Training in progress, step 490, checkpoint

last-checkpoint/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d1c5fc01647b6ed7ac5edaef7d0cbda83ee8d8813ee13b5876060f162313c9b4
+oid sha256:214d4c92e201996ef5000afb14d16c3a2cb53eba45c80910a30250ccf9fb671e
 size 194563400

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f7f3322b6279d066baac2977b90ec50d7b4e5fc6d034a5c6ee5ad537029ee610
+oid sha256:2b5a77bd4fe37d8b4c68f7fc3e886c3e283d1129eb3d38b56b63c170b448ff35
 size 100256339

last-checkpoint/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fb8d7ac45379fe8e70f4f956c9d628ef78eab8750412a42f9f7af2fde12cafbd
+oid sha256:4a993517fcad6b88249c3736cbc047e0728eb92a48102fcb394563ea5ea4fdaf
 size 14645

last-checkpoint/scaler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6884916bc3e83cb9f33c3ed9c11804720c2acc39b3f87549a97f4f007a89eb20
+oid sha256:631b854a8072c273e62f84ce9d54f2ab354ec712a745e46ebbee521a3aed2150
 size 1383

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d9671054f4efb882fe928a0a76ad9e3c53090dcbe026d23891f2bbb0f73c2434
+oid sha256:0085eb8f8087f7a151dc3cbff0d73149538e7fd3411aef06647570da27d3da03
 size 1465

last-checkpoint/trainer_state.json

@@ -2,9 +2,9 @@
   "best_global_step": null,
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 53.333333333333336,
+  "epoch": 54.44444444444444,
   "eval_steps": 500,
-  "global_step": 480,
+  "global_step": 490,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -10568,11 +10568,231 @@
       "rewards/compute_weighted_reward/mean": 0.9749999642372131,
       "rewards/compute_weighted_reward/std": 1.6546036005020142,
       "step": 480
+    },
+    {
+      "completion_length": 1400.0,
+      "completions/clipped_ratio": 0.25,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1201.0,
+      "completions/mean_length": 800.6666870117188,
+      "completions/mean_terminated_length": 600.888916015625,
+      "completions/min_length": 272.0,
+      "completions/min_terminated_length": 272.0,
+      "epoch": 53.44444444444444,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 6.093226432800293,
+      "kl": 0.011718678055331111,
+      "learning_rate": 1.2863734927012094e-09,
+      "loss": 0.0117,
+      "num_tokens": 13162020.0,
+      "reward": 0.6654167175292969,
+      "reward_std": 0.9684948921203613,
+      "rewards/compute_weighted_reward/mean": 0.6654166579246521,
+      "rewards/compute_weighted_reward/std": 1.245635747909546,
+      "step": 481
+    },
+    {
+      "completion_length": 1400.0,
+      "completions/clipped_ratio": 0.29166666666666663,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1368.0,
+      "completions/mean_length": 931.625,
+      "completions/mean_terminated_length": 738.7647094726562,
+      "completions/min_length": 344.0,
+      "completions/min_terminated_length": 344.0,
+      "epoch": 53.55555555555556,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 3.9402177333831787,
+      "kl": 0.007335595786571503,
+      "learning_rate": 1.1771618553447215e-09,
+      "loss": 0.0073,
+      "num_tokens": 13193259.0,
+      "reward": 0.056666698306798935,
+      "reward_std": 1.6200555562973022,
+      "rewards/compute_weighted_reward/mean": 0.05666669085621834,
+      "rewards/compute_weighted_reward/std": 1.5791513919830322,
+      "step": 482
+    },
+    {
+      "completion_length": 1400.0,
+      "completions/clipped_ratio": 0.04166666666666663,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1345.0,
+      "completions/mean_length": 604.5416870117188,
+      "completions/mean_terminated_length": 569.95654296875,
+      "completions/min_length": 256.0,
+      "completions/min_terminated_length": 256.0,
+      "epoch": 53.666666666666664,
+      "frac_reward_zero_std": 0.25,
+      "grad_norm": 5.48298978805542,
+      "kl": 0.00782405596692115,
+      "learning_rate": 1.0727667037011667e-09,
+      "loss": 0.0078,
+      "num_tokens": 13215070.0,
+      "reward": 1.693333387374878,
+      "reward_std": 0.9139549732208252,
+      "rewards/compute_weighted_reward/mean": 1.6933332681655884,
+      "rewards/compute_weighted_reward/std": 1.196791648864746,
+      "step": 483
+    },
+    {
+      "completion_length": 1400.0,
+      "completions/clipped_ratio": 0.20833333333333337,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1268.0,
+      "completions/mean_length": 825.7916870117188,
+      "completions/mean_terminated_length": 674.6842041015625,
+      "completions/min_length": 275.0,
+      "completions/min_terminated_length": 275.0,
+      "epoch": 53.77777777777778,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 3.4163742065429688,
+      "kl": 0.008350545074790716,
+      "learning_rate": 9.731931258429638e-10,
+      "loss": 0.0084,
+      "num_tokens": 13243193.0,
+      "reward": 0.880000114440918,
+      "reward_std": 0.974814772605896,
+      "rewards/compute_weighted_reward/mean": 0.8800000548362732,
+      "rewards/compute_weighted_reward/std": 1.2584704160690308,
+      "step": 484
+    },
+    {
+      "completion_length": 1400.0,
+      "completions/clipped_ratio": 0.08333333333333337,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1184.0,
+      "completions/mean_length": 619.4166870117188,
+      "completions/mean_terminated_length": 548.45458984375,
+      "completions/min_length": 200.0,
+      "completions/min_terminated_length": 200.0,
+      "epoch": 53.888888888888886,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 3.2675859928131104,
+      "kl": 0.015018837992101908,
+      "learning_rate": 8.784459748458317e-10,
+      "loss": 0.015,
+      "num_tokens": 13265445.0,
+      "reward": 1.5237501859664917,
+      "reward_std": 1.057565689086914,
+      "rewards/compute_weighted_reward/mean": 1.5237499475479126,
+      "rewards/compute_weighted_reward/std": 1.2061611413955688,
+      "step": 485
+    },
+    {
+      "completion_length": 1400.0,
+      "completions/clipped_ratio": 0.45833333333333337,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1119.0,
+      "completions/mean_length": 992.7916870117188,
+      "completions/mean_terminated_length": 648.2307739257812,
+      "completions/min_length": 417.0,
+      "completions/min_terminated_length": 417.0,
+      "epoch": 54.0,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 4.8129425048828125,
+      "kl": 0.006140718352980912,
+      "learning_rate": 7.885298685522235e-10,
+      "loss": 0.0061,
+      "num_tokens": 13297918.0,
+      "reward": -0.06458334624767303,
+      "reward_std": 0.5508111715316772,
+      "rewards/compute_weighted_reward/mean": -0.06458330154418945,
+      "rewards/compute_weighted_reward/std": 1.783735752105713,
+      "step": 486
+    },
+    {
+      "completion_length": 1400.0,
+      "completions/clipped_ratio": 0.33333333333333337,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1363.0,
+      "completions/mean_length": 882.2083740234375,
+      "completions/mean_terminated_length": 623.3125,
+      "completions/min_length": 284.0,
+      "completions/min_terminated_length": 284.0,
+      "epoch": 54.111111111111114,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 6.521482467651367,
+      "kl": 0.0026615530950948596,
+      "learning_rate": 7.034491893463057e-10,
+      "loss": 0.0027,
+      "num_tokens": 13326459.0,
+      "reward": 0.5283333659172058,
+      "reward_std": 1.406015157699585,
+      "rewards/compute_weighted_reward/mean": 0.5283333659172058,
+      "rewards/compute_weighted_reward/std": 1.6546525955200195,
+      "step": 487
+    },
+    {
+      "completion_length": 1400.0,
+      "completions/clipped_ratio": 0.25,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1293.0,
+      "completions/mean_length": 823.4583740234375,
+      "completions/mean_terminated_length": 631.2777709960938,
+      "completions/min_length": 342.0,
+      "completions/min_terminated_length": 342.0,
+      "epoch": 54.22222222222222,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 13.027168273925781,
+      "kl": 0.016307485639117658,
+      "learning_rate": 6.23208083940363e-10,
+      "loss": 0.0163,
+      "num_tokens": 13355240.0,
+      "reward": 0.9591667056083679,
+      "reward_std": 1.122135877609253,
+      "rewards/compute_weighted_reward/mean": 0.9591667056083679,
+      "rewards/compute_weighted_reward/std": 1.348125696182251,
+      "step": 488
+    },
+    {
+      "completion_length": 1400.0,
+      "completions/clipped_ratio": 0.04166666666666663,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1327.0,
+      "completions/mean_length": 658.2083740234375,
+      "completions/mean_terminated_length": 625.95654296875,
+      "completions/min_length": 262.0,
+      "completions/min_terminated_length": 262.0,
+      "epoch": 54.333333333333336,
+      "frac_reward_zero_std": 0.5,
+      "grad_norm": 5.863112926483154,
+      "kl": 0.009161067078821361,
+      "learning_rate": 5.47810463172671e-10,
+      "loss": 0.0092,
+      "num_tokens": 13378183.0,
+      "reward": 1.6708335876464844,
+      "reward_std": 0.5133538246154785,
+      "rewards/compute_weighted_reward/mean": 1.6708334684371948,
+      "rewards/compute_weighted_reward/std": 1.1477991342544556,
+      "step": 489
+    },
+    {
+      "completion_length": 1400.0,
+      "completions/clipped_ratio": 0.29166666666666663,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1060.0,
+      "completions/mean_length": 854.4583740234375,
+      "completions/mean_terminated_length": 629.8235473632812,
+      "completions/min_length": 146.0,
+      "completions/min_terminated_length": 146.0,
+      "epoch": 54.44444444444444,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 8.846403121948242,
+      "kl": 0.008679779479280114,
+      "learning_rate": 4.772600018168815e-10,
+      "loss": 0.0087,
+      "num_tokens": 13406868.0,
+      "reward": 0.15416672825813293,
+      "reward_std": 1.329119086265564,
+      "rewards/compute_weighted_reward/mean": 0.15416669845581055,
+      "rewards/compute_weighted_reward/std": 1.5062823295593262,
+      "step": 490
     }
   ],
   "logging_steps": 1,
   "max_steps": 500,
-  "num_input_tokens_seen": 13135406,
+  "num_input_tokens_seen": 13406868,
   "num_train_epochs": 56,
   "save_steps": 10,
   "stateful_callbacks": {