{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9991111111111111, "eval_steps": 100, "global_step": 562, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0017777777777777779, "grad_norm": 0.34293448639816615, "learning_rate": 8.771929824561403e-09, "logits/chosen": 13.316938400268555, "logits/rejected": 13.604828834533691, "logps/chosen": -366.0712890625, "logps/rejected": -384.89288330078125, "loss": 0.6931, "rewards/accuracies": 0.0, "rewards/chosen": 0.0, "rewards/margins": 0.0, "rewards/rejected": 0.0, "step": 1 }, { "epoch": 0.017777777777777778, "grad_norm": 0.38494137076048135, "learning_rate": 8.771929824561403e-08, "logits/chosen": 14.320709228515625, "logits/rejected": 12.689621925354004, "logps/chosen": -420.3371887207031, "logps/rejected": -431.1959228515625, "loss": 0.6929, "rewards/accuracies": 0.3194444477558136, "rewards/chosen": -0.0003511953691486269, "rewards/margins": 0.00028524029767140746, "rewards/rejected": -0.0006364354630932212, "step": 10 }, { "epoch": 0.035555555555555556, "grad_norm": 0.37882743094316385, "learning_rate": 1.7543859649122805e-07, "logits/chosen": 13.643712997436523, "logits/rejected": 11.55290412902832, "logps/chosen": -413.57061767578125, "logps/rejected": -409.67694091796875, "loss": 0.6937, "rewards/accuracies": 0.4437499940395355, "rewards/chosen": -0.0014681232860311866, "rewards/margins": -0.0013028818648308516, "rewards/rejected": -0.00016524126112926751, "step": 20 }, { "epoch": 0.05333333333333334, "grad_norm": 0.3955514954356053, "learning_rate": 2.631578947368421e-07, "logits/chosen": 13.529040336608887, "logits/rejected": 12.290833473205566, "logps/chosen": -371.78033447265625, "logps/rejected": -382.180419921875, "loss": 0.6925, "rewards/accuracies": 0.612500011920929, "rewards/chosen": -0.00034525356022641063, "rewards/margins": 0.004034269601106644, "rewards/rejected": -0.0043795229867100716, "step": 30 }, { "epoch": 0.07111111111111111, "grad_norm": 0.3113111173035772, "learning_rate": 3.508771929824561e-07, "logits/chosen": 13.795173645019531, "logits/rejected": 12.55676555633545, "logps/chosen": -389.0183410644531, "logps/rejected": -415.35247802734375, "loss": 0.6938, "rewards/accuracies": 0.5, "rewards/chosen": -0.003066176315769553, "rewards/margins": -0.0013317791745066643, "rewards/rejected": -0.0017343973740935326, "step": 40 }, { "epoch": 0.08888888888888889, "grad_norm": 0.32606516023193766, "learning_rate": 4.3859649122807013e-07, "logits/chosen": 14.106832504272461, "logits/rejected": 12.1203031539917, "logps/chosen": -422.3417053222656, "logps/rejected": -397.51385498046875, "loss": 0.6937, "rewards/accuracies": 0.4625000059604645, "rewards/chosen": -0.00344509887509048, "rewards/margins": -0.0021442542783915997, "rewards/rejected": -0.001300844713114202, "step": 50 }, { "epoch": 0.10666666666666667, "grad_norm": 0.33670626880499194, "learning_rate": 4.999564631597801e-07, "logits/chosen": 13.741307258605957, "logits/rejected": 11.687819480895996, "logps/chosen": -413.763427734375, "logps/rejected": -397.7774353027344, "loss": 0.6916, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -0.0033744447864592075, "rewards/margins": 0.003910833969712257, "rewards/rejected": -0.007285278290510178, "step": 60 }, { "epoch": 0.12444444444444444, "grad_norm": 0.34616943203548983, "learning_rate": 4.991828966534002e-07, "logits/chosen": 14.954236030578613, "logits/rejected": 12.991679191589355, "logps/chosen": -437.3307189941406, "logps/rejected": -431.5257263183594, "loss": 0.6914, "rewards/accuracies": 