| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 8.0, | |
| "eval_steps": 1, | |
| "global_step": 20, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 18.625, | |
| "learning_rate": 7.599999999999999e-06, | |
| "log_odds_chosen": 0.3293280005455017, | |
| "log_odds_ratio": -0.545608401298523, | |
| "logits/chosen": -0.22181883454322815, | |
| "logits/rejected": -0.2948111891746521, | |
| "logps/chosen": -1.9494528770446777, | |
| "logps/rejected": -2.2380290031433105, | |
| "loss": 1.8976, | |
| "nll_loss": 1.8430625200271606, | |
| "rewards/accuracies": 0.984375, | |
| "rewards/chosen": -0.1949452906847, | |
| "rewards/margins": 0.028857626020908356, | |
| "rewards/rejected": -0.22380293905735016, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_log_odds_chosen": 0.3650469183921814, | |
| "eval_log_odds_ratio": -0.5312943458557129, | |
| "eval_logits/chosen": -0.11938808858394623, | |
| "eval_logits/rejected": -0.15210816264152527, | |
| "eval_logps/chosen": -1.7196087837219238, | |
| "eval_logps/rejected": -2.026923179626465, | |
| "eval_loss": 1.6382209062576294, | |
| "eval_nll_loss": 1.585091471672058, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.17196084558963776, | |
| "eval_rewards/margins": 0.030731473118066788, | |
| "eval_rewards/rejected": -0.20269232988357544, | |
| "eval_runtime": 0.9093, | |
| "eval_samples_per_second": 19.795, | |
| "eval_steps_per_second": 9.898, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 8.125, | |
| "learning_rate": 7.2e-06, | |
| "log_odds_chosen": 0.3380122482776642, | |
| "log_odds_ratio": -0.5430496335029602, | |
| "logits/chosen": -0.07896450906991959, | |
| "logits/rejected": -0.11844252794981003, | |
| "logps/chosen": -1.65217125415802, | |
| "logps/rejected": -1.9326075315475464, | |
| "loss": 1.5573, | |
| "nll_loss": 1.5030204057693481, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -0.16521713137626648, | |
| "rewards/margins": 0.02804364264011383, | |
| "rewards/rejected": -0.19326075911521912, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_log_odds_chosen": 0.39069631695747375, | |
| "eval_log_odds_ratio": -0.5206953883171082, | |
| "eval_logits/chosen": -0.10498537868261337, | |
| "eval_logits/rejected": -0.136393204331398, | |
| "eval_logps/chosen": -1.5854017734527588, | |
| "eval_logps/rejected": -1.9058791399002075, | |
| "eval_loss": 1.5333527326583862, | |
| "eval_nll_loss": 1.4812833070755005, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.15854017436504364, | |
| "eval_rewards/margins": 0.032047729939222336, | |
| "eval_rewards/rejected": -0.19058789312839508, | |
| "eval_runtime": 0.9117, | |
| "eval_samples_per_second": 19.744, | |
| "eval_steps_per_second": 9.872, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 7.875, | |
| "learning_rate": 6.799999999999999e-06, | |
| "log_odds_chosen": 0.36618566513061523, | |
| "log_odds_ratio": -0.5316208004951477, | |
| "logits/chosen": -0.045309849083423615, | |
| "logits/rejected": -0.09037788212299347, | |
| "logps/chosen": -1.4718542098999023, | |
| "logps/rejected": -1.7647374868392944, | |
| "loss": 1.4427, | |
| "nll_loss": 1.3894941806793213, | |
| "rewards/accuracies": 0.984375, | |
| "rewards/chosen": -0.14718542993068695, | |
| "rewards/margins": 0.029288342222571373, | |
| "rewards/rejected": -0.17647376656532288, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "eval_log_odds_chosen": 0.41840454936027527, | |
| "eval_log_odds_ratio": -0.5097466111183167, | |
| "eval_logits/chosen": -0.10623180121183395, | |
| "eval_logits/rejected": -0.13603489100933075, | |
| "eval_logps/chosen": -1.4396543502807617, | |
| "eval_logps/rejected": -1.7711676359176636, | |
| "eval_loss": 1.4337514638900757, | |
| "eval_nll_loss": 1.3827767372131348, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.1439654380083084, | |
| "eval_rewards/margins": 0.03315134346485138, | |
| "eval_rewards/rejected": -0.1771167814731598, | |
| "eval_runtime": 0.9101, | |
| "eval_samples_per_second": 19.778, | |
| "eval_steps_per_second": 9.889, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "grad_norm": 7.78125, | |
| "learning_rate": 6.4e-06, | |
| "log_odds_chosen": 0.4042999744415283, | |
| "log_odds_ratio": -0.5167055726051331, | |
| "logits/chosen": -0.0431833378970623, | |
| "logits/rejected": -0.07951641082763672, | |
| "logps/chosen": -1.3549946546554565, | |
| "logps/rejected": -1.6684811115264893, | |
| "loss": 1.3493, | |
| "nll_loss": 1.2975877523422241, | |
| "rewards/accuracies": 0.984375, | |
| "rewards/chosen": -0.13549946248531342, | |
| "rewards/margins": 0.03134865313768387, | |
| "rewards/rejected": -0.1668480932712555, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "eval_log_odds_chosen": 0.44134366512298584, | |
| "eval_log_odds_ratio": -0.5008935332298279, | |
| "eval_logits/chosen": -0.104909747838974, | |
| "eval_logits/rejected": -0.13491111993789673, | |
| "eval_logps/chosen": -1.340553641319275, | |
| "eval_logps/rejected": -1.6810719966888428, | |
| "eval_loss": 1.3429497480392456, | |
| "eval_nll_loss": 1.292860507965088, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.13405534625053406, | |
| "eval_rewards/margins": 0.03405185043811798, | |
| "eval_rewards/rejected": -0.16810721158981323, | |
| "eval_runtime": 0.9119, | |
| "eval_samples_per_second": 19.738, | |
| "eval_steps_per_second": 9.869, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 6.96875, | |
| "learning_rate": 6e-06, | |
| "log_odds_chosen": 0.4279482960700989, | |
| "log_odds_ratio": -0.5070147514343262, | |
| "logits/chosen": -0.034176260232925415, | |
| "logits/rejected": -0.07206660509109497, | |
| "logps/chosen": -1.2959173917770386, | |
| "logps/rejected": -1.6209981441497803, | |
| "loss": 1.2683, | |
| "nll_loss": 1.2175886631011963, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -0.12959173321723938, | |
| "rewards/margins": 0.03250807896256447, | |
| "rewards/rejected": -0.16209980845451355, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_log_odds_chosen": 0.46249422430992126, | |
| "eval_log_odds_ratio": -0.49292659759521484, | |
| "eval_logits/chosen": -0.11007735878229141, | |
| "eval_logits/rejected": -0.14086602628231049, | |
| "eval_logps/chosen": -1.2711644172668457, | |
| "eval_logps/rejected": -1.6204769611358643, | |
| "eval_loss": 1.2642947435379028, | |
| "eval_nll_loss": 1.215002179145813, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.12711645662784576, | |
| "eval_rewards/margins": 0.034931257367134094, | |
| "eval_rewards/rejected": -0.16204769909381866, | |
| "eval_runtime": 0.9131, | |
| "eval_samples_per_second": 19.712, | |
| "eval_steps_per_second": 9.856, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "grad_norm": 6.65625, | |
| "learning_rate": 5.6e-06, | |
| "log_odds_chosen": 0.4413740336894989, | |
| "log_odds_ratio": -0.5025829672813416, | |
| "logits/chosen": -0.026908639818429947, | |
| "logits/rejected": -0.06323603540658951, | |
| "logps/chosen": -1.2035048007965088, | |
| "logps/rejected": -1.5302585363388062, | |
| "loss": 1.1736, | |
| "nll_loss": 1.1233787536621094, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -0.12035048753023148, | |
| "rewards/margins": 0.03267538174986839, | |
| "rewards/rejected": -0.15302586555480957, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "eval_log_odds_chosen": 0.4787115454673767, | |
| "eval_log_odds_ratio": -0.4869447946548462, | |
| "eval_logits/chosen": -0.11934076249599457, | |
| "eval_logits/rejected": -0.15017718076705933, | |
| "eval_logps/chosen": -1.2190965414047241, | |
| "eval_logps/rejected": -1.5743210315704346, | |
| "eval_loss": 1.1919617652893066, | |
| "eval_nll_loss": 1.1432671546936035, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.12190967053174973, | |
| "eval_rewards/margins": 0.03552243858575821, | |
| "eval_rewards/rejected": -0.15743210911750793, | |
| "eval_runtime": 0.9096, | |
| "eval_samples_per_second": 19.79, | |
| "eval_steps_per_second": 9.895, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "grad_norm": 6.96875, | |
| "learning_rate": 5.2e-06, | |
| "log_odds_chosen": 0.46652132272720337, | |
| "log_odds_ratio": -0.4945680797100067, | |
| "logits/chosen": -0.0492391437292099, | |
| "logits/rejected": -0.0838259607553482, | |
| "logps/chosen": -1.1840012073516846, | |
| "logps/rejected": -1.525024652481079, | |
| "loss": 1.1212, | |
| "nll_loss": 1.0717414617538452, | |
| "rewards/accuracies": 0.984375, | |
| "rewards/chosen": -0.11840011179447174, | |
| "rewards/margins": 0.03410235792398453, | |
| "rewards/rejected": -0.15250247716903687, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "eval_log_odds_chosen": 0.4977823495864868, | |
| "eval_log_odds_ratio": -0.4801054000854492, | |
| "eval_logits/chosen": -0.12862297892570496, | |
| "eval_logits/rejected": -0.16006678342819214, | |
| "eval_logps/chosen": -1.1753649711608887, | |
| "eval_logps/rejected": -1.5391558408737183, | |
| "eval_loss": 1.1233677864074707, | |
| "eval_nll_loss": 1.0753573179244995, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.1175365075469017, | |
| "eval_rewards/margins": 0.03637908399105072, | |
| "eval_rewards/rejected": -0.15391558408737183, | |
| "eval_runtime": 0.9109, | |
| "eval_samples_per_second": 19.76, | |
| "eval_steps_per_second": 9.88, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "grad_norm": 6.84375, | |
| "learning_rate": 4.8e-06, | |
| "log_odds_chosen": 0.4919642210006714, | |
| "log_odds_ratio": -0.48277872800827026, | |
| "logits/chosen": -0.05812246352434158, | |
| "logits/rejected": -0.0962948203086853, | |
| "logps/chosen": -1.1183116436004639, | |
| "logps/rejected": -1.4710171222686768, | |
| "loss": 1.0518, | |
| "nll_loss": 1.0035254955291748, | |
| "rewards/accuracies": 0.984375, | |
| "rewards/chosen": -0.11183115839958191, | |
| "rewards/margins": 0.03527054935693741, | |
| "rewards/rejected": -0.14710170030593872, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "eval_log_odds_chosen": 0.5149489045143127, | |
| "eval_log_odds_ratio": -0.47412246465682983, | |
| "eval_logits/chosen": -0.14218762516975403, | |
| "eval_logits/rejected": -0.17368356883525848, | |
| "eval_logps/chosen": -1.1381663084030151, | |
| "eval_logps/rejected": -1.5092874765396118, | |
| "eval_loss": 1.0610299110412598, | |
| "eval_nll_loss": 1.013617753982544, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.11381663382053375, | |
| "eval_rewards/margins": 0.03711211308836937, | |
| "eval_rewards/rejected": -0.15092875063419342, | |
| "eval_runtime": 0.9136, | |
| "eval_samples_per_second": 19.703, | |
| "eval_steps_per_second": 9.852, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "grad_norm": 6.71875, | |
| "learning_rate": 4.4e-06, | |
| "log_odds_chosen": 0.5135414004325867, | |
| "log_odds_ratio": -0.47593823075294495, | |
| "logits/chosen": -0.06834974884986877, | |
| "logits/rejected": -0.11046632379293442, | |
| "logps/chosen": -1.0532824993133545, | |
| "logps/rejected": -1.4117248058319092, | |
| "loss": 0.9805, | |
| "nll_loss": 0.9329336285591125, | |
| "rewards/accuracies": 0.984375, | |
| "rewards/chosen": -0.1053282618522644, | |
| "rewards/margins": 0.035844214260578156, | |
| "rewards/rejected": -0.14117246866226196, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "eval_log_odds_chosen": 0.5320238471031189, | |
| "eval_log_odds_ratio": -0.4681590497493744, | |
| "eval_logits/chosen": -0.1644686907529831, | |
| "eval_logits/rejected": -0.19686466455459595, | |
