End of training

all_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 4.229552814649651e-06,
-    "train_runtime": 224.5386,
+    "train_loss": 4.321709275245666e-06,
+    "train_runtime": 239.6333,
     "train_samples": 160,
-    "train_samples_per_second": 1.425,
-    "train_steps_per_second": 0.089
+    "train_samples_per_second": 1.335,
+    "train_steps_per_second": 0.083
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cd8fbc6de9c364fa5f5a500098b1b31f465b9c6da2caea520c2ac6afd2ea308c
+oid sha256:07101fdeff75e2bdfedc7595476932b673abf2c20f0e44baa5d3f24039884a84
 size 1976163472

train_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 4.229552814649651e-06,
-    "train_runtime": 224.5386,
+    "train_loss": 4.321709275245666e-06,
+    "train_runtime": 239.6333,
     "train_samples": 160,
-    "train_samples_per_second": 1.425,
-    "train_steps_per_second": 0.089
+    "train_samples_per_second": 1.335,
+    "train_steps_per_second": 0.083
 }

trainer_state.json

@@ -10,203 +10,203 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "completion_length": 302.125,
+      "completion_length": 344.25,
       "epoch": 0.1,
-      "grad_norm": 6.64198112487793,
+      "grad_norm": 5.403124809265137,
       "kl": 0.0,
       "learning_rate": 5e-07,
       "loss": -0.0,
-      "reward": 2.1132799829356372,
-      "reward_std": 2.0948233329690993,
-      "rewards/concensus_correctness_reward_func": 0.9079375043511391,
-      "rewards/consensus_reward_func": 0.3125,
+      "reward": 2.3487942395731807,
+      "reward_std": 1.331467665731907,
+      "rewards/concensus_correctness_reward_func": 0.6510000005364418,
+      "rewards/consensus_reward_func": 0.75,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.0625,
-      "rewards/question_recreation_reward_func": 0.4575299544958398,
-      "rewards/soft_format_reward_func": 0.0,
+      "rewards/final_correctness_reward_func": 0.0,
+      "rewards/question_recreation_reward_func": 0.46238797600381076,
+      "rewards/soft_format_reward_func": 0.015625,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.3728125037159771,
+      "rewards/xmlcount_reward_func": 0.46978125628083944,
       "step": 2
     },
     {
-      "completion_length": 257.84375,
+      "completion_length": 364.75,
       "epoch": 0.2,
-      "grad_norm": 5.92756462097168,
-      "kl": 0.0023351105592155363,
+      "grad_norm": 6.2670817375183105,
+      "kl": 0.0021806862114317482,
       "learning_rate": 4.864543104251586e-07,
       "loss": 0.0,
-      "reward": 2.712690506130457,
-      "reward_std": 1.4277536073932424,
-      "rewards/concensus_correctness_reward_func": 0.4374999962747097,
-      "rewards/consensus_reward_func": 0.4375,
+      "reward": 2.635450964793563,
+      "reward_std": 2.075526769272983,
+      "rewards/concensus_correctness_reward_func": 0.9166250005364418,
+      "rewards/consensus_reward_func": 0.5,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.375,
-      "rewards/question_recreation_reward_func": 0.6360654862364754,
+      "rewards/final_correctness_reward_func": 0.25,
+      "rewards/question_recreation_reward_func": 0.4957009544596076,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.8266249969601631,
+      "rewards/xmlcount_reward_func": 0.4731249902397394,
       "step": 4
     },
     {
-      "completion_length": 336.78125,
+      "completion_length": 404.71875,
       "epoch": 0.3,
-      "grad_norm": 5.250879764556885,
-      "kl": 0.0028426997378119268,
+      "grad_norm": 5.5379438400268555,
+      "kl": 0.0030734814645256847,
       "learning_rate": 4.472851273490984e-07,
       "loss": 0.0,
-      "reward": 2.5608479902148247,
-      "reward_std": 1.7966383500024676,
-      "rewards/concensus_correctness_reward_func": 0.4780624993145466,
-      "rewards/consensus_reward_func": 0.5625,
+      "reward": 2.522380673326552,
+      "reward_std": 2.0315658951876685,
+      "rewards/concensus_correctness_reward_func": 0.895562507212162,
+      "rewards/consensus_reward_func": 0.3125,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.25,
-      "rewards/question_recreation_reward_func": 0.5151917014736682,
+      "rewards/final_correctness_reward_func": 0.125,
+      "rewards/question_recreation_reward_func": 0.5827869249042124,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.7550937533378601,
+      "rewards/xmlcount_reward_func": 0.6065312507562339,
       "step": 6
     },
     {
-      "completion_length": 281.65625,
+      "completion_length": 354.0,
       "epoch": 0.4,
-      "grad_norm": 5.771662712097168,
-      "kl": 0.002982961232191883,
+      "grad_norm": 6.651269435882568,
+      "kl": 0.003421090141273453,
       "learning_rate": 3.867370395306068e-07,
       "loss": 0.0,
-      "reward": 1.6815071050077677,
-      "reward_std": 1.2123009855858982,
-      "rewards/concensus_correctness_reward_func": 0.20543750189244747,
-      "rewards/consensus_reward_func": 0.25,
+      "reward": 2.4581775926053524,
+      "reward_std": 1.335879288148135,
+      "rewards/concensus_correctness_reward_func": 0.6216250024735928,
+      "rewards/consensus_reward_func": 0.5625,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.1875,
-      "rewards/question_recreation_reward_func": 0.42456963809672743,
+      "rewards/final_correctness_reward_func": 0.125,
+      "rewards/question_recreation_reward_func": 0.4378338464302942,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.6140000028535724,
+      "rewards/xmlcount_reward_func": 0.7112187594175339,
       "step": 8
     },
     {
-      "completion_length": 319.875,
+      "completion_length": 300.0,
       "epoch": 0.5,
-      "grad_norm": 7.018432140350342,
-      "kl": 0.00529094669764163,
+      "grad_norm": 6.796236038208008,
+      "kl": 0.004635282557501341,
       "learning_rate": 3.1137137178519977e-07,
       "loss": 0.0,
-      "reward": 3.415244299918413,
-      "reward_std": 2.043720346176997,
-      "rewards/concensus_correctness_reward_func": 1.2149999849498272,
-      "rewards/consensus_reward_func": 0.5625,
+      "reward": 2.6668134666979313,
+      "reward_std": 1.702367587480694,
+      "rewards/concensus_correctness_reward_func": 0.6051250025629997,
+      "rewards/consensus_reward_func": 0.5,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.25,
-      "rewards/question_recreation_reward_func": 0.574900631327182,
+      "rewards/final_correctness_reward_func": 0.375,
+      "rewards/question_recreation_reward_func": 0.61225092317909,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.8128437548875809,
+      "rewards/xmlcount_reward_func": 0.5744375269860029,
       "step": 10
     },
     {
-      "completion_length": 204.90625,
+      "completion_length": 246.28125,
       "epoch": 0.6,
-      "grad_norm": 5.2989678382873535,
-      "kl": 0.00497750002250541,
+      "grad_norm": 4.856728553771973,
+      "kl": 0.006234622225747444,
       "learning_rate": 2.2935516363191693e-07,
       "loss": 0.0,
-      "reward": 1.8684281595051289,
-      "reward_std": 0.6075215226737782,
-      "rewards/concensus_correctness_reward_func": 0.23737500235438347,
-      "rewards/consensus_reward_func": 0.25,
+      "reward": 1.907209224998951,
+      "reward_std": 1.2923782205325551,
+      "rewards/concensus_correctness_reward_func": 0.2936249990016222,
+      "rewards/consensus_reward_func": 0.375,
       "rewards/cumulative_reward_2": 0.0,
       "rewards/final_correctness_reward_func": 0.0,
-      "rewards/question_recreation_reward_func": 0.437771886237897,
+      "rewards/question_recreation_reward_func": 0.48558421013876796,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.9432812370359898,
+      "rewards/xmlcount_reward_func": 0.7530000098049641,
       "step": 12
     },
     {
-      "completion_length": 324.53125,
+      "completion_length": 277.84375,
       "epoch": 0.7,
-      "grad_norm": 6.422370910644531,
-      "kl": 0.005500798411958385,
+      "grad_norm": 38.3473014831543,
+      "kl": 0.005429227188869845,
       "learning_rate": 1.4957614383675767e-07,
       "loss": 0.0,
-      "reward": 1.6825342141091824,
-      "reward_std": 0.7608490395359695,
-      "rewards/concensus_correctness_reward_func": 0.08293750137090683,
-      "rewards/consensus_reward_func": 0.125,
+      "reward": 2.3881534561514854,
+      "reward_std": 1.1518572303466499,
+      "rewards/concensus_correctness_reward_func": 0.43924999982118607,
