| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 20, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "completion_length": 226.9375, | |
| "epoch": 0.1, | |
| "grad_norm": 29.540603637695312, | |
| "kl": 0.0, | |
| "learning_rate": 5e-07, | |
| "loss": 0.0, | |
| "reward": 3.9297456964850426, | |
| "reward_std": 0.3364956162404269, | |
| "rewards/concensus_correctness_reward_func": 1.090124998241663, | |
| "rewards/consensus_reward_func": 1.0625, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.125, | |
| "rewards/question_recreation_reward_func": 0.6852769576944411, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.1875, | |
| "rewards/xmlcount_reward_func": 0.7793437507934868, | |
| "step": 2 | |
| }, | |
| { | |
| "completion_length": 140.875, | |
| "epoch": 0.2, | |
| "grad_norm": 21.954511642456055, | |
| "kl": 0.15643696358893067, | |
| "learning_rate": 4.864543104251586e-07, | |
| "loss": 0.0002, | |
| "reward": 6.30455806851387, | |
| "reward_std": 0.9327005930244923, | |
| "rewards/concensus_correctness_reward_func": 1.7068124860525131, | |
| "rewards/consensus_reward_func": 1.75, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.3125, | |
| "rewards/question_recreation_reward_func": 0.9510268457233906, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.40625, | |
| "rewards/xmlcount_reward_func": 1.1779687404632568, | |
| "step": 4 | |
| }, | |
| { | |
| "completion_length": 192.03125, | |
| "epoch": 0.3, | |
| "grad_norm": 734.108154296875, | |
| "kl": 19.806165897636674, | |
| "learning_rate": 4.472851273490984e-07, | |
| "loss": 0.0198, | |
| "reward": 5.56097885966301, | |
| "reward_std": 0.9280266016721725, | |
| "rewards/concensus_correctness_reward_func": 1.368249997496605, | |
| "rewards/consensus_reward_func": 1.625, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0625, | |
| "rewards/question_recreation_reward_func": 0.9666663929820061, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.34375, | |
| "rewards/xmlcount_reward_func": 1.1948125064373016, | |
| "step": 6 | |
| }, | |
| { | |
| "completion_length": 149.34375, | |
| "epoch": 0.4, | |
| "grad_norm": 7694.41552734375, | |
| "kl": 136.99966540414607, | |
| "learning_rate": 3.867370395306068e-07, | |
| "loss": 0.137, | |
| "reward": 6.051403075456619, | |
| "reward_std": 0.6295170672237873, | |
| "rewards/concensus_correctness_reward_func": 1.7816874752752483, | |
| "rewards/consensus_reward_func": 1.75, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.95365309715271, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.359375, | |
| "rewards/xmlcount_reward_func": 1.2066875100135803, | |
| "step": 8 | |
| }, | |
| { | |
| "completion_length": 135.21875, | |
| "epoch": 0.5, | |
| "grad_norm": 15.76472282409668, | |
| "kl": 3.0038471170701087, | |
| "learning_rate": 3.1137137178519977e-07, | |
| "loss": 0.003, | |
| "reward": 7.077707350254059, | |
| "reward_std": 0.28928810740762856, | |
| "rewards/concensus_correctness_reward_func": 2.1765625029802322, | |
| "rewards/consensus_reward_func": 1.9375, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.375, | |
| "rewards/question_recreation_reward_func": 0.9071761351078749, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.453125, | |
| "rewards/xmlcount_reward_func": 1.2283437550067902, | |
| "step": 10 | |
| }, | |
| { | |
| "completion_length": 129.0, | |
| "epoch": 0.6, | |
| "grad_norm": 20.300708770751953, | |
| "kl": 1.9722216473892331, | |
| "learning_rate": 2.2935516363191693e-07, | |
| "loss": 0.002, | |
| "reward": 6.014487326145172, | |
| "reward_std": 0.4830540050752461, | |
| "rewards/concensus_correctness_reward_func": 1.6554999873042107, | |
| "rewards/consensus_reward_func": 1.75, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.8902374170720577, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.484375, | |
| "rewards/xmlcount_reward_func": 1.234375, | |
| "step": 12 | |
| }, | |
| { | |
| "completion_length": 149.5, | |
| "epoch": 0.7, | |
| "grad_norm": 24576346.0, | |
| "kl": 367109.22633283085, | |
| "learning_rate": 1.4957614383675767e-07, | |
| "loss": 367.1093, | |
| "reward": 5.701199531555176, | |
| "reward_std": 0.5387236783280969, | |
| "rewards/concensus_correctness_reward_func": 1.585875004529953, | |
| "rewards/consensus_reward_func": 1.625, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.9483870044350624, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.359375, | |
| "rewards/xmlcount_reward_func": 1.1825625076889992, | |
| "step": 14 | |
| }, | |
| { | |
| "completion_length": 138.3125, | |
| "epoch": 0.8, | |
| "grad_norm": 20.45535659790039, | |
| "kl": 1.363151403144002, | |
| "learning_rate": 8.067960709356478e-08, | |
| "loss": 0.0014, | |
| "reward": 6.976692289113998, | |
| "reward_std": 0.3461971321667079, | |
| "rewards/concensus_correctness_reward_func": 2.1019374765455723, | |
| "rewards/consensus_reward_func": 1.9375, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.3125, | |
| "rewards/question_recreation_reward_func": 0.9495673254132271, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.4375, | |
| "rewards/xmlcount_reward_func": 1.23768749833107, | |
| "step": 16 | |
| }, | |
| { | |
| "completion_length": 142.03125, | |
| "epoch": 0.9, | |
| "grad_norm": 172.57008361816406, | |
| "kl": 22.81431655958295, | |
| "learning_rate": 3.013156219837776e-08, | |
| "loss": 0.0228, | |
| "reward": 5.793281584978104, | |
| "reward_std": 0.7580720736295916, | |
| "rewards/concensus_correctness_reward_func": 1.465375017374754, | |
| "rewards/consensus_reward_func": 1.625, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.25, | |
| "rewards/question_recreation_reward_func": 0.9237815849483013, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.359375, | |
| "rewards/xmlcount_reward_func": 1.1697500050067902, | |
| "step": 18 | |
| }, | |
| { | |
| "completion_length": 136.3125, | |
| "epoch": 1.0, | |
| "grad_norm": 24.609342575073242, | |
| "kl": 1.8361739106476307, | |
| "learning_rate": 3.4096741493194193e-09, | |
| "loss": 0.0018, | |
| "reward": 6.313207149505615, | |
| "reward_std": 0.8437363105767872, | |
| "rewards/concensus_correctness_reward_func": 1.7949999794363976, | |
| "rewards/consensus_reward_func": 1.75, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.25, | |
| "rewards/question_recreation_reward_func": 0.9263008404523134, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.375, | |
| "rewards/xmlcount_reward_func": 1.2169062495231628, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 20, | |
| "total_flos": 0.0, | |
| "train_loss": 36.72972322183632, | |
| "train_runtime": 157.1284, | |
| "train_samples_per_second": 2.037, | |
| "train_steps_per_second": 0.127 | |
| } | |
| ], | |
| "logging_steps": 2, | |
| "max_steps": 20, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 25, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |