| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 100, | |
| "global_step": 270, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 41.70182769977995, | |
| "learning_rate": 3.703703703703703e-09, | |
| "logits/chosen": -1.5467987060546875, | |
| "logits/rejected": -1.308578372001648, | |
| "logps/chosen": -37.12041473388672, | |
| "logps/rejected": -61.68842315673828, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 37.047557637653945, | |
| "learning_rate": 3.7037037037037036e-08, | |
| "logits/chosen": -0.9453011751174927, | |
| "logits/rejected": -1.405588984489441, | |
| "logps/chosen": -28.646324157714844, | |
| "logps/rejected": -40.27184295654297, | |
| "loss": 0.693, | |
| "rewards/accuracies": 0.4930555522441864, | |
| "rewards/chosen": -0.00016396207502111793, | |
| "rewards/margins": 0.001127531286329031, | |
| "rewards/rejected": -0.0012914934195578098, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 38.71986120422308, | |
| "learning_rate": 7.407407407407407e-08, | |
| "logits/chosen": -1.144275188446045, | |
| "logits/rejected": -1.4153748750686646, | |
| "logps/chosen": -38.298439025878906, | |
| "logps/rejected": -41.69558334350586, | |
| "loss": 0.6927, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.0009220183710567653, | |
| "rewards/margins": 0.0017043445259332657, | |
| "rewards/rejected": -0.0007823262130841613, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 39.765817458266895, | |
| "learning_rate": 9.99623976252115e-08, | |
| "logits/chosen": -0.8312944173812866, | |
| "logits/rejected": -1.4380576610565186, | |
| "logps/chosen": -32.726478576660156, | |
| "logps/rejected": -43.810638427734375, | |
| "loss": 0.6925, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": 0.001944520859979093, | |
| "rewards/margins": 0.0023826442193239927, | |
| "rewards/rejected": -0.00043812341755256057, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 35.39090968218046, | |
| "learning_rate": 9.929548316723981e-08, | |
| "logits/chosen": -1.0166804790496826, | |
| "logits/rejected": -1.4055747985839844, | |
| "logps/chosen": -31.433521270751953, | |
| "logps/rejected": -46.84731674194336, | |
| "loss": 0.6915, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.0015541177708655596, | |
| "rewards/margins": 0.002054566517472267, | |
| "rewards/rejected": -0.0005004488630220294, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 42.118488242592456, | |
| "learning_rate": 9.780577808869398e-08, | |
| "logits/chosen": -1.0024656057357788, | |
| "logits/rejected": -1.3138850927352905, | |
| "logps/chosen": -30.340801239013672, | |
| "logps/rejected": -43.08925247192383, | |
| "loss": 0.6905, | |
| "rewards/accuracies": 0.606249988079071, | |
| "rewards/chosen": 0.0033007909078150988, | |
| "rewards/margins": 0.003635294735431671, | |
| "rewards/rejected": -0.000334503740305081, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 36.92382364235435, | |
| "learning_rate": 9.551814704830733e-08, | |
| "logits/chosen": -1.1217851638793945, | |
| "logits/rejected": -1.4166934490203857, | |
| "logps/chosen": -31.992273330688477, | |
| "logps/rejected": -46.66759490966797, | |
| "loss": 0.6886, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.008065198548138142, | |
| "rewards/margins": 0.010263124480843544, | |
| "rewards/rejected": -0.0021979263983666897, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 39.24669749459966, | |
| "learning_rate": 9.247077288236486e-08, | |
| "logits/chosen": -0.9909595251083374, | |
| "logits/rejected": -1.514561653137207, | |
| "logps/chosen": -31.16408348083496, | |
| "logps/rejected": -57.032684326171875, | |
| "loss": 0.6863, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": 0.009227448143064976, | |
| "rewards/margins": 0.01421318482607603, | |
| "rewards/rejected": -0.004985739476978779, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 38.72478614258279, | |
| "learning_rate": 8.871451929520661e-08, | |
| "logits/chosen": -0.7630345225334167, | |
| "logits/rejected": -1.5050407648086548, | |
| "logps/chosen": -25.78152084350586, | |
| "logps/rejected": -38.437538146972656, | |
| "loss": 0.6839, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 0.013479883782565594, | |
| "rewards/margins": 0.018440086394548416, | |
| "rewards/rejected": -0.004960204008966684, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 35.63155458414573, | |
| "learning_rate": 8.431208189343668e-08, | |
| "logits/chosen": -0.8956977725028992, | |
| "logits/rejected": -1.463505506515503, | |
| "logps/chosen": -30.033924102783203, | |
| "logps/rejected": -56.11266326904297, | |
| "loss": 0.6827, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": 0.013467254117131233, | |
| "rewards/margins": 0.022690553218126297, | |
| "rewards/rejected": -0.00922329816967249, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 37.21064690820017, | |
| "learning_rate": 7.93369417339209e-08, | |
| "logits/chosen": -0.7459134459495544, | |
| "logits/rejected": -1.5061514377593994, | |
| "logps/chosen": -28.494699478149414, | |
| "logps/rejected": -49.45507049560547, | |
| "loss": 0.6794, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 0.015813838690519333, | |
| "rewards/margins": 0.02535676583647728, | |
| "rewards/rejected": -0.009542929008603096, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "eval_logits/chosen": -0.8265538215637207, | |
| "eval_logits/rejected": -1.4756767749786377, | |
| "eval_logps/chosen": -35.586036682128906, | |
| "eval_logps/rejected": -53.076499938964844, | |
| "eval_loss": 0.6905515193939209, | |
| "eval_rewards/accuracies": 0.5899999737739563, | |
| "eval_rewards/chosen": -0.004811165854334831, | |
| "eval_rewards/margins": 0.004816049709916115, | |
| "eval_rewards/rejected": -0.009627215564250946, | |
| "eval_runtime": 65.669, | |
| "eval_samples_per_second": 11.954, | |
| "eval_steps_per_second": 0.381, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 41.489362342224126, | |
| "learning_rate": 7.387213885189744e-08, | |
| "logits/chosen": -0.9582328796386719, | |
| "logits/rejected": -1.4757133722305298, | |
| "logps/chosen": -29.652179718017578, | |
| "logps/rejected": -51.44434356689453, | |
| "loss": 0.6783, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": 0.01538840215653181, | |
| "rewards/margins": 0.030315887182950974, | |
| "rewards/rejected": -0.014927484095096588, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 34.811996422375785, | |
| "learning_rate": 6.800888624023552e-08, | |
| "logits/chosen": -1.1017180681228638, | |
| "logits/rejected": -1.2910267114639282, | |
| "logps/chosen": -34.17676544189453, | |
| "logps/rejected": -47.27839660644531, | |
| "loss": 0.6752, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": 0.019322626292705536, | |
| "rewards/margins": 0.03619503602385521, | |
| "rewards/rejected": -0.01687241531908512, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 36.517002002282, | |
| "learning_rate": 6.184504741390595e-08, | |
| "logits/chosen": -1.0513687133789062, | |
| "logits/rejected": -1.4773876667022705, | |
| "logps/chosen": -30.05930519104004, | |
| "logps/rejected": -49.12663269042969, | |
| "loss": 0.6722, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": 0.017959367483854294, | |
| "rewards/margins": 0.038190923631191254, | |
| "rewards/rejected": -0.02023155428469181, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 45.92150266428131, | |
| "learning_rate": 5.5483502970626584e-08, | |
| "logits/chosen": -1.0855646133422852, | |
| "logits/rejected": -1.4738510847091675, | |
| "logps/chosen": -35.4333610534668, | |
| "logps/rejected": -48.60808563232422, | |
| "loss": 0.6713, | |
| "rewards/accuracies": 0.893750011920929, | |
| "rewards/chosen": 0.01625698432326317, | |
| "rewards/margins": 0.04006917029619217, | |
| "rewards/rejected": -0.023812180384993553, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 52.17233360297021, | |
| "learning_rate": 4.9030433411408785e-08, | |
| "logits/chosen": -1.0967937707901, | |
| "logits/rejected": -1.4282653331756592, | |
| "logps/chosen": -27.93039321899414, | |
| "logps/rejected": -48.556251525878906, | |
| "loss": 0.6666, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": 0.01784222200512886, | |
| "rewards/margins": 0.05219421908259392, | |
| "rewards/rejected": -0.03435199707746506, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 36.1012077969649, | |
| "learning_rate": 4.259354688243757e-08, | |
| "logits/chosen": -1.088585615158081, | |
| "logits/rejected": -1.4759548902511597, | |
| "logps/chosen": -30.474609375, | |
| "logps/rejected": -50.4084358215332, | |
| "loss": 0.668, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": 0.007844483479857445, | |
| "rewards/margins": 0.05492537096142769, | |
| "rewards/rejected": -0.04708089306950569, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 39.4057901333892, | |
| "learning_rate": 3.628028141903493e-08, | |
| "logits/chosen": -1.1803226470947266, | |
| "logits/rejected": -1.41080641746521, | |
| "logps/chosen": -32.83087158203125, | |
| "logps/rejected": -45.69880676269531, | |
| "loss": 0.6612, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": 0.01473814807832241, | |
| "rewards/margins": 0.06375907361507416, | |
| "rewards/rejected": -0.0490209236741066, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 38.966283555722626, | |
| "learning_rate": 3.0196011698042156e-08, | |
| "logits/chosen": -1.0210983753204346, | |
| "logits/rejected": -1.4852098226547241, | |
| "logps/chosen": -29.679798126220703, | |
| "logps/rejected": -50.24894332885742, | |
| "loss": 0.6648, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": 0.006426820065826178, | |
| "rewards/margins": 0.06615011394023895, | |
| "rewards/rejected": -0.059723298996686935, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 40.704072746648535, | |
| "learning_rate": 2.444229022970634e-08, | |
| "logits/chosen": -0.9898138046264648, | |
| "logits/rejected": -1.4493308067321777, | |
| "logps/chosen": -29.71084976196289, | |
| "logps/rejected": -52.1125602722168, | |
| "loss": 0.6587, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": 0.005389463622123003, | |
| "rewards/margins": 0.06682445108890533, | |
| "rewards/rejected": -0.06143498420715332, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 48.9298479981709, | |
| "learning_rate": 1.9115152345327152e-08, | |
| "logits/chosen": -1.0230295658111572, | |
| "logits/rejected": -1.5533816814422607, | |
| "logps/chosen": -30.142093658447266, | |
| "logps/rejected": -58.23160934448242, | |
| "loss": 0.6555, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 0.004340417217463255, | |
| "rewards/margins": 0.08527159690856934, | |
| "rewards/rejected": -0.08093118667602539, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "eval_logits/chosen": -0.8588806390762329, | |
| "eval_logits/rejected": -1.5130012035369873, | |
| "eval_logps/chosen": -39.04316711425781, | |
| "eval_logps/rejected": -58.72098159790039, | |
| "eval_loss": 0.6837000846862793, | |
| "eval_rewards/accuracies": 0.6399999856948853, | |
| "eval_rewards/chosen": -0.0393824502825737, | |
| "eval_rewards/margins": 0.026689546182751656, | |
| "eval_rewards/rejected": -0.0660719946026802, | |
| "eval_runtime": 65.4972, | |
| "eval_samples_per_second": 11.985, | |
| "eval_steps_per_second": 0.382, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 38.64475256798845, | |
| "learning_rate": 1.4303513272105055e-08, | |
| "logits/chosen": -0.9661673307418823, | |
| "logits/rejected": -1.5310702323913574, | |
| "logps/chosen": -31.798084259033203, | |
| "logps/rejected": -55.2061882019043, | |
| "loss": 0.6543, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": 0.008609571494162083, | |
| "rewards/margins": 0.0721936896443367, | |
| "rewards/rejected": -0.06358411908149719, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 46.11579146010557, | |
| "learning_rate": 1.0087684049605349e-08, | |
| "logits/chosen": -1.0623886585235596, | |
| "logits/rejected": -1.484731912612915, | |
| "logps/chosen": -33.05571365356445, | |
| "logps/rejected": -68.4007568359375, | |
| "loss": 0.6512, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": 0.002279428532347083, | |
| "rewards/margins": 0.09988276660442352, | |
| "rewards/rejected": -0.09760333597660065, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 42.366279277913364, | |
| "learning_rate": 6.538031058667609e-09, | |
| "logits/chosen": -0.9526311159133911, | |
| "logits/rejected": -1.5703321695327759, | |
| "logps/chosen": -32.05732727050781, | |
| "logps/rejected": -69.91355895996094, | |
| "loss": 0.6539, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": -0.0006113927811384201, | |
| "rewards/margins": 0.10779329389333725, | |
| "rewards/rejected": -0.10840468108654022, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 42.96686513067178, | |
| "learning_rate": 3.713801536555483e-09, | |
| "logits/chosen": -1.1161487102508545, | |
| "logits/rejected": -1.5149171352386475, | |
| "logps/chosen": -31.100393295288086, | |
| "logps/rejected": -60.296607971191406, | |
| "loss": 0.6558, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.0010345360497012734, | |
| "rewards/margins": 0.09006427228450775, | |
| "rewards/rejected": -0.09109880030155182, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 46.95737595476474, | |
| "learning_rate": 1.662134681666899e-09, | |
| "logits/chosen": -0.9997628927230835, | |
| "logits/rejected": -1.6180452108383179, | |
| "logps/chosen": -27.377477645874023, | |
| "logps/rejected": -54.080360412597656, | |
| "loss": 0.6512, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 0.0026857522316277027, | |
| "rewards/margins": 0.09887233376502991, | |
| "rewards/rejected": -0.0961865708231926, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 42.2077659611999, | |
| "learning_rate": 4.172748534499448e-10, | |
| "logits/chosen": -1.158440113067627, | |
| "logits/rejected": -1.4339531660079956, | |
| "logps/chosen": -32.833595275878906, | |
| "logps/rejected": -55.285552978515625, | |
| "loss": 0.6543, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -0.0024300869554281235, | |
| "rewards/margins": 0.09633214771747589, | |
| "rewards/rejected": -0.09876222908496857, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 44.84738059556724, | |
| "learning_rate": 0.0, | |
| "logits/chosen": -1.0697628259658813, | |
| "logits/rejected": -1.563280463218689, | |
| "logps/chosen": -33.97581481933594, | |
| "logps/rejected": -64.04183197021484, | |
| "loss": 0.6482, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": -0.0003928671940229833, | |
| "rewards/margins": 0.0867936760187149, | |
| "rewards/rejected": -0.0871865525841713, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 270, | |
| "total_flos": 0.0, | |
| "train_loss": 0.1692200431117305, | |
| "train_runtime": 798.9647, | |
| "train_samples_per_second": 21.608, | |
| "train_steps_per_second": 0.338 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 270, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |