End of training

all_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 2.6065856218338014e-06,
-    "train_runtime": 261.2528,
+    "train_loss": 2.6590918423607945e-06,
+    "train_runtime": 231.6034,
     "train_samples": 160,
-    "train_samples_per_second": 1.225,
-    "train_steps_per_second": 0.077
+    "train_samples_per_second": 1.382,
+    "train_steps_per_second": 0.086
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4e0a4b7e4a48b737071c4fbf382d2ce78a803d21a72c0402e9ca41dbd52300ae
+oid sha256:65870f1c21cfa9175e945181d5a7ffeb9be5b0b73626a4b955326cb5562c6bb4
 size 1976163472

train_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 2.6065856218338014e-06,
-    "train_runtime": 261.2528,
+    "train_loss": 2.6590918423607945e-06,
+    "train_runtime": 231.6034,
     "train_samples": 160,
-    "train_samples_per_second": 1.225,
-    "train_steps_per_second": 0.077
+    "train_samples_per_second": 1.382,
+    "train_steps_per_second": 0.086
 }

trainer_state.json

@@ -10,203 +10,203 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "completion_length": 455.09375,
+      "completion_length": 314.3125,
       "epoch": 0.1,
-      "grad_norm": 4.092990398406982,
+      "grad_norm": 9.280272483825684,
       "kl": 0.0,
       "learning_rate": 4.965903258506806e-07,
       "loss": -0.0,
-      "reward": 1.5772938383743167,
-      "reward_std": 1.2715626050485298,
-      "rewards/concensus_correctness_reward_func": 0.3084375001490116,
-      "rewards/consensus_reward_func": 0.3125,
+      "reward": 2.0575108302291483,
+      "reward_std": 0.8434620648458804,
+      "rewards/concensus_correctness_reward_func": 0.3087499961256981,
+      "rewards/consensus_reward_func": 0.25,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.25,
-      "rewards/question_recreation_reward_func": 0.47582510093343444,
+      "rewards/final_correctness_reward_func": 0.4375,
+      "rewards/question_recreation_reward_func": 0.5297296575736254,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.23053125850856304,
+      "rewards/xmlcount_reward_func": 0.5315312626771629,
       "step": 2
     },
     {
-      "completion_length": 393.1875,
+      "completion_length": 360.875,
       "epoch": 0.2,
-      "grad_norm": 4.268237113952637,
-      "kl": 0.0010177667390962597,
+      "grad_norm": 10.275147438049316,
+      "kl": 0.0013246262224129168,
       "learning_rate": 4.698684378016222e-07,
       "loss": 0.0,
-      "reward": 1.5314793214201927,
-      "reward_std": 1.1808095844462514,
-      "rewards/concensus_correctness_reward_func": 0.16712500527501106,
-      "rewards/consensus_reward_func": 0.125,
+      "reward": 2.3956077583134174,
+      "reward_std": 1.8462393523659557,
+      "rewards/concensus_correctness_reward_func": 0.48374999687075615,
+      "rewards/consensus_reward_func": 0.625,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.1875,
-      "rewards/question_recreation_reward_func": 0.43022931227460504,
+      "rewards/final_correctness_reward_func": 0.0625,
+      "rewards/question_recreation_reward_func": 0.5609202478080988,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.6216250043362379,
+      "rewards/xmlcount_reward_func": 0.6634375005960464,
       "step": 4
     },
     {
-      "completion_length": 489.53125,
+      "completion_length": 371.75,
       "epoch": 0.3,
-      "grad_norm": 7.274165630340576,
-      "kl": 0.0015757040673634037,
+      "grad_norm": 15.171697616577148,
+      "kl": 0.0013205935320002027,
       "learning_rate": 4.193203929064353e-07,
       "loss": 0.0,
-      "reward": 3.2696286998689175,
-      "reward_std": 3.0783104081638157,
-      "rewards/concensus_correctness_reward_func": 1.4845624985173345,
-      "rewards/consensus_reward_func": 0.375,
+      "reward": 2.12227426469326,
+      "reward_std": 1.2299888343550265,
+      "rewards/concensus_correctness_reward_func": 0.3491250006482005,
+      "rewards/consensus_reward_func": 0.5,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.375,
-      "rewards/question_recreation_reward_func": 0.593472408130765,
+      "rewards/final_correctness_reward_func": 0.1875,
+      "rewards/question_recreation_reward_func": 0.5709305738564581,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.44159374572336674,