| "eval_logps/chosen": -1.104932427406311, | |
| "eval_logps/rejected": -1.4836636781692505, | |
| "eval_loss": 1.0012433528900146, | |
| "eval_nll_loss": 0.9544275999069214, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.1104932650923729, | |
| "eval_rewards/margins": 0.03787311539053917, | |
| "eval_rewards/rejected": -0.14836637675762177, | |
| "eval_runtime": 0.9099, | |
| "eval_samples_per_second": 19.783, | |
| "eval_steps_per_second": 9.892, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 6.75, | |
| "learning_rate": 4e-06, | |
| "log_odds_chosen": 0.5470705032348633, | |
| "log_odds_ratio": -0.4658868908882141, | |
| "logits/chosen": -0.09963471442461014, | |
| "logits/rejected": -0.13952209055423737, | |
| "logps/chosen": -1.0679322481155396, | |
| "logps/rejected": -1.4557496309280396, | |
| "loss": 0.9299, | |
| "nll_loss": 0.8832955956459045, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.10679321736097336, | |
| "rewards/margins": 0.038781747221946716, | |
| "rewards/rejected": -0.14557495713233948, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_log_odds_chosen": 0.547681450843811, | |
| "eval_log_odds_ratio": -0.46276018023490906, | |
| "eval_logits/chosen": -0.1875133067369461, | |
| "eval_logits/rejected": -0.22008682787418365, | |
| "eval_logps/chosen": -1.07937490940094, | |
| "eval_logps/rejected": -1.4652737379074097, | |
| "eval_loss": 0.9495540261268616, | |
| "eval_nll_loss": 0.9032779335975647, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.10793750733137131, | |
| "eval_rewards/margins": 0.03858988359570503, | |
| "eval_rewards/rejected": -0.14652739465236664, | |
| "eval_runtime": 0.9118, | |
| "eval_samples_per_second": 19.742, | |
| "eval_steps_per_second": 9.871, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "grad_norm": 6.5625, | |
| "learning_rate": 3.6e-06, | |
| "log_odds_chosen": 0.5115205645561218, | |
| "log_odds_ratio": -0.4781632423400879, | |
| "logits/chosen": -0.11988667398691177, | |
| "logits/rejected": -0.15675179660320282, | |
| "logps/chosen": -1.0269582271575928, | |
| "logps/rejected": -1.3790360689163208, | |
| "loss": 0.8761, | |
| "nll_loss": 0.8282526135444641, | |
| "rewards/accuracies": 0.984375, | |
| "rewards/chosen": -0.10269583761692047, | |
| "rewards/margins": 0.03520777449011803, | |
| "rewards/rejected": -0.1379036009311676, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "eval_log_odds_chosen": 0.560771107673645, | |
| "eval_log_odds_ratio": -0.45837968587875366, | |
| "eval_logits/chosen": -0.21046772599220276, | |
| "eval_logits/rejected": -0.2431277632713318, | |
| "eval_logps/chosen": -1.0591222047805786, | |
| "eval_logps/rejected": -1.4509668350219727, | |
| "eval_loss": 0.9070050120353699, | |
| "eval_nll_loss": 0.8611669540405273, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.10591220110654831, | |
| "eval_rewards/margins": 0.03918447345495224, | |
| "eval_rewards/rejected": -0.14509668946266174, | |
| "eval_runtime": 0.9119, | |
| "eval_samples_per_second": 19.739, | |
| "eval_steps_per_second": 9.869, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "grad_norm": 6.5625, | |
| "learning_rate": 3.2e-06, | |
| "log_odds_chosen": 0.6133227944374084, | |
| "log_odds_ratio": -0.4412252902984619, | |
| "logits/chosen": -0.14904728531837463, | |
| "logits/rejected": -0.1963028907775879, | |
| "logps/chosen": -0.9898001551628113, | |
| "logps/rejected": -1.4105660915374756, | |
| "loss": 0.8337, | |
| "nll_loss": 0.7895629405975342, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.09898000955581665, | |
| "rewards/margins": 0.04207659140229225, | |
| "rewards/rejected": -0.1410566121339798, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "eval_log_odds_chosen": 0.566861093044281, | |
| "eval_log_odds_ratio": -0.45656245946884155, | |
| "eval_logits/chosen": -0.22315236926078796, | |