0.550000011920929, "rewards/chosen": -0.0038897425401955843, "rewards/margins": 0.0035481129307299852, "rewards/rejected": -0.007437856402248144, "step": 70 }, { "epoch": 0.14222222222222222, "grad_norm": 0.4091997164465182, "learning_rate": 4.974452899279291e-07, "logits/chosen": 12.901407241821289, "logits/rejected": 11.605668067932129, "logps/chosen": -391.2544250488281, "logps/rejected": -403.60089111328125, "loss": 0.692, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": -0.009723111987113953, "rewards/margins": 0.000866956717800349, "rewards/rejected": -0.010590068995952606, "step": 80 }, { "epoch": 0.16, "grad_norm": 0.41253119841994956, "learning_rate": 4.947503654462276e-07, "logits/chosen": 13.798151016235352, "logits/rejected": 12.685941696166992, "logps/chosen": -403.194580078125, "logps/rejected": -421.2926330566406, "loss": 0.6888, "rewards/accuracies": 0.5062500238418579, "rewards/chosen": -0.012594172731041908, "rewards/margins": 0.004347759298980236, "rewards/rejected": -0.01694193109869957, "step": 90 }, { "epoch": 0.17777777777777778, "grad_norm": 0.3696840111639605, "learning_rate": 4.911085493475802e-07, "logits/chosen": 14.222944259643555, "logits/rejected": 12.481651306152344, "logps/chosen": -427.9515075683594, "logps/rejected": -426.4696350097656, "loss": 0.6885, "rewards/accuracies": 0.606249988079071, "rewards/chosen": -0.01204882562160492, "rewards/margins": 0.010180080309510231, "rewards/rejected": -0.02222890593111515, "step": 100 }, { "epoch": 0.17777777777777778, "eval_logits/chosen": 14.262085914611816, "eval_logits/rejected": 12.830548286437988, "eval_logps/chosen": -407.780517578125, "eval_logps/rejected": -402.74957275390625, "eval_loss": 0.6883670687675476, "eval_rewards/accuracies": 0.6190476417541504, "eval_rewards/chosen": -0.015755515545606613, "eval_rewards/margins": 0.008608067408204079, "eval_rewards/rejected": -0.024363582953810692, "eval_runtime": 91.0668, "eval_samples_per_second": 10.981, "eval_steps_per_second": 0.692, "step": 100 }, { "epoch": 0.19555555555555557, "grad_norm": 0.387876184968801, "learning_rate": 4.865339311109869e-07, "logits/chosen": 13.772363662719727, "logits/rejected": 12.22777271270752, "logps/chosen": -405.01214599609375, "logps/rejected": -403.9130859375, "loss": 0.688, "rewards/accuracies": 0.6875, "rewards/chosen": -0.0224502794444561, "rewards/margins": 0.012338267639279366, "rewards/rejected": -0.03478854522109032, "step": 110 }, { "epoch": 0.21333333333333335, "grad_norm": 0.3899509909292818, "learning_rate": 4.810442090457072e-07, "logits/chosen": 13.653238296508789, "logits/rejected": 12.279303550720215, "logps/chosen": -402.15704345703125, "logps/rejected": -404.1790466308594, "loss": 0.6866, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -0.024786310270428658, "rewards/margins": 0.014094953425228596, "rewards/rejected": -0.03888126462697983, "step": 120 }, { "epoch": 0.2311111111111111, "grad_norm": 0.3557450823856152, "learning_rate": 4.746606218199385e-07, "logits/chosen": 14.775594711303711, "logits/rejected": 12.874218940734863, "logps/chosen": -390.8421630859375, "logps/rejected": -400.7987365722656, "loss": 0.6854, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -0.03051091358065605, "rewards/margins": 0.017188305035233498, "rewards/rejected": -0.0476992204785347, "step": 130 }, { "epoch": 0.24888888888888888, "grad_norm": 0.36125778338353265, "learning_rate": 4.674078662925359e-07, "logits/chosen": 13.882017135620117, "logits/rejected": 