+      "rewards/consensus_reward_func": 0.5,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.25,
-      "rewards/question_recreation_reward_func": 0.4649092424660921,
+      "rewards/final_correctness_reward_func": 0.125,
+      "rewards/question_recreation_reward_func": 0.557153444038704,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.7596875075250864,
+      "rewards/xmlcount_reward_func": 0.7667500115931034,
       "step": 14
     },
     {
-      "completion_length": 322.84375,
+      "completion_length": 289.0,
       "epoch": 0.8,
-      "grad_norm": 4.691440582275391,
-      "kl": 0.007061985146719962,
+      "grad_norm": 8.100913047790527,
+      "kl": 0.006314124533673748,
       "learning_rate": 8.067960709356478e-08,
       "loss": 0.0,
-      "reward": 3.5417994149029255,
-      "reward_std": 2.206683919765055,
-      "rewards/concensus_correctness_reward_func": 1.647125005722046,
-      "rewards/consensus_reward_func": 0.4375,
+      "reward": 2.4665270633995533,
+      "reward_std": 1.5265621307771653,
+      "rewards/concensus_correctness_reward_func": 0.5556250028312206,
+      "rewards/consensus_reward_func": 0.5625,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.25,
-      "rewards/question_recreation_reward_func": 0.5895181887317449,
+      "rewards/final_correctness_reward_func": 0.125,
+      "rewards/question_recreation_reward_func": 0.5798395671881735,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.6176562458276749,
+      "rewards/xmlcount_reward_func": 0.6435625050216913,
       "step": 16
     },
     {
-      "completion_length": 319.28125,
+      "completion_length": 378.90625,
       "epoch": 0.9,
-      "grad_norm": 5.079524517059326,
-      "kl": 0.0047352922993013635,
+      "grad_norm": 4.619319915771484,
+      "kl": 0.0059225473523838446,
       "learning_rate": 3.013156219837776e-08,
       "loss": 0.0,
-      "reward": 3.7184752356261015,
-      "reward_std": 3.4974633325473405,
-      "rewards/concensus_correctness_reward_func": 1.7295625060796738,
-      "rewards/consensus_reward_func": 0.4375,
+      "reward": 2.774918533861637,
+      "reward_std": 1.6789401592686772,
+      "rewards/concensus_correctness_reward_func": 0.5491875000298023,
+      "rewards/consensus_reward_func": 0.5,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.125,
-      "rewards/question_recreation_reward_func": 0.6388814090169035,
+      "rewards/final_correctness_reward_func": 0.3125,
+      "rewards/question_recreation_reward_func": 0.5945435594767332,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.787531275767833,
+      "rewards/xmlcount_reward_func": 0.8186875078827143,
       "step": 18
     },
     {
-      "completion_length": 441.875,
+      "completion_length": 296.0625,
       "epoch": 1.0,
-      "grad_norm": 5.213651657104492,
-      "kl": 0.0048062925998237915,
+      "grad_norm": 6.017106056213379,
+      "kl": 0.006086832385335583,
       "learning_rate": 3.4096741493194193e-09,
       "loss": 0.0,
-      "reward": 3.076372131705284,
-      "reward_std": 1.007006654748693,
-      "rewards/concensus_correctness_reward_func": 0.7903749961405993,
-      "rewards/consensus_reward_func": 0.8125,
+      "reward": 2.033856872469187,
+      "reward_std": 1.216237832326442,
+      "rewards/concensus_correctness_reward_func": 0.262687498703599,
+      "rewards/consensus_reward_func": 0.3125,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.25,
-      "rewards/question_recreation_reward_func": 0.6119345831684768,
+      "rewards/final_correctness_reward_func": 0.125,
+      "rewards/question_recreation_reward_func": 0.5571068990975618,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.611562492325902,
+      "rewards/xmlcount_reward_func": 0.7765624914318323,
       "step": 20
     },
     {
       "epoch": 1.0,
       "step": 20,
       "total_flos": 0.0,
-      "train_loss": 4.229552814649651e-06,
-      "train_runtime": 224.5386,
-      "train_samples_per_second": 1.425,
-      "train_steps_per_second": 0.089
+      "train_loss": 4.321709275245666e-06,
+      "train_runtime": 239.6333,
+      "train_samples_per_second": 1.335,
+      "train_steps_per_second": 0.083
     }
   ],
   "logging_steps": 2,