| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.9881235154394297, | |
| "eval_steps": 500, | |
| "global_step": 312, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06334125098970704, | |
| "grad_norm": 440.0, | |
| "learning_rate": 7.8125e-06, | |
| "log_odds_chosen": 3.1748764514923096, | |
| "log_odds_ratio": -9.884220123291016, | |
| "logps/chosen": -23.121124267578125, | |
| "logps/rejected": -26.29541015625, | |
| "loss": 178.8174, | |
| "nll_loss": 10.681877136230469, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -1.1560561656951904, | |
| "rewards/margins": 0.1587144136428833, | |
| "rewards/rejected": -1.3147705793380737, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.12668250197941408, | |
| "grad_norm": 340.0, | |
| "learning_rate": 1.5625e-05, | |
| "log_odds_chosen": 6.453858852386475, | |
| "log_odds_ratio": -9.807241439819336, | |
| "logps/chosen": -22.413124084472656, | |
| "logps/rejected": -28.864864349365234, | |
| "loss": 178.3701, | |
| "nll_loss": 10.657767295837402, | |
| "rewards/accuracies": 0.578125, | |
| "rewards/chosen": -1.1206562519073486, | |
| "rewards/margins": 0.32258713245391846, | |
| "rewards/rejected": -1.443243384361267, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.19002375296912113, | |
| "grad_norm": 276.0, | |
| "learning_rate": 2.34375e-05, | |
| "log_odds_chosen": 6.200798988342285, | |
| "log_odds_ratio": -6.756936073303223, | |
| "logps/chosen": -19.177919387817383, | |
| "logps/rejected": -25.377355575561523, | |
| "loss": 148.0439, | |
| "nll_loss": 8.914896011352539, | |
| "rewards/accuracies": 0.559374988079071, | |
| "rewards/chosen": -0.9588959813117981, | |
| "rewards/margins": 0.3099718987941742, | |
| "rewards/rejected": -1.26886785030365, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.25336500395882816, | |
| "grad_norm": 900.0, | |
| "learning_rate": 3.125e-05, | |
| "log_odds_chosen": 2.6450843811035156, | |
| "log_odds_ratio": -5.436456203460693, | |
| "logps/chosen": -13.422134399414062, | |
| "logps/rejected": -16.063411712646484, | |
| "loss": 101.9164, | |
| "nll_loss": 6.097954750061035, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": -0.6711066961288452, | |
| "rewards/margins": 0.1320638358592987, | |
| "rewards/rejected": -0.8031705617904663, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.3167062549485352, | |
| "grad_norm": 93.5, | |
| "learning_rate": 3.90625e-05, | |
| "log_odds_chosen": 0.6309150457382202, | |
| "log_odds_ratio": -1.286942481994629, | |
| "logps/chosen": -3.097271680831909, | |
| "logps/rejected": -3.7033658027648926, | |
| "loss": 42.6767, | |
| "nll_loss": 2.6029443740844727, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.15486359596252441, | |
| "rewards/margins": 0.03030467964708805, | |
| "rewards/rejected": -0.18516826629638672, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.38004750593824227, | |
| "grad_norm": 152.0, | |
| "learning_rate": 4.6875e-05, | |
| "log_odds_chosen": 0.14747688174247742, | |
| "log_odds_ratio": -0.863192081451416, | |
| "logps/chosen": -1.753333330154419, | |
| "logps/rejected": -1.8908967971801758, | |
| "loss": 33.7883, | |
| "nll_loss": 2.068608283996582, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.08766667544841766, | |
| "rewards/margins": 0.006878170184791088, | |
| "rewards/rejected": -0.09454484283924103, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.4433887569279493, | |
| "grad_norm": 320.0, | |
| "learning_rate": 4.998583895833834e-05, | |
| "log_odds_chosen": 0.0781320109963417, | |
| "log_odds_ratio": -0.8847883939743042, | |
| "logps/chosen": -1.601119041442871, | |
| "logps/rejected": -1.677050232887268, | |
| "loss": 30.4438, | |
| "nll_loss": 1.8584957122802734, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.08005593717098236, | |
| "rewards/margins": 0.003796567674726248, | |
| "rewards/rejected": -0.08385250717401505, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.5067300079176563, | |
| "grad_norm": 79.5, | |
| "learning_rate": 4.989935734988098e-05, | |
| "log_odds_chosen": 0.17777779698371887, | |
| "log_odds_ratio": -0.7818454504013062, | |
| "logps/chosen": -1.434004545211792, | |
| "logps/rejected": -1.5908782482147217, | |
| "loss": 29.0365, | |
| "nll_loss": 1.7756904363632202, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.07170023769140244, | |
| "rewards/margins": 0.00784367136657238, | |
| "rewards/rejected": -0.07954390347003937, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.5700712589073634, | |
| "grad_norm": 46.0, | |
| "learning_rate": 4.973453315132592e-05, | |
| "log_odds_chosen": 0.12557587027549744, | |
| "log_odds_ratio": -0.7924664616584778, | |
| "logps/chosen": -1.366229772567749, | |
| "logps/rejected": -1.46891450881958, | |
| "loss": 27.2712, | |
| "nll_loss": 1.6648271083831787, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.06831149756908417, | |
| "rewards/margins": 0.00513423141092062, | |
| "rewards/rejected": -0.07344572246074677, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.6334125098970704, | |
| "grad_norm": 322.0, | |
| "learning_rate": 4.9491884960580894e-05, | |
| "log_odds_chosen": 0.2877430319786072, | |
| "log_odds_ratio": -0.712502121925354, | |
| "logps/chosen": -1.2252671718597412, | |
| "logps/rejected": -1.4585528373718262, | |
| "loss": 25.6522, | |
| "nll_loss": 1.5676389932632446, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": -0.06126336008310318, | |
| "rewards/margins": 0.011664283461868763, | |
| "rewards/rejected": -0.07292764633893967, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.6967537608867775, | |
| "grad_norm": 74.0, | |
| "learning_rate": 4.9172176238625164e-05, | |
| "log_odds_chosen": 0.12641319632530212, | |
| "log_odds_ratio": -0.7623997926712036, | |
| "logps/chosen": -1.275046467781067, | |
| "logps/rejected": -1.3824188709259033, | |
| "loss": 25.7811, | |
| "nll_loss": 1.5732007026672363, | |
| "rewards/accuracies": 0.5218750238418579, | |
| "rewards/chosen": -0.06375232338905334, | |
| "rewards/margins": 0.005368619225919247, | |
| "rewards/rejected": -0.06912095099687576, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.7600950118764845, | |
| "grad_norm": 49.25, | |
| "learning_rate": 4.877641290737884e-05, | |
| "log_odds_chosen": 0.19897113740444183, | |
| "log_odds_ratio": -0.7030541300773621, | |
| "logps/chosen": -1.1849174499511719, | |
| "logps/rejected": -1.3446745872497559, | |
| "loss": 25.5335, | |
| "nll_loss": 1.5606908798217773, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": -0.05924587324261665, | |
| "rewards/margins": 0.007987860590219498, | |
| "rewards/rejected": -0.06723372638225555, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.8234362628661916, | |
| "grad_norm": 73.0, | |
| "learning_rate": 4.8305840184700356e-05, | |
| "log_odds_chosen": 0.2491961419582367, | |
| "log_odds_ratio": -0.693637490272522, | |
| "logps/chosen": -1.1569511890411377, | |
| "logps/rejected": -1.3474953174591064, | |
| "loss": 25.4002, | |
| "nll_loss": 1.5528333187103271, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.057847559452056885, | |
| "rewards/margins": 0.009527203626930714, | |
| "rewards/rejected": -0.06737476587295532, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.8867775138558986, | |
| "grad_norm": 62.75, | |
| "learning_rate": 4.7761938666470403e-05, | |
| "log_odds_chosen": 0.1128716841340065, | |
| "log_odds_ratio": -0.76786869764328, | |
| "logps/chosen": -1.2037312984466553, | |
| "logps/rejected": -1.294019103050232, | |
| "loss": 25.6293, | |
| "nll_loss": 1.5634351968765259, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.060186564922332764, | |
| "rewards/margins": 0.004514397121965885, | |
| "rewards/rejected": -0.06470096111297607, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.9501187648456056, | |
| "grad_norm": 43.25, | |
| "learning_rate": 4.71464196680895e-05, | |
| "log_odds_chosen": 0.15465858578681946, | |
| "log_odds_ratio": -0.7334991693496704, | |
| "logps/chosen": -1.1338040828704834, | |
| "logps/rejected": -1.2355806827545166, | |
| "loss": 23.9048, | |
| "nll_loss": 1.457373023033142, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -0.05669020861387253, | |
| "rewards/margins": 0.005088826175779104, | |
| "rewards/rejected": -0.06177903339266777, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.0253365003958828, | |
| "grad_norm": 39.25, | |
| "learning_rate": 4.6461219840046654e-05, | |
| "log_odds_chosen": 0.339913547039032, | |
| "log_odds_ratio": -0.6567232608795166, | |
| "logps/chosen": -1.0351102352142334, | |
| "logps/rejected": -1.276901364326477, | |
| "loss": 26.0349, | |
| "nll_loss": 1.4135476350784302, | |
| "rewards/accuracies": 0.6111111044883728, | |
| "rewards/chosen": -0.05175551399588585, | |