12.225305557250977, "logps/chosen": -420.66033935546875, "logps/rejected": -413.9700622558594, "loss": 0.6837, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -0.040414221584796906, "rewards/margins": 0.021218815818428993, "rewards/rejected": -0.06163303926587105, "step": 140 }, { "epoch": 0.26666666666666666, "grad_norm": 0.42414032297199183, "learning_rate": 4.593140019656625e-07, "logits/chosen": 13.458653450012207, "logits/rejected": 11.939371109008789, "logps/chosen": -395.70318603515625, "logps/rejected": -426.5689392089844, "loss": 0.6815, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -0.04780956357717514, "rewards/margins": 0.022214924916625023, "rewards/rejected": -0.07002449035644531, "step": 150 }, { "epoch": 0.28444444444444444, "grad_norm": 0.31038758853221154, "learning_rate": 4.504103424280266e-07, "logits/chosen": 14.602518081665039, "logits/rejected": 12.48104476928711, "logps/chosen": -445.41900634765625, "logps/rejected": -427.4073791503906, "loss": 0.6785, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -0.04993806034326553, "rewards/margins": 0.03340305760502815, "rewards/rejected": -0.08334111422300339, "step": 160 }, { "epoch": 0.3022222222222222, "grad_norm": 0.5166732651791626, "learning_rate": 4.407313342086905e-07, "logits/chosen": 14.151707649230957, "logits/rejected": 12.189191818237305, "logps/chosen": -463.04193115234375, "logps/rejected": -419.1507263183594, "loss": 0.6757, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -0.06344902515411377, "rewards/margins": 0.031504396349191666, "rewards/rejected": -0.09495342522859573, "step": 170 }, { "epoch": 0.32, "grad_norm": 0.351499350492249, "learning_rate": 4.3031442351014115e-07, "logits/chosen": 13.737968444824219, "logits/rejected": 12.326078414916992, "logps/chosen": -399.82208251953125, "logps/rejected": -415.14105224609375, "loss": 0.6754, "rewards/accuracies": 0.75, "rewards/chosen": -0.08228292316198349, "rewards/margins": 0.03530151769518852, "rewards/rejected": -0.11758442968130112, "step": 180 }, { "epoch": 0.3377777777777778, "grad_norm": 0.4342072416786669, "learning_rate": 4.19199911336207e-07, "logits/chosen": 13.775731086730957, "logits/rejected": 11.498677253723145, "logps/chosen": -436.33734130859375, "logps/rejected": -418.3775939941406, "loss": 0.6717, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -0.08298386633396149, "rewards/margins": 0.04400986433029175, "rewards/rejected": -0.12699371576309204, "step": 190 }, { "epoch": 0.35555555555555557, "grad_norm": 0.4572664943304227, "learning_rate": 4.0743079757530443e-07, "logits/chosen": 14.21821403503418, "logits/rejected": 12.080196380615234, "logps/chosen": -428.86083984375, "logps/rejected": -398.98223876953125, "loss": 0.6712, "rewards/accuracies": 0.75, "rewards/chosen": -0.08896859735250473, "rewards/margins": 0.042290668934583664, "rewards/rejected": -0.1312592625617981, "step": 200 }, { "epoch": 0.35555555555555557, "eval_logits/chosen": 14.084491729736328, "eval_logits/rejected": 12.648215293884277, "eval_logps/chosen": -415.914794921875, "eval_logps/rejected": -414.950439453125, "eval_loss": 0.6679643988609314, "eval_rewards/accuracies": 0.7936508059501648, "eval_rewards/chosen": -0.09709871560335159, "eval_rewards/margins": 0.04927373677492142, "eval_rewards/rejected": -0.146372452378273, "eval_runtime": 91.0693, "eval_samples_per_second": 10.981, "eval_steps_per_second": 0.692, "step": 200 }, { "epoch": 0.37333333333333335, "grad_norm": 0.3933409306767686, "learning_rate": 3.9505261464222127e-07, "logits/chosen": 