Training in progress, step 500, checkpoint

last-checkpoint/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:219d1c5d99bb64f89da33d738edff3877126afcdc42bb3868f9cadfba05afabf
+oid sha256:40e781c129f82797c0a6821bcdc7e7c877cdc491a2e369fbbdc5f4d8f8555217
 size 194563400

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3bedb66f52cb59f1661032cabc537435aa1ee41c45a5af728f608df074e799de
+oid sha256:6804192c71eec3b4a512a29b5211006e9dc557c1b052e87f96ec0a401d77f542
 size 100256339

last-checkpoint/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0c337337254f5e2543e7e0082217448503a00a475a88eb70e26cc114ee5abafd
+oid sha256:2d61dbc3f2c2b23c059c6b8721c9652c292e77f8e6bca5c2f6d12181ba94bfe4
 size 14645

last-checkpoint/scaler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0b73d4cd42a94a1103ceefa33eaa6ef4fb3cf4f32efbe8707ea89780e8b5d2e9
+oid sha256:43f429e2475c017e0929ea1a41fab783edaaa7552eb6003f1b53b0de645a5e96
 size 1383

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:873eeaba556ee4cab82e58f0b7fbe7b6cd39f0bcccd0b0238d64f7db5dd50f9c
+oid sha256:9d0ddb482cfa0bc26b42f43c3031f7f23c3b5c7fc0c791438490253e4dc5338f
 size 1465

