End of training

all_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 2.6590918423607945e-06,
-    "train_runtime": 231.6034,
+    "train_loss": 4.049018025398255e-06,
+    "train_runtime": 208.6037,
     "train_samples": 160,
-    "train_samples_per_second": 1.382,
-    "train_steps_per_second": 0.086
+    "train_samples_per_second": 1.534,
+    "train_steps_per_second": 0.096
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:65870f1c21cfa9175e945181d5a7ffeb9be5b0b73626a4b955326cb5562c6bb4
+oid sha256:c27a1c54960fca5bab3fc331df9494949a6460f90a93f41e51e5a8a053712a07
 size 1976163472

train_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 2.6590918423607945e-06,
-    "train_runtime": 231.6034,
+    "train_loss": 4.049018025398255e-06,
+    "train_runtime": 208.6037,
     "train_samples": 160,
-    "train_samples_per_second": 1.382,
-    "train_steps_per_second": 0.086
+    "train_samples_per_second": 1.534,
+    "train_steps_per_second": 0.096
 }

trainer_state.json

@@ -10,203 +10,203 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "completion_length": 314.3125,
+      "completion_length": 353.875,
       "epoch": 0.1,
-      "grad_norm": 9.280272483825684,
+      "grad_norm": 5.031754970550537,
       "kl": 0.0,
       "learning_rate": 4.965903258506806e-07,
-      "loss": -0.0,
-      "reward": 2.0575108302291483,
-      "reward_std": 0.8434620648458804,
-      "rewards/concensus_correctness_reward_func": 0.3087499961256981,
-      "rewards/consensus_reward_func": 0.25,
+      "loss": 0.0,
+      "reward": 2.7734319660812616,
+      "reward_std": 2.4145944046613295,
+      "rewards/concensus_correctness_reward_func": 1.25,
+      "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.4375,
-      "rewards/question_recreation_reward_func": 0.5297296575736254,
+      "rewards/final_correctness_reward_func": 0.375,
+      "rewards/question_recreation_reward_func": 0.5963382440531859,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.5315312626771629,
+      "rewards/xmlcount_reward_func": 0.5520937647670507,
       "step": 2
     },
     {
-      "completion_length": 360.875,
+      "completion_length": 334.46875,
       "epoch": 0.2,
-      "grad_norm": 10.275147438049316,
-      "kl": 0.0013246262224129168,
+      "grad_norm": 5.746564865112305,
+      "kl": 0.0010456620839249808,
       "learning_rate": 4.698684378016222e-07,
       "loss": 0.0,
-      "reward": 2.3956077583134174,
-      "reward_std": 1.8462393523659557,
-      "rewards/concensus_correctness_reward_func": 0.48374999687075615,
-      "rewards/consensus_reward_func": 0.625,
+      "reward": 1.7992137037217617,
+      "reward_std": 0.7727034210693091,
+      "rewards/concensus_correctness_reward_func": 0.08506250008940697,
+      "rewards/consensus_reward_func": 0.25,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.0625,
-      "rewards/question_recreation_reward_func": 0.5609202478080988,
-      "rewards/soft_format_reward_func": 0.0,
+      "rewards/final_correctness_reward_func": 0.1875,
+      "rewards/question_recreation_reward_func": 0.5819949674187228,
+      "rewards/soft_format_reward_func": 0.015625,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.6634375005960464,
+      "rewards/xmlcount_reward_func": 0.6790312509983778,
       "step": 4
     },
     {
-      "completion_length": 371.75,
+      "completion_length": 361.59375,
       "epoch": 0.3,
-      "grad_norm": 15.171697616577148,
-      "kl": 0.0013205935320002027,
+      "grad_norm": 3.3398702144622803,
+      "kl": 0.001574008900206536,
       "learning_rate": 4.193203929064353e-07,
       "loss": 0.0,
-      "reward": 2.12227426469326,
-      "reward_std": 1.2299888343550265,
-      "rewards/concensus_correctness_reward_func": 0.3491250006482005,
-      "rewards/consensus_reward_func": 0.5,
+      "reward": 2.9710799902677536,
+      "reward_std": 0.6227061338722706,
+      "rewards/concensus_correctness_reward_func": 1.2891874983906746,
+      "rewards/consensus_reward_func": 0.0625,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.1875,
-      "rewards/question_recreation_reward_func": 0.5709305738564581,
+      "rewards/final_correctness_reward_func": 0.3125,
+      "rewards/question_recreation_reward_func": 0.6594237750396132,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.5147187504917383,
+      "rewards/xmlcount_reward_func": 0.6474687531590462,
       "step": 6
     },
     {
-      "completion_length": 298.125,
+      "completion_length": 319.0625,
       "epoch": 0.4,
-      "grad_norm": 4.224185466766357,
-      "kl": 0.002366176384384744,
+      "grad_norm": 4.177555561065674,
+      "kl": 0.002741159994911868,
       "learning_rate": 3.5042385616324236e-07,
       "loss": 0.0,
-      "reward": 2.0370243564248085,
-      "reward_std": 1.1827143937116489,
-      "rewards/concensus_correctness_reward_func": 0.2841249965131283,
-      "rewards/consensus_reward_func": 0.4375,
+      "reward": 2.2837948873639107,
+      "reward_std": 1.5259181885048747,
+      "rewards/concensus_correctness_reward_func": 0.29250000044703484,
+      "rewards/consensus_reward_func": 0.375,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.0625,
-      "rewards/question_recreation_reward_func": 0.4238368161022663,
+      "rewards/final_correctness_reward_func": 0.25,
+      "rewards/question_recreation_reward_func": 0.611263575963676,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.8290624935179949,
+      "rewards/xmlcount_reward_func": 0.7550312634557486,
       "step": 8
     },
     {
-      "completion_length": 321.1875,
+      "completion_length": 315.1875,
       "epoch": 0.5,
-      "grad_norm": 4.544751167297363,
-      "kl": 0.003099232639215188,
+      "grad_norm": 27.93036651611328,
+      "kl": 0.0033685116504784673,
       "learning_rate": 2.706448363680831e-07,
       "loss": 0.0,
-      "reward": 1.6763035412877798,
-      "reward_std": 0.9356762454845011,
-      "rewards/concensus_correctness_reward_func": 0.11650000140070915,
-      "rewards/consensus_reward_func": 0.1875,
+      "reward": 1.7934808060526848,
+      "reward_std": 1.1033665230497718,
+      "rewards/concensus_correctness_reward_func": 0.23343750089406967,
+      "rewards/consensus_reward_func": 0.25,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.1875,
-      "rewards/question_recreation_reward_func": 0.4396473169326782,
+      "rewards/final_correctness_reward_func": 0.25,
+      "rewards/question_recreation_reward_func": 0.46682453801622614,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.7451562508940697,
+      "rewards/xmlcount_reward_func": 0.5932187382131815,
       "step": 10
     },
     {
-      "completion_length": 313.96875,
+      "completion_length": 294.90625,
       "epoch": 0.6,
-      "grad_norm": 5.193901538848877,
-      "kl": 0.0036189594757161103,
+      "grad_norm": 5.095856666564941,
+      "kl": 0.004714873259217711,
       "learning_rate": 1.886286282148002e-07,
       "loss": 0.0,
-      "reward": 2.294987767934799,
-      "reward_std": 1.8568786369287409,
-      "rewards/concensus_correctness_reward_func": 0.7528750021010637,
-      "rewards/consensus_reward_func": 0.125,
+      "reward": 1.7133545614778996,
+      "reward_std": 0.9104429350118153,
+      "rewards/concensus_correctness_reward_func": 0.1314375028014183,
+      "rewards/consensus_reward_func": 0.1875,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.0625,
-      "rewards/question_recreation_reward_func": 0.6182377617806196,
+      "rewards/final_correctness_reward_func": 0.125,
+      "rewards/question_recreation_reward_func": 0.5073858371470124,
       "rewards/soft_format_reward_func": 0.0,
-      "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.7363750152289867,
+      "rewards/strict_format_reward_func": 0.015625,
+      "rewards/xmlcount_reward_func": 0.7464062590152025,
       "step": 12
     },
     {
-      "completion_length": 307.15625,
+      "completion_length": 255.84375,
       "epoch": 0.7,
-      "grad_norm": 5.295619487762451,
-      "kl": 0.004047992595587857,
+      "grad_norm": 5.774594306945801,
+      "kl": 0.005561068042879924,
       "learning_rate": 1.1326296046939333e-07,
       "loss": 0.0,
-      "reward": 2.109817750751972,
-      "reward_std": 0.8735323960427195,
-      "rewards/concensus_correctness_reward_func": 0.23537499830126762,
-      "rewards/consensus_reward_func": 0.375,
+      "reward": 2.31035278737545,
+      "reward_std": 0.9512235811562277,
+      "rewards/concensus_correctness_reward_func": 0.10537499934434891,
+      "rewards/consensus_reward_func": 0.3125,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.1875,