| "rewards/margins": 0.012089554220438004, | |
| "rewards/rejected": -0.06384507566690445, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.0886777513855899, | |
| "grad_norm": 39.5, | |
| "learning_rate": 4.5708495074500414e-05, | |
| "log_odds_chosen": 0.5737000703811646, | |
| "log_odds_ratio": -0.5414241552352905, | |
| "logps/chosen": -0.8744705319404602, | |
| "logps/rejected": -1.2483097314834595, | |
| "loss": 19.751, | |
| "nll_loss": 1.2073694467544556, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.04372352734208107, | |
| "rewards/margins": 0.018691958859562874, | |
| "rewards/rejected": -0.062415480613708496, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.152019002375297, | |
| "grad_norm": 41.25, | |
| "learning_rate": 4.489061372204453e-05, | |
| "log_odds_chosen": 0.6578646302223206, | |
| "log_odds_ratio": -0.5210095047950745, | |
| "logps/chosen": -0.8167413473129272, | |
| "logps/rejected": -1.2592405080795288, | |
| "loss": 19.3149, | |
| "nll_loss": 1.181132197380066, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.04083707183599472, | |
| "rewards/margins": 0.02212495729327202, | |
| "rewards/rejected": -0.06296203285455704, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.2153602533650039, | |
| "grad_norm": 27.0, | |
| "learning_rate": 4.401014914000078e-05, | |
| "log_odds_chosen": 0.5433515310287476, | |
| "log_odds_ratio": -0.5697063207626343, | |
| "logps/chosen": -0.8254748582839966, | |
| "logps/rejected": -1.1539433002471924, | |
| "loss": 19.3225, | |
| "nll_loss": 1.179168701171875, | |
| "rewards/accuracies": 0.690625011920929, | |
| "rewards/chosen": -0.04127374291419983, | |
| "rewards/margins": 0.016423424705863, | |
| "rewards/rejected": -0.05769716575741768, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.278701504354711, | |
| "grad_norm": 79.5, | |
| "learning_rate": 4.306987159568479e-05, | |
| "log_odds_chosen": 0.592140793800354, | |
| "log_odds_ratio": -0.5446491241455078, | |
| "logps/chosen": -0.8890258073806763, | |
| "logps/rejected": -1.2678449153900146, | |
| "loss": 19.6595, | |
| "nll_loss": 1.2014886140823364, | |
| "rewards/accuracies": 0.703125, | |
| "rewards/chosen": -0.044451288878917694, | |
| "rewards/margins": 0.018940959125757217, | |
| "rewards/rejected": -0.06339225172996521, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.342042755344418, | |
| "grad_norm": 26.25, | |
| "learning_rate": 4.207273955012018e-05, | |
| "log_odds_chosen": 0.5168221592903137, | |
| "log_odds_ratio": -0.5695356130599976, | |
| "logps/chosen": -0.8439987897872925, | |
| "logps/rejected": -1.158204197883606, | |
| "loss": 18.8408, | |
| "nll_loss": 1.1490731239318848, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.04219993203878403, | |
| "rewards/margins": 0.01571027934551239, | |
| "rewards/rejected": -0.05791021138429642, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.405384006334125, | |
| "grad_norm": 32.0, | |
| "learning_rate": 4.10218903496256e-05, | |
| "log_odds_chosen": 0.5098962187767029, | |
| "log_odds_ratio": -0.5828264355659485, | |
| "logps/chosen": -0.8689939379692078, | |
| "logps/rejected": -1.1937323808670044, | |
| "loss": 20.1778, | |
| "nll_loss": 1.231968641281128, | |
| "rewards/accuracies": 0.6781250238418579, | |
| "rewards/chosen": -0.04344969242811203, | |
| "rewards/margins": 0.0162369217723608, | |
| "rewards/rejected": -0.05968661978840828, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.468725257323832, | |
| "grad_norm": 34.0, | |
| "learning_rate": 3.992063035456259e-05, | |
| "log_odds_chosen": 0.6211134195327759, | |
| "log_odds_ratio": -0.5309170484542847, | |
| "logps/chosen": -0.7933008074760437, | |
| "logps/rejected": -1.1834070682525635, | |
| "loss": 18.7885, | |
| "nll_loss": 1.14773428440094, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.039665043354034424, | |
| "rewards/margins": 0.01950531266629696, | |
| "rewards/rejected": -0.059170354157686234, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.5320665083135392, | |
| "grad_norm": 28.5, | |
| "learning_rate": 3.8772424536302564e-05, | |
| "log_odds_chosen": 0.6670337915420532, | |
| "log_odds_ratio": -0.5194506049156189, | |
| "logps/chosen": -0.7793987989425659, | |
| "logps/rejected": -1.1679680347442627, | |
| "loss": 18.5035, | |
| "nll_loss": 1.1304987668991089, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": -0.038969945162534714, | |
