{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 100, "global_step": 1359, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "grad_norm": 0.416015625, "learning_rate": 3.676470588235294e-09, "logits/chosen": 0.3823161721229553, "logits/rejected": 0.49768388271331787, "logps/chosen": -143.86793518066406, "logps/rejected": -156.11575317382812, "loss": 0.6931, "rewards/accuracies": 0.0, "rewards/chosen": 0.0, "rewards/margins": 0.0, "rewards/margins_max": 0.0, "rewards/margins_min": 0.0, "rewards/margins_std": 0.0, "rewards/rejected": 0.0, "step": 1 }, { "epoch": 0.01, "grad_norm": 0.5, "learning_rate": 3.676470588235294e-08, "logits/chosen": -0.03673278167843819, "logits/rejected": 0.7153948545455933, "logps/chosen": -280.6231689453125, "logps/rejected": -194.65892028808594, "loss": 0.6933, "rewards/accuracies": 0.3888888955116272, "rewards/chosen": -0.0011863290565088391, "rewards/margins": -0.00027060293359681964, "rewards/margins_max": 0.0016658814856782556, "rewards/margins_min": -0.002207087352871895, "rewards/margins_std": 0.0027386024594306946, "rewards/rejected": -0.0009157261229120195, "step": 10 }, { "epoch": 0.01, "grad_norm": 0.37109375, "learning_rate": 7.352941176470588e-08, "logits/chosen": 0.007032014429569244, "logits/rejected": 0.7839561104774475, "logps/chosen": -332.548828125, "logps/rejected": -255.7633819580078, "loss": 0.6932, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": -0.0009580751066096127, "rewards/margins": 0.00016147561836987734, "rewards/margins_max": 0.0031394611578434706, "rewards/margins_min": -0.0028165101539343596, "rewards/margins_std": 0.004211507737636566, "rewards/rejected": -0.0011195506667718291, "step": 20 }, { "epoch": 0.02, "grad_norm": 0.51171875, "learning_rate": 1.1029411764705881e-07, "logits/chosen": -0.04039617255330086, "logits/rejected": 0.607696533203125, "logps/chosen": -272.3553161621094, "logps/rejected": -256.77227783203125, "loss": 0.6932, "rewards/accuracies": 0.4625000059604645, "rewards/chosen": -0.000774255720898509, "rewards/margins": 8.788631384959444e-05, "rewards/margins_max": 0.001997451763600111, "rewards/margins_min": -0.0018216792959719896, "rewards/margins_std": 0.002700533252209425, "rewards/rejected": -0.0008621420711278915, "step": 30 }, { "epoch": 0.03, "grad_norm": 0.42578125, "learning_rate": 1.4705882352941175e-07, "logits/chosen": -0.06750641763210297, "logits/rejected": 0.6604307889938354, "logps/chosen": -295.84283447265625, "logps/rejected": -260.4482421875, "loss": 0.693, "rewards/accuracies": 0.4625000059604645, "rewards/chosen": -0.0007127780700102448, "rewards/margins": 0.00020934769418090582, "rewards/margins_max": 0.0021955964621156454, "rewards/margins_min": -0.0017769013065844774, "rewards/margins_std": 0.0028089801780879498, "rewards/rejected": -0.0009221257641911507, "step": 40 }, { "epoch": 0.04, "grad_norm": 0.470703125, "learning_rate": 1.8382352941176472e-07, "logits/chosen": -0.015060709789395332, "logits/rejected": 0.6939523220062256, "logps/chosen": -287.7695617675781, "logps/rejected": -237.29794311523438, "loss": 0.6927, "rewards/accuracies": 0.5375000238418579, "rewards/chosen": 0.0005769954295828938, "rewards/margins": 0.0007454338483512402, "rewards/margins_max": 0.0031318417750298977, "rewards/margins_min": -0.0016409741947427392, "rewards/margins_std": 0.0033748901914805174, "rewards/rejected": -0.0001684384187683463, "step": 50 }, { "epoch": 0.04, "grad_norm": 0.49609375, "learning_rate": 2.2058823529411763e-07, "logits/chosen": -0.0018195391166955233, "logits/rejected": 0.7142434120178223, "logps/chosen": -287.01068115234375, "logps/rejected": -256.10614013671875, "loss": 0.6923, "rewards/accuracies": 0.625, "rewards/chosen": 0.0012248576385900378, "rewards/margins": 0.0020126227755099535, "rewards/margins_max": 0.004514734260737896, "rewards/margins_min": -0.0004894885933026671, "rewards/margins_std": 0.0035385198425501585, "rewards/rejected": -0.0007877650787122548, "step": 60 }, { "epoch": 0.05, "grad_norm": 0.421875, "learning_rate": 2.5735294117647057e-07, "logits/chosen": 0.11708433926105499, "logits/rejected": 0.5532101988792419, "logps/chosen": -279.79791259765625, "logps/rejected": -243.6463623046875, "loss": 0.692, "rewards/accuracies": 0.6625000238418579, "rewards/chosen": 0.0009990496328100562, "rewards/margins": 0.0019516434986144304, "rewards/margins_max": 0.004653914365917444, "rewards/margins_min": -0.0007506275433115661, "rewards/margins_std": 0.0038215883541852236, "rewards/rejected": -0.0009525938075967133, "step": 70 }, { "epoch": 0.06, "grad_norm": 0.50390625, "learning_rate": 2.941176470588235e-07, "logits/chosen": -0.1581064760684967, "logits/rejected": 0.7092851996421814, "logps/chosen": -311.269775390625, "logps/rejected": -251.24081420898438, "loss": 0.6915, "rewards/accuracies": 0.762499988079071, "rewards/chosen": 0.002858319552615285, "rewards/margins": 0.0039031144697219133, "rewards/margins_max": 0.006772955413907766, "rewards/margins_min": 0.001033273758366704, "rewards/margins_std": 0.004058566875755787, "rewards/rejected": -0.001044794567860663, "step": 80 }, { "epoch": 0.07, "grad_norm": 0.390625, "learning_rate": 3.3088235294117644e-07, "logits/chosen": -0.045442212373018265, "logits/rejected": 0.7158526182174683, "logps/chosen": -265.78192138671875, "logps/rejected": -218.0985870361328, "loss": 0.691, "rewards/accuracies": 0.75, "rewards/chosen": 0.003086745273321867, "rewards/margins": 0.0036944295279681683, "rewards/margins_max": 0.007027274928987026, "rewards/margins_min": 0.0003615838068071753, "rewards/margins_std": 0.0047133550979197025, "rewards/rejected": -0.0006076841382309794, "step": 90 }, { "epoch": 0.07, "grad_norm": 0.458984375, "learning_rate": 3.6764705882352943e-07, "logits/chosen": 0.0736229196190834, "logits/rejected": 0.7526043057441711, "logps/chosen": -253.97119140625, "logps/rejected": -216.62966918945312, "loss": 0.6907, "rewards/accuracies": 0.762499988079071, "rewards/chosen": 0.004654365126043558, "rewards/margins": 0.005358897615224123, "rewards/margins_max": 0.008878113701939583, "rewards/margins_min": 0.0018396815285086632, "rewards/margins_std": 0.004976923577487469, "rewards/rejected": -0.0007045331294648349, "step": 100 }, { "epoch": 0.08, "grad_norm": 0.42578125, "learning_rate": 4.044117647058823e-07, "logits/chosen": -0.09733302891254425, "logits/rejected": 0.6439273953437805, "logps/chosen": -289.07769775390625, "logps/rejected": -234.5022735595703, "loss": 0.6899, "rewards/accuracies": 0.800000011920929, "rewards/chosen": 0.005760089494287968, "rewards/margins": 0.006674544420093298, "rewards/margins_max": 0.010509292595088482, "rewards/margins_min": 0.002839796245098114, "rewards/margins_std": 0.005423153750598431, "rewards/rejected": -0.0009144555078819394, "step": 110 }, { "epoch": 0.09, "grad_norm": 0.423828125, "learning_rate": 4.4117647058823526e-07, "logits/chosen": 0.018124500289559364, "logits/rejected": 0.7241690754890442, "logps/chosen": -311.22100830078125, "logps/rejected": -241.0907440185547, "loss": 0.6888, "rewards/accuracies": 0.8500000238418579, "rewards/chosen": 0.008410803973674774, "rewards/margins": 0.007946285419166088, "rewards/margins_max": 0.013005959801375866, "rewards/margins_min": 0.002886611968278885, "rewards/margins_std": 0.0071554603055119514, "rewards/rejected": 0.000464518612716347, "step": 120 }, { "epoch": 0.1, "grad_norm": 0.494140625, "learning_rate": 4.779411764705882e-07, "logits/chosen": -0.1438770443201065, "logits/rejected": 