14.01300048828125, "logits/rejected": 12.405731201171875, "logps/chosen": -414.12432861328125, "logps/rejected": -435.1546936035156, "loss": 0.6667, "rewards/accuracies": 0.78125, "rewards/chosen": -0.10342928022146225, "rewards/margins": 0.06037931516766548, "rewards/rejected": -0.16380859911441803, "step": 210 }, { "epoch": 0.39111111111111113, "grad_norm": 0.48372655964524586, "learning_rate": 3.821132513220511e-07, "logits/chosen": 13.346511840820312, "logits/rejected": 11.28764820098877, "logps/chosen": -409.07830810546875, "logps/rejected": -409.73162841796875, "loss": 0.6614, "rewards/accuracies": 0.8125, "rewards/chosen": -0.11821464449167252, "rewards/margins": 0.068251833319664, "rewards/rejected": -0.18646648526191711, "step": 220 }, { "epoch": 0.4088888888888889, "grad_norm": 0.4254357046966758, "learning_rate": 3.6866276749778575e-07, "logits/chosen": 14.086942672729492, "logits/rejected": 12.073091506958008, "logps/chosen": -421.7301330566406, "logps/rejected": -425.4147033691406, "loss": 0.6613, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -0.1326219141483307, "rewards/margins": 0.06661403924226761, "rewards/rejected": -0.1992359459400177, "step": 230 }, { "epoch": 0.4266666666666667, "grad_norm": 0.5227754305907224, "learning_rate": 3.5475320047835385e-07, "logits/chosen": 13.673505783081055, "logits/rejected": 11.437660217285156, "logps/chosen": -427.7577209472656, "logps/rejected": -431.9249572753906, "loss": 0.6557, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -0.14575575292110443, "rewards/margins": 0.08081929385662079, "rewards/rejected": -0.22657504677772522, "step": 240 }, { "epoch": 0.4444444444444444, "grad_norm": 0.5124383191463783, "learning_rate": 3.4043836367638084e-07, "logits/chosen": 14.250768661499023, "logits/rejected": 12.134879112243652, "logps/chosen": -437.49945068359375, "logps/rejected": -460.59295654296875, "loss": 0.656, "rewards/accuracies": 0.8500000238418579, "rewards/chosen": -0.15989777445793152, "rewards/margins": 0.09294405579566956, "rewards/rejected": -0.25284186005592346, "step": 250 }, { "epoch": 0.4622222222222222, "grad_norm": 0.46067856126558626, "learning_rate": 3.257736384145506e-07, "logits/chosen": 14.0197172164917, "logits/rejected": 12.032342910766602, "logps/chosen": -459.6131286621094, "logps/rejected": -442.3273010253906, "loss": 0.6506, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -0.17320798337459564, "rewards/margins": 0.0869670957326889, "rewards/rejected": -0.26017507910728455, "step": 260 }, { "epoch": 0.48, "grad_norm": 0.5260206224619711, "learning_rate": 3.1081575966602624e-07, "logits/chosen": 12.930171012878418, "logits/rejected": 11.695451736450195, "logps/chosen": -399.03790283203125, "logps/rejected": -436.02532958984375, "loss": 0.6438, "rewards/accuracies": 0.75, "rewards/chosen": -0.21191565692424774, "rewards/margins": 0.0779983252286911, "rewards/rejected": -0.28991398215293884, "step": 270 }, { "epoch": 0.49777777777777776, "grad_norm": 0.49006540494490103, "learning_rate": 2.9562259655786065e-07, "logits/chosen": 14.197479248046875, "logits/rejected": 12.239548683166504, "logps/chosen": -450.38623046875, "logps/rejected": -440.9632873535156, "loss": 0.6414, "rewards/accuracies": 0.793749988079071, "rewards/chosen": -0.21933193504810333, "rewards/margins": 0.11443768441677094, "rewards/rejected": -0.33376961946487427, "step": 280 }, { "epoch": 0.5155555555555555, "grad_norm": 0.43019209624061255, "learning_rate": 2.8025292848658625e-07, "logits/chosen": 