Training in progress, step 100, checkpoint

last-checkpoint/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d397d32e454caeeede6ba6b3ea396f8062976cc199ed8b908146eae390366621
+oid sha256:9c52debc52282412e2aa8b8c03dc882545c7506e2e04de9cd91fd713990b5e35
 size 194563400

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:93482f2dd4a14e514cd0c572dcac758f3ea7e1906cd2e33efeea332929bac31c
+oid sha256:62872fd99c0f0b06843bfc20a4fc537acd025aefddb1fd31af42d21a9f5429f8
 size 99240837

last-checkpoint/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:81a2d733348528c1d1612632839661eb622f740721f36359c94febe0ae478617
+oid sha256:c116c2a8c602abd0e289a35bc9bbb94a973d913f8f26addccbf2f5e68e78dbad
 size 14645

last-checkpoint/scaler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9fcfce7e32b31bead77e9e963d8e602c77a653a75c54ecbe876542fbeeba3fdc
+oid sha256:c4393a84a3109995aa1202073b039b12062e3189ed89aa0b94ef0510ba843009
 size 1383

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5fa48ce1412cb8d3fc995a86958161f32b67a285ffa5bec336b75ac13d073aba
+oid sha256:f5db2a1ee16d0f9a24fdf7419d14686c02d87df1ccf008ff752eb59f9d58772d
 size 1465