last-checkpoint/trainer_state.json

@@ -2,9 +2,9 @@
   "best_global_step": null,
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 15.3125,
+  "epoch": 15.625,
   "eval_steps": 500,
-  "global_step": 490,
+  "global_step": 500,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10788,11 +10788,231 @@
       "rewards/compute_weighted_reward/mean": 1.4883333444595337,
       "rewards/compute_weighted_reward/std": 1.16587233543396,
       "step": 490
+    },
+    {
+      "completion_length": 583.5833587646484,
+      "completions/clipped_ratio": 0.04166666666666663,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1331.0,
+      "completions/mean_length": 583.5833740234375,
+      "completions/mean_terminated_length": 548.0869750976562,
+      "completions/min_length": 207.0,
+      "completions/min_terminated_length": 207.0,
+      "epoch": 15.34375,
+      "frac_reward_zero_std": 0.25,
+      "grad_norm": 0.647857666015625,
+      "kl": 0.02400309219956398,
+      "learning_rate": 1.0289003460074165e-09,
+      "loss": 0.0489,
+      "num_tokens": 12713527.0,
+      "reward": 1.3562501668930054,
+      "reward_std": 0.5569199323654175,
+      "rewards/compute_weighted_reward/mean": 1.3562501668930054,
+      "rewards/compute_weighted_reward/std": 0.889302670955658,
+      "step": 491
+    },
+    {
+      "completion_length": 728.3750076293945,
+      "completions/clipped_ratio": 0.20833333333333337,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1028.0,
+      "completions/mean_length": 731.0416870117188,
+      "completions/mean_terminated_length": 555.0,
+      "completions/min_length": 282.0,
+      "completions/min_terminated_length": 282.0,
+      "epoch": 15.375,
+      "frac_reward_zero_std": 0.25,
+      "grad_norm": 0.8000059723854065,
+      "kl": 0.023584565613418818,
+      "learning_rate": 8.767851876239074e-10,
+      "loss": 0.0869,
+      "num_tokens": 12738974.0,
+      "reward": 1.4620835781097412,
+      "reward_std": 0.9179548025131226,
+      "rewards/compute_weighted_reward/mean": 1.4620834589004517,
+      "rewards/compute_weighted_reward/std": 1.6754546165466309,
+      "step": 492
+    },
+    {
+      "completion_length": 594.0000305175781,
+      "completions/clipped_ratio": 0.20833333333333337,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 733.0,
+      "completions/mean_length": 594.0,
+      "completions/mean_terminated_length": 381.8947448730469,
+      "completions/min_length": 158.0,
+      "completions/min_terminated_length": 158.0,
+      "epoch": 15.40625,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.656679630279541,
+      "kl": 0.019565630238503218,
+      "learning_rate": 7.368119432699382e-10,
+      "loss": 0.1084,
+      "num_tokens": 12760364.0,
+      "reward": 0.9337500333786011,
+      "reward_std": 1.1254143714904785,
+      "rewards/compute_weighted_reward/mean": 0.9337499737739563,
+      "rewards/compute_weighted_reward/std": 1.215303897857666,
+      "step": 493
+    },
+    {
+      "completion_length": 919.7500305175781,
+      "completions/clipped_ratio": 0.45833333333333337,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1107.0,
+      "completions/mean_length": 919.75,
+      "completions/mean_terminated_length": 513.3846435546875,
+      "completions/min_length": 259.0,
+      "completions/min_terminated_length": 259.0,
+      "epoch": 15.4375,
+      "frac_reward_zero_std": 0.25,
+      "grad_norm": 0.7352123260498047,
+      "kl": 0.022763641085475683,
+      "learning_rate": 6.089874350439505e-10,
+      "loss": 0.1486,
+      "num_tokens": 12789854.0,
+      "reward": 0.534583330154419,
+      "reward_std": 1.1599116325378418,
+      "rewards/compute_weighted_reward/mean": 0.534583330154419,
+      "rewards/compute_weighted_reward/std": 1.7653746604919434,
+      "step": 494
+    },
+    {
+      "completion_length": 816.6666870117188,
+      "completions/clipped_ratio": 0.33333333333333337,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1353.0,
+      "completions/mean_length": 821.7916870117188,
+      "completions/mean_terminated_length": 532.6875,
+      "completions/min_length": 285.0,
+      "completions/min_terminated_length": 285.0,
+      "epoch": 15.46875,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 1.3775248527526855,
+      "kl": 0.03886409197002649,
+      "learning_rate": 4.933178929321102e-10,
+      "loss": 0.1886,
+      "num_tokens": 12817605.0,
+      "reward": 0.5058333873748779,
+      "reward_std": 1.149335265159607,
+      "rewards/compute_weighted_reward/mean": 0.5058333277702332,
+      "rewards/compute_weighted_reward/std": 1.560180902481079,
+      "step": 495
+    },
+    {
+      "completion_length": 809.1250305175781,
+      "completions/clipped_ratio": 0.16666666666666663,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1317.0,
+      "completions/mean_length": 809.125,
+      "completions/mean_terminated_length": 690.9500122070312,
+      "completions/min_length": 253.0,
+      "completions/min_terminated_length": 253.0,
+      "epoch": 15.5,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 1.130301594734192,
+      "kl": 0.024388552643358707,
+      "learning_rate": 3.898089545047445e-10,
+      "loss": 0.0514,
+      "num_tokens": 12844950.0,
+      "reward": 1.0716667175292969,
+      "reward_std": 1.2712818384170532,
+      "rewards/compute_weighted_reward/mean": 1.0716667175292969,
+      "rewards/compute_weighted_reward/std": 1.4085689783096313,
+      "step": 496
+    },
+    {
+      "completion_length": 1121.9583435058594,
+      "completions/clipped_ratio": 0.625,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 930.0,
+      "completions/mean_length": 1121.9583740234375,
+      "completions/mean_terminated_length": 658.5555419921875,
+      "completions/min_length": 392.0,
+      "completions/min_terminated_length": 392.0,
+      "epoch": 15.53125,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.9501820206642151,
+      "kl": 0.040883428417146206,
+      "learning_rate": 2.9846566464150626e-10,
+      "loss": 0.1841,
+      "num_tokens": 12879269.0,
+      "reward": -0.44583332538604736,
+      "reward_std": 1.1888765096664429,
+      "rewards/compute_weighted_reward/mean": -0.445833295583725,
+      "rewards/compute_weighted_reward/std": 1.4850410223007202,
+      "step": 497
+    },
+    {
+      "completion_length": 645.6250076293945,
+      "completions/clipped_ratio": 0.125,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1119.0,
+      "completions/mean_length": 645.625,
+      "completions/mean_terminated_length": 537.857177734375,
+      "completions/min_length": 185.0,
+      "completions/min_terminated_length": 185.0,
+      "epoch": 15.5625,
+      "frac_reward_zero_std": 0.5,
+      "grad_norm": 0.46877413988113403,
+      "kl": 0.013578799553215504,
+      "learning_rate": 2.1929247528540418e-10,
+      "loss": 0.0527,
+      "num_tokens": 12901676.0,
+      "reward": 2.0500001907348633,
+      "reward_std": 0.5961085557937622,
+      "rewards/compute_weighted_reward/mean": 2.0500001907348633,
+      "rewards/compute_weighted_reward/std": 1.0217292308807373,
+      "step": 498
+    },
+    {
+      "completion_length": 947.8333740234375,
+      "completions/clipped_ratio": 0.45833333333333337,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 915.0,
+      "completions/mean_length": 947.8333740234375,
+      "completions/mean_terminated_length": 565.2307739257812,
+      "completions/min_length": 307.0,
+      "completions/min_terminated_length": 307.0,
+      "epoch": 15.59375,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.9576979279518127,
+      "kl": 0.03686598874628544,
+      "learning_rate": 1.5229324522605947e-10,
+      "loss": 0.2095,
+      "num_tokens": 12932380.0,
+      "reward": 0.09458336234092712,
+      "reward_std": 1.4334077835083008,
+      "rewards/compute_weighted_reward/mean": 0.09458336979150772,
+      "rewards/compute_weighted_reward/std": 1.6301053762435913,
+      "step": 499
+    },
+    {
+      "completion_length": 1035.250015258789,
+      "completions/clipped_ratio": 0.5416666666666667,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1097.0,
+      "completions/mean_length": 1035.25,
+      "completions/mean_terminated_length": 604.1818237304688,
+      "completions/min_length": 294.0,
+      "completions/min_terminated_length": 294.0,
+      "epoch": 15.625,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 2.101461172103882,
+      "kl": 0.04244990833103657,
+      "learning_rate": 9.747123991141193e-11,
+      "loss": 0.1304,
+      "num_tokens": 12965248.0,
+      "reward": 0.10583341121673584,
+      "reward_std": 1.346000075340271,
+      "rewards/compute_weighted_reward/mean": 0.10583335161209106,
+      "rewards/compute_weighted_reward/std": 1.7509969472885132,
+      "step": 500
     }
   ],
   "logging_steps": 1,
   "max_steps": 500,
-  "num_input_tokens_seen": 12691799,
+  "num_input_tokens_seen": 12965248,
   "num_train_epochs": 16,
   "save_steps": 10,
   "stateful_callbacks": {
@@ -10802,7 +11022,7 @@
         "should_evaluate": false,
         "should_log": false,
         "should_save": true,
-        "should_training_stop": false
+        "should_training_stop": true
       },
       "attributes": {}
     }