13.873621940612793, "logits/rejected": 11.836385726928711, "logps/chosen": -467.55108642578125, "logps/rejected": -473.07818603515625, "loss": 0.6342, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -0.23439328372478485, "rewards/margins": 0.13239440321922302, "rewards/rejected": -0.3667876720428467, "step": 290 }, { "epoch": 0.5333333333333333, "grad_norm": 0.46075707476643984, "learning_rate": 2.647662177121486e-07, "logits/chosen": 13.895523071289062, "logits/rejected": 11.491305351257324, "logps/chosen": -416.88592529296875, "logps/rejected": -417.43243408203125, "loss": 0.6339, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -0.2427409142255783, "rewards/margins": 0.10592355579137802, "rewards/rejected": -0.3486644923686981, "step": 300 }, { "epoch": 0.5333333333333333, "eval_logits/chosen": 14.174385070800781, "eval_logits/rejected": 12.855634689331055, "eval_logps/chosen": -432.1300354003906, "eval_logps/rejected": -437.43072509765625, "eval_loss": 0.6389195322990417, "eval_rewards/accuracies": 0.7539682388305664, "eval_rewards/chosen": -0.2592509388923645, "eval_rewards/margins": 0.11192431300878525, "eval_rewards/rejected": -0.37117522954940796, "eval_runtime": 91.0722, "eval_samples_per_second": 10.98, "eval_steps_per_second": 0.692, "step": 300 }, { "epoch": 0.5511111111111111, "grad_norm": 0.5603386362578662, "learning_rate": 2.492223793099743e-07, "logits/chosen": 13.52906608581543, "logits/rejected": 12.464258193969727, "logps/chosen": -425.34600830078125, "logps/rejected": -457.48968505859375, "loss": 0.6345, "rewards/accuracies": 0.8062499761581421, "rewards/chosen": -0.26422396302223206, "rewards/margins": 0.11599861085414886, "rewards/rejected": -0.3802226185798645, "step": 310 }, { "epoch": 0.5688888888888889, "grad_norm": 0.5899560581428063, "learning_rate": 2.3368154937118352e-07, "logits/chosen": 14.568766593933105, "logits/rejected": 12.381284713745117, "logps/chosen": -482.33648681640625, "logps/rejected": -460.39959716796875, "loss": 0.6352, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -0.2616764307022095, "rewards/margins": 0.11895668506622314, "rewards/rejected": -0.3806331157684326, "step": 320 }, { "epoch": 0.5866666666666667, "grad_norm": 0.5671140786141569, "learning_rate": 2.1820385234773604e-07, "logits/chosen": 13.732978820800781, "logits/rejected": 11.89165210723877, "logps/chosen": -432.77435302734375, "logps/rejected": -449.07147216796875, "loss": 0.6348, "rewards/accuracies": 0.78125, "rewards/chosen": -0.28699907660484314, "rewards/margins": 0.1393791139125824, "rewards/rejected": -0.42637819051742554, "step": 330 }, { "epoch": 0.6044444444444445, "grad_norm": 0.494895870546235, "learning_rate": 2.0284916844260697e-07, "logits/chosen": 14.013145446777344, "logits/rejected": 12.176668167114258, "logps/chosen": -466.3651428222656, "logps/rejected": -457.82196044921875, "loss": 0.6271, "rewards/accuracies": 0.78125, "rewards/chosen": -0.2999596893787384, "rewards/margins": 0.150094673037529, "rewards/rejected": -0.4500543177127838, "step": 340 }, { "epoch": 0.6222222222222222, "grad_norm": 0.44235392764613574, "learning_rate": 1.876769019449141e-07, "logits/chosen": 13.546051025390625, "logits/rejected": 11.55055046081543, "logps/chosen": -455.70892333984375, "logps/rejected": -447.29156494140625, "loss": 0.6205, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -0.32695135474205017, "rewards/margins": 0.1475459635257721, "rewards/rejected": -0.47449731826782227, "step": 350 }, { "epoch": 0.64, "grad_norm": 0.4906124921285579, "learning_rate": 