last-checkpoint/trainer_state.json

@@ -2,9 +2,9 @@
   "best_global_step": null,
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 4.090909090909091,
+  "epoch": 4.545454545454545,
   "eval_steps": 500,
-  "global_step": 90,
+  "global_step": 100,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -1988,11 +1988,231 @@
       "rewards/compute_weighted_reward/mean": -0.41999998688697815,
       "rewards/compute_weighted_reward/std": 0.9323742389678955,
       "step": 90
+    },
+    {
+      "completion_length": 974.8333333333334,
+      "completions/clipped_ratio": 0.5,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 908.0,
+      "completions/mean_length": 976.8333740234375,
+      "completions/mean_terminated_length": 553.6666870117188,
+      "completions/min_length": 371.0,
+      "completions/min_terminated_length": 371.0,
+      "epoch": 4.136363636363637,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.6058995127677917,
+      "kl": 0.0036607286747312173,
+      "learning_rate": 1.2e-06,
+      "loss": 0.1045,
+      "num_tokens": 3333182.0,
+      "reward": 0.26875004172325134,
+      "reward_std": 1.0469281673431396,
+      "rewards/compute_weighted_reward/mean": 0.26874998211860657,
+      "rewards/compute_weighted_reward/std": 1.6799956560134888,
+      "step": 91
+    },
+    {
+      "completion_length": 1188.9583333333333,
+      "completions/clipped_ratio": 0.5416666666666667,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1387.0,
+      "completions/mean_length": 1188.9583740234375,
+      "completions/mean_terminated_length": 939.5454711914062,
+      "completions/min_length": 488.0,
+      "completions/min_terminated_length": 488.0,
+      "epoch": 4.181818181818182,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.5676044821739197,
+      "kl": 0.003291421138177005,
+      "learning_rate": 1.2133333333333333e-06,
+      "loss": 0.1006,
+      "num_tokens": 3370363.0,
+      "reward": -0.2941666841506958,
+      "reward_std": 1.2139606475830078,
+      "rewards/compute_weighted_reward/mean": -0.2941666543483734,
+      "rewards/compute_weighted_reward/std": 1.2161625623703003,
+      "step": 92
+    },
+    {
+      "completion_length": 1085.4583333333333,
+      "completions/clipped_ratio": 0.5,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1255.0,
+      "completions/mean_length": 1085.4583740234375,
+      "completions/mean_terminated_length": 770.9166870117188,
+      "completions/min_length": 297.0,
+      "completions/min_terminated_length": 297.0,
+      "epoch": 4.2272727272727275,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.6693485379219055,
+      "kl": 0.0045636411814484745,
+      "learning_rate": 1.2266666666666664e-06,
+      "loss": 0.1505,
+      "num_tokens": 3404622.0,
+      "reward": -0.3425000309944153,
+      "reward_std": 1.0440089702606201,
+      "rewards/compute_weighted_reward/mean": -0.3424999713897705,
+      "rewards/compute_weighted_reward/std": 0.9814950823783875,
+      "step": 93
+    },
+    {
+      "completion_length": 1044.0416666666667,
+      "completions/clipped_ratio": 0.5416666666666667,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1037.0,
+      "completions/mean_length": 1044.041748046875,
+      "completions/mean_terminated_length": 623.3636474609375,
+      "completions/min_length": 48.0,
+      "completions/min_terminated_length": 48.0,
+      "epoch": 4.2727272727272725,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.6476233005523682,
+      "kl": 0.005175514816073701,
+      "learning_rate": 1.24e-06,
+      "loss": 0.1359,
+      "num_tokens": 3437971.0,
+      "reward": -0.47833335399627686,
+      "reward_std": 0.8860996961593628,
+      "rewards/compute_weighted_reward/mean": -0.47833332419395447,
+      "rewards/compute_weighted_reward/std": 0.9345524907112122,
+      "step": 94
+    },
+    {
+      "completion_length": 1078.0833333333333,
+      "completions/clipped_ratio": 0.5416666666666667,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1215.0,
+      "completions/mean_length": 1078.0833740234375,
+      "completions/mean_terminated_length": 697.6364135742188,
+      "completions/min_length": 352.0,
+      "completions/min_terminated_length": 352.0,
+      "epoch": 4.318181818181818,
+      "frac_reward_zero_std": 0.25,
+      "grad_norm": 0.5297577381134033,
+      "kl": 0.005124464087809126,
+      "learning_rate": 1.2533333333333333e-06,
+      "loss": 0.0919,
+      "num_tokens": 3471915.0,
+      "reward": -0.5570833683013916,
+      "reward_std": 0.743101954460144,
+      "rewards/compute_weighted_reward/mean": -0.5570833086967468,
+      "rewards/compute_weighted_reward/std": 0.9160429835319519,
+      "step": 95
+    },
+    {
+      "completion_length": 1209.375,
+      "completions/clipped_ratio": 0.7083333333333333,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1287.0,
+      "completions/mean_length": 1209.375,
+      "completions/mean_terminated_length": 746.4285888671875,
+      "completions/min_length": 51.0,
+      "completions/min_terminated_length": 51.0,
+      "epoch": 4.363636363636363,
+      "frac_reward_zero_std": 0.25,
+      "grad_norm": 0.6037044525146484,
+      "kl": 0.005498341323497395,
+      "learning_rate": 1.2666666666666665e-06,
+      "loss": 0.0587,
+      "num_tokens": 3509178.0,
+      "reward": -0.7341666221618652,
+      "reward_std": 0.5504173040390015,
+      "rewards/compute_weighted_reward/mean": -0.73416668176651,
+      "rewards/compute_weighted_reward/std": 0.8664366006851196,
+      "step": 96
+    },
+    {
+      "completion_length": 1162.0,
+      "completions/clipped_ratio": 0.625,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1223.0,
+      "completions/mean_length": 1162.0,
+      "completions/mean_terminated_length": 765.3333129882812,
+      "completions/min_length": 462.0,
+      "completions/min_terminated_length": 462.0,
+      "epoch": 4.409090909090909,
+      "frac_reward_zero_std": 0.25,
+      "grad_norm": 0.6083352565765381,
+      "kl": 0.007090592368816336,
+      "learning_rate": 1.28e-06,
+      "loss": 0.1455,
+      "num_tokens": 3545394.0,
+      "reward": -0.5695833563804626,
+      "reward_std": 0.7859379053115845,
+      "rewards/compute_weighted_reward/mean": -0.5695833563804626,
+      "rewards/compute_weighted_reward/std": 0.9432252049446106,
+      "step": 97
+    },
+    {
+      "completion_length": 1135.9166666666667,
+      "completions/clipped_ratio": 0.625,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1101.0,
+      "completions/mean_length": 1159.041748046875,
+      "completions/mean_terminated_length": 757.4444580078125,
+      "completions/min_length": 496.0,
+      "completions/min_terminated_length": 496.0,
+      "epoch": 4.454545454545454,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.7465776801109314,
+      "kl": 0.0063691134079514695,
+      "learning_rate": 1.2933333333333332e-06,
+      "loss": 0.1078,
+      "num_tokens": 3582205.0,
+      "reward": -0.36000001430511475,
+      "reward_std": 1.0805914402008057,
+      "rewards/compute_weighted_reward/mean": -0.35999998450279236,
+      "rewards/compute_weighted_reward/std": 1.0936933755874634,
+      "step": 98
+    },
+    {
+      "completion_length": 1235.5833333333333,
+      "completions/clipped_ratio": 0.6666666666666667,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1126.0,
+      "completions/mean_length": 1235.5833740234375,
+      "completions/mean_terminated_length": 906.75,
+      "completions/min_length": 315.0,
+      "completions/min_terminated_length": 315.0,
+      "epoch": 4.5,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.7566038966178894,
+      "kl": 0.00896931966417469,
+      "learning_rate": 1.3066666666666665e-06,
+      "loss": 0.1517,
+      "num_tokens": 3620709.0,
+      "reward": -0.8066667318344116,
+      "reward_std": 0.6958996057510376,
+      "rewards/compute_weighted_reward/mean": -0.8066666722297668,
+      "rewards/compute_weighted_reward/std": 0.7807168960571289,
+      "step": 99
+    },
+    {
+      "completion_length": 1026.3333333333333,
+      "completions/clipped_ratio": 0.5416666666666667,
+      "completions/max_length": 1400.0,
+      "completions/max_terminated_length": 1100.0,
+      "completions/mean_length": 1026.3333740234375,
+      "completions/mean_terminated_length": 584.727294921875,
+      "completions/min_length": 149.0,
+      "completions/min_terminated_length": 149.0,
+      "epoch": 4.545454545454545,
+      "frac_reward_zero_std": 0.0,
+      "grad_norm": 0.7430728673934937,
+      "kl": 0.00684088855147517,
+      "learning_rate": 1.32e-06,
+      "loss": 0.1119,
+      "num_tokens": 3653873.0,
+      "reward": -0.3241666555404663,
+      "reward_std": 0.7433856725692749,
+      "rewards/compute_weighted_reward/mean": -0.3241666555404663,
+      "rewards/compute_weighted_reward/std": 1.124342918395996,
+      "step": 100
     }
   ],
   "logging_steps": 1,
   "max_steps": 1500,
-  "num_input_tokens_seen": 3300702,
+  "num_input_tokens_seen": 3653873,
   "num_train_epochs": 69,
   "save_steps": 10,
   "stateful_callbacks": {