| "eval_logits/rejected": -0.2561546862125397, | |
| "eval_logps/chosen": -1.0492280721664429, | |
| "eval_logps/rejected": -1.443569302558899, | |
| "eval_loss": 0.8863641023635864, | |
| "eval_nll_loss": 0.8407078981399536, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.1049228087067604, | |
| "eval_rewards/margins": 0.03943413123488426, | |
| "eval_rewards/rejected": -0.14435693621635437, | |
| "eval_runtime": 0.9164, | |
| "eval_samples_per_second": 19.642, | |
| "eval_steps_per_second": 9.821, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "grad_norm": 6.5, | |
| "learning_rate": 2.8e-06, | |
| "log_odds_chosen": 0.49919161200523376, | |
| "log_odds_ratio": -0.4824044108390808, | |
| "logits/chosen": -0.15154853463172913, | |
| "logits/rejected": -0.18894340097904205, | |
| "logps/chosen": -1.0274688005447388, | |
| "logps/rejected": -1.3725156784057617, | |
| "loss": 0.7975, | |
| "nll_loss": 0.7492961883544922, | |
| "rewards/accuracies": 0.953125, | |
| "rewards/chosen": -0.1027468740940094, | |
| "rewards/margins": 0.034504685550928116, | |
| "rewards/rejected": -0.13725155591964722, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "eval_log_odds_chosen": 0.57403564453125, | |
| "eval_log_odds_ratio": -0.45407184958457947, | |
| "eval_logits/chosen": -0.23968791961669922, | |
| "eval_logits/rejected": -0.2726818025112152, | |
| "eval_logps/chosen": -1.0405869483947754, | |
| "eval_logps/rejected": -1.4385521411895752, | |
| "eval_loss": 0.8664105534553528, | |
| "eval_nll_loss": 0.8210033774375916, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.10405868291854858, | |
| "eval_rewards/margins": 0.03979651629924774, | |
| "eval_rewards/rejected": -0.14385519921779633, | |
| "eval_runtime": 0.9125, | |
| "eval_samples_per_second": 19.726, | |
| "eval_steps_per_second": 9.863, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "grad_norm": 6.40625, | |
| "learning_rate": 2.4e-06, | |
| "log_odds_chosen": 0.6314184665679932, | |
| "log_odds_ratio": -0.4354270100593567, | |
| "logits/chosen": -0.1857105791568756, | |
| "logits/rejected": -0.22656874358654022, | |
| "logps/chosen": -0.9448862075805664, | |
| "logps/rejected": -1.3670051097869873, | |
| "loss": 0.788, | |
| "nll_loss": 0.7444556951522827, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.09448862820863724, | |
| "rewards/margins": 0.04221189767122269, | |
| "rewards/rejected": -0.13670052587985992, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "eval_log_odds_chosen": 0.5794126987457275, | |
| "eval_log_odds_ratio": -0.4523561894893646, | |
| "eval_logits/chosen": -0.2507224678993225, | |
| "eval_logits/rejected": -0.2837482690811157, | |
| "eval_logps/chosen": -1.0325850248336792, | |
| "eval_logps/rejected": -1.4329302310943604, | |
| "eval_loss": 0.8492475152015686, | |
| "eval_nll_loss": 0.8040118217468262, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.10325851291418076, | |
| "eval_rewards/margins": 0.04003452509641647, | |
| "eval_rewards/rejected": -0.14329302310943604, | |
| "eval_runtime": 0.9113, | |
| "eval_samples_per_second": 19.752, | |
| "eval_steps_per_second": 9.876, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "grad_norm": 6.25, | |
| "learning_rate": 2e-06, | |
| "log_odds_chosen": 0.5837588906288147, | |
| "log_odds_ratio": -0.45432496070861816, | |
| "logits/chosen": -0.18676218390464783, | |
| "logits/rejected": -0.23436766862869263, | |
| "logps/chosen": -1.0190256834030151, | |
| "logps/rejected": -1.4252986907958984, | |
| "loss": 0.78, | |
| "nll_loss": 0.7345477342605591, | |
| "rewards/accuracies": 0.984375, | |
| "rewards/chosen": -0.1019025593996048, | |
| "rewards/margins": 0.040627315640449524, | |
| "rewards/rejected": -0.14252987504005432, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_log_odds_chosen": 0.5839331150054932, | |
| "eval_log_odds_ratio": -0.45080792903900146, | |
| "eval_logits/chosen": -0.26143890619277954, | |
| "eval_logits/rejected": -0.29441285133361816, | |
| "eval_logps/chosen": -1.0263959169387817, | |
| "eval_logps/rejected": -1.4286550283432007, | |
| "eval_loss": 0.8333800435066223, | |
| "eval_nll_loss": 0.7882992625236511, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.10263960063457489, | |
| "eval_rewards/margins": 0.040225885808467865, | |
| "eval_rewards/rejected": -0.14286547899246216, | |
| "eval_runtime": 0.9128, | |
| "eval_samples_per_second": 19.72, | |
| "eval_steps_per_second": 9.86, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 6.4, | |
| "grad_norm": 6.25, | |
| "learning_rate": 1.6e-06, | |
| "log_odds_chosen": 0.6216727495193481, | |
| "log_odds_ratio": -0.4399999976158142, | |
| "logits/chosen": -0.19661211967468262, | |
| "logits/rejected": -0.24193710088729858, | |
| "logps/chosen": -0.949080228805542, | |
| "logps/rejected": -1.3682817220687866, | |
| "loss": 0.7395, | |
| "nll_loss": 0.6955283880233765, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.09490802884101868, | |
| "rewards/margins": 0.04192016273736954, | |
| "rewards/rejected": -0.13682818412780762, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 6.4, | |
| "eval_log_odds_chosen": 0.5865243673324585, | |
| "eval_log_odds_ratio": -0.44995343685150146, | |
| "eval_logits/chosen": -0.2722480595111847, | |
| "eval_logits/rejected": -0.30540305376052856, | |
| "eval_logps/chosen": -1.0213567018508911, | |
| "eval_logps/rejected": -1.4244111776351929, | |
| "eval_loss": 0.821089506149292, | |
| "eval_nll_loss": 0.7760941386222839, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.10213566571474075, | |
| "eval_rewards/margins": 0.040305450558662415, | |
| "eval_rewards/rejected": -0.14244110882282257, | |
| "eval_runtime": 0.9109, | |
| "eval_samples_per_second": 19.76, | |
| "eval_steps_per_second": 9.88, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 6.8, | |
| "grad_norm": 6.15625, | |
| "learning_rate": 1.2e-06, | |
| "log_odds_chosen": 0.5441011786460876, | |
| "log_odds_ratio": -0.4661071002483368, | |
| "logits/chosen": -0.20644214749336243, | |
| "logits/rejected": -0.2489599585533142, | |
| "logps/chosen": -1.002626657485962, | |
| "logps/rejected": -1.3740018606185913, | |
| "loss": 0.7446, | |
| "nll_loss": 0.6980130076408386, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -0.10026266425848007, | |
| "rewards/margins": 0.037137530744075775, | |
| "rewards/rejected": -0.13740019500255585, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 6.8, | |
| "eval_log_odds_chosen": 0.5888115763664246, | |
| "eval_log_odds_ratio": -0.44922754168510437, | |
| "eval_logits/chosen": -0.2722025513648987, | |
| "eval_logits/rejected": -0.30542224645614624, | |
| "eval_logps/chosen": -1.0186693668365479, | |
| "eval_logps/rejected": -1.4229329824447632, | |
| "eval_loss": 0.8164036870002747, | |
| "eval_nll_loss": 0.7714808583259583, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.10186693072319031, | |
| "eval_rewards/margins": 0.040426358580589294, | |
| "eval_rewards/rejected": -0.1422932893037796, | |
| "eval_runtime": 0.9099, | |
| "eval_samples_per_second": 19.782, | |
| "eval_steps_per_second": 9.891, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "grad_norm": 6.125, | |
| "learning_rate": 8e-07, | |
| "log_odds_chosen": 0.603852391242981, | |
| "log_odds_ratio": -0.44559258222579956, | |
| "logits/chosen": -0.21281519532203674, | |
| "logits/rejected": -0.2561994791030884, | |
| "logps/chosen": -0.9362643957138062, | |
| "logps/rejected": -1.3377454280853271, | |
| "loss": 0.7518, | |
| "nll_loss": 0.7072104811668396, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.09362644702196121, | |
| "rewards/margins": 0.040148116648197174, | |