| "rewards/margins": 0.01942845992743969, | |
| "rewards/rejected": -0.058398403227329254, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.5954077593032463, | |
| "grad_norm": 24.375, | |
| "learning_rate": 3.758088557514501e-05, | |
| "log_odds_chosen": 0.5610660314559937, | |
| "log_odds_ratio": -0.552033543586731, | |
| "logps/chosen": -0.795187771320343, | |
| "logps/rejected": -1.1401742696762085, | |
| "loss": 18.7339, | |
| "nll_loss": 1.1432664394378662, | |
| "rewards/accuracies": 0.684374988079071, | |
| "rewards/chosen": -0.03975939005613327, | |
| "rewards/margins": 0.017249327152967453, | |
| "rewards/rejected": -0.05700872093439102, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.6587490102929534, | |
| "grad_norm": 29.625, | |
| "learning_rate": 3.634976249348867e-05, | |
| "log_odds_chosen": 0.4703447222709656, | |
| "log_odds_ratio": -0.5870831608772278, | |
| "logps/chosen": -0.7940428256988525, | |
| "logps/rejected": -1.0778682231903076, | |
| "loss": 18.7796, | |
| "nll_loss": 1.1443703174591064, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.039702143520116806, | |
| "rewards/margins": 0.014191272668540478, | |
| "rewards/rejected": -0.05389341711997986, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.7220902612826603, | |
| "grad_norm": 28.875, | |
| "learning_rate": 3.508292886002013e-05, | |
| "log_odds_chosen": 0.5749887228012085, | |
| "log_odds_ratio": -0.5368548631668091, | |
| "logps/chosen": -0.8258606195449829, | |
| "logps/rejected": -1.1885448694229126, | |
| "loss": 18.9446, | |
| "nll_loss": 1.1571968793869019, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.04129302501678467, | |
| "rewards/margins": 0.018134210258722305, | |
| "rewards/rejected": -0.05942724272608757, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.7854315122723674, | |
| "grad_norm": 26.75, | |
| "learning_rate": 3.378437060203357e-05, | |
| "log_odds_chosen": 0.5498477220535278, | |
| "log_odds_ratio": -0.5687417984008789, | |
| "logps/chosen": -0.81671142578125, | |
| "logps/rejected": -1.1441940069198608, | |
| "loss": 18.5467, | |
| "nll_loss": 1.130732774734497, | |
| "rewards/accuracies": 0.6781250238418579, | |
| "rewards/chosen": -0.04083556681871414, | |
| "rewards/margins": 0.0163741298019886, | |
| "rewards/rejected": -0.05720970779657364, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.8487727632620743, | |
| "grad_norm": 36.5, | |
| "learning_rate": 3.24581734642289e-05, | |
| "log_odds_chosen": 0.7345950603485107, | |
| "log_odds_ratio": -0.5215600728988647, | |
| "logps/chosen": -0.7965916991233826, | |
| "logps/rejected": -1.2689704895019531, | |
| "loss": 18.424, | |
| "nll_loss": 1.125421404838562, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.03982958942651749, | |
| "rewards/margins": 0.02361893653869629, | |
| "rewards/rejected": -0.06344853341579437, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.9121140142517814, | |
| "grad_norm": 30.5, | |
| "learning_rate": 3.110851015344735e-05, | |
| "log_odds_chosen": 0.5858422517776489, | |
| "log_odds_ratio": -0.5331650376319885, | |
| "logps/chosen": -0.8209154009819031, | |
| "logps/rejected": -1.1717817783355713, | |
| "loss": 18.0192, | |
| "nll_loss": 1.0995434522628784, | |
| "rewards/accuracies": 0.7281249761581421, | |
| "rewards/chosen": -0.04104577749967575, | |
| "rewards/margins": 0.017543310299515724, | |
| "rewards/rejected": -0.058589089661836624, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.9754552652414885, | |
| "grad_norm": 31.5, | |
| "learning_rate": 2.9739627209791965e-05, | |
| "log_odds_chosen": 0.6027151346206665, | |
| "log_odds_ratio": -0.5478641390800476, | |
| "logps/chosen": -0.8092836141586304, | |
| "logps/rejected": -1.1755163669586182, | |
| "loss": 18.8905, | |
| "nll_loss": 1.1532632112503052, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.04046417772769928, | |
| "rewards/margins": 0.0183116365224123, | |
| "rewards/rejected": -0.05877581983804703, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 2.0506730007917655, | |
| "grad_norm": 26.625, | |
| "learning_rate": 2.8355831645441388e-05, | |
| "log_odds_chosen": 0.9976516962051392, | |
| "log_odds_ratio": -0.412594735622406, | |
| "logps/chosen": -0.6453574299812317, | |
| "logps/rejected": -1.2032675743103027, | |
| "loss": 17.6716, | |
| "nll_loss": 0.9611276984214783, | |
| "rewards/accuracies": 0.8333333134651184, | |
| "rewards/chosen": -0.032267872244119644, | |
| "rewards/margins": 0.027895506471395493, | |
| "rewards/rejected": -0.06016337499022484, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.1140142517814726, | |
| "grad_norm": 25.125, | |
| "learning_rate": 2.6961477393196126e-05, | |
| "log_odds_chosen": 1.592689871788025, | |
| "log_odds_ratio": -0.27941542863845825, | |
| "logps/chosen": -0.5471819639205933, | |
| "logps/rejected": -1.4586117267608643, | |
| "loss": 13.9962, | |
| "nll_loss": 0.8607938885688782, | |
| "rewards/accuracies": 0.9281250238418579, | |
| "rewards/chosen": -0.027359094470739365, | |
| "rewards/margins": 0.04557149484753609, | |
| "rewards/rejected": -0.07293058931827545, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 2.1773555027711797, | |
| "grad_norm": 23.125, | |
| "learning_rate": 2.556095160739513e-05, | |
| "log_odds_chosen": 1.3663856983184814, | |
| "log_odds_ratio": -0.3143647313117981, | |
| "logps/chosen": -0.5476871728897095, | |
| "logps/rejected": -1.304485559463501, | |
| "loss": 13.0544, | |
| "nll_loss": 0.800182044506073, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": -0.027384359389543533, | |
| "rewards/margins": 0.037839919328689575, | |
| "rewards/rejected": -0.06522427499294281, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.240696753760887, | |
| "grad_norm": 26.75, | |
| "learning_rate": 2.415866086030516e-05, | |
| "log_odds_chosen": 1.4820417165756226, | |
| "log_odds_ratio": -0.29762935638427734, | |
| "logps/chosen": -0.5468381643295288, | |
| "logps/rejected": -1.3758976459503174, | |
| "loss": 13.3913, | |
| "nll_loss": 0.8220744132995605, | |
| "rewards/accuracies": 0.921875, | |
| "rewards/chosen": -0.02734190784394741, | |
| "rewards/margins": 0.04145296663045883, | |
| "rewards/rejected": -0.06879488378763199, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 2.304038004750594, | |
| "grad_norm": 26.25, | |
| "learning_rate": 2.2759017277414166e-05, | |
| "log_odds_chosen": 1.4541139602661133, | |
| "log_odds_ratio": -0.30560511350631714, | |
| "logps/chosen": -0.5400351285934448, | |
| "logps/rejected": -1.3224079608917236, | |
| "loss": 13.3401, | |
| "nll_loss": 0.818473219871521, | |
| "rewards/accuracies": 0.8968750238418579, | |
| "rewards/chosen": -0.0270017571747303, | |
| "rewards/margins": 0.03911864012479782, | |
| "rewards/rejected": -0.06612040102481842, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.367379255740301, | |
| "grad_norm": 45.0, | |
| "learning_rate": 2.136642465525219e-05, | |
| "log_odds_chosen": 1.3566925525665283, | |
| "log_odds_ratio": -0.312085896730423, | |
| "logps/chosen": -0.5589757561683655, | |
| "logps/rejected": -1.2790095806121826, | |
| "loss": 13.2166, | |
| "nll_loss": 0.8104313015937805, | |
| "rewards/accuracies": 0.8968750238418579, | |
| "rewards/chosen": -0.027948787435889244, | |
| "rewards/margins": 0.03600168973207474, | |
| "rewards/rejected": -0.06395047903060913, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 2.4307205067300077, | |
| "grad_norm": 31.0, | |
| "learning_rate": 1.9985264605418183e-05, | |
| "log_odds_chosen": 1.4590914249420166, | |
| "log_odds_ratio": -0.3126595616340637, | |
| "logps/chosen": -0.5508120059967041, | |
| "logps/rejected": -1.3448957204818726, | |
| "loss": 13.3402, | |
| "nll_loss": 0.8181284666061401, | |
| "rewards/accuracies": 0.890625, | |
| "rewards/chosen": -0.027540603652596474, | |
| "rewards/margins": 0.03970418870449066, | |
| "rewards/rejected": -0.06724478304386139, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.494061757719715, | |
| "grad_norm": 25.375, | |
| "learning_rate": 1.8619882768408935e-05, | |
| "log_odds_chosen": 1.5256664752960205, | |
| "log_odds_ratio": -0.29186081886291504, | |
| "logps/chosen": -0.5439263582229614, | |
| "logps/rejected": -1.4009654521942139, | |
| "loss": 13.2504, | |
| "nll_loss": 0.8135560750961304, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": -0.02719631791114807, | |
| "rewards/margins": 0.04285195842385292, | |
| "rewards/rejected": -0.07004828751087189, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 2.557403008709422, | |
| "grad_norm": 24.25, | |
| "learning_rate": 1.7274575140626318e-05, | |
| "log_odds_chosen": 1.4801175594329834, | |
| "log_odds_ratio": -0.3182724118232727, | |
| "logps/chosen": -0.5556176900863647, | |
| "logps/rejected": -1.355858564376831, | |
| "loss": 13.7318, | |
| "nll_loss": 0.8423222303390503, | |
| "rewards/accuracies": 0.8968750238418579, | |
| "rewards/chosen": -0.027780886739492416, | |
| "rewards/margins": 0.04001203924417496, | |
| "rewards/rejected": -0.06779292970895767, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.620744259699129, | |
| "grad_norm": 21.25, | |
| "learning_rate": 1.59535745575832e-05, | |
| "log_odds_chosen": 1.538774847984314, | |
| "log_odds_ratio": -0.2739660441875458, | |
| "logps/chosen": -0.5107089281082153, | |
| "logps/rejected": -1.3470090627670288, | |
| "loss": 12.7697, | |
| "nll_loss": 0.7844074368476868, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -0.025535445660352707, | |
| "rewards/margins": 0.041815001517534256, | |
| "rewards/rejected": -0.06735045462846756, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 2.684085510688836, | |
| "grad_norm": 26.5, | |
| "learning_rate": 1.466103737583699e-05, | |
| "log_odds_chosen": 1.6077247858047485, | |
| "log_odds_ratio": -0.27898281812667847, | |
| "logps/chosen": -0.5174364447593689, | |
| "logps/rejected": -1.3971433639526367, | |
| "loss": 12.9226, | |
| "nll_loss": 0.7937139272689819, | |
| "rewards/accuracies": 0.909375011920929, | |
| "rewards/chosen": -0.025871824473142624, | |
| "rewards/margins": 0.04398534819483757, | |
| "rewards/rejected": -0.0698571726679802, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 2.7474267616785433, | |
| "grad_norm": 37.5, | |
| "learning_rate": 1.340103039555415e-05, | |
| "log_odds_chosen": 1.4663047790527344, | |
| "log_odds_ratio": -0.3185238242149353, | |
| "logps/chosen": -0.5488572716712952, | |
| "logps/rejected": -1.336905837059021, | |
| "loss": 13.052, | |
| "nll_loss": 0.7998241782188416, | |
| "rewards/accuracies": 0.903124988079071, | |
| "rewards/chosen": -0.02744286321103573, | |
| "rewards/margins": 0.03940243273973465, | |
| "rewards/rejected": -0.06684528291225433, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 2.81076801266825, | |
| "grad_norm": 33.25, | |
| "learning_rate": 1.217751806485235e-05, | |
| "log_odds_chosen": 1.548749327659607, | |
| "log_odds_ratio": -0.2925991415977478, | |
| "logps/chosen": -0.5320712327957153, | |
| "logps/rejected": -1.400617241859436, | |
| "loss": 13.2408, | |
| "nll_loss": 0.8129196166992188, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": -0.026603559032082558, | |
| "rewards/margins": 0.043427299708127975, | |
| "rewards/rejected": -0.07003085315227509, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 2.8741092636579575, | |
| "grad_norm": 28.25, | |
| "learning_rate": 1.0994350006179931e-05, | |
| "log_odds_chosen": 1.4858758449554443, | |
| "log_odds_ratio": -0.29021698236465454, | |
| "logps/chosen": -0.5439127683639526, | |
| "logps/rejected": -1.3659403324127197, | |
| "loss": 13.1169, | |
| "nll_loss": 0.8052948117256165, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": -0.027195638045668602, | |
| "rewards/margins": 0.04110138118267059, | |
| "rewards/rejected": -0.06829701364040375, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 2.937450514647664, | |
| "grad_norm": 23.125, | |
| "learning_rate": 9.855248903979506e-06, | |
| "log_odds_chosen": 1.4533650875091553, | |
| "log_odds_ratio": -0.2951867878437042, | |
| "logps/chosen": -0.5295075178146362, | |
| "logps/rejected": -1.2960364818572998, | |
| "loss": 13.1057, | |
| "nll_loss": 0.8043462038040161, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": -0.026475373655557632, | |
| "rewards/margins": 0.0383264534175396, | |
| "rewards/rejected": -0.06480182707309723, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 3.0126682501979416, | |
| "grad_norm": 29.375, | |
| "learning_rate": 8.763798791745411e-06, | |
| "log_odds_chosen": 1.5862672328948975, | |
| "log_odds_ratio": -0.2758994698524475, | |
| "logps/chosen": -0.5054166316986084, | |
| "logps/rejected": -1.356021761894226, | |
| "loss": 14.3167, | |
| "nll_loss": 0.7815780639648438, | |
| "rewards/accuracies": 0.9222221970558167, | |
| "rewards/chosen": -0.02527083270251751, | |
| "rewards/margins": 0.042530253529548645, | |
| "rewards/rejected": -0.0678010806441307, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 3.0760095011876483, | |
| "grad_norm": 21.25, | |
| "learning_rate": 7.723433775328384e-06, | |
| "log_odds_chosen": 2.0705292224884033, | |
| "log_odds_ratio": -0.19949007034301758, | |
| "logps/chosen": -0.4086482524871826, | |
| "logps/rejected": -1.4631153345108032, | |
| "loss": 10.5377, | |
| "nll_loss": 0.6486285924911499, | |
| "rewards/accuracies": 0.965624988079071, | |
| "rewards/chosen": -0.02043241262435913, | |
| "rewards/margins": 0.05272335559129715, | |
| "rewards/rejected": -0.07315576076507568, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 3.1393507521773554, | |
| "grad_norm": 30.0, | |
| "learning_rate": 6.737427227968063e-06, | |
| "log_odds_chosen": 2.158939838409424, | |
| "log_odds_ratio": -0.17655470967292786, | |
| "logps/chosen": -0.36677083373069763, | |
| "logps/rejected": -1.4107964038848877, | |
| "loss": 10.2002, | |
| "nll_loss": 0.6286830902099609, | |
| "rewards/accuracies": 0.9781249761581421, | |
| "rewards/chosen": -0.01833854243159294, | |
| "rewards/margins": 0.0522012785077095, | |
| "rewards/rejected": -0.07053982466459274, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 3.2026920031670625, | |
| "grad_norm": 24.375, | |
| "learning_rate": 5.808881491049723e-06, | |
| "log_odds_chosen": 2.1467604637145996, | |
| "log_odds_ratio": -0.21470721065998077, | |
| "logps/chosen": -0.45862817764282227, | |
| "logps/rejected": -1.5745422840118408, | |
| "loss": 10.6424, | |
| "nll_loss": 0.6544127464294434, | |
| "rewards/accuracies": 0.953125, | |
| "rewards/chosen": -0.022931408137083054, | |
| "rewards/margins": 0.05579570680856705, | |
| "rewards/rejected": -0.0787271112203598, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 3.2660332541567696, | |
| "grad_norm": 28.375, | |
| "learning_rate": 4.940718112990553e-06, | |
| "log_odds_chosen": 2.224365711212158, | |
| "log_odds_ratio": -0.18038669228553772, | |
| "logps/chosen": -0.38552433252334595, | |
| "logps/rejected": -1.5169192552566528, | |
| "loss": 10.064, | |
| "nll_loss": 0.6199790835380554, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -0.019276214763522148, | |
| "rewards/margins": 0.056569743901491165, | |
| "rewards/rejected": -0.07584595680236816, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 3.3293745051464767, | |
| "grad_norm": 22.875, | |
| "learning_rate": 4.135668656967434e-06, | |
| "log_odds_chosen": 2.1139309406280518, | |
| "log_odds_ratio": -0.18317027390003204, | |
| "logps/chosen": -0.42459002137184143, | |
| "logps/rejected": -1.4972463846206665, | |
| "loss": 10.2835, | |
| "nll_loss": 0.6335573792457581, | |
| "rewards/accuracies": 0.965624988079071, | |
| "rewards/chosen": -0.02122950181365013, | |
| "rewards/margins": 0.053632818162441254, | |
| "rewards/rejected": -0.07486231625080109, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 3.392715756136184, | |
| "grad_norm": 26.5, | |
| "learning_rate": 3.3962661064086354e-06, | |
| "log_odds_chosen": 2.135136842727661, | |
| "log_odds_ratio": -0.18204157054424286, | |
| "logps/chosen": -0.40990591049194336, | |
| "logps/rejected": -1.4916514158248901, | |
| "loss": 10.287, | |
| "nll_loss": 0.6338369250297546, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -0.020495299249887466, | |
| "rewards/margins": 0.05408727377653122, | |
| "rewards/rejected": -0.07458256930112839, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 3.456057007125891, | |
| "grad_norm": 22.625, | |
| "learning_rate": 2.7248368952908053e-06, | |
| "log_odds_chosen": 2.2748658657073975, | |
| "log_odds_ratio": -0.17176181077957153, | |
| "logps/chosen": -0.41875600814819336, | |
| "logps/rejected": -1.62033212184906, | |
| "loss": 10.5217, | |
| "nll_loss": 0.6490176916122437, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.020937800407409668, | |
| "rewards/margins": 0.060078807175159454, | |
| "rewards/rejected": -0.08101661503314972, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 3.519398258115598, | |
| "grad_norm": 21.25, | |
| "learning_rate": 2.1234935883170047e-06, | |
| "log_odds_chosen": 2.311600685119629, | |
| "log_odds_ratio": -0.17423459887504578, | |
| "logps/chosen": -0.3958762288093567, | |
| "logps/rejected": -1.5846991539001465, | |
| "loss": 9.5479, | |
| "nll_loss": 0.5880299806594849, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -0.019793810322880745, | |
| "rewards/margins": 0.05944114923477173, | |
| "rewards/rejected": -0.07923495769500732, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 3.5827395091053047, | |
| "grad_norm": 25.875, | |
| "learning_rate": 1.59412823400657e-06, | |
| "log_odds_chosen": 2.3392279148101807, | |
| "log_odds_ratio": -0.17188401520252228, | |
| "logps/chosen": -0.40492624044418335, | |
| "logps/rejected": -1.6255334615707397, | |
| "loss": 10.1858, | |
| "nll_loss": 0.628017783164978, | |
| "rewards/accuracies": 0.9593750238418579, | |
| "rewards/chosen": -0.020246315747499466, | |
| "rewards/margins": 0.061030369251966476, | |
| "rewards/rejected": -0.08127669245004654, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 3.646080760095012, | |
| "grad_norm": 26.625, | |
| "learning_rate": 1.138406411610482e-06, | |
| "log_odds_chosen": 2.3245720863342285, | |
| "log_odds_ratio": -0.16916075348854065, | |
| "logps/chosen": -0.37623462080955505, | |
| "logps/rejected": -1.5502853393554688, | |
| "loss": 9.6503, | |
| "nll_loss": 0.5946874618530273, | |
| "rewards/accuracies": 0.9468749761581421, | |
| "rewards/chosen": -0.018811732530593872, | |
| "rewards/margins": 0.058702535927295685, | |
| "rewards/rejected": -0.07751427590847015, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 3.709422011084719, | |
| "grad_norm": 21.875, | |
| "learning_rate": 7.577619905828282e-07, | |
| "log_odds_chosen": 2.351839065551758, | |
| "log_odds_ratio": -0.16314415633678436, | |
| "logps/chosen": -0.3871734142303467, | |
| "logps/rejected": -1.6283880472183228, | |
| "loss": 9.8769, | |
| "nll_loss": 0.6091512441635132, | |
| "rewards/accuracies": 0.971875011920929, | |
| "rewards/chosen": -0.019358674064278603, | |
| "rewards/margins": 0.06206073611974716, | |
| "rewards/rejected": -0.08141940087080002, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 3.772763262074426, | |
| "grad_norm": 23.0, | |
| "learning_rate": 4.5339261909704e-07, | |
| "log_odds_chosen": 2.405986785888672, | |
| "log_odds_ratio": -0.16327905654907227, | |
| "logps/chosen": -0.4028860628604889, | |
| "logps/rejected": -1.711242914199829, | |
| "loss": 10.0405, | |
| "nll_loss": 0.6193649172782898, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -0.020144306123256683, | |
| "rewards/margins": 0.06541784107685089, | |
| "rewards/rejected": -0.08556214720010757, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 3.836104513064133, | |
| "grad_norm": 23.375, | |
| "learning_rate": 2.262559558016325e-07, | |
| "log_odds_chosen": 2.3016839027404785, | |
| "log_odds_ratio": -0.16708031296730042, | |
| "logps/chosen": -0.38780677318573, | |
| "logps/rejected": -1.5517771244049072, | |
| "loss": 9.6048, | |
| "nll_loss": 0.5919488072395325, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.01939033903181553, | |
| "rewards/margins": 0.058198511600494385, | |
| "rewards/rejected": -0.07758884876966476, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 3.89944576405384, | |
| "grad_norm": 24.75, | |
| "learning_rate": 7.706665667180091e-08, | |
| "log_odds_chosen": 2.3498542308807373, | |
| "log_odds_ratio": -0.1596686691045761, | |
| "logps/chosen": -0.39143842458724976, | |
| "logps/rejected": -1.6413482427597046, | |
| "loss": 10.6228, | |
| "nll_loss": 0.6559435725212097, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -0.019571922719478607, | |
| "rewards/margins": 0.06249549239873886, | |
| "rewards/rejected": -0.08206741511821747, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 3.9627870150435474, | |
| "grad_norm": 23.625, | |
| "learning_rate": 6.294126437336734e-09, | |
| "log_odds_chosen": 2.3062288761138916, | |
| "log_odds_ratio": -0.17084279656410217, | |
| "logps/chosen": -0.3852413594722748, | |
| "logps/rejected": -1.5706732273101807, | |
| "loss": 10.5082, | |
| "nll_loss": 0.6482213139533997, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -0.019262069836258888, | |
| "rewards/margins": 0.059271592646837234, | |
| "rewards/rejected": -0.07853366434574127, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 3.9881235154394297, | |
| "step": 312, | |
| "total_flos": 0.0, | |
| "train_loss": 25.76104594499637, | |
| "train_runtime": 3640.7426, | |
| "train_samples_per_second": 5.548, | |
| "train_steps_per_second": 0.086 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 312, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 100, | |
| "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": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |