Training in progress, step 350, checkpoint

last-checkpoint/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:290c186538be67cf4326d42aff57ad27407d235602524f2fe8bc6a5e0efc65a7
+oid sha256:4486f35279323679530cf73f57d30330d896d78adba3db63862bf00fd96a3426
 size 194563400

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5cf1cf37fe6514dd57202fdd0e8c9f79a55858e2c5e0b4a073a68f8661a25002
+oid sha256:bb7b055ac92ba714cd690e0a196dd8b598d7d11d17591f4d7df3379cba6e1df8
 size 100256339

last-checkpoint/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d3ab52318c90ec94201baa5c717b95295f0a756d9c55b2a616930c39bcde242f
+oid sha256:e43cfc9b78bf4fcc2c474982c0b41c03250d3b56f391c4d232a2e2b1a37ff8d3
 size 14645

last-checkpoint/scaler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7dba4adec89b0959eafff0fd51bce4a65fd8662d186b5d39f7914f1e1d1d535e
+oid sha256:e7a3b7c7f6ae3d4b9e2fcb0774d91cea73c152177e3f6818dfc11db3a9a74275
 size 1383

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3bcaaaf2d3b1c41963e23ab37a9fe26f5b59148bd68b0e85ec53f1d5e9927805
+oid sha256:3f0606b2be3dd375d9eeb91bd17f7f75fb345244906b8d79f255436b74591a97
 size 1465

last-checkpoint/trainer_state.json

@@ -2,9 +2,9 @@
   "best_global_step": null,
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.7199999999999998,
+  "epoch": 2.8,
   "eval_steps": 500,
-  "global_step": 340,
+  "global_step": 350,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -7488,11 +7488,231 @@
       "rewards/compute_weighted_reward/mean": 1.880416750907898,
       "rewards/compute_weighted_reward/std": 1.6687992811203003,
       "step": 340
+    },
+    {
+      "completion_length": 675.5833333333334,
+      "completions/clipped_ratio": 0.125,
+      "completions/max_length": 2000.0,
+      "completions/max_terminated_length": 1690.0,
+      "completions/mean_length": 675.5833740234375,
+      "completions/mean_terminated_length": 486.3809509277344,
+      "completions/min_length": 232.0,
+      "completions/min_terminated_length": 232.0,
+      "epoch": 2.7279999999999998,
+      "frac_reward_zero_std": 0.25,
+      "grad_norm": 0.4411628544330597,
+      "kl": 0.0001766148191488052,
+      "learning_rate": 4.9003041134365705e-08,
+      "loss": 0.1452,
+      "num_tokens": 10454794.0,
+      "reward": 1.412083387374878,
+      "reward_std": 1.055234432220459,
+      "rewards/compute_weighted_reward/mean": 1.4120832681655884,
+      "rewards/compute_weighted_reward/std": 1.6767903566360474,
+      "step": 341
+    },
+    {
+      "completion_length": 960.5833333333334,
+      "completions/clipped_ratio": 0.125,
+      "completions/max_length": 1737.0,
+      "completions/max_terminated_length": 1578.0,
+      "completions/mean_length": 960.5833740234375,
+      "completions/mean_terminated_length": 849.6666870117188,
+      "completions/min_length": 349.0,
+      "completions/min_terminated_length": 349.0,
+      "epoch": 2.7359999999999998,
+      "frac_reward_zero_std": 0.25,
+      "grad_norm": 0.3251778483390808,
+      "kl": 0.00015508719335836454,
+      "learning_rate": 4.898574603995548e-08,
+      "loss": 0.0453,
+      "num_tokens": 10486392.0,
+      "reward": 0.7616666555404663,
+      "reward_std": 0.807964026927948,
+      "rewards/compute_weighted_reward/mean": 0.7616665959358215,
+      "rewards/compute_weighted_reward/std": 1.4406177997589111,
+      "step": 342
+    },
+    {
+      "completion_length": 1160.0,
+      "completions/clipped_ratio": 0.41666666666666663,
+      "completions/max_length": 2000.0,
+      "completions/max_terminated_length": 1064.0,
+      "completions/mean_length": 1165.3333740234375,
+      "completions/mean_terminated_length": 569.1428833007812,
+      "completions/min_length": 188.0,
+      "completions/min_terminated_length": 188.0,
+      "epoch": 2.7439999999999998,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.5680258870124817,
+      "kl": 0.00021060577925406201,
+      "learning_rate": 4.896830532173602e-08,
+      "loss": 0.1619,
+      "num_tokens": 10522532.0,
+      "reward": 0.5170833468437195,
+      "reward_std": 1.2685867547988892,
+      "rewards/compute_weighted_reward/mean": 0.5170833468437195,
+      "rewards/compute_weighted_reward/std": 1.8057841062545776,
+      "step": 343
+    },
+    {
+      "completion_length": 743.375,
+      "completions/clipped_ratio": 0.16666666666666663,
+      "completions/max_length": 1949.0,
+      "completions/max_terminated_length": 1716.0,
+      "completions/mean_length": 743.375,
+      "completions/mean_terminated_length": 502.25,
+      "completions/min_length": 286.0,
+      "completions/min_terminated_length": 286.0,
+      "epoch": 2.752,
+      "frac_reward_zero_std": 0.25,
+      "grad_norm": 0.2251403033733368,
+      "kl": 0.00010871579085384535,
+      "learning_rate": 4.895071908559452e-08,
+      "loss": 0.0623,
+      "num_tokens": 10548191.0,
+      "reward": 2.2512502670288086,
+      "reward_std": 0.5843120813369751,
+      "rewards/compute_weighted_reward/mean": 2.2512500286102295,
+      "rewards/compute_weighted_reward/std": 1.9334185123443604,
+      "step": 344
+    },
+    {
+      "completion_length": 646.6666666666666,
+      "completions/clipped_ratio": 0.08333333333333337,
+      "completions/max_length": 2000.0,
+      "completions/max_terminated_length": 1780.0,
+      "completions/mean_length": 646.6666870117188,
+      "completions/mean_terminated_length": 523.6363525390625,
+      "completions/min_length": 250.0,
+      "completions/min_terminated_length": 250.0,
+      "epoch": 2.76,
+      "frac_reward_zero_std": 0.5,
+      "grad_norm": 0.2939091920852661,
+      "kl": 0.00013193290965318738,
+      "learning_rate": 4.8932987438301677e-08,
+      "loss": 0.0348,
+      "num_tokens": 10571151.0,
+      "reward": 1.0816667079925537,
+      "reward_std": 0.6287143230438232,
+      "rewards/compute_weighted_reward/mean": 1.0816667079925537,
+      "rewards/compute_weighted_reward/std": 1.3874456882476807,
+      "step": 345
+    },
+    {
+      "completion_length": 862.1666666666666,
+      "completions/clipped_ratio": 0.20833333333333337,
+      "completions/max_length": 2000.0,
+      "completions/max_terminated_length": 1420.0,
+      "completions/mean_length": 862.1666870117188,
+      "completions/mean_terminated_length": 562.73681640625,
+      "completions/min_length": 223.0,
+      "completions/min_terminated_length": 223.0,
+      "epoch": 2.768,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.375482976436615,
+      "kl": 0.00013598250078909283,
+      "learning_rate": 4.891511048751102e-08,
+      "loss": 0.1055,
+      "num_tokens": 10599601.0,
+      "reward": 1.0750000476837158,
+      "reward_std": 0.5297895669937134,
+      "rewards/compute_weighted_reward/mean": 1.0749999284744263,
+      "rewards/compute_weighted_reward/std": 1.3649685382843018,
+      "step": 346
+    },
+    {
+      "completion_length": 1025.125,
+      "completions/clipped_ratio": 0.20833333333333337,
+      "completions/max_length": 2000.0,
+      "completions/max_terminated_length": 1771.0,
+      "completions/mean_length": 1025.125,
+      "completions/mean_terminated_length": 768.5789794921875,
+      "completions/min_length": 408.0,
+      "completions/min_terminated_length": 408.0,
+      "epoch": 2.776,
+      "frac_reward_zero_std": 0.25,
+      "grad_norm": 0.35587796568870544,
+      "kl": 0.0001902539578016634,
+      "learning_rate": 4.889708834175823e-08,
+      "loss": 0.1037,
+      "num_tokens": 10631908.0,
+      "reward": 0.41458332538604736,
+      "reward_std": 0.8512474894523621,
+      "rewards/compute_weighted_reward/mean": 0.41458332538604736,
+      "rewards/compute_weighted_reward/std": 1.1446017026901245,
+      "step": 347
+    },
+    {
+      "completion_length": 760.375,
+      "completions/clipped_ratio": 0.16666666666666663,
+      "completions/max_length": 2000.0,
+      "completions/max_terminated_length": 1163.0,
+      "completions/mean_length": 760.375,
+      "completions/mean_terminated_length": 512.4500122070312,
+      "completions/min_length": 279.0,
+      "completions/min_terminated_length": 279.0,
+      "epoch": 2.784,
+      "frac_reward_zero_std": 0.25,
+      "grad_norm": 0.3131572902202606,
+      "kl": 0.00019689797333436823,
+      "learning_rate": 4.8878921110460536e-08,
+      "loss": 0.0717,
+      "num_tokens": 10658035.0,
+      "reward": 0.4520833194255829,
+      "reward_std": 0.8793530464172363,
+      "rewards/compute_weighted_reward/mean": 0.4520833492279053,
+      "rewards/compute_weighted_reward/std": 1.1837999820709229,
+      "step": 348
+    },
+    {
+      "completion_length": 1075.4166666666667,
+      "completions/clipped_ratio": 0.33333333333333337,
+      "completions/max_length": 2000.0,
+      "completions/max_terminated_length": 1009.0,
+      "completions/mean_length": 1118.0,
+      "completions/mean_terminated_length": 677.0,
+      "completions/min_length": 352.0,
+      "completions/min_terminated_length": 352.0,
+      "epoch": 2.792,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.5552175045013428,
+      "kl": 0.0001898959982706098,
+      "learning_rate": 4.8860608903916e-08,
+      "loss": 0.2009,
+      "num_tokens": 10693225.0,
+      "reward": 0.3566667437553406,
+      "reward_std": 1.2520555257797241,
+      "rewards/compute_weighted_reward/mean": 0.3566666841506958,
+      "rewards/compute_weighted_reward/std": 1.4636603593826294,
+      "step": 349
+    },
+    {
+      "completion_length": 1150.9166666666667,
+      "completions/clipped_ratio": 0.375,
+      "completions/max_length": 2000.0,
+      "completions/max_terminated_length": 1194.0,
+      "completions/mean_length": 1150.916748046875,
+      "completions/mean_terminated_length": 641.4666748046875,
+      "completions/min_length": 273.0,
+      "completions/min_terminated_length": 273.0,
+      "epoch": 2.8,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.36329564452171326,
+      "kl": 0.00010918113108952336,
+      "learning_rate": 4.8842151833302874e-08,
+      "loss": 0.1516,
+      "num_tokens": 10728071.0,
+      "reward": -0.19333335757255554,
+      "reward_std": 0.5602983236312866,
+      "rewards/compute_weighted_reward/mean": -0.19333334267139435,
+      "rewards/compute_weighted_reward/std": 0.6461839079856873,
+      "step": 350
     }
   ],
   "logging_steps": 1,
   "max_steps": 1500,
-  "num_input_tokens_seen": 10431398,
+  "num_input_tokens_seen": 10728071,
   "num_train_epochs": 12,
   "save_steps": 10,
   "stateful_callbacks": {