+      "rewards/xmlcount_reward_func": 0.5147187504917383,
       "step": 6
     },
     {
-      "completion_length": 391.28125,
+      "completion_length": 298.125,
       "epoch": 0.4,
-      "grad_norm": 4.208556175231934,
-      "kl": 0.001971208315808326,
+      "grad_norm": 4.224185466766357,
+      "kl": 0.002366176384384744,
       "learning_rate": 3.5042385616324236e-07,
       "loss": 0.0,
-      "reward": 2.182107772678137,
-      "reward_std": 1.9605439552105963,
-      "rewards/concensus_correctness_reward_func": 0.773624999448657,
-      "rewards/consensus_reward_func": 0.25,
+      "reward": 2.0370243564248085,
+      "reward_std": 1.1827143937116489,
+      "rewards/concensus_correctness_reward_func": 0.2841249965131283,
+      "rewards/consensus_reward_func": 0.4375,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.0,
-      "rewards/question_recreation_reward_func": 0.4976702528074384,
+      "rewards/final_correctness_reward_func": 0.0625,
+      "rewards/question_recreation_reward_func": 0.4238368161022663,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.6608125120401382,
+      "rewards/xmlcount_reward_func": 0.8290624935179949,
       "step": 8
     },
     {
-      "completion_length": 354.875,
+      "completion_length": 321.1875,
       "epoch": 0.5,
-      "grad_norm": 4.434547424316406,
-      "kl": 0.003326518097310327,
+      "grad_norm": 4.544751167297363,
+      "kl": 0.003099232639215188,
       "learning_rate": 2.706448363680831e-07,
       "loss": 0.0,
-      "reward": 2.21128673478961,
-      "reward_std": 1.2140152475330979,
-      "rewards/concensus_correctness_reward_func": 0.2775624953210354,
-      "rewards/consensus_reward_func": 0.5625,
+      "reward": 1.6763035412877798,
+      "reward_std": 0.9356762454845011,
+      "rewards/concensus_correctness_reward_func": 0.11650000140070915,
+      "rewards/consensus_reward_func": 0.1875,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.125,
-      "rewards/question_recreation_reward_func": 0.5232867355225608,
+      "rewards/final_correctness_reward_func": 0.1875,
+      "rewards/question_recreation_reward_func": 0.4396473169326782,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.7229375082533807,
+      "rewards/xmlcount_reward_func": 0.7451562508940697,
       "step": 10
     },
     {
-      "completion_length": 361.8125,
+      "completion_length": 313.96875,
       "epoch": 0.6,
-      "grad_norm": 4.211982250213623,
-      "kl": 0.003115422310656868,
+      "grad_norm": 5.193901538848877,
+      "kl": 0.0036189594757161103,
       "learning_rate": 1.886286282148002e-07,
       "loss": 0.0,
-      "reward": 1.6850453205406666,
-      "reward_std": 0.9321421514177928,
-      "rewards/concensus_correctness_reward_func": 0.13374999910593033,
-      "rewards/consensus_reward_func": 0.25,
+      "reward": 2.294987767934799,
+      "reward_std": 1.8568786369287409,
+      "rewards/concensus_correctness_reward_func": 0.7528750021010637,
+      "rewards/consensus_reward_func": 0.125,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.0,
-      "rewards/question_recreation_reward_func": 0.678232790902257,
+      "rewards/final_correctness_reward_func": 0.0625,
+      "rewards/question_recreation_reward_func": 0.6182377617806196,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.6230625174939632,
+      "rewards/xmlcount_reward_func": 0.7363750152289867,
       "step": 12
     },
     {
-      "completion_length": 376.125,
+      "completion_length": 307.15625,
       "epoch": 0.7,
-      "grad_norm": 5.003601551055908,
-      "kl": 0.0039603932164027356,
+      "grad_norm": 5.295619487762451,
+      "kl": 0.004047992595587857,
       "learning_rate": 1.1326296046939333e-07,
       "loss": 0.0,
-      "reward": 1.172058342024684,
-      "reward_std": 1.0172426409553736,
-      "rewards/concensus_correctness_reward_func": 0.08587499894201756,
-      "rewards/consensus_reward_func": 0.125,
+      "reward": 2.109817750751972,
+      "reward_std": 0.8735323960427195,
+      "rewards/concensus_correctness_reward_func": 0.23537499830126762,
+      "rewards/consensus_reward_func": 0.375,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.125,
-      "rewards/question_recreation_reward_func": 0.36012083862442523,
+      "rewards/final_correctness_reward_func": 0.1875,
+      "rewards/question_recreation_reward_func": 0.6091302447021008,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.47606251761317253,
+      "rewards/xmlcount_reward_func": 0.7028124947100878,
       "step": 14
     },
     {
-      "completion_length": 385.25,
+      "completion_length": 353.65625,
       "epoch": 0.8,
-      "grad_norm": 4.454174995422363,
-      "kl": 0.003436649523791857,
+      "grad_norm": 4.300991535186768,
+      "kl": 0.0034836612321669236,
       "learning_rate": 5.271487265090163e-08,
       "loss": 0.0,
-      "reward": 4.223175309598446,
-      "reward_std": 3.490735905012116,
-      "rewards/concensus_correctness_reward_func": 2.179250003769994,
-      "rewards/consensus_reward_func": 0.4375,
+      "reward": 3.4648909382522106,
+      "reward_std": 2.4452727790921926,
+      "rewards/concensus_correctness_reward_func": 1.3952499981969595,
+      "rewards/consensus_reward_func": 0.1875,
       "rewards/cumulative_reward_2": 0.0,
       "rewards/final_correctness_reward_func": 0.4375,
-      "rewards/question_recreation_reward_func": 0.6258939877152443,
+      "rewards/question_recreation_reward_func": 0.6780784213915467,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.5430312510579824,
+      "rewards/xmlcount_reward_func": 0.7665625056251884,
       "step": 16
     },
     {
-      "completion_length": 380.46875,
+      "completion_length": 367.625,
       "epoch": 0.9,
-      "grad_norm": 4.106444835662842,
-      "kl": 0.0034525456139817834,
+      "grad_norm": 5.286797046661377,
+      "kl": 0.004129875829676166,
       "learning_rate": 1.3545689574841341e-08,
       "loss": 0.0,
-      "reward": 1.6361163649708033,
-      "reward_std": 1.052419565152377,
-      "rewards/concensus_correctness_reward_func": 0.28506250120699406,
-      "rewards/consensus_reward_func": 0.25,
+      "reward": 2.059478480485268,
+      "reward_std": 1.6678909536276478,
+      "rewards/concensus_correctness_reward_func": 0.737125001847744,
+      "rewards/consensus_reward_func": 0.125,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.0,
-      "rewards/question_recreation_reward_func": 0.561085123103112,
-      "rewards/soft_format_reward_func": 0.0,
+      "rewards/final_correctness_reward_func": 0.1875,
+      "rewards/question_recreation_reward_func": 0.6268222468206659,
+      "rewards/soft_format_reward_func": 0.015625,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.5399687550961971,
+      "rewards/xmlcount_reward_func": 0.36740624997764826,
       "step": 18
     },
     {
-      "completion_length": 393.1875,
+      "completion_length": 329.34375,
       "epoch": 1.0,
-      "grad_norm": 3.7455756664276123,
-      "kl": 0.004411991249071434,
+      "grad_norm": 5.023986339569092,
+      "kl": 0.003600221549277194,
       "learning_rate": 0.0,
       "loss": 0.0,
-      "reward": 2.0187183283269405,
-      "reward_std": 0.8147448373201769,
-      "rewards/concensus_correctness_reward_func": 0.0521249994635582,
-      "rewards/consensus_reward_func": 0.1875,
+      "reward": 1.976121924817562,
+      "reward_std": 1.5817541421856731,
+      "rewards/concensus_correctness_reward_func": 0.3413750007748604,
+      "rewards/consensus_reward_func": 0.5625,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.5,
-      "rewards/question_recreation_reward_func": 0.6972183138132095,
+      "rewards/final_correctness_reward_func": 0.0625,
+      "rewards/question_recreation_reward_func": 0.6549656670540571,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.5818750094622374,
+      "rewards/xmlcount_reward_func": 0.3547812607139349,
       "step": 20
     },
     {
       "epoch": 1.0,
       "step": 20,
       "total_flos": 0.0,
-      "train_loss": 2.6065856218338014e-06,
-      "train_runtime": 261.2528,
-      "train_samples_per_second": 1.225,
-      "train_steps_per_second": 0.077
+      "train_loss": 2.6590918423607945e-06,
+      "train_runtime": 231.6034,
+      "train_samples_per_second": 1.382,
+      "train_steps_per_second": 0.086
     }
   ],
   "logging_steps": 2,