| "rewards/rejected": -0.1337745636701584, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "eval_log_odds_chosen": 0.5903448462486267, | |
| "eval_log_odds_ratio": -0.4487246870994568, | |
| "eval_logits/chosen": -0.2774750292301178, | |
| "eval_logits/rejected": -0.3105829358100891, | |
| "eval_logps/chosen": -1.0175344944000244, | |
| "eval_logps/rejected": -1.4226274490356445, | |
| "eval_loss": 0.812524139881134, | |
| "eval_nll_loss": 0.7676517963409424, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.10175344347953796, | |
| "eval_rewards/margins": 0.040509287267923355, | |
| "eval_rewards/rejected": -0.1422627568244934, | |
| "eval_runtime": 0.9145, | |
| "eval_samples_per_second": 19.683, | |
| "eval_steps_per_second": 9.841, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "grad_norm": 6.21875, | |
| "learning_rate": 4e-07, | |
| "log_odds_chosen": 0.6179953217506409, | |
| "log_odds_ratio": -0.4413047730922699, | |
| "logits/chosen": -0.22949087619781494, | |
| "logits/rejected": -0.27671945095062256, | |
| "logps/chosen": -0.9877333045005798, | |
| "logps/rejected": -1.4103630781173706, | |
| "loss": 0.7431, | |
| "nll_loss": 0.6990159749984741, | |
| "rewards/accuracies": 0.984375, | |
| "rewards/chosen": -0.09877333790063858, | |
| "rewards/margins": 0.0422629676759243, | |
| "rewards/rejected": -0.14103631675243378, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "eval_log_odds_chosen": 0.5911502838134766, | |
| "eval_log_odds_ratio": -0.44839149713516235, | |
| "eval_logits/chosen": -0.276792049407959, | |
| "eval_logits/rejected": -0.31038832664489746, | |
| "eval_logps/chosen": -1.0161813497543335, | |
| "eval_logps/rejected": -1.4216675758361816, | |
| "eval_loss": 0.8106683492660522, | |
| "eval_nll_loss": 0.7658291459083557, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.10161812603473663, | |
| "eval_rewards/margins": 0.04054862633347511, | |
| "eval_rewards/rejected": -0.14216677844524384, | |
| "eval_runtime": 0.9137, | |
| "eval_samples_per_second": 19.701, | |
| "eval_steps_per_second": 9.85, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "grad_norm": 6.15625, | |
| "learning_rate": 0.0, | |
| "log_odds_chosen": 0.578275740146637, | |
| "log_odds_ratio": -0.45558279752731323, | |
| "logits/chosen": -0.210488423705101, | |
| "logits/rejected": -0.24888469278812408, | |
| "logps/chosen": -0.9616943001747131, | |
| "logps/rejected": -1.3508220911026, | |
| "loss": 0.726, | |
| "nll_loss": 0.6804530620574951, | |
| "rewards/accuracies": 0.984375, | |
| "rewards/chosen": -0.09616944193840027, | |
| "rewards/margins": 0.038912780582904816, | |
| "rewards/rejected": -0.1350822150707245, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_log_odds_chosen": 0.5904660224914551, | |
| "eval_log_odds_ratio": -0.44868627190589905, | |
| "eval_logits/chosen": -0.27975308895111084, | |
| "eval_logits/rejected": -0.3132215440273285, | |
| "eval_logps/chosen": -1.0163097381591797, | |
| "eval_logps/rejected": -1.4213618040084839, | |
| "eval_loss": 0.8110137581825256, | |
| "eval_nll_loss": 0.7661450505256653, | |
| "eval_rewards/accuracies": 1.0, | |
| "eval_rewards/chosen": -0.10163097083568573, | |
| "eval_rewards/margins": 0.04050520807504654, | |
| "eval_rewards/rejected": -0.14213618636131287, | |
| "eval_runtime": 0.9114, | |
| "eval_samples_per_second": 19.75, | |
| "eval_steps_per_second": 9.875, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "step": 20, | |
| "total_flos": 0.0, | |
| "train_loss": 1.0276277005672454, | |
| "train_runtime": 261.791, | |
| "train_samples_per_second": 6.074, | |
| "train_steps_per_second": 0.076 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 20, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "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 | |
| } | |