End of training

all_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 4.617630838765763e-06,
-    "train_runtime": 463.4946,
-    "train_samples": 599,
-    "train_samples_per_second": 0.69,
-    "train_steps_per_second": 0.043
+    "train_loss": 6.580724198101962e-06,
+    "train_runtime": 297.5347,
+    "train_samples": 160,
+    "train_samples_per_second": 1.076,
+    "train_steps_per_second": 0.067
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4a57fd08d5b5d0ca9650c6d1a6a1f2bdab037daed303c9a4ec056d64690a2823
+oid sha256:f279d5f96ef311bfa26853ace4483c311734470b5d555484ca3bdbee3eb1046e
 size 1976163472

train_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 4.617630838765763e-06,
-    "train_runtime": 463.4946,
-    "train_samples": 599,
-    "train_samples_per_second": 0.69,
-    "train_steps_per_second": 0.043
+    "train_loss": 6.580724198101962e-06,
+    "train_runtime": 297.5347,
+    "train_samples": 160,
+    "train_samples_per_second": 1.076,
+    "train_steps_per_second": 0.067
 }

trainer_state.json

@@ -2,7 +2,7 @@
   "best_global_step": null,
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.2671118530884808,
+  "epoch": 1.0,
   "eval_steps": 500,
   "global_step": 20,
   "is_hyper_param_search": false,
@@ -10,203 +10,203 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "completion_length": 299.34375,
-      "epoch": 0.02671118530884808,
-      "grad_norm": 94.53134155273438,
+      "completion_length": 428.96875,
+      "epoch": 0.1,
+      "grad_norm": 19.05489730834961,
       "kl": 0.0,
       "learning_rate": 4.965903258506806e-07,
       "loss": 0.0,
-      "reward": 0.4095814856700599,
-      "reward_std": 0.24939912266563624,
-      "rewards/concensus_correctness_reward_func": 0.0,
-      "rewards/consensus_reward_func": 0.0625,
+      "reward": 0.7318255892023444,
+      "reward_std": 1.2992593459784985,
+      "rewards/concensus_correctness_reward_func": 0.625,
+      "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.25,
-      "rewards/question_recreation_reward_func": 0.17958148289471865,
+      "rewards/final_correctness_reward_func": 0.0,
+      "rewards/question_recreation_reward_func": 0.21735686634201556,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": -0.08249999745748937,
+      "rewards/xmlcount_reward_func": -0.11053124722093344,
       "step": 2
     },
     {
-      "completion_length": 316.9375,
-      "epoch": 0.05342237061769616,
-      "grad_norm": 9.352435111999512,
-      "kl": 0.0014936136085452745,
+      "completion_length": 359.3125,
+      "epoch": 0.2,
+      "grad_norm": 9.828889846801758,
+      "kl": 0.0023647180787520483,
       "learning_rate": 4.698684378016222e-07,
       "loss": 0.0,
-      "reward": 1.0824220394715667,
-      "reward_std": 1.2490277905308176,
-      "rewards/concensus_correctness_reward_func": 0.625,
-      "rewards/consensus_reward_func": 0.0625,
+      "reward": 0.4143540321383625,
+      "reward_std": 0.4350623367281514,
+      "rewards/concensus_correctness_reward_func": 0.0,
+      "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.0625,
-      "rewards/question_recreation_reward_func": 0.1588594950735569,
+      "rewards/final_correctness_reward_func": 0.0,
+      "rewards/question_recreation_reward_func": 0.3118227787781507,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.1735625034198165,
+      "rewards/xmlcount_reward_func": 0.10253124916926026,
       "step": 4
     },
     {
-      "completion_length": 432.5,
-      "epoch": 0.08013355592654424,
-      "grad_norm": 72.27102661132812,
-      "kl": 0.0026759140364447376,
+      "completion_length": 367.9375,
+      "epoch": 0.3,
+      "grad_norm": 17.577388763427734,
+      "kl": 0.0013310233516676817,
       "learning_rate": 4.193203929064353e-07,
       "loss": 0.0,
-      "reward": 0.26975517414393835,
-      "reward_std": 0.23289441590895876,
+      "reward": 0.25027111219242215,
+      "reward_std": 0.45022515289019793,
       "rewards/concensus_correctness_reward_func": 0.0,
       "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
       "rewards/final_correctness_reward_func": 0.0,
-      "rewards/question_recreation_reward_func": 0.16150517805363052,
+      "rewards/question_recreation_reward_func": 0.178021113446448,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.10824999865144491,
+      "rewards/xmlcount_reward_func": 0.07224999833852053,
       "step": 6
     },
     {
-      "completion_length": 424.3125,
-      "epoch": 0.10684474123539232,
-      "grad_norm": 57.67626190185547,
-      "kl": 0.003170746138493996,
+      "completion_length": 426.25,
+      "epoch": 0.4,
+      "grad_norm": 5.737998008728027,
+      "kl": 0.0014973582146922126,
       "learning_rate": 3.5042385616324236e-07,
       "loss": 0.0,
-      "reward": 0.38037113891914487,
-      "reward_std": 0.40154265408637,
+      "reward": 0.24784131965134293,
+      "reward_std": 0.27939651551423594,
       "rewards/concensus_correctness_reward_func": 0.0,
       "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.125,
-      "rewards/question_recreation_reward_func": 0.14277737913653255,
+      "rewards/final_correctness_reward_func": 0.0,
+      "rewards/question_recreation_reward_func": 0.2965288240229711,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.11259375081135659,
+      "rewards/xmlcount_reward_func": -0.04868750323657878,
       "step": 8
     },
     {
-      "completion_length": 388.5,
-      "epoch": 0.1335559265442404,
-      "grad_norm": 5.881807804107666,
-      "kl": 0.008150014091370394,
+      "completion_length": 397.0,
+      "epoch": 0.5,
+      "grad_norm": 100.46427917480469,
+      "kl": 0.012153907351603266,
       "learning_rate": 2.706448363680831e-07,
       "loss": 0.0,
-      "reward": 0.009647406113799661,
-      "reward_std": 0.48192066283809254,
-      "rewards/concensus_correctness_reward_func": 0.0,
+      "reward": 0.9702369113219902,
+      "reward_std": 1.4984166798894876,
+      "rewards/concensus_correctness_reward_func": 0.625,
       "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.0,
-      "rewards/question_recreation_reward_func": 0.11724113661330193,
+      "rewards/final_correctness_reward_func": 0.0625,
+      "rewards/question_recreation_reward_func": 0.3326743959914893,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": -0.10759374493500218,
+      "rewards/xmlcount_reward_func": -0.04993749689310789,
       "step": 10
     },
     {
-      "completion_length": 329.3125,
-      "epoch": 0.16026711185308848,
-      "grad_norm": 32.46157455444336,
-      "kl": 0.0016044293206505245,
+      "completion_length": 395.9375,
+      "epoch": 0.6,
+      "grad_norm": 77.1864013671875,
+      "kl": 0.01412309450097382,
       "learning_rate": 1.886286282148002e-07,
       "loss": 0.0,
-      "reward": 0.3704580693738535,
-      "reward_std": 0.46280903373553883,
+      "reward": 0.5360800982452929,
+      "reward_std": 0.544823872827692,
       "rewards/concensus_correctness_reward_func": 0.0,
       "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.0625,
-      "rewards/question_recreation_reward_func": 0.19455181124794763,
+      "rewards/final_correctness_reward_func": 0.1875,
+      "rewards/question_recreation_reward_func": 0.3132363584591076,
       "rewards/soft_format_reward_func": 0.0,
-      "rewards/strict_format_reward_func": 0.015625,
-      "rewards/xmlcount_reward_func": 0.09778124839067459,
+      "rewards/strict_format_reward_func": 0.0,
+      "rewards/xmlcount_reward_func": 0.03534375294111669,
       "step": 12
     },
     {
-      "completion_length": 385.28125,
-      "epoch": 0.18697829716193656,
-      "grad_norm": 63.96397018432617,
-      "kl": 0.019191041095837136,
+      "completion_length": 331.75,
+      "epoch": 0.7,
+      "grad_norm": 114.8893814086914,
+      "kl": 0.02116644348097907,
       "learning_rate": 1.1326296046939333e-07,
       "loss": 0.0,
-      "reward": 0.7466699600918218,
-      "reward_std": 1.2343180573134873,
-      "rewards/concensus_correctness_reward_func": 0.625,
+      "reward": 0.5505048093036748,
+      "reward_std": 0.5207882627146319,
+      "rewards/concensus_correctness_reward_func": 0.016312500461935997,
       "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
       "rewards/final_correctness_reward_func": 0.0625,
-      "rewards/question_recreation_reward_func": 0.1140449601225555,
+      "rewards/question_recreation_reward_func": 0.3140360583201982,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": -0.05487499665468931,
+      "rewards/xmlcount_reward_func": 0.1576562486588955,
       "step": 14
     },
     {
-      "completion_length": 381.03125,
-      "epoch": 0.21368948247078465,
-      "grad_norm": 12.218117713928223,
-      "kl": 0.003359480255312519,
+      "completion_length": 370.84375,
+      "epoch": 0.8,
+      "grad_norm": 40.09614562988281,
+      "kl": 0.005846220530656865,
       "learning_rate": 5.271487265090163e-08,
       "loss": 0.0,
-      "reward": 0.4025605304632336,
-      "reward_std": 0.7095250401762314,
-      "rewards/concensus_correctness_reward_func": 0.0,
-      "rewards/consensus_reward_func": 0.0,
+      "reward": 0.8011319562792778,
+      "reward_std": 0.8181897932081483,
+      "rewards/concensus_correctness_reward_func": 0.0078125,
+      "rewards/consensus_reward_func": 0.0625,
       "rewards/cumulative_reward_2": 0.0,
       "rewards/final_correctness_reward_func": 0.125,
-      "rewards/question_recreation_reward_func": 0.23962301563005894,
+      "rewards/question_recreation_reward_func": 0.3886632148642093,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.037937495624646544,
+      "rewards/xmlcount_reward_func": 0.21715625002980232,
       "step": 16
     },
     {
-      "completion_length": 261.03125,
-      "epoch": 0.24040066777963273,
-      "grad_norm": 9.750836372375488,
-      "kl": 0.0017970694570976775,
+      "completion_length": 441.90625,
+      "epoch": 0.9,
+      "grad_norm": 6.093602657318115,
+      "kl": 0.003782330619287677,
       "learning_rate": 1.3545689574841341e-08,
       "loss": 0.0,
-      "reward": 0.4833010680740699,
-      "reward_std": 0.5299922852864256,
+      "reward": 0.5890867845155299,
+      "reward_std": 0.5183219794707838,
       "rewards/concensus_correctness_reward_func": 0.0,
       "rewards/consensus_reward_func": 0.0625,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.0625,
-      "rewards/question_recreation_reward_func": 0.2038323215674609,
+      "rewards/final_correctness_reward_func": 0.125,
+      "rewards/question_recreation_reward_func": 0.2310555315343663,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.15446875197812915,
+      "rewards/xmlcount_reward_func": 0.1705312468111515,
       "step": 18
     },
     {
-      "completion_length": 358.25,
-      "epoch": 0.2671118530884808,
-      "grad_norm": 30.307376861572266,
-      "kl": 0.0023762249620631337,
+      "completion_length": 353.71875,
+      "epoch": 1.0,
+      "grad_norm": 8.690644264221191,
+      "kl": 0.0031965336165740155,
       "learning_rate": 0.0,
       "loss": 0.0,
-      "reward": 0.5694868991849944,
-      "reward_std": 0.5472512795531657,
-      "rewards/concensus_correctness_reward_func": 0.05999999865889549,
-      "rewards/consensus_reward_func": 0.0625,
+      "reward": 2.057790318969637,
+      "reward_std": 0.697710811640718,
+      "rewards/concensus_correctness_reward_func": 1.2657500000059372,
+      "rewards/consensus_reward_func": 0.1875,
       "rewards/cumulative_reward_2": 0.0,
       "rewards/final_correctness_reward_func": 0.0625,
-      "rewards/question_recreation_reward_func": 0.25226815952919424,
+      "rewards/question_recreation_reward_func": 0.31100902207253966,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.13221875578165054,
+      "rewards/xmlcount_reward_func": 0.23103125020861626,
       "step": 20
     },
     {
-      "epoch": 0.2671118530884808,
+      "epoch": 1.0,
       "step": 20,
       "total_flos": 0.0,
-      "train_loss": 4.617630838765763e-06,
-      "train_runtime": 463.4946,
-      "train_samples_per_second": 0.69,
-      "train_steps_per_second": 0.043
+      "train_loss": 6.580724198101962e-06,
+      "train_runtime": 297.5347,
+      "train_samples_per_second": 1.076,
+      "train_steps_per_second": 0.067
     }
   ],
   "logging_steps": 2,