.999--s42","configuration":"sync-wsd-h80--0.002--0.9--0.999","method":"sync","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.393733781347953,"jobId":"2459027","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"d8a303b6c8b5420ce0f369c110a74ec463a381fbc4d4e455efb904d9c3cede2f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.9--0.999--s43","configuration":"sync-wsd-h80--0.002--0.9--0.999","method":"sync","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.389741085295071,"jobId":"2459030","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"3b657e3115ac0f8e09be03cef695c271750ea0e401dc2b6f69ee254fba77d5c5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.9--0.999--s44","configuration":"sync-wsd-h80--0.002--0.9--0.999","method":"sync","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.392415910235665,"jobId":"2459035","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"8d22563491db6a416885f837b83d89b37c97bfaf3250019121820fdc197fd480","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.95--0.9--s42","configuration":"sync-wsd-h80--0.004--0.95--0.9","method":"sync","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":42,"loss":3.383647938076112,"jobId":"2459036","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"4057a4f77f29b4be0ae14bd3925242e1abae5c22463536e95355fb228a28afbf","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.95--0.9--s43","configuration":"sync-wsd-h80--0.004--0.95--0.9","method":"sync","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":43,"loss":3.388870167473736,"jobId":"2459037","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"d41d73830b615b9b1501b4892779c6c95fde1eecfeac901e0e750af7ca95cc5d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.95--0.9--s44","configuration":"sync-wsd-h80--0.004--0.95--0.9","method":"sync","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":44,"loss":3.389124034119998,"jobId":"2459038","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"14d6383e74b9904ff2e2906bc6084d61dc053dc86edf2b92c931f106be0a95f8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.003--0.95--0.9--s42","configuration":"sync-wsd-h80--0.003--0.95--0.9","method":"sync","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.3818844564960417,"jobId":"2459039","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"5dfc9c58461fa095ae4664bbae626305214a7ce854b4e4378e8bd94c48e1a24f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.003--0.95--0.9--s43","configuration":"sync-wsd-h80--0.003--0.95--0.9","method":"sync","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.387048817222482,"jobId":"2459040","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"29ca2b93585cea1c31a0d677cc69cd895985f21be578b47cfa5b4fa05440f109","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.003--0.95--0.9--s44","configuration":"sync-wsd-h80--0.003--0.95--0.9","method":"sync","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.389963399789087,"jobId":"2459059","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"1e32e7011609320f54750282acc761e329c9f33c3d20baf200a297c8202c8407","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.95--0.9--s42","configuration":"sync-wsd-h80--0.002--0.95--0.9","method":"sync","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.3841094680773423,"jobId":"2459060","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"e3bf5dc98f3c46c45ac05f37d34fadd30149aac5e05ffaac7af7875905aca867","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.95--0.9--s43","configuration":"sync-wsd-h80--0.002--0.95--0.9","method":"sync","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.3883398592945184,"jobId":"2459061","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"6d7daa3f0add95ada140751476fbb446152dbe780ce74d3a968cd6ba89413c6e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.95--0.9--s44","configuration":"sync-wsd-h80--0.002--0.95--0.9","method":"sync","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.3902548419548144,"jobId":"2459062","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"57f548fad9b2b8a1fbcf0ee712aecbcdfbf49f5c8a9e8258ba84be7807dda428","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.95--0.98--s42","configuration":"sync-wsd-h80--0.004--0.95--0.98","method":"sync","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":42,"loss":3.3759750879053594,"jobId":"2459063","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"419346579e38e853c98fff96f257f549653fe4f98ad3d1d7e3dc1cbdc5dafb3c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.95--0.98--s43","configuration":"sync-wsd-h80--0.004--0.95--0.98","method":"sync","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":43,"loss":3.381783574629176,"jobId":"2459064","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"aa8d78c1c0b4ec723668394ce49d028710b53c3d28e6380cad9dd82a73fd3691","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.95--0.98--s44","configuration":"sync-wsd-h80--0.004--0.95--0.98","method":"sync","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":44,"loss":3.380806040847772,"jobId":"2459065","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"d58af4b92c28fe96189d6924e0634e7e69d9662dcb06cedc73651f8a99ec07bd","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.003--0.95--0.98--s42","configuration":"sync-wsd-h80--0.003--0.95--0.98","method":"sync","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.375018283755546,"jobId":"2459066","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"818447c2f231b63d5c0473a2d8bf1492d0a0dcebec675000de052a91d3ddf033","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.003--0.95--0.98--s43","configuration":"sync-wsd-h80--0.003--0.95--0.98","method":"sync","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.379192834151603,"jobId":"2459067","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"996d6c7bdf53f9f65e8863e49fd7fb95772e9f3c14d02e5510089380c2122c50","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.003--0.95--0.98--s44","configuration":"sync-wsd-h80--0.003--0.95--0.98","method":"sync","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.380375006930562,"jobId":"2459068","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"17ec99c05e99be6edd20011c1b70454181f2c313f6fa39f9611c24a659373764","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.95--0.98--s42","configuration":"sync-wsd-h80--0.002--0.95--0.98","method":"sync","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.3775112392312505,"jobId":"2459069","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"46fbf7a7cf74bad757dcf7a92e540de95a04c73d3bc141f8141c1e15488d6961","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.95--0.98--s43","configuration":"sync-wsd-h80--0.002--0.95--0.98","method":"sync","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.387363717423831,"jobId":"2459070","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"11f5d721c9373ef666d9ef665b9f7e553eac1bb554271f3f4550e0642e1bdc5f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.95--0.98--s44","configuration":"sync-wsd-h80--0.002--0.95--0.98","method":"sync","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.377901445556874,"jobId":"2459071","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"0ec87355cbe8f4d0d2c44abd0a376622ab52dd7ba8fb547e521ff86f97b02e99","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.95--0.999--s42","configuration":"sync-wsd-h80--0.004--0.95--0.999","method":"sync","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":42,"loss":3.3758311076744336,"jobId":"2459072","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"35d9d14d81528cc0d4f3a0a5ab43367d58d8e87362e13d8dac66005ee741d96a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.95--0.999--s43","configuration":"sync-wsd-h80--0.004--0.95--0.999","method":"sync","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":43,"loss":3.3782369164252612,"jobId":"2459073","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"b9f377e12cd43d5257e4c6535b9d1ef3801f7967eb1e56884db48e8e86ac6578","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.004--0.95--0.999--s44","configuration":"sync-wsd-h80--0.004--0.95--0.999","method":"sync","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":44,"loss":3.37813549995855,"jobId":"2459074","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"d9dc9f344256d968910cdccf22793b28c24b4c527335beeaf0999474e0681b0d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.003--0.95--0.999--s42","configuration":"sync-wsd-h80--0.003--0.95--0.999","method":"sync","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.371351940317979,"jobId":"2460653","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"9fa14213ce387d491e6b37eab5bde48ded0bec1e48b4aa75b3b259a2e6e877f8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/final.pt"},{"id":"sync-wsd-h80--0.003--0.95--0.999--s43","configuration":"sync-wsd-h80--0.003--0.95--0.999","method":"sync","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.3773290357226062,"jobId":"2460654","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"082cdd3da7c4895e3529d6514715e4efa016c70241bd47bab2bf90298d8a98fd","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/final.pt"},{"id":"sync-wsd-h80--0.003--0.95--0.999--s44","configuration":"sync-wsd-h80--0.003--0.95--0.999","method":"sync","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.3756280799544163,"jobId":"2460783","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"09b98b23bff6095db3084f2b7820643bf7eb063817e7948414c7cb57947891cb","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/final.pt"},{"id":"sync-wsd-h80--0.002--0.95--0.999--s42","configuration":"sync-wsd-h80--0.002--0.95--0.999","method":"sync","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.373592376293663,"jobId":"2460784","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"4acf9a5ac805fe1bc856d5c46fde88fd028d15646353cd7060f587432d3cf68b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.95--0.999--s43","configuration":"sync-wsd-h80--0.002--0.95--0.999","method":"sync","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.3816051022697224,"jobId":"2460785","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"3ca6645273253308dc0974eab53881c87ef9c5d43d5966e7b1636fa0cd0cad3d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h80--0.002--0.95--0.999--s44","configuration":"sync-wsd-h80--0.002--0.95--0.999","method":"sync","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.3754881626222772,"jobId":"2460786","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"53a4d08086815092f0617bfc5e77dc956b4628bda4c50356d2405f5de4019e67","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.003--0.9--0.9--s42","configuration":"sync-wsd-h120--0.003--0.9--0.9","method":"sync","horizon":120,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.3691511967457104,"jobId":"2461519","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"e98adfe88807b82f846d5985354d418a2109d4dd982a203100d8b0dc2eb2230a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.003--0.9--0.9--s43","configuration":"sync-wsd-h120--0.003--0.9--0.9","method":"sync","horizon":120,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.364713668866867,"jobId":"2461523","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"723e3947164c972bd9a84c0544c97971b9457914078d31a9341cf91dba93203f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.003--0.9--0.9--s44","configuration":"sync-wsd-h120--0.003--0.9--0.9","method":"sync","horizon":120,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.368625202189785,"jobId":"2461524","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"f4ba144d1b59a693f5c7de8e60102bcc63839b7d88127850a54f7e370dcab36f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.9--0.9--s42","configuration":"sync-wsd-h120--0.002--0.9--0.9","method":"sync","horizon":120,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.374923689294086,"jobId":"2461526","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"a9486b29f35d32670ce9148ad946a15c630bbb859dbdce0dec20e61af611d513","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.9--0.9--s43","configuration":"sync-wsd-h120--0.002--0.9--0.9","method":"sync","horizon":120,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.367432421913799,"jobId":"2461527","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"ca6a6cc701ffae059a2069539e5c0da265bf206e1d0a1381622541e6b1fc6f6e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.9--0.9--s44","configuration":"sync-wsd-h120--0.002--0.9--0.9","method":"sync","horizon":120,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.3699094124197315,"jobId":"2461528","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"268a9a4c6bdf736ca3979d01d5aad949dde85fe8f93efda5995bb980edb9d1e1","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.003--0.9--0.98--s42","configuration":"sync-wsd-h120--0.003--0.9--0.98","method":"sync","horizon":120,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.3589789181908407,"jobId":"2461529","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"fe161ab59b1da4324bf8a436c924204a0cf882f05f3305226efc7809a554ea61","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.003--0.9--0.98--s43","configuration":"sync-wsd-h120--0.003--0.9--0.98","method":"sync","horizon":120,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.353192517783011,"jobId":"2461548","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"d83bd101187d23ee2c530a48d81c9e6473d07a4c3c48783a6129b33fb0b4337d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.003--0.9--0.98--s44","configuration":"sync-wsd-h120--0.003--0.9--0.98","method":"sync","horizon":120,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.3573516337837934,"jobId":"2461549","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"247aa962ed21e97713a7560db7217e8936a7bff4ca0f52fedd30c1890b6b47bc","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.9--0.98--s42","configuration":"sync-wsd-h120--0.002--0.9--0.98","method":"sync","horizon":120,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.356640648651833,"jobId":"2461550","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"6de3e6db8edada0c1e424c37fde5f1bd4367288f3194346b7c1ea451a9d81654","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.9--0.98--s43","configuration":"sync-wsd-h120--0.002--0.9--0.98","method":"sync","horizon":120,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.350406513061352,"jobId":"2461551","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"21a390b5b7040f7ead6db06e2dda44050a69638169b0a62e0db960793e24a1bb","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.9--0.98--s44","configuration":"sync-wsd-h120--0.002--0.9--0.98","method":"sync","horizon":120,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.3560097341079382,"jobId":"2461552","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"79e7048281e5d51fbff5c98b29e57ce6052e170049fc2d31d6fe609447dbb200","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.003--0.9--0.999--s42","configuration":"sync-wsd-h120--0.003--0.9--0.999","method":"sync","horizon":120,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.3505589714755843,"jobId":"2461553","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"5d9de60c4339178f8dba4408fff8face1812b8349e9ee188c7f567cb1772b615","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.003--0.9--0.999--s43","configuration":"sync-wsd-h120--0.003--0.9--0.999","method":"sync","horizon":120,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.3500551367449667,"jobId":"2461554","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"1de442cfa5ee313630365e69acc0df9f29d10a1495446caa77d00d874b5dc872","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.003--0.9--0.999--s44","configuration":"sync-wsd-h120--0.003--0.9--0.999","method":"sync","horizon":120,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.350563310440662,"jobId":"2461555","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"fb69e174dda123e8dce12f0ed96c3f9773f75894f42825bd9327097db7b0f9e0","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.9--0.999--s42","configuration":"sync-wsd-h120--0.002--0.9--0.999","method":"sync","horizon":120,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.353615725642847,"jobId":"2461556","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"851ff4781d03ef4df9a3d6e9de4a0af53afd894828008e2909a355f8507f5095","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.9--0.999--s43","configuration":"sync-wsd-h120--0.002--0.9--0.999","method":"sync","horizon":120,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.3500319262251748,"jobId":"2461557","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"9324a71daee969af51a0e006c6c18a308ebc49d9d426e66d2b364dc83500f4b8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.9--0.999--s44","configuration":"sync-wsd-h120--0.002--0.9--0.999","method":"sync","horizon":120,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.3538007354347514,"jobId":"2461558","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"b927c62af4c3b1d479bc828d23b03c370c13a90ae9d07aefd5743ff07e4a1122","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.003--0.95--0.9--s42","configuration":"sync-wsd-h120--0.003--0.95--0.9","method":"sync","horizon":120,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.347097920415326,"jobId":"2461559","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"f5bdffede29907151712f3d116494eb8738fbf9aa71abd412c97e6c45712b66c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.003--0.95--0.9--s43","configuration":"sync-wsd-h120--0.003--0.95--0.9","method":"sync","horizon":120,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.3516200603362734,"jobId":"2461560","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"86f77c96222b09e31869ab0b4ecf0dd889917f31bb3a90bbf8a735fbb4a9cd11","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.003--0.95--0.9--s44","configuration":"sync-wsd-h120--0.003--0.95--0.9","method":"sync","horizon":120,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.353726081056074,"jobId":"2461561","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"24f65db8a35e4dc278e9bf70200b5f5e9cde065ce3275728f994547d64514669","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.95--0.9--s42","configuration":"sync-wsd-h120--0.002--0.95--0.9","method":"sync","horizon":120,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.348942577352643,"jobId":"2461562","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"5fd83c340a2af6b98372cbfc076970e0e217376c0f8f6b3d658339387ac44fa1","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.95--0.9--s43","configuration":"sync-wsd-h120--0.002--0.95--0.9","method":"sync","horizon":120,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.352916446734371,"jobId":"2461563","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"408ae6b36e36352c63407fa97274e1a8f2fc1146de32a9cbef9c4e1292d30021","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.95--0.9--s44","configuration":"sync-wsd-h120--0.002--0.95--0.9","method":"sync","horizon":120,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.3537109313600366,"jobId":"2461564","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"11d66426e6961c59da6f7420074a2c0fca0dd52221ad92773acf0a2f499c4445","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.003--0.95--0.98--s42","configuration":"sync-wsd-h120--0.003--0.95--0.98","method":"sync","horizon":120,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.340056066217504,"jobId":"2461565","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"0df5f00389d8db3faa01e1d498ad5afb378b3a037ca53246167f332986851a1d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.003--0.95--0.98--s43","configuration":"sync-wsd-h120--0.003--0.95--0.98","method":"sync","horizon":120,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.34416150046102,"jobId":"2461566","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"179e6cf55329a778c1fd076e5900699a117963dbd26fbd91da71971f38641593","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.003--0.95--0.98--s44","configuration":"sync-wsd-h120--0.003--0.95--0.98","method":"sync","horizon":120,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.344916610736368,"jobId":"2461567","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"bfa888f0f8be12dcebdc78f9997c195f29eeb1f6b8ead2fa0c306fe655aedcf8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.95--0.98--s42","configuration":"sync-wsd-h120--0.002--0.95--0.98","method":"sync","horizon":120,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.3428652603456044,"jobId":"2461568","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"ca74eeffa1309f84eae15331922196dfe7300b4dce96c52ae4e7fed76ad97a0e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.95--0.98--s43","configuration":"sync-wsd-h120--0.002--0.95--0.98","method":"sync","horizon":120,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.351132205130257,"jobId":"2461569","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"e34d1a33d6f759d6f390c5b6ef53ee349cdd9f95abbc42a03810a61a7d18ecc6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.95--0.98--s44","configuration":"sync-wsd-h120--0.002--0.95--0.98","method":"sync","horizon":120,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.3427909079978417,"jobId":"2461570","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"d11cd4a00bda5016533ea67044ec83a768d288ef2b0763c834bc0ed5ae0c399b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.003--0.95--0.999--s42","configuration":"sync-wsd-h120--0.003--0.95--0.999","method":"sync","horizon":120,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.3358776806003805,"jobId":"2461571","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"b7c465db5c2e8407b878538541c8b36ac874bd9de262b682d4a2bc1376a6834a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/final.pt"},{"id":"sync-wsd-h120--0.003--0.95--0.999--s43","configuration":"sync-wsd-h120--0.003--0.95--0.999","method":"sync","horizon":120,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.342626461286864,"jobId":"2461572","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"5ae69c1395311365e60244b94ec1edad2f1d9e1795234f0cf2478b8af5dcc180","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/final.pt"},{"id":"sync-wsd-h120--0.003--0.95--0.999--s44","configuration":"sync-wsd-h120--0.003--0.95--0.999","method":"sync","horizon":120,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.3389874721956683,"jobId":"2461573","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"84c3092ed4e4a0c32ec5f2f1b7c260bfc60da483835f5dca951fc5d86fed2299","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/final.pt"},{"id":"sync-wsd-h120--0.002--0.95--0.999--s42","configuration":"sync-wsd-h120--0.002--0.95--0.999","method":"sync","horizon":120,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.3372535258342455,"jobId":"2461574","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"d2f482a3e2d0182a43255ef0a86e2682d60bfe98a21ec1b4c32d4d7ca8c2bdb8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.95--0.999--s43","configuration":"sync-wsd-h120--0.002--0.95--0.999","method":"sync","horizon":120,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.344210140111933,"jobId":"2461575","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"42a550ef7dc3bf3186f508f9727fb9647c7924ad44e936583beae470cf9e964e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h120--0.002--0.95--0.999--s44","configuration":"sync-wsd-h120--0.002--0.95--0.999","method":"sync","horizon":120,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.3385842410922026,"jobId":"2461576","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"03a0deac21eb5680d309131f90d4177fa7c69de3bab1a59b0d02e8c2976e024e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.003--0.9--0.9--s42","configuration":"sync-wsd-h160--0.003--0.9--0.9","method":"sync","horizon":160,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.3454572941441034,"jobId":"2462419","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"a984d2033c5db528d67fa7b73e715c0812cd76b5c967747cf7195bd046d42ee3","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.003--0.9--0.9--s43","configuration":"sync-wsd-h160--0.003--0.9--0.9","method":"sync","horizon":160,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.341659362961695,"jobId":"2462420","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"46631432aa9befeb8ed70754a6b2ec4de797c5e98aebc271dc140c771d963543","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.003--0.9--0.9--s44","configuration":"sync-wsd-h160--0.003--0.9--0.9","method":"sync","horizon":160,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.344833484559326,"jobId":"2462421","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"65a49504fa3a36ad6b04ee10e6143ee4feca1c8372f694a6f6b579cce50a0bf5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.9--0.9--s42","configuration":"sync-wsd-h160--0.002--0.9--0.9","method":"sync","horizon":160,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.3511317501123843,"jobId":"2462422","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"62572c7f0f9288f3bfc57b95e8228074048c84ceab67e0a6715f36db2c04fccd","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.9--0.9--s43","configuration":"sync-wsd-h160--0.002--0.9--0.9","method":"sync","horizon":160,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.344096787293426,"jobId":"2462423","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"15e90b835282af221334b5852cdccfe393b7a97e4bd9ffc65f46e2dd077b254f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.9--0.9--s44","configuration":"sync-wsd-h160--0.002--0.9--0.9","method":"sync","horizon":160,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.3467681063344408,"jobId":"2462424","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"e2931c61a19b774342c3ee6f121993af3d247d8993362aa4911da74117d3ca06","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.003--0.9--0.98--s42","configuration":"sync-wsd-h160--0.003--0.9--0.98","method":"sync","horizon":160,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.3353408910110067,"jobId":"2462425","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"8363fd14adb7c19c4d2f8cbbf2f80aa2b3737ad0296e6fa0898eb6286aefc393","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.003--0.9--0.98--s43","configuration":"sync-wsd-h160--0.003--0.9--0.98","method":"sync","horizon":160,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.329703427837872,"jobId":"2462426","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"e3f075b58362aabf76cc5787bc35f6542096549af65a351fde56847c0e1f9b13","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.003--0.9--0.98--s44","configuration":"sync-wsd-h160--0.003--0.9--0.98","method":"sync","horizon":160,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.333176981799873,"jobId":"2462427","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"27aee36d77786b54a3adf62653b5790b61e89f0560bd21d5124e8e4fd7425575","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.9--0.98--s42","configuration":"sync-wsd-h160--0.002--0.9--0.98","method":"sync","horizon":160,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.332162405103496,"jobId":"2462428","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"9cabb5e606835305ff57e13aac69477f486463416ea7937a4f1c645b287c5e84","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.9--0.98--s43","configuration":"sync-wsd-h160--0.002--0.9--0.98","method":"sync","horizon":160,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.3273526546602206,"jobId":"2462429","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"a5cb379de5fd2d960d75b134648b1a33223d7443acb176cd30f3413be60fc7c6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.9--0.98--s44","configuration":"sync-wsd-h160--0.002--0.9--0.98","method":"sync","horizon":160,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.3324995921913207,"jobId":"2462430","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"b1103c60c8ccb4fb2e8826de46f48ddd72e7e59bd5b57cc86803915cd89207ba","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.003--0.9--0.999--s42","configuration":"sync-wsd-h160--0.003--0.9--0.999","method":"sync","horizon":160,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.3288749161129143,"jobId":"2462431","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"9a9935237f12cf74143188f32d5dce737f4c75b6c907e2a25037497f3f5633ba","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.003--0.9--0.999--s43","configuration":"sync-wsd-h160--0.003--0.9--0.999","method":"sync","horizon":160,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.326803840448947,"jobId":"2462432","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"dcedb75f9a23ef031fcaf5361d3e516a75fc8157aab00b01d6f935878fe61292","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.003--0.9--0.999--s44","configuration":"sync-wsd-h160--0.003--0.9--0.999","method":"sync","horizon":160,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.326943668256754,"jobId":"2462433","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"fb00f4cc8574c8e2e32bc8ac66c5bbceb505db019402f9afc8db9d0ce062d6f8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.9--0.999--s42","configuration":"sync-wsd-h160--0.002--0.9--0.999","method":"sync","horizon":160,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.3289655426657414,"jobId":"2462434","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"1137d89542b5cf38d0abaf20d101d9ac7766b5101f17ef1e798d52dece4e4794","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.9--0.999--s43","configuration":"sync-wsd-h160--0.002--0.9--0.999","method":"sync","horizon":160,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.3259936856136956,"jobId":"2462436","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"149d10e143117367645e30751331ff7724787603017158ef92fb3083271da8d5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.9--0.999--s44","configuration":"sync-wsd-h160--0.002--0.9--0.999","method":"sync","horizon":160,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.330497737937856,"jobId":"2462437","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"24c370cb36b7cbbe867bc747cbc3610f16518bc4901ba93169c245ddd29e8fac","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.003--0.95--0.9--s42","configuration":"sync-wsd-h160--0.003--0.95--0.9","method":"sync","horizon":160,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.3256956317066297,"jobId":"2462438","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"cafe06780dca307d1181ce964cb7e02aa1aebc05afd427ed074eeaea5dc66b3b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.003--0.95--0.9--s43","configuration":"sync-wsd-h160--0.003--0.95--0.9","method":"sync","horizon":160,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.329253608321027,"jobId":"2462439","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"33a93bd6c1968dff1d3e7a6b597c71a09addd1f60bdf51ea217429b55932f96b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.003--0.95--0.9--s44","configuration":"sync-wsd-h160--0.003--0.95--0.9","method":"sync","horizon":160,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.3312332817107655,"jobId":"2462440","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"61fb1e9b63d2b891eeb96611eef0b058c79742cc8ea82afbb3ce7b5f038799b6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.95--0.9--s42","configuration":"sync-wsd-h160--0.002--0.95--0.9","method":"sync","horizon":160,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.3268325459296793,"jobId":"2462441","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"7f6447cd47fcc6fe2100ea9a424bff5232d3b049be3dd5d976df77ff195cec2e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.95--0.9--s43","configuration":"sync-wsd-h160--0.002--0.95--0.9","method":"sync","horizon":160,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.330710784572413,"jobId":"2462442","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"32cdede875b97c0db8c983896499262fe3c0c82ed6b1df23a317dfd586579294","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.95--0.9--s44","configuration":"sync-wsd-h160--0.002--0.95--0.9","method":"sync","horizon":160,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.3312423701441385,"jobId":"2462443","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"6c185218dab7c4192740219d8b2f8c91cd1d7c30a18685b452608b12f2111aca","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.003--0.95--0.98--s42","configuration":"sync-wsd-h160--0.003--0.95--0.98","method":"sync","horizon":160,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.3181250881633693,"jobId":"2462444","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"46a565df52bfc93dbaad3569aebb5a542ecb4a20c3b6710ddf352c53750ae9d6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.003--0.95--0.98--s43","configuration":"sync-wsd-h160--0.003--0.95--0.98","method":"sync","horizon":160,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.321727266794994,"jobId":"2462445","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"68b1ca5963792cfeda2a3eb945a848a0473c7e08fffd76285fd59abf4b03bc14","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.003--0.95--0.98--s44","configuration":"sync-wsd-h160--0.003--0.95--0.98","method":"sync","horizon":160,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.3226821783562257,"jobId":"2462446","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"e9a92bef1729a462a0e161f80bc704c4cf92a47dca36ba3271d8b864ec70dc6f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.95--0.98--s42","configuration":"sync-wsd-h160--0.002--0.95--0.98","method":"sync","horizon":160,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.3210442227505452,"jobId":"2462447","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"4ebcd6d09eab37fd01c31c6d19a0b505739af0fd1845c0cdb8c14534be9a8e11","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.95--0.98--s43","configuration":"sync-wsd-h160--0.002--0.95--0.98","method":"sync","horizon":160,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.3280224159050658,"jobId":"2462448","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"14610ecdb84b64a9afdb08827c7c4c863ec15ed3d7aa44ae58f0afbe9a0ceae5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.95--0.98--s44","configuration":"sync-wsd-h160--0.002--0.95--0.98","method":"sync","horizon":160,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.3205936185545912,"jobId":"2462449","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"08a52e8cae530a555cd1b5fdf675f859f806b116d213f10e13dcda0d32b50681","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.003--0.95--0.999--s42","configuration":"sync-wsd-h160--0.003--0.95--0.999","method":"sync","horizon":160,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.313441292120476,"jobId":"2462450","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"225beeea7659341d73d81bb4d4229810db4fc8a35b038c565581fa3588c0cb25","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/final.pt"},{"id":"sync-wsd-h160--0.003--0.95--0.999--s43","configuration":"sync-wsd-h160--0.003--0.95--0.999","method":"sync","horizon":160,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.3209763697506856,"jobId":"2462451","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"30ce54bde1cb6364aae7467495694e02ce5ecb77f344ef87d9101bc651d7e9e9","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/final.pt"},{"id":"sync-wsd-h160--0.003--0.95--0.999--s44","configuration":"sync-wsd-h160--0.003--0.95--0.999","method":"sync","horizon":160,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.316854902655363,"jobId":"2462452","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"91e18b54ccdb718e27de172c3f8177b9bd4a87cddf785bbcb2ac9145cad5f8c0","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/final.pt"},{"id":"sync-wsd-h160--0.002--0.95--0.999--s42","configuration":"sync-wsd-h160--0.002--0.95--0.999","method":"sync","horizon":160,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.3143719220945855,"jobId":"2462453","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"9b8a20615b97ef8dfbb3113a089981b1848b7724e5314880ddfe3f571d8ebc93","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.95--0.999--s43","configuration":"sync-wsd-h160--0.002--0.95--0.999","method":"sync","horizon":160,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.32156990315758,"jobId":"2462454","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"9c1a4f521cb7918eb2447254a63dbd16f1afb1ebe86de2a5335c7e1039030eba","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"sync-wsd-h160--0.002--0.95--0.999--s44","configuration":"sync-wsd-h160--0.002--0.95--0.999","method":"sync","horizon":160,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.3156719160106447,"jobId":"2462455","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"537aaa0750c752af25d395b08ee641c5a03d116f79f0fa8da77cb228737ab36b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/sync-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":[],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.9--0.9--s42","configuration":"awc8-wsd-h20--0.01--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.01,"seed":42,"loss":3.7399308415913497,"jobId":"2478551","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"2208f6080e8c42a6ad8ca966d04070c3b7ec4b832b14e8bc52a04b9d7821f419","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.9--0.9--s43","configuration":