-      "rewards/question_recreation_reward_func": 0.6091302447021008,
+      "rewards/final_correctness_reward_func": 0.3125,
+      "rewards/question_recreation_reward_func": 0.6544152703136206,
       "rewards/soft_format_reward_func": 0.0,
-      "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.7028124947100878,
+      "rewards/strict_format_reward_func": 0.015625,
+      "rewards/xmlcount_reward_func": 0.9099375084042549,
       "step": 14
     },
     {
-      "completion_length": 353.65625,
+      "completion_length": 274.25,
       "epoch": 0.8,
-      "grad_norm": 4.300991535186768,
-      "kl": 0.0034836612321669236,
+      "grad_norm": 4.460197448730469,
+      "kl": 0.006159632313938346,
       "learning_rate": 5.271487265090163e-08,
       "loss": 0.0,
-      "reward": 3.4648909382522106,
-      "reward_std": 2.4452727790921926,
-      "rewards/concensus_correctness_reward_func": 1.3952499981969595,
-      "rewards/consensus_reward_func": 0.1875,
+      "reward": 3.217790398746729,
+      "reward_std": 1.8388326059503015,
+      "rewards/concensus_correctness_reward_func": 0.8442499972879887,
+      "rewards/consensus_reward_func": 0.3125,
       "rewards/cumulative_reward_2": 0.0,
       "rewards/final_correctness_reward_func": 0.4375,
-      "rewards/question_recreation_reward_func": 0.6780784213915467,
+      "rewards/question_recreation_reward_func": 0.7221341663971543,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.7665625056251884,
+      "rewards/xmlcount_reward_func": 0.9014062602072954,
       "step": 16
     },
     {
-      "completion_length": 367.625,
+      "completion_length": 301.46875,
       "epoch": 0.9,
-      "grad_norm": 5.286797046661377,
-      "kl": 0.004129875829676166,
+      "grad_norm": 5.027771472930908,
+      "kl": 0.006903290901391301,
       "learning_rate": 1.3545689574841341e-08,
       "loss": 0.0,
-      "reward": 2.059478480485268,
-      "reward_std": 1.6678909536276478,
-      "rewards/concensus_correctness_reward_func": 0.737125001847744,
-      "rewards/consensus_reward_func": 0.125,
+      "reward": 2.116435706615448,
+      "reward_std": 0.9236437288418529,
+      "rewards/concensus_correctness_reward_func": 0.2777499984949827,
+      "rewards/consensus_reward_func": 0.25,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.1875,
-      "rewards/question_recreation_reward_func": 0.6268222468206659,
-      "rewards/soft_format_reward_func": 0.015625,
+      "rewards/final_correctness_reward_func": 0.0625,
+      "rewards/question_recreation_reward_func": 0.6349669927731156,
+      "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.36740624997764826,
+      "rewards/xmlcount_reward_func": 0.8912187665700912,
       "step": 18
     },
     {
-      "completion_length": 329.34375,
+      "completion_length": 251.90625,
       "epoch": 1.0,
-      "grad_norm": 5.023986339569092,
-      "kl": 0.003600221549277194,
+      "grad_norm": 5.603349208831787,
+      "kl": 0.008129101392114535,
       "learning_rate": 0.0,
       "loss": 0.0,
-      "reward": 1.976121924817562,
-      "reward_std": 1.5817541421856731,
-      "rewards/concensus_correctness_reward_func": 0.3413750007748604,
-      "rewards/consensus_reward_func": 0.5625,
+      "reward": 1.8469502702355385,
+      "reward_std": 0.9816727490979247,
+      "rewards/concensus_correctness_reward_func": 0.1430624984204769,
+      "rewards/consensus_reward_func": 0.1875,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.0625,
-      "rewards/question_recreation_reward_func": 0.6549656670540571,
+      "rewards/final_correctness_reward_func": 0.25,
+      "rewards/question_recreation_reward_func": 0.6068252576515079,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.3547812607139349,
+      "rewards/xmlcount_reward_func": 0.6595625132322311,
       "step": 20
     },
     {
       "epoch": 1.0,
       "step": 20,
       "total_flos": 0.0,
-      "train_loss": 2.6590918423607945e-06,
-      "train_runtime": 231.6034,
-      "train_samples_per_second": 1.382,
-      "train_steps_per_second": 0.086
+      "train_loss": 4.049018025398255e-06,
+      "train_runtime": 208.6037,
+      "train_samples_per_second": 1.534,
+      "train_steps_per_second": 0.096
     }
   ],
   "logging_steps": 2,