1.7274575140626315e-07, "logits/chosen": 13.1173095703125, "logits/rejected": 10.774048805236816, "logps/chosen": -454.48797607421875, "logps/rejected": -431.7547912597656, "loss": 0.6226, "rewards/accuracies": 0.793749988079071, "rewards/chosen": -0.30864274501800537, "rewards/margins": 0.155232235789299, "rewards/rejected": -0.4638749659061432, "step": 360 }, { "epoch": 0.6577777777777778, "grad_norm": 0.632117867399185, "learning_rate": 1.5811348254745572e-07, "logits/chosen": 13.715913772583008, "logits/rejected": 12.367976188659668, "logps/chosen": -395.98883056640625, "logps/rejected": -431.3076171875, "loss": 0.6198, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -0.3397838771343231, "rewards/margins": 0.15032809972763062, "rewards/rejected": -0.49011197686195374, "step": 370 }, { "epoch": 0.6755555555555556, "grad_norm": 0.49281947117085684, "learning_rate": 1.4383670477413674e-07, "logits/chosen": 13.90184211730957, "logits/rejected": 11.279874801635742, "logps/chosen": -461.05078125, "logps/rejected": -442.17449951171875, "loss": 0.6203, "rewards/accuracies": 0.8187500238418579, "rewards/chosen": -0.3398580551147461, "rewards/margins": 0.16354627907276154, "rewards/rejected": -0.5034043192863464, "step": 380 }, { "epoch": 0.6933333333333334, "grad_norm": 0.42880409636253375, "learning_rate": 1.2997065216600178e-07, "logits/chosen": 13.401517868041992, "logits/rejected": 12.245413780212402, "logps/chosen": -430.324462890625, "logps/rejected": -473.84893798828125, "loss": 0.6211, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -0.3511757552623749, "rewards/margins": 0.1507997363805771, "rewards/rejected": -0.5019755363464355, "step": 390 }, { "epoch": 0.7111111111111111, "grad_norm": 0.40477681812860294, "learning_rate": 1.1656896978687259e-07, "logits/chosen": 13.779144287109375, "logits/rejected": 11.85938835144043, "logps/chosen": -440.189208984375, "logps/rejected": -442.59576416015625, "loss": 0.6203, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -0.3478553891181946, "rewards/margins": 0.1676493138074875, "rewards/rejected": -0.5155047178268433, "step": 400 }, { "epoch": 0.7111111111111111, "eval_logits/chosen": 13.944354057312012, "eval_logits/rejected": 12.625584602355957, "eval_logps/chosen": -443.5887451171875, "eval_logps/rejected": -453.4456787109375, "eval_loss": 0.6203290224075317, "eval_rewards/accuracies": 0.7539682388305664, "eval_rewards/chosen": -0.37383800745010376, "eval_rewards/margins": 0.15748701989650726, "eval_rewards/rejected": -0.531325101852417, "eval_runtime": 91.0982, "eval_samples_per_second": 10.977, "eval_steps_per_second": 0.692, "step": 400 }, { "epoch": 0.7288888888888889, "grad_norm": 0.4976401369848891, "learning_rate": 1.0368350614236685e-07, "logits/chosen": 14.432429313659668, "logits/rejected": 12.734024047851562, "logps/chosen": -437.3631286621094, "logps/rejected": -450.3123474121094, "loss": 0.6156, "rewards/accuracies": 0.8062499761581421, "rewards/chosen": -0.37959882616996765, "rewards/margins": 0.17338675260543823, "rewards/rejected": -0.5529855489730835, "step": 410 }, { "epoch": 0.7466666666666667, "grad_norm": 0.49003467823924163, "learning_rate": 9.136411258810229e-08, "logits/chosen": 14.124117851257324, "logits/rejected": 12.797445297241211, "logps/chosen": -429.11767578125, "logps/rejected": -457.7822265625, "loss": 0.6164, "rewards/accuracies": 0.824999988079071, "rewards/chosen": -0.3667467534542084, "rewards/margins": 0.17160849273204803, "rewards/rejected": -0.5383552312850952, "step": 420 }, { "epoch": 0.7644444444444445, "grad_norm": 0.5544257285079301, "learning_rate": 7.965845046448657e-08, "logits/chosen": 14.171014785766602, "logits/rejected": 11.860380172729492, "logps/chosen": -460.3724670410156, "logps/rejected": -464.814208984375, "loss": 0.6111, "rewards/accuracies": 0.84375, "rewards/chosen": -0.3682584762573242, "rewards/margins": 0.21057240664958954, "rewards/rejected": -0.578830897808075, "step": 430 }, { "epoch": 0.7822222222222223, "grad_norm": 0.4893812986921009, "learning_rate": 6.861180670424982e-08, "logits/chosen": 14.053049087524414, "logits/rejected": 12.361749649047852, "logps/chosen": -458.22113037109375, "logps/rejected": -473.56927490234375, "loss": 0.6182, "rewards/accuracies": 0.84375, "rewards/chosen": -0.38496845960617065, "rewards/margins": 0.1849747598171234, "rewards/rejected": -0.5699432492256165, "step": 440 }, { "epoch": 0.8, "grad_norm": 0.451783793255229, "learning_rate": 5.826691862609986e-08, "logits/chosen": 14.11127758026123, "logits/rejected": 12.265645027160645, "logps/chosen": -446.2030334472656, "logps/rejected": -448.130859375, "loss": 0.6154, "rewards/accuracies": 0.793749988079071, "rewards/chosen": -0.3934079110622406, "rewards/margins": 0.20773550868034363, "rewards/rejected": -0.601143479347229, "step": 450 }, { "epoch": 0.8177777777777778, "grad_norm": 0.5757576180607112, "learning_rate": 4.86638085923389e-08, "logits/chosen": 13.425259590148926, "logits/rejected": 11.878642082214355, "logps/chosen": -452.25042724609375, "logps/rejected": -480.77154541015625, "loss": 0.6026, "rewards/accuracies": 0.8125, "rewards/chosen": -0.42305225133895874, "rewards/margins": 0.19301000237464905, "rewards/rejected": -0.6160622835159302, "step": 460 }, { "epoch": 0.8355555555555556, "grad_norm": 0.6028404780293103, "learning_rate": 3.983962917011829e-08, "logits/chosen": 13.737588882446289, "logits/rejected": 11.253530502319336, "logps/chosen": -453.21466064453125, "logps/rejected": -455.3412170410156, "loss": 0.6099, "rewards/accuracies": 0.793749988079071, "rewards/chosen": -0.40479034185409546, "rewards/margins": 0.23396852612495422, "rewards/rejected": -0.6387587785720825, "step": 470 }, { "epoch": 0.8533333333333334, "grad_norm": 0.4822754579818044, "learning_rate": 3.182851939537409e-08, "logits/chosen": 13.7351655960083, "logits/rejected": 12.563407897949219, "logps/chosen": -420.5502014160156, "logps/rejected": -467.66143798828125, "loss": 0.6121, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -0.4305228590965271, "rewards/margins": 0.1863301545381546, "rewards/rejected": -0.6168529987335205, "step": 480 }, { "epoch": 0.8711111111111111, "grad_norm": 0.49761272281985847, "learning_rate": 2.466147269552893e-08, "logits/chosen": 13.694448471069336, "logits/rejected": 12.38302230834961, "logps/chosen": -429.01910400390625, "logps/rejected": -461.55718994140625, "loss": 0.6128, "rewards/accuracies": 0.793749988079071, "rewards/chosen": -0.4084044396877289, "rewards/margins": 0.16671887040138245, "rewards/rejected": -0.5751233100891113, "step": 490 }, { "epoch": 0.8888888888888888, "grad_norm": 0.6004438030491424, "learning_rate": 1.8366216981942628e-08, "logits/chosen": 13.4122896194458, "logits/rejected": 11.679363250732422, "logps/chosen": -436.2166442871094, "logps/rejected": -459.05841064453125, "loss": 0.6102, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -0.4055519104003906, "rewards/margins": 0.21610493957996368, "rewards/rejected": -0.6216568946838379, "step": 500 }, { "epoch": 0.8888888888888888, "eval_logits/chosen": 13.842653274536133, "eval_logits/rejected": 12.531366348266602, "eval_logps/chosen": -447.70013427734375, "eval_logps/rejected": -459.2376403808594, "eval_loss": 0.6131365299224854, "eval_rewards/accuracies": 0.7539682388305664, "eval_rewards/chosen": -0.4149521589279175, "eval_rewards/margins": 0.17429186403751373, "eval_rewards/rejected": -0.5892440676689148, "eval_runtime": 91.0747, "eval_samples_per_second": 10.98, "eval_steps_per_second": 0.692, "step": 500 }, { "epoch": 0.9066666666666666, "grad_norm": 0.512528339625896, "learning_rate": 1.296710737600934e-08, "logits/chosen": 14.248278617858887, "logits/rejected": 11.994184494018555, "logps/chosen": -501.92333984375, "logps/rejected": -475.35723876953125, "loss": 0.6136, "rewards/accuracies": 0.78125, "rewards/chosen": -0.40441107749938965, "rewards/margins": 0.22114601731300354, "rewards/rejected": -0.6255571246147156, "step": 510 }, { "epoch": 0.9244444444444444, "grad_norm": 0.46717306963218175, "learning_rate": 8.485031983924557e-09, "logits/chosen": 13.698689460754395, "logits/rejected": 11.797239303588867, "logps/chosen": -473.1220703125, "logps/rejected": -475.6698303222656, "loss": 0.609, "rewards/accuracies": 0.824999988079071, "rewards/chosen": -0.40674418210983276, "rewards/margins": 0.2259480506181717, "rewards/rejected": -0.6326922178268433, "step": 520 }, { "epoch": 0.9422222222222222, "grad_norm": 0.46210848243091895, "learning_rate": 4.937331084660129e-09, "logits/chosen": 13.86840534210205, "logits/rejected": 11.828570365905762, "logps/chosen": -465.6477966308594, "logps/rejected": -473.4181213378906, "loss": 0.607, "rewards/accuracies": 0.793749988079071, "rewards/chosen": -0.4186909794807434, "rewards/margins": 0.19602572917938232, "rewards/rejected": -0.614716649055481, "step": 530 }, { "epoch": 0.96, "grad_norm": 0.48983149292621647, "learning_rate": 2.337730043793423e-09, "logits/chosen": 13.412510871887207, "logits/rejected": 11.925616264343262, "logps/chosen": -441.61871337890625, "logps/rejected": -453.08489990234375, "loss": 0.6106, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -0.42602628469467163, "rewards/margins": 0.19294434785842896, "rewards/rejected": -0.6189705729484558, "step": 540 }, { "epoch": 0.9777777777777777, "grad_norm": 0.46702206339372576, "learning_rate": 6.962862127343205e-10, "logits/chosen": 14.47253131866455, "logits/rejected": 12.454832077026367, "logps/chosen": -464.0704650878906, "logps/rejected": -453.48187255859375, "loss": 0.6073, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -0.4236368238925934, "rewards/margins": 0.20897309482097626, "rewards/rejected": -0.6326099038124084, "step": 550 }, { "epoch": 0.9955555555555555, "grad_norm": 0.620631975977973, "learning_rate": 1.9350018786556956e-11, "logits/chosen": 13.917093276977539, "logits/rejected": 12.686841011047363, "logps/chosen": -464.35980224609375, "logps/rejected": -498.465576171875, "loss": 0.6102, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -0.4351399838924408, "rewards/margins": 0.1616152673959732, "rewards/rejected": -0.5967552661895752, "step": 560 }, { "epoch": 0.9991111111111111, "step": 562, "total_flos": 0.0, "train_loss": 0.6475430355801701, "train_runtime": 7558.4306, "train_samples_per_second": 4.763, "train_steps_per_second": 0.074 } ], "logging_steps": 10, "max_steps": 562, "num_input_tokens_seen": 0, "num_train_epochs": 1, "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": 4, "trial_name": null, "trial_params": null }