"awc8-wsd-h20--0.01--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.01,"seed":43,"loss":3.7291007105153855,"jobId":"2478552","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"f86bfca3c81979e32900237ab02bcf26c30877fcbabde43f67651ddb704197f1","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.9--0.9--s44","configuration":"awc8-wsd-h20--0.01--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.01,"seed":44,"loss":3.7392321073116714,"jobId":"2478553","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"55c81c1fc0c3354bc198bd8be1c7a769d09950befcf3fb056ac9f6f254d2e587","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.01-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.9--0.9--s42","configuration":"awc8-wsd-h20--0.008--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.008,"seed":42,"loss":3.7272634253307966,"jobId":"2478554","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"ee478d981f22d0edb82fe86f1d81a81d5d3d964d20e88b600d237fc6f8997f69","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.9--0.9--s43","configuration":"awc8-wsd-h20--0.008--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.008,"seed":43,"loss":3.731839734352052,"jobId":"2478555","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"6dcd3c70e5864e86822834e44eef6ed90375133c38e8625a3ec96627566f50c1","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.9--0.9--s44","configuration":"awc8-wsd-h20--0.008--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.008,"seed":44,"loss":3.7365845429544415,"jobId":"2478556","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"007ad94a8b0d6d958780fa201a5c779af61ad0e8f116860cc6d2401f896dce5e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.008-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.006--0.9--0.9--s42","configuration":"awc8-wsd-h20--0.006--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.006,"seed":42,"loss":3.7211869087289102,"jobId":"2478557","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"fbe6baad5480f18d5b18edd60930c22d58698fe4e567d2df10160d9e4e57708a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.006--0.9--0.9--s43","configuration":"awc8-wsd-h20--0.006--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.006,"seed":43,"loss":3.7336587401190826,"jobId":"2478558","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"b4c77120ba532ea4f6cf8c3483edab4f9ebffa4d896debcb92476ab831d0415e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.006--0.9--0.9--s44","configuration":"awc8-wsd-h20--0.006--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.006,"seed":44,"loss":3.732928391411746,"jobId":"2478559","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"644f24cce52cf9030794a8fed5b953ef7db5f4553d037db2b688fecd23484a62","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.9--0.9--s42","configuration":"awc8-wsd-h20--0.004--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":42,"loss":3.759674159466903,"jobId":"2478560","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"164f329e060f85ef8de53bc26fc2eff9f1d2f6e968a366edec608ea4ade3a681","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.9--0.9--s43","configuration":"awc8-wsd-h20--0.004--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":43,"loss":3.75243314709979,"jobId":"2478561","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"e7f152d5c8309b386b1459fd261dd80984b2e568bc3f67fcab3cf959799f123e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.9--0.9--s44","configuration":"awc8-wsd-h20--0.004--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":44,"loss":3.7489837221859847,"jobId":"2478563","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"96b6b8c88d123eda8edde5f1d04feafc35d3c77bd9a333235a0ed45e0f36baa7","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.9--0.9--s42","configuration":"awc8-wsd-h20--0.003--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.780051696458729,"jobId":"2478564","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"505d4ad6fea35b0fdb5d27c120a60d0c39f0ef02613f24408565de13430633ba","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.9--0.9--s43","configuration":"awc8-wsd-h20--0.003--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.754887184131928,"jobId":"2478565","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"126e62c588aced9e33b09ed779a231d459e4f6da554815c361db44a8ee3491a8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.9--0.9--s44","configuration":"awc8-wsd-h20--0.003--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.771339394482213,"jobId":"2478566","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"740e18fdf93eb26f3f127270de2b77e776ddb481af2ad2a935a9c05c720e8c05","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.9--0.9--s42","configuration":"awc8-wsd-h20--0.002--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.7996437607319002,"jobId":"2478567","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"fbdb252a29c004aea59977f2f7c71c3fac7de8155afa51e4f3a30221f1354e3e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.9--0.9--s43","configuration":"awc8-wsd-h20--0.002--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.782585271397604,"jobId":"2478568","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"1c0ddfed5a4208f3b522085649b3af43774c2d5da9919d67e30a699bc064edff","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.9--0.9--s44","configuration":"awc8-wsd-h20--0.002--0.9--0.9","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.79766998233501,"jobId":"2478569","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"057cc957cdee874c0065a2729ebd9cdecf4c4470f182181820a6907dfa54819f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.9--0.98--s42","configuration":"awc8-wsd-h20--0.01--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.01,"seed":42,"loss":3.649899225139402,"jobId":"2478570","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"ca0dd55ecca8f65e346b7b1d3d21885494d200e424eef30b169b8920178b79aa","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.9--0.98--s43","configuration":"awc8-wsd-h20--0.01--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.01,"seed":43,"loss":3.6421674317637716,"jobId":"2478571","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"469326e6ab25253f555df041b4eac5965bdac2b08e763ff1e8c3a5a927e6eb4f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.9--0.98--s44","configuration":"awc8-wsd-h20--0.01--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.01,"seed":44,"loss":3.64566813186026,"jobId":"2478572","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"c38d133e3ccf5eaf165ffa092691b8865ae9718655dc9bbf213e28aab7d0cfbb","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.01-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.9--0.98--s42","configuration":"awc8-wsd-h20--0.008--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.008,"seed":42,"loss":3.641203605182922,"jobId":"2478573","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"f0aa37f8934b23a76b65e3637e65c81433f389bfccde7b05c9aa393c341a3179","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.9--0.98--s43","configuration":"awc8-wsd-h20--0.008--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.008,"seed":43,"loss":3.6390867063940076,"jobId":"2478574","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"89b14fc31e96e03c7205d06464e8a6364620752df08efe84b67e5c1da9858895","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.9--0.98--s44","configuration":"awc8-wsd-h20--0.008--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.008,"seed":44,"loss":3.6430117021673647,"jobId":"2478575","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"9937a39fe89884c8b12eb87aae5c1f6c2a462aaf67754d0f6bef2ddf7fa62619","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.008-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.006--0.9--0.98--s42","configuration":"awc8-wsd-h20--0.006--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.006,"seed":42,"loss":3.6348810709061317,"jobId":"2478576","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"a29152e224b781b11bb46e27222621f1bd112b093d0392208bce0f98de87854a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.006--0.9--0.98--s43","configuration":"awc8-wsd-h20--0.006--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.006,"seed":43,"loss":3.64469187089275,"jobId":"2478577","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"f8e540bd529d78a9aa98e1a098e25093fa316ae3e501786e893c9fd29511d3a2","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.006--0.9--0.98--s44","configuration":"awc8-wsd-h20--0.006--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.006,"seed":44,"loss":3.639960767141114,"jobId":"2478578","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"9038d74f3bda179852dbddbd7b19e2b1dbf3baf8cdb77c401a89ec202f26a8e9","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.9--0.98--s42","configuration":"awc8-wsd-h20--0.004--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":42,"loss":3.6576849139433754,"jobId":"2477538","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"e49e7bce358bbb6d5a2c3bface61567b683447f5f5327383ae2d294544e4d9a3","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.9--0.98--s43","configuration":"awc8-wsd-h20--0.004--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":43,"loss":3.6645932975147146,"jobId":"2478579","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"41c3151fe8f5257a4a7760ea42c8e5ef5bd39fd081f9465c7211acf69ea8c54f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.9--0.98--s44","configuration":"awc8-wsd-h20--0.004--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":44,"loss":3.6504553762937646,"jobId":"2478580","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"75e0318012134a782cf50c0e2b8345ff6ce93fd2c5f2131b614bbae2bdc872f1","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.9--0.98--s42","configuration":"awc8-wsd-h20--0.003--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.673098105839217,"jobId":"2478581","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"c943918421b102d2e481525cb4c834f6cca701e138b10341dc89a1ef286c574c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.9--0.98--s43","configuration":"awc8-wsd-h20--0.003--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.67417623649733,"jobId":"2478582","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"89613778618199ab0e5b8bc0c6530a469115347253d7ba45b58deb8fe40fc7dd","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.9--0.98--s44","configuration":"awc8-wsd-h20--0.003--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.671167278966557,"jobId":"2478583","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"d38853ed86b24cebcfbe30f227a1c3bb504a61f4d08c00883b7c0aaa9a77db6a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.9--0.98--s42","configuration":"awc8-wsd-h20--0.002--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.7095806459825638,"jobId":"2478584","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"ed97b3688b0b8348cc06e2f8ab2e1541f97e102336525510a940e10b3c7b0865","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.9--0.98--s43","configuration":"awc8-wsd-h20--0.002--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.692611871122898,"jobId":"2478585","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"8f7b2a79acb36ded68e42cadf3663aa961233597d49bcf57a9a000d6d8f675a5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.9--0.98--s44","configuration":"awc8-wsd-h20--0.002--0.9--0.98","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.7079620559754964,"jobId":"2478586","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"7de9da81981e85dc44d6f7055902dff02bd98c5427ac496120d6a20373635c05","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.9--0.999--s42","configuration":"awc8-wsd-h20--0.01--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.01,"seed":42,"loss":3.621028274060837,"jobId":"2478587","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"d3d1561e4651e5b1c6592339f2497b4be88ba26702ea1a4d41876b7404d5c09c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.9--0.999--s43","configuration":"awc8-wsd-h20--0.01--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.01,"seed":43,"loss":3.6182242029862475,"jobId":"2478588","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"187fa2ae77a1a30f470c6147b8f039752ee73a8b9c446bcc096f010720fe10c9","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.9--0.999--s44","configuration":"awc8-wsd-h20--0.01--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.01,"seed":44,"loss":3.6197059886215275,"jobId":"2478589","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"1c1a5a2e56ae968c92d36ebf042ef59d6190527c806acd0958abe7fa088206ae","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.01-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.9--0.999--s42","configuration":"awc8-wsd-h20--0.008--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.008,"seed":42,"loss":3.619966783967022,"jobId":"2478590","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"921f1d5570709f41dc23f8b19fa1906222b3d15fdba0f02bdae395df286a4aed","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.9--0.999--s43","configuration":"awc8-wsd-h20--0.008--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.008,"seed":43,"loss":3.613442201490631,"jobId":"2478591","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"a9f36b087ac3ee9099e4c89c4e76080c25ff1a1f064d873e9ac0414987b62a57","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.9--0.999--s44","configuration":"awc8-wsd-h20--0.008--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.008,"seed":44,"loss":3.6279922378386114,"jobId":"2478592","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"c2222749be73cb548d71bc8e075a7ca8cc018ef4945520d1f44bcefc093d040e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.008-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.006--0.9--0.999--s42","configuration":"awc8-wsd-h20--0.006--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.006,"seed":42,"loss":3.6054565734078152,"jobId":"2478593","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"06ad20af0ff3ffbb2fd364f492421711810cc1a4f80afaadb39c7e2cf3889aee","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.006--0.9--0.999--s43","configuration":"awc8-wsd-h20--0.006--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.006,"seed":43,"loss":3.6113107910124556,"jobId":"2478594","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"5723a0bfb8869dc896b9c9e3e8d79ab87ae77f24d4df9dec52f5f4c31dd2ae50","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.006--0.9--0.999--s44","configuration":"awc8-wsd-h20--0.006--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.006,"seed":44,"loss":3.608697753371398,"jobId":"2478595","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"68caec269fa82ef5c07ebcc94335bfa05e89c7cd8a5e658a7dc35dd0f4ec9df1","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.9--0.999--s42","configuration":"awc8-wsd-h20--0.004--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":42,"loss":3.6178593849568252,"jobId":"2478596","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"781180035e2e2711e98ce9d06f866738d71de2f60fe24d580cf05e3e1d7e42af","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.9--0.999--s43","configuration":"awc8-wsd-h20--0.004--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":43,"loss":3.6142982909380508,"jobId":"2478597","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"e479db68ff6798c28fa3580719e409bf05295d69a39f209bdc4662970ec8ad3e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.9--0.999--s44","configuration":"awc8-wsd-h20--0.004--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":44,"loss":3.6211837708759087,"jobId":"2478598","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"2e0db0f9e70fd9825b791ee2c896c70483b784534d7244e5bde733968d3f8aa0","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.9--0.999--s42","configuration":"awc8-wsd-h20--0.003--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.6250492355354593,"jobId":"2479238","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"28f0a2c5daa3a77c6456d0dc46c63b18f45ec3682f820bf7cb18235c839089bf","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.9--0.999--s43","configuration":"awc8-wsd-h20--0.003--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.645326117693374,"jobId":"2479239","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"160c47a62c51bb917b630c7f133d21bec932323a84cc6a13998dba298a27888b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.9--0.999--s44","configuration":"awc8-wsd-h20--0.003--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.638544595733517,"jobId":"2479240","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"990409b96ab3f9c2730ad7907396b27ee3b5727f8873c5a1f594e8fd7cba28f1","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.9--0.999--s42","configuration":"awc8-wsd-h20--0.002--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.6699918494163555,"jobId":"2479241","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"28f8c4d8ba90b0b28e2122924b1062e1e676d7ebef74618b1a1219c4be59ec01","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.9--0.999--s43","configuration":"awc8-wsd-h20--0.002--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.660008063284621,"jobId":"2479242","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"4b50df550a7f6bac0c9d74c43fee721ad30ed0ad14c9cb3a12e6b5b68d017960","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.9--0.999--s44","configuration":"awc8-wsd-h20--0.002--0.9--0.999","method":"awc8","horizon":20,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.6688880052118806,"jobId":"2479243","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"33c4376b47d94666e35d5a454a932ef1f8d73199f984b3fa26591e2f236772a1","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.95--0.9--s42","configuration":"awc8-wsd-h20--0.01--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.01,"seed":42,"loss":3.6797678497909576,"jobId":"2479244","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"1f54d0c1a67cf6262d630d5e2f81f4a69ad23dfd221db25616b96a6dc35616da","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.95--0.9--s43","configuration":"awc8-wsd-h20--0.01--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.01,"seed":43,"loss":3.6754679553898604,"jobId":"2479245","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"d47fe12540978624ce056b401488c9bdf980676b7c41f183415ad56c851d7e12","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.95--0.9--s44","configuration":"awc8-wsd-h20--0.01--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.01,"seed":44,"loss":3.6844085278999286,"jobId":"2479246","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"ce18197bcfd898f4ac8d438f4e2f530608410ff31df6354edc895df1a2ad26d3","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.01-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.95--0.9--s42","configuration":"awc8-wsd-h20--0.008--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.008,"seed":42,"loss":3.6750836838361454,"jobId":"2479247","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"3aa8ea4d09604cd103b38cd09aa09be1a21fdf6651c3c0033e06aa19e711caf5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.95--0.9--s43","configuration":"awc8-wsd-h20--0.008--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.008,"seed":43,"loss":3.674464403217774,"jobId":"2479248","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"74014c085d3da72e76ed7b9caa9625f8af740fb3cfb81a21fdcbc3435bd32d3c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.95--0.9--s44","configuration":"awc8-wsd-h20--0.008--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.008,"seed":44,"loss":3.6784932157365837,"jobId":"2479249","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"80ac34b8fb9bfc9aa463ca2c892ccf42517921bcafac7c36530cd9d2736b3a86","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.008-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.006--0.95--0.9--s42","configuration":"awc8-wsd-h20--0.006--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.006,"seed":42,"loss":3.674766800128119,"jobId":"2479250","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"0d7f5f55f924dafac11872c31147447743b41ef634edf94c7d3e5388c3736b69","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.006--0.95--0.9--s43","configuration":"awc8-wsd-h20--0.006--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.006,"seed":43,"loss":3.6807401945053195,"jobId":"2479251","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"9dc0b6e9e3012ec3ed0a1ebba8a81e1e83e9f9eed143f7b7bab89edcdcf2bec4","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.006--0.95--0.9--s44","configuration":"awc8-wsd-h20--0.006--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.006,"seed":44,"loss":3.684643268230684,"jobId":"2479252","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"0bf305808a5cb5b2424ac98b9e8a74f0e921ff024f4e1c706cc87021f7d48fcc","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.95--0.9--s42","configuration":"awc8-wsd-h20--0.004--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":42,"loss":3.6706539413275823,"jobId":"2479253","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"9b78a60c8436fd0ef5fbd3106d51ba6701583f3309db5b935cee444396966e95","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.95--0.9--s43","configuration":"awc8-wsd-h20--0.004--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":43,"loss":3.6821017310288564,"jobId":"2479254","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"400eb85d6d24e6ea605b857e408d6f79853cd856ae6fc343b57abf2758d03d92","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.95--0.9--s44","configuration":"awc8-wsd-h20--0.004--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":44,"loss":3.6828935666665137,"jobId":"2479255","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"d55be4bcee03c6237a0713ec344b406528cc71085f5425f2ef20a9a44c0d4a10","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.95--0.9--s42","configuration":"awc8-wsd-h20--0.003--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.6842535597515753,"jobId":"2479256","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"e6b21d3c15eee2a62124cd752243a1ef45eb7af94e6eeb53cd7c8accfe6ca955","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.95--0.9--s43","configuration":"awc8-wsd-h20--0.003--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.6910077991873904,"jobId":"2479257","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"984cd34056777b030a527392f27423dd95d8a5c048cea43e319f4c5daf4b2b29","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.95--0.9--s44","configuration":"awc8-wsd-h20--0.003--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.695541996941523,"jobId":"2479259","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"8b6e4f0bd34392e9bad8cf5928daa86ccc1295a6432d2de1d0b7afece399fa86","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.95--0.9--s42","configuration":"awc8-wsd-h20--0.002--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.71129650006366,"jobId":"2479260","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"d5b46747651389a2a7f04e3ddcd101dce69b17fb70b28f58d902eef02e9b5dbf","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.95--0.9--s43","configuration":"awc8-wsd-h20--0.002--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.704439264163356,"jobId":"2479261","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"13ae69530f08a05e0c545074d75d9660a704d0027a3806b2b14854190b28727a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.95--0.9--s44","configuration":"awc8-wsd-h20--0.002--0.95--0.9","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.718250021278996,"jobId":"2479262","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"dd8f643ce2d9f90cbbf8fa069747b24192f156edd67b6c1fd7ae65a478f2fc7c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.95--0.98--s42","configuration":"awc8-wsd-h20--0.01--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.01,"seed":42,"loss":3.6172776136154665,"jobId":"2479263","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"fcaa4f2b5fc5e8a577ceaee8cf8ceca4ddca0ee685e421d0f5447acfc1d1a680","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.95--0.98--s43","configuration":"awc8-wsd-h20--0.01--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.01,"seed":43,"loss":3.6107255052433764,"jobId":"2479264","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"8336918481f3aef1505c647049f6cfb40d6680e510f853650117f485ce606ff3","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.95--0.98--s44","configuration":"awc8-wsd-h20--0.01--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.01,"seed":44,"loss":3.621138881242901,"jobId":"2479265","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"bab7c9101bc738eb149316dd500ace5d6c2fc1d28c4472aa7aec21cf671f2529","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.01-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.95--0.98--s42","configuration":"awc8-wsd-h20--0.008--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.008,"seed":42,"loss":3.616572170045111,"jobId":"2479266","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"ca8743c63b8ce2e52488a209e6ce1f60eb4313311e7366fb203225e697bc1cdf","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.95--0.98--s43","configuration":"awc8-wsd-h20--0.008--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.008,"seed":43,"loss":3.6115444370452328,"jobId":"2479267","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"75c9eba9cfa850e9b1ca1cde2336270bad2041cd056fc483e4e74565809cd587","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.95--0.98--s44","configuration":"awc8-wsd-h20--0.008--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.008,"seed":44,"loss":3.6192615827092975,"jobId":"2479268","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"ed9ff209fdebf9687610c7465eb547684b7aa1bf5887e62a2d3f3a594bbca199","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.008-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.006--0.95--0.98--s42","configuration":"awc8-wsd-h20--0.006--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.006,"seed":42,"loss":3.620306398628966,"jobId":"2479269","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"bdaeb31ffed42734d1a9e48c48f46aac1d6b892f6893e8337a9d7f49a2b839c5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.006--0.95--0.98--s43","configuration":"awc8-wsd-h20--0.006--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.006,"seed":43,"loss":3.6155477850644306,"jobId":"2479270","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"7f04c528708390400b71e322377f8d9f217433c9bd045bd7728148241154f57e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.006--0.95--0.98--s44","configuration":"awc8-wsd-h20--0.006--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.006,"seed":44,"loss":3.6195543470806535,"jobId":"2479271","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"5bdbb9c80974b08ba729a57cdff4f03b58be4dcc61e81aa45775d48dc588efd1","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.95--0.98--s42","configuration":"awc8-wsd-h20--0.004--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":42,"loss":3.622101986832102,"jobId":"2479272","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"f7163331ba61f1e9c7934a822571d34e39a5d017d29de9a7f48ffde9c0883f56","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.95--0.98--s43","configuration":"awc8-wsd-h20--0.004--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":43,"loss":3.6267173488083095,"jobId":"2479273","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"69d46d065b831692a48e7a7e44b15086074f96d72203ed94d893bed0a3a67224","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.95--0.98--s44","configuration":"awc8-wsd-h20--0.004--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":44,"loss":3.628047020082648,"jobId":"2479274","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"8f414e39923d9737d9f61c46b9e7c53382a8504fa035cdaa39fdcce5de0ec68d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.95--0.98--s42","configuration":"awc8-wsd-h20--0.003--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.6329989668651663,"jobId":"2479308","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"e3d02fd6cd61acba47adaf3e0c32f4f793383ec17128d711763b64a1ee20f8de","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.95--0.98--s43","configuration":"awc8-wsd-h20--0.003--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.6417305349872295,"jobId":"2479309","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"d3c1dbb8d488baea40bc613953ce010f2239aee209b001773709f1180442a548","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.95--0.98--s44","configuration":"awc8-wsd-h20--0.003--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.6354018805587303,"jobId":"2479310","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"58fc4b5a07c72065c270177ca9a8818e281fc8e54c6ac6e8798579b51cd74dbc","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.95--0.98--s42","configuration":"awc8-wsd-h20--0.002--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.677922958460426,"jobId":"2479311","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"d28361c031e0ed7003897ec62a41523dc30d8f06ee73fa1cbb831f4c2cb8173a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.95--0.98--s43","configuration":"awc8-wsd-h20--0.002--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.6730007590378038,"jobId":"2479312","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"ff926f67516efd77af709427c6cfbb873551e0254f0318aead147f87b4604ae4","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.95--0.98--s44","configuration":"awc8-wsd-h20--0.002--0.95--0.98","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.6717062149905817,"jobId":"2479313","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"399f27c41585d31356643a3ef5f9e5295438ce334a92f0a3c16c6c44a824d908","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.95--0.999--s42","configuration":"awc8-wsd-h20--0.01--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.01,"seed":42,"loss":3.5950639300065372,"jobId":"2479314","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"68124984b08d6ca50fc86c13c1ef15e151088e83e1883c00d321498dd28d1b0f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.95--0.999--s43","configuration":"awc8-wsd-h20--0.01--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.01,"seed":43,"loss":3.5922687209068687,"jobId":"2479315","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"22ea9ee9d373b2a80040203397266fe8b4fcb032fbc1f977a1282e4196549c44","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.01--0.95--0.999--s44","configuration":"awc8-wsd-h20--0.01--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.01,"seed":44,"loss":3.6077726903230727,"jobId":"2479316","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"f608bbdf7016f91dadaefd58370addb32759cd65fae1cea2b7cf3b8a44e34821","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.01-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.95--0.999--s42","configuration":"awc8-wsd-h20--0.008--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.008,"seed":42,"loss":3.602557061427824,"jobId":"2479317","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"83ca7d76bf5df5896c395069bc6116a79cc3f36afc5dbf132298ac22eb56860a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.95--0.999--s43","configuration":"awc8-wsd-h20--0.008--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.008,"seed":43,"loss":3.589014955110382,"jobId":"2479318","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"1701e84c843595e224da362d0e354ad6ecf4762680f02f8eb85af7b457203e87","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.008--0.95--0.999--s44","configuration":"awc8-wsd-h20--0.008--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.008,"seed":44,"loss":3.5940841008689515,"jobId":"2479319","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"d3f1de2b7185223efbd27694347e4f488dd4b5e13e97f0b4e62ff7b4393cb8d6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.008-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.006--0.95--0.999--s42","configuration":"awc8-wsd-h20--0.006--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.006,"seed":42,"loss":3.5881057930940465,"jobId":"2479658","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"96123d836996f5486880cee3654a7e6105fc12ee498db3b6b4cb6198ba32fbb2","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/final.pt"},{"id":"awc8-wsd-h20--0.006--0.95--0.999--s43","configuration":"awc8-wsd-h20--0.006--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.006,"seed":43,"loss":3.595977846361822,"jobId":"2479659","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"17b9da036f44b92b3f9b774b83f43a84cbd323aedac1ef1f2469fca87fa3cd25","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/final.pt"},{"id":"awc8-wsd-h20--0.006--0.95--0.999--s44","configuration":"awc8-wsd-h20--0.006--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.006,"seed":44,"loss":3.5953002432009535,"jobId":"2479660","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"80800b13f26159de620f6538a469f84add13c4ed58f965ea6e3fdc61018d57b8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/final.pt"},{"id":"awc8-wsd-h20--0.004--0.95--0.999--s42","configuration":"awc8-wsd-h20--0.004--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":42,"loss":3.5957114146924805,"jobId":"2479661","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"efbc0fbecc6cecd34089980b75314ce6b374da0ae3e094366c22a621bcb9eb96","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.95--0.999--s43","configuration":"awc8-wsd-h20--0.004--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":43,"loss":3.5953545512546574,"jobId":"2479662","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"c40cb7a9f9ba296534c588c9cf0845876f79e406bb939c8f930d580c1452a7e9","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.004--0.95--0.999--s44","configuration":"awc8-wsd-h20--0.004--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":44,"loss":3.603553165567626,"jobId":"2479663","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"c060d02fb5238057d2ab4eec69a11852b15b5fe6e2de99f36ea8c876fe5f104c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.95--0.999--s42","configuration":"awc8-wsd-h20--0.003--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.5982387195327967,"jobId":"2479664","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"c2692992567ea6b3fa916bc28c202b0d89d28e3579cc707c1695d6b5260a3a30","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.95--0.999--s43","configuration":"awc8-wsd-h20--0.003--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.613193303011285,"jobId":"2479665","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"387cfe8165347ac163a74a0316325cee9154bebdc8446e2f7125f5e0da24ba97","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.003--0.95--0.999--s44","configuration":"awc8-wsd-h20--0.003--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.608397527065818,"jobId":"2479666","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"dba98e8ef258152b440ef9af80a2e12411c8eaeaff62703246ec15bbac7e2f15","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.95--0.999--s42","configuration":"awc8-wsd-h20--0.002--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.649128402327742,"jobId":"2479667","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"f810df0244f56fd9aae0688279ff5949bd8ee7595b5e82a82ff4d739136dbe01","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.95--0.999--s43","configuration":"awc8-wsd-h20--0.002--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.6382752088299606,"jobId":"2479668","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"f4e2c782f1bdceb4fdd606b0dc0f5506e7d46d28d7d048898a248bcbb5e6daca","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h20--0.002--0.95--0.999--s44","configuration":"awc8-wsd-h20--0.002--0.95--0.999","method":"awc8","horizon":20,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.640122381938774,"jobId":"2479669","tokens":408027136,"startTokens":0,"incrementalTokens":408027136,"recipeIdentity":"1b7211e3f29624e396b9c48fd1875bfabb6286cfd260c8af322367fd030273dd","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-20/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.006--0.9--0.9--s42","configuration":"awc8-wsd-h40--0.006--0.9--0.9","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.006,"seed":42,"loss":3.6209855104109328,"jobId":"2480625","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"a320bd6bc38c54e847c3074dcfee6a7b5c49d978afb5ce7eb90ac0574d9dd497","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.006--0.9--0.9--s43","configuration":"awc8-wsd-h40--0.006--0.9--0.9","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.006,"seed":43,"loss":3.6323833602012727,"jobId":"2480626","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"226e88acfb25e532f321fb3f5aa910cf80f1f0f25cbcf82ee3f0f281ded3a2ba","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.006--0.9--0.9--s44","configuration":"awc8-wsd-h40--0.006--0.9--0.9","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.006,"seed":44,"loss":3.6347050611070006,"jobId":"2480627","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"14cfe52532738e879450ff01defeb4d491386b29c1589e5d19e5896da721a759","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.9--0.9--s42","configuration":"awc8-wsd-h40--0.004--0.9--0.9","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":42,"loss":3.648053087155755,"jobId":"2480628","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"6c946f13bdb7d58068672c95356b77c1e0a2a13a8eedd488308a7a04ea6a9a6c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.9--0.9--s43","configuration":"awc8-wsd-h40--0.004--0.9--0.9","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":43,"loss":3.64414889483402,"jobId":"2480629","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"863ce256468e42d5d33283d28a961a886f4f12f77c122368c4c24e767fbf28f6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.9--0.9--s44","configuration":"awc8-wsd-h40--0.004--0.9--0.9","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":44,"loss":3.6421183747577417,"jobId":"2480630","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"05ffdc6f00a8cf2b69a2cccf11ba828b2a92b0a144b707559359e67e3b519660","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.9--0.9--s42","configuration":"awc8-wsd-h40--0.003--0.9--0.9","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.6621054579254735,"jobId":"2480632","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"a9b11b7ed0ea6407a5d7e5028402cbea7c01167fd5a2b9f86ed23da930774f3f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.9--0.9--s43","configuration":"awc8-wsd-h40--0.003--0.9--0.9","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.640312957176013,"jobId":"2480633","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"f70909a0deab7b2a89e2803a689f9f1bbc7f23bd7f1a54f587f05584ca0967d0","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.9--0.9--s44","configuration":"awc8-wsd-h40--0.003--0.9--0.9","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.6506803655939053,"jobId":"2480634","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"738ac68007a8ee6c02ad29b366f5d25fe1096a4f56d7c54cd57737ef6a39b401","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.9--0.9--s42","configuration":"awc8-wsd-h40--0.002--0.9--0.9","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.6792063167065248,"jobId":"2480635","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"7a37cc718c440a30a395a6916856fcb90b91372c0b77c5e62151bbb41405be7e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.9--0.9--s43","configuration":"awc8-wsd-h40--0.002--0.9--0.9","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.6598426073959938,"jobId":"2480636","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"634ded3f2f4e15474f6f79f49ceb386ee373806292a187a000a35af3500bc1d2","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.9--0.9--s44","configuration":"awc8-wsd-h40--0.002--0.9--0.9","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.6759053224992413,"jobId":"2480637","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"e328d55db899d770cc42ef2d7ad0d89d73b6641d9411aaa03965bc32e6db42e7","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.006--0.9--0.98--s42","configuration":"awc8-wsd-h40--0.006--0.9--0.98","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.006,"seed":42,"loss":3.5283966212746614,"jobId":"2480638","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"b5a45cbfc7efed828568898c64d16ca598b6725dcd40838c76102abaa78365f6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.006--0.9--0.98--s43","configuration":"awc8-wsd-h40--0.006--0.9--0.98","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.006,"seed":43,"loss":3.5349761858346045,"jobId":"2480639","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"a5ac5094a6c0730d756cc681f27e85e69943bb37ad7bec445a8e2cdcd6c60859","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.006--0.9--0.98--s44","configuration":"awc8-wsd-h40--0.006--0.9--0.98","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.006,"seed":44,"loss":3.529488725798317,"jobId":"2480640","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"06cab3017878f770defb07f5660f005af2725ecaac12d93f5119e139f2de5fe2","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.9--0.98--s42","configuration":"awc8-wsd-h40--0.004--0.9--0.98","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":42,"loss":3.539456550603085,"jobId":"2480641","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"40cd3959b6054bc530c7b0da8f9bf2c50c3944d4667e0bcd1f78d65fce15dde2","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.9--0.98--s43","configuration":"awc8-wsd-h40--0.004--0.9--0.98","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":43,"loss":3.544250028278091,"jobId":"2480642","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"b7d05813332bd3605d4660ea6c2061638407af8f52f2a9a69a56eeedf1480bc6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.9--0.98--s44","configuration":"awc8-wsd-h40--0.004--0.9--0.98","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":44,"loss":3.5346538530127796,"jobId":"2480643","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"b9364d9d515758793624c2b37e80ac05d11c84d377a74d28d30e976dff013e0c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.9--0.98--s42","configuration":"awc8-wsd-h40--0.003--0.9--0.98","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.554913063305192,"jobId":"2480644","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"fc05ae5daacc786e30849162ed387493f61093acacf9a2d241e0813f1d1711d0","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.9--0.98--s43","configuration":"awc8-wsd-h40--0.003--0.9--0.98","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.5514979037753567,"jobId":"2480645","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"4e5e636520f55be752973fcad4fc6a8567a9f2b5993a6c550323f9a5d6cb506f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.9--0.98--s44","configuration":"awc8-wsd-h40--0.003--0.9--0.98","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.544381138406371,"jobId":"2480646","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"e534528ed3fc9cf6a874bcff9ee18f430ad364aebf21828163f5e5085f8fb619","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.9--0.98--s42","configuration":"awc8-wsd-h40--0.002--0.9--0.98","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.575773738630764,"jobId":"2480647","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"54ce58b2f5b617c15cc60e48a79fa4097dfbcc3f0580e39822985c0dacd72a0a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.9--0.98--s43","configuration":"awc8-wsd-h40--0.002--0.9--0.98","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.5570104143639076,"jobId":"2480648","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"9713dd1d21764c88c0284f32a64f5d17fe5f01b1ef3e433bada2698ddcf2410d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.9--0.98--s44","configuration":"awc8-wsd-h40--0.002--0.9--0.98","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.568927825449281,"jobId":"2480649","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"cc0b0fb43a4ef294aa2f9a0ffe5cd07bfebccafbba86f0fcbad43f26f35f10f7","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.006--0.9--0.999--s42","configuration":"awc8-wsd-h40--0.006--0.9--0.999","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.006,"seed":42,"loss":3.4943263633001638,"jobId":"2480650","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"8c904158e05ead49242cd34183b9d21a57fbde80b95f3ffdf67ebef3de44ca62","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.006--0.9--0.999--s43","configuration":"awc8-wsd-h40--0.006--0.9--0.999","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.006,"seed":43,"loss":3.496827753624678,"jobId":"2480651","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"b1d844e65cf0c89fb7933824a07ec7becaedc61aa4025ab5949b13be3cc491e9","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.006--0.9--0.999--s44","configuration":"awc8-wsd-h40--0.006--0.9--0.999","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.006,"seed":44,"loss":3.495527070617188,"jobId":"2480652","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"364db1f3c6359689033ecddc4b42ed28136ca00140d40ae269c0221606520b1f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.9--0.999--s42","configuration":"awc8-wsd-h40--0.004--0.9--0.999","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":42,"loss":3.4996621550603053,"jobId":"2480653","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"ad987437e6b33209382bdba23710e07151e0b22463653cafc755d92b2c162b16","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.9--0.999--s43","configuration":"awc8-wsd-h40--0.004--0.9--0.999","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":43,"loss":3.497683933870879,"jobId":"2480654","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"9b653e8a1d33580b8b0293b5e799565b12348e69e1d31ef18ac5f0fd23d2f424","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.9--0.999--s44","configuration":"awc8-wsd-h40--0.004--0.9--0.999","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":44,"loss":3.497369623464941,"jobId":"2480655","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"a9945daf6cebba4f76234027768a16c6d468523c119b4a67217f540300a81739","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.9--0.999--s42","configuration":"awc8-wsd-h40--0.003--0.9--0.999","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.503650241445448,"jobId":"2480656","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"78117f741c30ec87e30424f4532f1cedffdac6668d753b64e01deb13106cf933","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.9--0.999--s43","configuration":"awc8-wsd-h40--0.003--0.9--0.999","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.515859640584827,"jobId":"2480657","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"a0d9ef891426920a1f3ed909503ea2591418cb5e784e86f9c3da53cb0b331fe4","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.9--0.999--s44","configuration":"awc8-wsd-h40--0.003--0.9--0.999","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.50444245450166,"jobId":"2480658","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"f2b83f54f951a4da8b44a9dc4be85d6cda32f886c9038286d8cc2969e6545f83","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.9--0.999--s42","configuration":"awc8-wsd-h40--0.002--0.9--0.999","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.531035744998508,"jobId":"2480659","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"29ce3b688613d52c5a5c02076ab29b18b563223abe3993b727cb048382f37020","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.9--0.999--s43","configuration":"awc8-wsd-h40--0.002--0.9--0.999","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.526083773394841,"jobId":"2480660","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"ca2cb3489a93c46ca47fc74ead8a62e12a62a786492db842351b464a5902e599","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.9--0.999--s44","configuration":"awc8-wsd-h40--0.002--0.9--0.999","method":"awc8","horizon":40,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.528276419588812,"jobId":"2480661","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"afb44ca80dc09a20f57270b7ff4e65a83385214f3b90f67c75a098b6c7ceeca5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.006--0.95--0.9--s42","configuration":"awc8-wsd-h40--0.006--0.95--0.9","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.006,"seed":42,"loss":3.56617539654266,"jobId":"2480662","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"20c76679f4b2d44ca5bd95fba39d9aee80845441079045c5299f7df4011baa13","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.006--0.95--0.9--s43","configuration":"awc8-wsd-h40--0.006--0.95--0.9","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.006,"seed":43,"loss":3.5698809137446466,"jobId":"2480663","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"78471eb74e19389b4ed65846ca13504be2ef89a61344ed670c7c67c3cbaafa80","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.006--0.95--0.9--s44","configuration":"awc8-wsd-h40--0.006--0.95--0.9","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.006,"seed":44,"loss":3.5757832803340976,"jobId":"2480664","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"3abcb8eb5b42802b8fbd9f214a14d9e4633920aeb824cde591a1e21958a66381","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.95--0.9--s42","configuration":"awc8-wsd-h40--0.004--0.95--0.9","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":42,"loss":3.5598033404702436,"jobId":"2480665","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"c62f1e9c3aa5c978aa3f4b969c5481a5cc3e1c89fb29d640991b936d1cfff9b4","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.95--0.9--s43","configuration":"awc8-wsd-h40--0.004--0.95--0.9","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":43,"loss":3.5696122454449406,"jobId":"2480666","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"b201b665385ce6b61f50f6a3280d284d30af2d6c1b37d19a25221cb2082eef38","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.95--0.9--s44","configuration":"awc8-wsd-h40--0.004--0.95--0.9","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":44,"loss":3.5710064930191603,"jobId":"2480667","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"05e108a2ce9345a4ae2ba41eff7a1d08bc4e76a41d7f192204706bf074b757ab","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.95--0.9--s42","configuration":"awc8-wsd-h40--0.003--0.95--0.9","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.5714906390281143,"jobId":"2480668","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"8d720a3e981bd355b6f1f21443093cde8b3217905f5cbce3935c8d946cc2aa06","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.95--0.9--s43","configuration":"awc8-wsd-h40--0.003--0.95--0.9","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.576699536579847,"jobId":"2480669","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"a3088b8abb561806032d0f55e8cd6d6e6cd56b0566940ad7dbadd49b9551050e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.95--0.9--s44","configuration":"awc8-wsd-h40--0.003--0.95--0.9","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.578922651370372,"jobId":"2480670","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"2a60dc08718f51d6341882510ad4d120ca781d12ee4c3c118c3d3f061bc9eb48","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.95--0.9--s42","configuration":"awc8-wsd-h40--0.002--0.95--0.9","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.587098174154891,"jobId":"2480671","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"0695af5503ef72ac20eb9d69187842c955baa5331ed727f16df81c59243e5c54","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.95--0.9--s43","configuration":"awc8-wsd-h40--0.002--0.95--0.9","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.5880995088563816,"jobId":"2480672","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"8dbec3e35359f2783ab9822db33ab167effb71c8414780394a77534afe98605b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.95--0.9--s44","configuration":"awc8-wsd-h40--0.002--0.95--0.9","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.5912551195912434,"jobId":"2480673","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"6fe24bf5d539f7a65af01f604ecd2c1be81af9138f22af7d41f756c284425315","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.006--0.95--0.98--s42","configuration":"awc8-wsd-h40--0.006--0.95--0.98","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.006,"seed":42,"loss":3.508575881067268,"jobId":"2481085","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"021c54f018081dbbda4334d104054ea341460e292ad87a9e173395b6b3b382ba","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.006--0.95--0.98--s43","configuration":"awc8-wsd-h40--0.006--0.95--0.98","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.006,"seed":43,"loss":3.5078378357030546,"jobId":"2481086","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"2f282ec91df84c979b87c0a2dbfadf9fcb4e72c325287d21a9685a34361d9fc6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.006--0.95--0.98--s44","configuration":"awc8-wsd-h40--0.006--0.95--0.98","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.006,"seed":44,"loss":3.507598741533378,"jobId":"2481087","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"3f71596b1cc4d6996d7a4ee25d608392b7f74b62127556f45fd94ccbfeb453aa","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.95--0.98--s42","configuration":"awc8-wsd-h40--0.004--0.95--0.98","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":42,"loss":3.5070463037187873,"jobId":"2481088","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"909400517bf7ddc62282f2ad1d63e5fbf0cc91b2ceaecbee3bd13e1197259bb8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.95--0.98--s43","configuration":"awc8-wsd-h40--0.004--0.95--0.98","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":43,"loss":3.5105572381927987,"jobId":"2481089","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"082147df37807ff31bca126f8782c53f2024960b7a167f460f75ef9fb323a5fd","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.95--0.98--s44","configuration":"awc8-wsd-h40--0.004--0.95--0.98","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":44,"loss":3.5119876925955005,"jobId":"2481090","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"92c453d654c0fdbd4946631bfbfc854fa8d0d2fe44309184590ec7d5477fd1e1","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.95--0.98--s42","configuration":"awc8-wsd-h40--0.003--0.95--0.98","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.5088096238591273,"jobId":"2481091","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"1742112f5e28acd09d94ba266a91758d5adfa0b4ed52209df512d82292f23ec7","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.95--0.98--s43","configuration":"awc8-wsd-h40--0.003--0.95--0.98","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.518283526200785,"jobId":"2481092","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"eb8853a22264bd174573eee50bf8d7aa9d960fc095673061c41ab126f8f2a331","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.95--0.98--s44","configuration":"awc8-wsd-h40--0.003--0.95--0.98","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.513770410851948,"jobId":"2481093","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"f485ff21655ebc865c517f4b7ebe593a9044daac83058cb3b7c6c027c365140e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.95--0.98--s42","configuration":"awc8-wsd-h40--0.002--0.95--0.98","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.5451323272193713,"jobId":"2481094","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"59c1f97dec45d437e3ced0d08d7e9cf10d557010dd900ec9a422a82df2ee5e7c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.95--0.98--s43","configuration":"awc8-wsd-h40--0.002--0.95--0.98","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.5433717236466116,"jobId":"2481095","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"dd7156a6509ab9eb2075a92ce9f55e748dae4ff9b97b82887a505386b955e910","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.95--0.98--s44","configuration":"awc8-wsd-h40--0.002--0.95--0.98","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.538147812999349,"jobId":"2481096","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"ea414fb12070c56c8608295408e865169a0d8d1a27426f7d95c0a26c3a68c350","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.006--0.95--0.999--s42","configuration":"awc8-wsd-h40--0.006--0.95--0.999","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.006,"seed":42,"loss":3.4735672746780732,"jobId":"2481097","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"e6f2593db657136a1a8e50acaef613c801fa83625316613394a80c83980e1c4d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/final.pt"},{"id":"awc8-wsd-h40--0.006--0.95--0.999--s43","configuration":"awc8-wsd-h40--0.006--0.95--0.999","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.006,"seed":43,"loss":3.4789400999554836,"jobId":"2481098","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"43dae8aafe05e36d73781226b5d94144d1b899d74dcf461905204cc1f8cf9de6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/final.pt"},{"id":"awc8-wsd-h40--0.006--0.95--0.999--s44","configuration":"awc8-wsd-h40--0.006--0.95--0.999","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.006,"seed":44,"loss":3.4763383999425486,"jobId":"2481099","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"3db56bb253af1abee8ac4f594bb4cc7934f1ea50afd9ee571e1a13d8e7f4ba2e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/final.pt"},{"id":"awc8-wsd-h40--0.004--0.95--0.999--s42","configuration":"awc8-wsd-h40--0.004--0.95--0.999","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":42,"loss":3.4738049895081455,"jobId":"2481100","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"d1835c98fe6ab1ae8a9112c94dffe867f135802ff1c1bb5f76d1d2a9820d4da0","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.95--0.999--s43","configuration":"awc8-wsd-h40--0.004--0.95--0.999","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":43,"loss":3.4773626034714615,"jobId":"2481101","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"fddd30b411d0c9f1902a7d91d8d0e1eeecc9d195be7506545203d87876e9ea76","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.004--0.95--0.999--s44","configuration":"awc8-wsd-h40--0.004--0.95--0.999","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":44,"loss":3.4839971104532155,"jobId":"2481102","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"5017ad899ec20cdea7253aed987bf7acd3424b65a75506480d52571ca4ac221a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.95--0.999--s42","configuration":"awc8-wsd-h40--0.003--0.95--0.999","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.4750812273482987,"jobId":"2481103","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"ab277b0ae79c238e590aaee645ff1ed1a6de612cad610d4c4cced23aef522150","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.95--0.999--s43","configuration":"awc8-wsd-h40--0.003--0.95--0.999","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.4859453615064666,"jobId":"2481104","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"916802bdb54c05fdf34bce89371003ba29b1ab7dbbe16e56946184db4bd70b39","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.003--0.95--0.999--s44","configuration":"awc8-wsd-h40--0.003--0.95--0.999","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.482726184281099,"jobId":"2481105","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"c844d0a29cea7542c33f01c70c5eaa88d94f47b11345237b6579635b556d590c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.95--0.999--s42","configuration":"awc8-wsd-h40--0.002--0.95--0.999","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.5072350134506016,"jobId":"2481106","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"095acec6c28d4db570adbf8aafc6d3ec619a58c775cae51c5d7c865de6cd51bd","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.95--0.999--s43","configuration":"awc8-wsd-h40--0.002--0.95--0.999","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.5055125890623753,"jobId":"2481107","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"e9ee41b7330175910f22cf2570f2f3bff8208bd387bf18336e9456431a44c54e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h40--0.002--0.95--0.999--s44","configuration":"awc8-wsd-h40--0.002--0.95--0.999","method":"awc8","horizon":40,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.506473711511068,"jobId":"2481108","tokens":816185344,"startTokens":357040128,"incrementalTokens":459145216,"recipeIdentity":"a365bf3a61e7c80eab29f61e0e52cdf05644b0a835a131d7c459ea03e5ddfce5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-40/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.9--0.9--s42","configuration":"awc8-wsd-h80--0.006--0.9--0.9","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.006,"seed":42,"loss":3.548122137708731,"jobId":"2481774","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"c2f544e2c66f5674e8518845a445ba9bf92fc14256aff00bb2483317aa005ce7","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.9--0.9--s43","configuration":"awc8-wsd-h80--0.006--0.9--0.9","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.006,"seed":43,"loss":3.5535805514107945,"jobId":"2481775","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"bface5c1fcff393cc96ea5b170448e3d0e81ba45d980dcc93063fb3dd46d2880","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.9--0.9--s44","configuration":"awc8-wsd-h80--0.006--0.9--0.9","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.006,"seed":44,"loss":3.557084646358441,"jobId":"2481776","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"14c7d5939ce93fbcee6c6da7de29a9b2d332d67a1f42e78fcf32b592c2aa8dab","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.9--0.9--s42","configuration":"awc8-wsd-h80--0.004--0.9--0.9","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":42,"loss":3.571304422519585,"jobId":"2481777","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"8b9d3e6341754ac3fbb7e37a4de8ceef881d9f454c4db6fd9605a4d8161daada","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.9--0.9--s43","configuration":"awc8-wsd-h80--0.004--0.9--0.9","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":43,"loss":3.5640851911404656,"jobId":"2481778","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"fb7b4e1d5b21826480a1b1f827bf340bb8a244d2e14acab2287526502d404cab","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.9--0.9--s44","configuration":"awc8-wsd-h80--0.004--0.9--0.9","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.004,"seed":44,"loss":3.562859040919509,"jobId":"2481779","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"c8a2d2bf04e6632cab7028709ab302f9cec34a8ed8441dd751ba96f9d0f091a9","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.003--0.9--0.9--s42","configuration":"awc8-wsd-h80--0.003--0.9--0.9","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.5737734380320734,"jobId":"2481780","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"b4a8835d20266caa506a44d44d36ebc8ec5aec2a9c926f371458e0fea045fff6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.003--0.9--0.9--s43","configuration":"awc8-wsd-h80--0.003--0.9--0.9","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.5600984304816894,"jobId":"2481781","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"d78484fbdbbfed0fce0a944a30c0725e37d82d083d81e98013c0f3e6b1754891","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.003--0.9--0.9--s44","configuration":"awc8-wsd-h80--0.003--0.9--0.9","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.5680965025025033,"jobId":"2481782","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"726114c13eb4db67a0b450a5336e5e1eb52e3ce013b73a721033f9921c2c9a19","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.9--0.9--s42","configuration":"awc8-wsd-h80--0.002--0.9--0.9","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.5877496646091758,"jobId":"2481783","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"d42f2c93ff946079f0b6aaec500543cccf45b203a75ab839f259150cde1f39d7","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.9--0.9--s43","configuration":"awc8-wsd-h80--0.002--0.9--0.9","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.5747625272557997,"jobId":"2481784","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"ffcf85665df13ce512af4d02b8a298aee9a77c498a8d08127e34adfe487f1330","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.9--0.9--s44","configuration":"awc8-wsd-h80--0.002--0.9--0.9","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.582056787771072,"jobId":"2481785","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"bbe93010ec1db7eeb1b69352d308c493b4b2686a6ef81bd08faa0df9e28006af","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.9--0.98--s42","configuration":"awc8-wsd-h80--0.006--0.9--0.98","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.006,"seed":42,"loss":3.4527778053217584,"jobId":"2481786","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"82cb9ebe0a4c528ccca5b3ebd451b922c89d5a2242edfc7dbe62230cfd2aa1c2","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.9--0.98--s43","configuration":"awc8-wsd-h80--0.006--0.9--0.98","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.006,"seed":43,"loss":3.4597880102890275,"jobId":"2481787","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"947e6da995d5cde436e065c4367d604ebf1dfc4d9b1cd3f75b00cd4ac7ca768e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.9--0.98--s44","configuration":"awc8-wsd-h80--0.006--0.9--0.98","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.006,"seed":44,"loss":3.4537285134678437,"jobId":"2481788","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"19644bae128ed05a53293480ee643f1d5b9982325f699f3cb5f85e6f033c3b87","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.9--0.98--s42","configuration":"awc8-wsd-h80--0.004--0.9--0.98","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":42,"loss":3.4553320907057965,"jobId":"2481789","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"eeda0fcd35d48383ab961e582cefa068c2438442f04840aaaf0c4d3f1f794515","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.9--0.98--s43","configuration":"awc8-wsd-h80--0.004--0.9--0.98","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":43,"loss":3.4620790641990338,"jobId":"2481790","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"47541316a0cab77f8ba8b633115d432e8b5ca52a7ebe0ac92c0054ff16a28eae","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.9--0.98--s44","configuration":"awc8-wsd-h80--0.004--0.9--0.98","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.004,"seed":44,"loss":3.4536922939559127,"jobId":"2481791","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"4aff13aa2e8f6fbd2fc1fb7772a27d262d29a8c333daf8fe43723de5ca5710b5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.003--0.9--0.98--s42","configuration":"awc8-wsd-h80--0.003--0.9--0.98","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.463526002061837,"jobId":"2481792","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"5a456ccebca83a3f9e12281ebd63334c1406b131d943e2a788ea5de9c8fa0466","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.003--0.9--0.98--s43","configuration":"awc8-wsd-h80--0.003--0.9--0.98","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.4653852986364355,"jobId":"2481793","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"d48039310e2d4e202c0eb5b27c4b65aac3f316fb26be1cdbe4439bb97a88de28","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.003--0.9--0.98--s44","configuration":"awc8-wsd-h80--0.003--0.9--0.98","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.4570630963915265,"jobId":"2481794","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"58c2b208deb6e68c14ae88380404eb05da069cb948be94254ece10179f323772","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.9--0.98--s42","configuration":"awc8-wsd-h80--0.002--0.9--0.98","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.4822807341736066,"jobId":"2481795","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"03caa69fc2a2a3ad674bd7f4d5f32a722d08e5d5ae9a9be7e5f6ecedfd46db96","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.9--0.98--s43","configuration":"awc8-wsd-h80--0.002--0.9--0.98","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.467179883755508,"jobId":"2481796","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"944a04c9a3f189941e4794e5483996f108182266d8dbabde0450f8c758bdcf10","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.9--0.98--s44","configuration":"awc8-wsd-h80--0.002--0.9--0.98","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.4738058380975665,"jobId":"2481797","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"9530a1585cd21cfb097163755126210cf89e860866d192875e180c1e015b13f7","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.9--0.999--s42","configuration":"awc8-wsd-h80--0.006--0.9--0.999","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.006,"seed":42,"loss":3.415323920062671,"jobId":"2481798","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"14f600e3d2cec74d65fdf5d93a86e0dbaaba3e13a9671b2ab8defc723c11c468","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.9--0.999--s43","configuration":"awc8-wsd-h80--0.006--0.9--0.999","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.006,"seed":43,"loss":3.4161165450787174,"jobId":"2481799","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"917572a4d28e11c125d22aba6ab6f5346b195679f6e052afcffde207928f084a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.9--0.999--s44","configuration":"awc8-wsd-h80--0.006--0.9--0.999","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.006,"seed":44,"loss":3.416442924180625,"jobId":"2481800","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"81cc338e98aa7d5de419f75428f68a57f4fbdb6e7a95acd23d865b20d1a5eeea","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.9--0.999--s42","configuration":"awc8-wsd-h80--0.004--0.9--0.999","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":42,"loss":3.4165495016981873,"jobId":"2481801","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"9f24f742bc898be04a074dab0265723bde7f5ca97f37c55c3d71d412d454c7db","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.9--0.999--s43","configuration":"awc8-wsd-h80--0.004--0.9--0.999","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":43,"loss":3.414990330896003,"jobId":"2481802","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"1ccc5059b9c7666d0515f2714395af256149610bbef62f61e6745090033db110","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.9--0.999--s44","configuration":"awc8-wsd-h80--0.004--0.9--0.999","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.004,"seed":44,"loss":3.4110390871190117,"jobId":"2481803","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"68a11f13fbcf0ff1bf047d2ec63d59cac44b09edc8127bf7fed909cfc826dd8b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.003--0.9--0.999--s42","configuration":"awc8-wsd-h80--0.003--0.9--0.999","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.41746784427521,"jobId":"2481804","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"e642983b4093fd618803c2e6cc1d3c6de534ae2e0cc2cb09f1c1d30c06aa728f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.003--0.9--0.999--s43","configuration":"awc8-wsd-h80--0.003--0.9--0.999","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.4265044628690835,"jobId":"2481805","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"1eb91a952f4cae71e2d946688da5580e324d33a06aca7b85e6cfcea5d424a357","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.003--0.9--0.999--s44","configuration":"awc8-wsd-h80--0.003--0.9--0.999","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.4145587232316856,"jobId":"2481806","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"ea2bab4899467f51e038fbb2f95d9b6be5a71355d311fe0ab7e0ba3a17f1418a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.9--0.999--s42","configuration":"awc8-wsd-h80--0.002--0.9--0.999","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.433559710226971,"jobId":"2481807","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"6a3e866907f811331edbdcfa10bd78447d6b5ec53457ad94cb50b22945353943","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.9--0.999--s43","configuration":"awc8-wsd-h80--0.002--0.9--0.999","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.431684296599056,"jobId":"2481808","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"658c9cc7a95570e5cf725494839ae854f47c13ee2328532a9f591d2f8cc079d9","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.9--0.999--s44","configuration":"awc8-wsd-h80--0.002--0.9--0.999","method":"awc8","horizon":80,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.4305317166636695,"jobId":"2481809","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"3b645335dc09d815f1443d8d170c2a61a5eb999899713f8f2d6bf32aa047a211","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.95--0.9--s42","configuration":"awc8-wsd-h80--0.006--0.95--0.9","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.006,"seed":42,"loss":3.491702525862739,"jobId":"2481810","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"216d668b7ef33c7acb11f9b716395ef05e1e1bc5b2ba298a447fbfadc77291a4","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.95--0.9--s43","configuration":"awc8-wsd-h80--0.006--0.95--0.9","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.006,"seed":43,"loss":3.495890585457421,"jobId":"2481811","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"95613d0a72af7a660027f745623185bf8d778fad5d4067f8ca41a9272fb0001d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.95--0.9--s44","configuration":"awc8-wsd-h80--0.006--0.95--0.9","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.006,"seed":44,"loss":3.5002686205334763,"jobId":"2481812","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"2307a214ffd389c3293adde39d86c534b84167f84253e55219563c389e8cbabd","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.95--0.9--s42","configuration":"awc8-wsd-h80--0.004--0.95--0.9","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":42,"loss":3.4847878325903787,"jobId":"2481813","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"49c74d8b050423808b0da4cc32b6ade41e0764e61dc6ded47f1f173510058c6f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.95--0.9--s43","configuration":"awc8-wsd-h80--0.004--0.95--0.9","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":43,"loss":3.49333101963653,"jobId":"2481814","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"ee774196f9aa83fc3e112fd757c98e96795ed4b3505214539822e1c038a02460","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.95--0.9--s44","configuration":"awc8-wsd-h80--0.004--0.95--0.9","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.004,"seed":44,"loss":3.493220159276739,"jobId":"2481815","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"0479129ece57f04975245211d6d532fc5b04db4d36cdaf128f7699b8e173014f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.003--0.95--0.9--s42","configuration":"awc8-wsd-h80--0.003--0.95--0.9","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.4940040759130357,"jobId":"2481816","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"aa4302c20a66f16330f92efe19f6fd6bb037b2e17f79e2ad53ced1b0932ff01b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.003--0.95--0.9--s43","configuration":"awc8-wsd-h80--0.003--0.95--0.9","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.494852123273075,"jobId":"2481817","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"5e11520544593a2addcb4241d09a18ac6602a2bf5d87ae824ed9a8c31ae58c0c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.003--0.95--0.9--s44","configuration":"awc8-wsd-h80--0.003--0.95--0.9","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.4974370610999816,"jobId":"2481818","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"9bf3b2e4247426c07cf9670d2da942ed6a66266f2e0b0699b70ed4c2e93e7c52","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.95--0.9--s42","configuration":"awc8-wsd-h80--0.002--0.95--0.9","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.5042184075782337,"jobId":"2481819","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"19e59a263a3cf218909883df36532d62341c3c0c35d73514b93c834483a52898","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.95--0.9--s43","configuration":"awc8-wsd-h80--0.002--0.95--0.9","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.508940097905032,"jobId":"2481820","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"31081c272aee34a76083bf9e7bd699f7687fa51b1bd59f696bf40303962ed24d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.95--0.9--s44","configuration":"awc8-wsd-h80--0.002--0.95--0.9","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.509875353364249,"jobId":"2481821","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"4bbdc4414ee734a474c5d212a885011b38ba26d398b7ecc29b6288a363b62ab8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.95--0.98--s42","configuration":"awc8-wsd-h80--0.006--0.95--0.98","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.006,"seed":42,"loss":3.429070065939246,"jobId":"2482106","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"c434bab36169de13bbad361f3fb762b0a7fc291068de0e1f00927046b77c1af0","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.95--0.98--s43","configuration":"awc8-wsd-h80--0.006--0.95--0.98","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.006,"seed":43,"loss":3.431562036894717,"jobId":"2482107","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"197e54dddf64fee322942d02376a0d392ac7294001161ad0247e1fbc5e107692","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.95--0.98--s44","configuration":"awc8-wsd-h80--0.006--0.95--0.98","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.006,"seed":44,"loss":3.4307619388913335,"jobId":"2482108","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"cb28c3e43fd58a885eea8c6f01f60f25115f110938379d8ed68b674122dff8b1","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.95--0.98--s42","configuration":"awc8-wsd-h80--0.004--0.95--0.98","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":42,"loss":3.4282375474222753,"jobId":"2482109","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"73ff79d52cb6fcdf88db094defcbe3f092ba3d9aea2af07437b1fbd35a6f5e22","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.95--0.98--s43","configuration":"awc8-wsd-h80--0.004--0.95--0.98","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":43,"loss":3.4317096205957034,"jobId":"2482110","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"84b524d14bacc8e89ca530a64b7658c90be8c5a41d55763d67695b4508f5830c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.95--0.98--s44","configuration":"awc8-wsd-h80--0.004--0.95--0.98","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.004,"seed":44,"loss":3.431566621085528,"jobId":"2482111","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"6c56d93122ce190e094bdcc52e7d33b8c324e6af0f0c6de5f2f75275340d2235","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.003--0.95--0.98--s42","configuration":"awc8-wsd-h80--0.003--0.95--0.98","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.425987926326014,"jobId":"2482112","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"9685eb3d5e8d5c50551f5eeee2731bd7263f0af96e8469299c84c49ac7a7f32a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.003--0.95--0.98--s43","configuration":"awc8-wsd-h80--0.003--0.95--0.98","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.4343659598054344,"jobId":"2482113","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"208df82acf96e75064e44629ea708657712f17990880b3896646b07865841da5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.003--0.95--0.98--s44","configuration":"awc8-wsd-h80--0.003--0.95--0.98","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.43147567038545,"jobId":"2482114","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"66124a8942fa1e4b467eb3132973809dbefde0a4f93e6399462369a9a8c3dcc8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.95--0.98--s42","configuration":"awc8-wsd-h80--0.002--0.95--0.98","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.451563449735407,"jobId":"2482115","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"e92e02e398e7ca62a11f30e1d904a947e893219e583a2316e7da4f1863bf7774","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.95--0.98--s43","configuration":"awc8-wsd-h80--0.002--0.95--0.98","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.454236601633877,"jobId":"2482116","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"3e3b7bef8628c4bb881c7c66752c238287669d6a9ed63e9f2c20c5cca951b854","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.95--0.98--s44","configuration":"awc8-wsd-h80--0.002--0.95--0.98","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.448060505913961,"jobId":"2482117","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"6028f48984e92aa1337ed658409ab8fd39c48128b96aaf69adc8615b99a9d14d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.95--0.999--s42","configuration":"awc8-wsd-h80--0.006--0.95--0.999","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.006,"seed":42,"loss":3.394842061696633,"jobId":"2482118","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"71e913d641d991d6835a22dc1b9ae2f7ff786a4b9f675e3fe1ea1b79208257fa","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.95--0.999--s43","configuration":"awc8-wsd-h80--0.006--0.95--0.999","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.006,"seed":43,"loss":3.3995739058385386,"jobId":"2482119","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"4f4f86ec8317e2dab01fea347fa5db4aa305b71e290bad98391d07e831d2db0a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.006--0.95--0.999--s44","configuration":"awc8-wsd-h80--0.006--0.95--0.999","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.006,"seed":44,"loss":3.3955923765626985,"jobId":"2482120","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"0b0bcdd1a071a39307e226d32a320e0e37de8d87d37fed63ec0614411405cdc0","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.006-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.95--0.999--s42","configuration":"awc8-wsd-h80--0.004--0.95--0.999","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":42,"loss":3.389211589365249,"jobId":"2482121","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"138744e39b0243bc794b66ab5af9da07e9b6de0aeff44552a5de72bf45f87f43","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.95--0.999--s43","configuration":"awc8-wsd-h80--0.004--0.95--0.999","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":43,"loss":3.3958276538213052,"jobId":"2482122","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"1fa2f89a26cbbed64cb297df6248e4ca7da3b1e0652b25e6b43c79c513a591ac","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.004--0.95--0.999--s44","configuration":"awc8-wsd-h80--0.004--0.95--0.999","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.004,"seed":44,"loss":3.4009676760500986,"jobId":"2482123","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"f219c554de6146cebcbd107939d10041be34e0ba920866d2745143913d44cb1a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.004-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.003--0.95--0.999--s42","configuration":"awc8-wsd-h80--0.003--0.95--0.999","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.3901543558150875,"jobId":"2482124","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"76f519163ac233eb6964fbf33c7e709daea62607161e95d3ee4d204b68e661c0","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/final.pt"},{"id":"awc8-wsd-h80--0.003--0.95--0.999--s43","configuration":"awc8-wsd-h80--0.003--0.95--0.999","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.3991775153865236,"jobId":"2482125","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"b38cafabd16cf8321a81eb37d669a14db06d964aa68e49834a7797228514ba44","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/final.pt"},{"id":"awc8-wsd-h80--0.003--0.95--0.999--s44","configuration":"awc8-wsd-h80--0.003--0.95--0.999","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.3966294039874243,"jobId":"2482126","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"03cab47858b55e02e969f32ba79b9b866e2a4ffab84e8ee29e5784a82f3df72a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/final.pt"},{"id":"awc8-wsd-h80--0.002--0.95--0.999--s42","configuration":"awc8-wsd-h80--0.002--0.95--0.999","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.4125951797725493,"jobId":"2482127","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"703838c8661c7739dda29096ec11c1ab9ac0b9463c57398b830e6a5ef54531af","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.95--0.999--s43","configuration":"awc8-wsd-h80--0.002--0.95--0.999","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.4147759569561096,"jobId":"2482128","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"97ebdce4b9b90a0b5a45a1dfb2ea127a43c8b64f1e01c41c53aa46551053811f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h80--0.002--0.95--0.999--s44","configuration":"awc8-wsd-h80--0.002--0.95--0.999","method":"awc8","horizon":80,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.41218367876118,"jobId":"2482129","tokens":1632239616,"startTokens":734527488,"incrementalTokens":897712128,"recipeIdentity":"590a0070fd123dcc19744fb2ad01a50aa8cca1f92488804ece0544bc50bdfbe5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-80/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.003--0.9--0.9--s42","configuration":"awc8-wsd-h120--0.003--0.9--0.9","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.5302565996856616,"jobId":"2482589","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"e14d9486565665d2ea54939cb9db23cbf3e05b7c089029d6d31a61b2009a07fe","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.003--0.9--0.9--s43","configuration":"awc8-wsd-h120--0.003--0.9--0.9","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.523836016022851,"jobId":"2482590","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"b7a902ca136f3ec5d27cc19f53ebab0cda0a26f7c173112f61551c1ba30e5b9b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.003--0.9--0.9--s44","configuration":"awc8-wsd-h120--0.003--0.9--0.9","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.53144581791852,"jobId":"2482591","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"8ebf89405196743b4d09cf2bba223566fa8ae360cd2447f1b45d37eb31fcb2aa","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.9--0.9--s42","configuration":"awc8-wsd-h120--0.002--0.9--0.9","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.5409533873909926,"jobId":"2482592","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"21f1491540b88f787c4189e1e3696c9b13e6d281a771ed884926fc595670357d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.9--0.9--s43","configuration":"awc8-wsd-h120--0.002--0.9--0.9","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.537012955573432,"jobId":"2482593","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"ecdceb353dee6714840128e40f5e4f54fa6e974af82c66fb2c5ac2223214f1d6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.9--0.9--s44","configuration":"awc8-wsd-h120--0.002--0.9--0.9","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.5427450870510637,"jobId":"2482594","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"5fc7942e76a196423554e6b438618b4f9ba58b246580b99d2cafbd3d5ec31f4c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.003--0.9--0.98--s42","configuration":"awc8-wsd-h120--0.003--0.9--0.98","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.4219033785359185,"jobId":"2482595","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"95a69f33712f12107949a8e3f873b9a7e15ba4b4050b4c67d166f4930e0add42","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.003--0.9--0.98--s43","configuration":"awc8-wsd-h120--0.003--0.9--0.98","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.426957445863731,"jobId":"2482596","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"ec11577a5cc4f74475bffa54661304f8c8d9d21c4567a0001809463d8d892e39","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.003--0.9--0.98--s44","configuration":"awc8-wsd-h120--0.003--0.9--0.98","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.4174817738686873,"jobId":"2482597","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"b8c41d9534f8bce0bdda4614fa428ab5cdb9dc03b25dc4399eeb70aedc252dd7","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.9--0.98--s42","configuration":"awc8-wsd-h120--0.002--0.9--0.98","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.4369610570896363,"jobId":"2482598","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"4c0567ca492a2d7b60d1c8fbcb75abeab397537c2f8a26a15a5019da7c33da5e","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.9--0.98--s43","configuration":"awc8-wsd-h120--0.002--0.9--0.98","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.425501630079509,"jobId":"2482599","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"8d483290942b186734c93c48692b178961bca881fd92251c6f3e15e44bf4de55","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.9--0.98--s44","configuration":"awc8-wsd-h120--0.002--0.9--0.98","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.4301407902846344,"jobId":"2482600","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"e27ed699c1199f0e866a57bec7f05eb05c665d880748ff2d824569fa6645af04","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.003--0.9--0.999--s42","configuration":"awc8-wsd-h120--0.003--0.9--0.999","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.3777699664737395,"jobId":"2482601","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"068ff7ed9340c26aa55ec2db14309c51afb30dac74875affd32bc26110133bc4","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.003--0.9--0.999--s43","configuration":"awc8-wsd-h120--0.003--0.9--0.999","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.385437664644348,"jobId":"2482602","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"40f9eef31dd0f2deeae7139ade500078d79fdb61354a99e01e4f7d45a46280bd","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.003--0.9--0.999--s44","configuration":"awc8-wsd-h120--0.003--0.9--0.999","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.3730460376330753,"jobId":"2482603","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"0c3a60b25c9262f6a013dae39ff66ab9413b6fd0137f10057971688b95e7d2d5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.9--0.999--s42","configuration":"awc8-wsd-h120--0.002--0.9--0.999","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.3875030628144307,"jobId":"2482604","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"e4dc93c704e9170517a38b072bd7c723a99e8665b42c1e021f8b133fefe33ef9","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.9--0.999--s43","configuration":"awc8-wsd-h120--0.002--0.9--0.999","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.386587167274686,"jobId":"2482605","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"afb0cd3807eb251d04fb89becc78a6b8ba8d2d8126f38d101f47f7d069bfadfe","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.9--0.999--s44","configuration":"awc8-wsd-h120--0.002--0.9--0.999","method":"awc8","horizon":120,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.386050278247549,"jobId":"2482606","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"2bcb0ec3776202a9a5f3de54d3107f13cc0f0bf929daf7a882cd6c97f35c1b65","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.003--0.95--0.9--s42","configuration":"awc8-wsd-h120--0.003--0.95--0.9","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.4565324680088088,"jobId":"2482607","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"d3fb75594397f749f86faaac9d8c6a9441c757aaa1299c24c52f625133a37ef9","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.003--0.95--0.9--s43","configuration":"awc8-wsd-h120--0.003--0.95--0.9","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.4592033680350687,"jobId":"2482608","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"dedf1d797344e95f44f5cd1d265acb62189451c7fd4a6f1085cb88f67a8c5bd5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.003--0.95--0.9--s44","configuration":"awc8-wsd-h120--0.003--0.95--0.9","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.460333828039905,"jobId":"2482609","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"4309c1d25bb5875d3c3fb05cf12be9850b6400b3bbcd07ad62d6916d04312699","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.95--0.9--s42","configuration":"awc8-wsd-h120--0.002--0.95--0.9","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.466457894755254,"jobId":"2482610","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"16144d3a7f152c2daaa9b461d29985e1805916d5b3fb1ac8dfcf0c5605c6c439","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.95--0.9--s43","configuration":"awc8-wsd-h120--0.002--0.95--0.9","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.4723190263726935,"jobId":"2482611","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"8a260e64d501acfe57c71485ae753a6489e8d722eaa5eb9a519451987bb28767","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.95--0.9--s44","configuration":"awc8-wsd-h120--0.002--0.95--0.9","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.472421223193334,"jobId":"2482612","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"f220667c2cebcee4598446f10850c61c6309e2588f4e9976e2d6697446bdc424","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.003--0.95--0.98--s42","configuration":"awc8-wsd-h120--0.003--0.95--0.98","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.38850917424433,"jobId":"2482613","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"c28fc938b8e56f9bd5d55fd60708f0da25825703e15fc4e8e7b6fe34c75406bf","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.003--0.95--0.98--s43","configuration":"awc8-wsd-h120--0.003--0.95--0.98","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.396818354457583,"jobId":"2482614","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"bf5c6c37216ee4f346976b7c06ac2383f6759a52513e8d5736f38ac428e5f3da","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.003--0.95--0.98--s44","configuration":"awc8-wsd-h120--0.003--0.95--0.98","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.3946147154764614,"jobId":"2482615","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"55cae209132436c2ebf0712742eff9cf12cc8f43b16de75c683c6419d49c6661","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.95--0.98--s42","configuration":"awc8-wsd-h120--0.002--0.95--0.98","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.408977548963366,"jobId":"2482617","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"264866f8998f2397795ea697994cf8fbd621da67605308f631e1ddd2fda56873","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.95--0.98--s43","configuration":"awc8-wsd-h120--0.002--0.95--0.98","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.413505536810266,"jobId":"2482618","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"05376aed83bf802b53f7cd41f3660d9a87be4c3e2c01855c72bc02933444dc73","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.95--0.98--s44","configuration":"awc8-wsd-h120--0.002--0.95--0.98","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.4067449574333812,"jobId":"2482619","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"6a48c879e3c73bd0914fc1aacc6b292a16299252e07e96cbf0019d222e7ac1df","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.003--0.95--0.999--s42","configuration":"awc8-wsd-h120--0.003--0.95--0.999","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.351924329591714,"jobId":"2482620","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"8d056b802156d1eef5ef9ee9728ac889d0e2f2f724fe77f7990084c25263c4f7","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/final.pt"},{"id":"awc8-wsd-h120--0.003--0.95--0.999--s43","configuration":"awc8-wsd-h120--0.003--0.95--0.999","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.359696544611667,"jobId":"2482621","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"0b86a5f6d8887085fc19a165db95b23f154d99f32e80d2f44a2ba97a063f037b","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/final.pt"},{"id":"awc8-wsd-h120--0.003--0.95--0.999--s44","configuration":"awc8-wsd-h120--0.003--0.95--0.999","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.357420486152484,"jobId":"2482622","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"d8a416a40fcda3f5941119ba37864b1b5c91bd70344855dac9a2d9a5a1b09589","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/final.pt"},{"id":"awc8-wsd-h120--0.002--0.95--0.999--s42","configuration":"awc8-wsd-h120--0.002--0.95--0.999","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.370138168708808,"jobId":"2482623","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"b66b61f02e00729b78d54de3e5614281e0401911f630f4c0d280d701c3e4f956","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.95--0.999--s43","configuration":"awc8-wsd-h120--0.002--0.95--0.999","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.3736219692634295,"jobId":"2482624","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"9c2f048e5f5b28c44e993287fbdd7bf42efd10dbd25d932c0eccb60e5454f1bb","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h120--0.002--0.95--0.999--s44","configuration":"awc8-wsd-h120--0.002--0.95--0.999","method":"awc8","horizon":120,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.3688452132349234,"jobId":"2482625","tokens":2448424960,"startTokens":1458831360,"incrementalTokens":989593600,"recipeIdentity":"e27c90e1e90959565f648d41149532628af523873b645d237789a82bba9aa15f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-120/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.003--0.9--0.9--s42","configuration":"awc8-wsd-h160--0.003--0.9--0.9","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.5063614500926086,"jobId":"2487090","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"fd627886ba06a1601ecc7e3ac52d91fba700acc099b2a4b06923fbbd14988169","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.003--0.9--0.9--s43","configuration":"awc8-wsd-h160--0.003--0.9--0.9","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.5025594780223552,"jobId":"2487091","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"a54649c9e11b2eda4dc185a6bdf4fd1a6dd713e1c20956d6989673dd7f089b7c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.003--0.9--0.9--s44","configuration":"awc8-wsd-h160--0.003--0.9--0.9","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.5105157180067326,"jobId":"2487092","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"19aef0661a52f446bbbad2a0eaff2457715126d8e5561dd81904a6a0b2d9c7f8","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.9--0.9--s42","configuration":"awc8-wsd-h160--0.002--0.9--0.9","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.514450208243573,"jobId":"2487093","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"9d17a9807435af56a4079a23762d50486db2723bf3c305fe14a1a52812780860","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.9--0.9--s43","configuration":"awc8-wsd-h160--0.002--0.9--0.9","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.5094643790826194,"jobId":"2487094","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"2130dd8b1b23cdfcf1802fb432b9f78425d36f77864d2837b124192305b03ab1","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.9--0.9--s44","configuration":"awc8-wsd-h160--0.002--0.9--0.9","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.517926365905667,"jobId":"2487095","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"414a4abca0354aa2d24eb6545d5e938cbd0ff645db331a79814c3c2b353dbc73","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.9-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.003--0.9--0.98--s42","configuration":"awc8-wsd-h160--0.003--0.9--0.98","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.3969474862469444,"jobId":"2487096","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"0c863ed6ff17bdc587881077b64b8266c58f51ebb279c27fcfef429410b98f13","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.003--0.9--0.98--s43","configuration":"awc8-wsd-h160--0.003--0.9--0.98","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.402039755179342,"jobId":"2487099","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"d6308fe6700cac259d2a9178fdf3635e2161efebf1266279702945bc2f364d0a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.003--0.9--0.98--s44","configuration":"awc8-wsd-h160--0.003--0.9--0.98","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.392649371669298,"jobId":"2487100","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"e3cab499e7daa6d1c6e86ba8a5b86f04e8b462598477a3e78ef7b648a7a14226","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.9--0.98--s42","configuration":"awc8-wsd-h160--0.002--0.9--0.98","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.40863865097916,"jobId":"2487102","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"17cb948e796a9a3ce5180526073581c635984b3adb2dd292ffc98df3b6d8eb59","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.9--0.98--s43","configuration":"awc8-wsd-h160--0.002--0.9--0.98","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.398285878117938,"jobId":"2487103","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"5e9e85153b210842bba444ce79657b104640dda9cc007b19c9406722629eef15","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.9--0.98--s44","configuration":"awc8-wsd-h160--0.002--0.9--0.98","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.4035677666090716,"jobId":"2487104","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"ad851acce0e3f9d9d43c22bf0f564a48db93c0058171561333848609d72a1974","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.98-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.003--0.9--0.999--s42","configuration":"awc8-wsd-h160--0.003--0.9--0.999","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.3529225996304977,"jobId":"2487105","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"728b18dfafd014f17788eaca69736b6d33b69f7d14e140c646b5c291f9369604","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.003--0.9--0.999--s43","configuration":"awc8-wsd-h160--0.003--0.9--0.999","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.359074351485768,"jobId":"2487106","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"dbee3b1c109b4476264a6965e0c375a334b3d1fdb78b03bd3790eacdf2469ecb","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.003--0.9--0.999--s44","configuration":"awc8-wsd-h160--0.003--0.9--0.999","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.347574796598013,"jobId":"2487107","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"1d5f481ea767210f95a722307e8bda6ee98d4aa4f00a68dfebfeff9b0e02f8d5","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.9--0.999--s42","configuration":"awc8-wsd-h160--0.002--0.9--0.999","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.358529524385071,"jobId":"2487108","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"9a20fca10acbbf5365c34d995961cdb3a88824c01fd9928c868022905b6a0f69","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.9--0.999--s43","configuration":"awc8-wsd-h160--0.002--0.9--0.999","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.3577636468280074,"jobId":"2487109","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"c11e40c118f6a25831a29763617a834d8ac8c57c2a476518e3c8b653634a726f","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.9--0.999--s44","configuration":"awc8-wsd-h160--0.002--0.9--0.999","method":"awc8","horizon":160,"beta1":0.9,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.358507563420065,"jobId":"2487110","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"24c886aaed0102614e4a011752720f239dd5820500540122774b324f971621da","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.9-b2-0.999-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.003--0.95--0.9--s42","configuration":"awc8-wsd-h160--0.003--0.95--0.9","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":42,"loss":3.437422778839181,"jobId":"2487111","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"944fa97d6eea38485952d5cd40148cfefb3485a99741a705791da1df91687425","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.003--0.95--0.9--s43","configuration":"awc8-wsd-h160--0.003--0.95--0.9","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":43,"loss":3.43905257227758,"jobId":"2487112","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"1ee5623d844ba1636050b8f600d355b996c76371ec589481aafe7fa309664ac4","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.003--0.95--0.9--s44","configuration":"awc8-wsd-h160--0.003--0.95--0.9","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.9,"lr":0.003,"seed":44,"loss":3.4385674619181477,"jobId":"2487113","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"8b61a66814f097697b33e9347dbb06c919ed750764d35cdc89cfea64ad81968c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.95--0.9--s42","configuration":"awc8-wsd-h160--0.002--0.95--0.9","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":42,"loss":3.441580831168483,"jobId":"2487114","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"855f7c6f85eb08b7f5e23751202eab85c114bbdf0227d1f9ee0c0f7bbabd3b2d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.95--0.9--s43","configuration":"awc8-wsd-h160--0.002--0.95--0.9","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":43,"loss":3.449206524816929,"jobId":"2487115","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"75427cd09730007e4bbfdcd62d06451bd11e607039e5f96ff1f3b6dd0cc1295c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.95--0.9--s44","configuration":"awc8-wsd-h160--0.002--0.95--0.9","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.9,"lr":0.002,"seed":44,"loss":3.4497888872279514,"jobId":"2487116","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"79b36b6613ede1817bf65ab1d1a8fde69174e5ae4b5db056994b39ab727f75b6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.9-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.003--0.95--0.98--s42","configuration":"awc8-wsd-h160--0.003--0.95--0.98","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":42,"loss":3.3660593553930678,"jobId":"2487117","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"837bd634da48e8c8e67d833568f6f3643aaf9de7c4c7d0cf76ac2933717a67e7","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.003--0.95--0.98--s43","configuration":"awc8-wsd-h160--0.003--0.95--0.98","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":43,"loss":3.3746523820282865,"jobId":"2487118","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"a4c1ed5169056a99c5c196c41414612767e792485941404cc80c8bd6520d255c","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.003--0.95--0.98--s44","configuration":"awc8-wsd-h160--0.003--0.95--0.98","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.98,"lr":0.003,"seed":44,"loss":3.372256653384429,"jobId":"2487119","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"08d0b1068a4f96e20ccd32695f9a3452634e75d2b7915b8de80c64b2f079d1e9","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.95--0.98--s42","configuration":"awc8-wsd-h160--0.002--0.95--0.98","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":42,"loss":3.3823749589014698,"jobId":"2487120","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"3b05d8aa02fba09ce8afb971778fb97fdf77d1137b2095bd206cc705379c8889","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.95--0.98--s43","configuration":"awc8-wsd-h160--0.002--0.95--0.98","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":43,"loss":3.387902159371044,"jobId":"2487121","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"fccd712c18d17f750382ef5a5cb1b4b3dcb4884e39cb0773ee80b11a6ad637be","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.95--0.98--s44","configuration":"awc8-wsd-h160--0.002--0.95--0.98","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.98,"lr":0.002,"seed":44,"loss":3.3818383573839106,"jobId":"2487122","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"56137731fc5d677f4412f95592cdb7f6328a7e1cf6bbceb1d13d30494bab802a","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.98-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.003--0.95--0.999--s42","configuration":"awc8-wsd-h160--0.003--0.95--0.999","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":42,"loss":3.3282830529138687,"jobId":"2487123","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"3d66f3cda284dd562a274a6f3ebe5b5a73b8ca577fb5e34320ecb6149147ea40","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/node-007/final.safetensors"],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s42/attempt-001/final.pt"},{"id":"awc8-wsd-h160--0.003--0.95--0.999--s43","configuration":"awc8-wsd-h160--0.003--0.95--0.999","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":43,"loss":3.3353299228975146,"jobId":"2487124","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"bcc51bb1a622b69ce26e7be417b5285f21f9bc1d463a5b44b35a392565bbab0d","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/node-007/final.safetensors"],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s43/attempt-001/final.pt"},{"id":"awc8-wsd-h160--0.003--0.95--0.999--s44","configuration":"awc8-wsd-h160--0.003--0.95--0.999","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.999,"lr":0.003,"seed":44,"loss":3.3331716778774796,"jobId":"2487125","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"f6e67fae44fe327c259e8ed4628dafde40b0fbcc8684938824caa5d0b30f78a6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/node-007/final.safetensors"],"finalState":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.003-s44/attempt-001/final.pt"},{"id":"awc8-wsd-h160--0.002--0.95--0.999--s42","configuration":"awc8-wsd-h160--0.002--0.95--0.999","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":42,"loss":3.3434955812164113,"jobId":"2487127","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"4a6d0463909b44a3f9a51875c5336c1e2b2e3afd1cce535c728f2e412ddabab1","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s42/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.95--0.999--s43","configuration":"awc8-wsd-h160--0.002--0.95--0.999","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":43,"loss":3.3479205570656556,"jobId":"2487128","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"a73689e57dda16885878c6f30b1bd4b243416b95677169b76c25ab0a6f576417","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s43/attempt-001/node-007/final.safetensors"],"finalState":null},{"id":"awc8-wsd-h160--0.002--0.95--0.999--s44","configuration":"awc8-wsd-h160--0.002--0.95--0.999","method":"awc8","horizon":160,"beta1":0.95,"beta2":0.999,"lr":0.002,"seed":44,"loss":3.3417755724939053,"jobId":"2487129","tokens":3264610304,"startTokens":2203582464,"incrementalTokens":1061027840,"recipeIdentity":"a60fbe6402aa0fa1656a4265d0310d0dcb0fb787524f657bea5e45a6c1ad99c6","source":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001","finalWeights":"/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/final.safetensors","workerWeights":["/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-000/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-001/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-002/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-003/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-004/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-005/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-006/final.safetensors","/nobackup/proj/disk/naiss2024-23-12/personal/zesen/repo/tiny-llm/runs/packed8-awc-wsd-20m-horizon-tuning/experiments/horizon-160/b1-0.95-b2-0.999-lr-0.002-s44/attempt-001/node-007/final.safetensors"],"finalState":null}]