0.6687305569648743, "logps/chosen": -265.34161376953125, "logps/rejected": -225.00424194335938, "loss": 0.6875, "rewards/accuracies": 0.862500011920929, "rewards/chosen": 0.011373260989785194, "rewards/margins": 0.01142856478691101, "rewards/margins_max": 0.018687166273593903, "rewards/margins_min": 0.004169962834566832, "rewards/margins_std": 0.010265212506055832, "rewards/rejected": -5.530398993869312e-05, "step": 130 }, { "epoch": 0.1, "grad_norm": 0.423828125, "learning_rate": 4.999868030671756e-07, "logits/chosen": 0.05402200296521187, "logits/rejected": 0.6528539657592773, "logps/chosen": -292.96112060546875, "logps/rejected": -259.50225830078125, "loss": 0.6862, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": 0.013445700518786907, "rewards/margins": 0.014345765113830566, "rewards/margins_max": 0.022081127390265465, "rewards/margins_min": 0.006610400043427944, "rewards/margins_std": 0.010939454659819603, "rewards/rejected": -0.0009000620921142399, "step": 140 }, { "epoch": 0.11, "grad_norm": 0.462890625, "learning_rate": 4.998383535732973e-07, "logits/chosen": -0.16393637657165527, "logits/rejected": 0.45549267530441284, "logps/chosen": -290.1100158691406, "logps/rejected": -285.8172302246094, "loss": 0.6856, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.016459070146083832, "rewards/margins": 0.016152694821357727, "rewards/margins_max": 0.02362522855401039, "rewards/margins_min": 0.008680161088705063, "rewards/margins_std": 0.010567759163677692, "rewards/rejected": 0.00030637450981885195, "step": 150 }, { "epoch": 0.12, "grad_norm": 0.494140625, "learning_rate": 4.995250566954361e-07, "logits/chosen": 0.018339941278100014, "logits/rejected": 0.7964209914207458, "logps/chosen": -283.98577880859375, "logps/rejected": -241.65353393554688, "loss": 0.6842, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": 0.01847967877984047, "rewards/margins": 0.01731337048113346, "rewards/margins_max": 0.028514724224805832, "rewards/margins_min": 0.006112017668783665, "rewards/margins_std": 0.015841105952858925, "rewards/rejected": 0.0011663079494610429, "step": 160 }, { "epoch": 0.13, "grad_norm": 0.5546875, "learning_rate": 4.990471191519357e-07, "logits/chosen": 0.03216102719306946, "logits/rejected": 0.7764664888381958, "logps/chosen": -279.2666931152344, "logps/rejected": -237.353515625, "loss": 0.6827, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.023826319724321365, "rewards/margins": 0.02013733796775341, "rewards/margins_max": 0.03089248575270176, "rewards/margins_min": 0.009382191114127636, "rewards/margins_std": 0.015210075303912163, "rewards/rejected": 0.0036889780312776566, "step": 170 }, { "epoch": 0.13, "grad_norm": 0.4453125, "learning_rate": 4.984048562937129e-07, "logits/chosen": -0.034670569002628326, "logits/rejected": 0.5347191095352173, "logps/chosen": -253.5879364013672, "logps/rejected": -252.5656280517578, "loss": 0.6827, "rewards/accuracies": 0.9375, "rewards/chosen": 0.021980373188853264, "rewards/margins": 0.020695097744464874, "rewards/margins_max": 0.03351113945245743, "rewards/margins_min": 0.007879055105149746, "rewards/margins_std": 0.018124621361494064, "rewards/rejected": 0.0012852762592956424, "step": 180 }, { "epoch": 0.14, "grad_norm": 0.37109375, "learning_rate": 4.975986918961825e-07, "logits/chosen": -0.0010436892043799162, "logits/rejected": 0.5325745344161987, "logps/chosen": -258.70465087890625, "logps/rejected": -254.8517608642578, "loss": 0.6811, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.02575596049427986, "rewards/margins": 0.0235120952129364, "rewards/margins_max": 0.037295907735824585, "rewards/margins_min": 0.009728283621370792, "rewards/margins_std": 0.019493255764245987, "rewards/rejected": 0.0022438622545450926, "step": 190 }, { "epoch": 0.15, "grad_norm": 0.447265625, "learning_rate": 4.966291578796448e-07, "logits/chosen": -0.08659172803163528, "logits/rejected": 0.8055790066719055, "logps/chosen": -268.54559326171875, "logps/rejected": -214.8706512451172, "loss": 0.68, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.02814793959259987, "rewards/margins": 0.02799884043633938, "rewards/margins_max": 0.042513541877269745, "rewards/margins_min": 0.013484140858054161, "rewards/margins_std": 0.020526882261037827, "rewards/rejected": 0.00014909838500898331, "step": 200 }, { "epoch": 0.15, "grad_norm": 0.3515625, "learning_rate": 4.954968939583149e-07, "logits/chosen": -0.15794704854488373, "logits/rejected": 0.6090285181999207, "logps/chosen": -289.7901306152344, "logps/rejected": -238.60183715820312, "loss": 0.6789, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.03186321258544922, "rewards/margins": 0.03159675747156143, "rewards/margins_max": 0.04762989282608032, "rewards/margins_min": 0.015563614666461945, "rewards/margins_std": 0.022674284875392914, "rewards/rejected": 0.0002664585190359503, "step": 210 }, { "epoch": 0.16, "grad_norm": 0.44921875, "learning_rate": 4.942026472182297e-07, "logits/chosen": 0.012926039285957813, "logits/rejected": 0.6630520820617676, "logps/chosen": -269.59930419921875, "logps/rejected": -248.7339324951172, "loss": 0.6778, "rewards/accuracies": 0.987500011920929, "rewards/chosen": 0.0339738167822361, "rewards/margins": 0.03334929794073105, "rewards/margins_max": 0.04980397969484329, "rewards/margins_min": 0.016894619911909103, "rewards/margins_std": 0.023270433768630028, "rewards/rejected": 0.0006245142430998385, "step": 220 }, { "epoch": 0.17, "grad_norm": 0.43359375, "learning_rate": 4.92747271624308e-07, "logits/chosen": -0.10246638208627701, "logits/rejected": 0.8088499903678894, "logps/chosen": -293.9709167480469, "logps/rejected": -229.0161895751953, "loss": 0.6776, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.03623218461871147, "rewards/margins": 0.03605912998318672, "rewards/margins_max": 0.05394510179758072, "rewards/margins_min": 0.018173154443502426, "rewards/margins_std": 0.025294587016105652, "rewards/rejected": 0.0001730593794491142, "step": 230 }, { "epoch": 0.18, "grad_norm": 0.50390625, "learning_rate": 4.911317274568909e-07, "logits/chosen": -0.1043270006775856, "logits/rejected": 0.7763813138008118, "logps/chosen": -290.52032470703125, "logps/rejected": -238.03591918945312, "loss": 0.6748, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.04272182658314705, "rewards/margins": 0.04140879958868027, "rewards/margins_max": 0.0613585002720356, "rewards/margins_min": 0.021459100767970085, "rewards/margins_std": 0.0282131377607584, "rewards/rejected": 0.0013130264123901725, "step": 240 }, { "epoch": 0.18, "grad_norm": 0.4765625, "learning_rate": 4.89357080678133e-07, "logits/chosen": 0.042627763003110886, "logits/rejected": 0.5050796270370483, "logps/chosen": -225.12393188476562, "logps/rejected": -208.74288940429688, "loss": 0.6767, "rewards/accuracies": 0.875, "rewards/chosen": 0.02972862496972084, "rewards/margins": 0.026549100875854492, "rewards/margins_max": 0.04413127154111862, "rewards/margins_min": 0.008966929279267788, "rewards/margins_std": 0.02486494556069374, "rewards/rejected": 0.00317952549085021, "step": 250 }, { "epoch": 0.19, "grad_norm": 0.6015625, "learning_rate": 4.874245022286637e-07, "logits/chosen": 0.061648208647966385, "logits/rejected": 0.6458357572555542, "logps/chosen": -262.064208984375, "logps/rejected": -268.61444091796875, "loss": 0.677, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.03966163843870163, "rewards/margins": 0.036633603274822235, "rewards/margins_max": 0.05438825488090515, "rewards/margins_min": 0.01887894608080387, "rewards/margins_std": 0.02510887011885643, "rewards/rejected": 0.0030280412174761295, "step": 260 }, { "epoch": 0.2, "grad_norm": 0.3984375, "learning_rate": 4.853352672549815e-07, "logits/chosen": 0.05668522045016289, "logits/rejected": 0.7193032503128052, "logps/chosen": -261.01708984375, "logps/rejected": -228.1970672607422, "loss": 0.6732, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.038796816021203995, "rewards/margins": 0.038500867784023285, "rewards/margins_max": 0.05987783148884773, "rewards/margins_min": 0.01712390035390854, "rewards/margins_std": 0.030231595039367676, "rewards/rejected": 0.0002959508856292814, "step": 270 }, { "epoch": 0.21, "grad_norm": 0.3984375, "learning_rate": 4.830907542680918e-07, "logits/chosen": -0.022307882085442543, "logits/rejected": 0.5763358473777771, "logps/chosen": -259.81414794921875, "logps/rejected": -243.997802734375, "loss": 0.6746, "rewards/accuracies": 0.9375, "rewards/chosen": 0.04487748071551323, "rewards/margins": 0.04191521182656288, "rewards/margins_max": 0.06111626699566841, "rewards/margins_min": 0.022714145481586456, "rewards/margins_std": 0.027154404670000076, "rewards/rejected": 0.0029622733127325773, "step": 280 }, { "epoch": 0.21, "grad_norm": 0.337890625, "learning_rate": 4.806924442339425e-07, "logits/chosen": -0.14441627264022827, "logits/rejected": 0.5065394639968872, "logps/chosen": -243.6468505859375, "logps/rejected": -224.4588165283203, "loss": 0.6735, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": 0.045001424849033356, "rewards/margins": 0.04354133456945419, "rewards/margins_max": 0.062143802642822266, "rewards/margins_min": 0.02493886649608612, "rewards/margins_std": 0.026307862251996994, "rewards/rejected": 0.0014600924914702773, "step": 290 }, { "epoch": 0.22, "grad_norm": 0.4140625, "learning_rate": 4.781419195962598e-07, "logits/chosen": -0.09291721135377884, "logits/rejected": 0.8421966433525085, "logps/chosen": -294.4910583496094, "logps/rejected": -235.1195526123047, "loss": 0.6669, "rewards/accuracies": 0.9375, "rewards/chosen": 0.05570490285754204, "rewards/margins": 0.05652566999197006, "rewards/margins_max": 0.0832170695066452, "rewards/margins_min": 0.02983427606523037, "rewards/margins_std": 0.0377473346889019, "rewards/rejected": -0.0008207682403735816, "step": 300 }, { "epoch": 0.23, "grad_norm": 0.427734375, "learning_rate": 4.754408632324253e-07, "logits/chosen": 0.08360351622104645, "logits/rejected": 0.9180965423583984, "logps/chosen": -274.46221923828125, "logps/rejected": -218.58261108398438, "loss": 0.6709, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.04704751446843147, "rewards/margins": 0.04507958143949509, "rewards/margins_max": 0.07067537307739258, "rewards/margins_min": 0.019483786076307297, "rewards/margins_std": 0.03619792312383652, "rewards/rejected": 0.0019679281394928694, "step": 310 }, { "epoch": 0.24, "grad_norm": 0.392578125, "learning_rate": 4.725910573430866e-07, "logits/chosen": -0.03652850538492203, "logits/rejected": 0.7102953195571899, "logps/chosen": -259.0872497558594, "logps/rejected": -220.06640625, "loss": 0.6666, "rewards/accuracies": 0.9375, "rewards/chosen": 0.052086759358644485, "rewards/margins": 0.05210639908909798, "rewards/margins_max": 0.07583759725093842, "rewards/margins_min": 0.02837519347667694, "rewards/margins_std": 0.033560991287231445, "rewards/rejected": -1.9636423530755565e-05, "step": 320 }, { "epoch": 0.24, "grad_norm": 0.482421875, "learning_rate": 4.6959438227623293e-07, "logits/chosen": -0.07948180288076401, "logits/rejected": 0.6266988515853882, "logps/chosen": -291.2080078125, "logps/rejected": -269.5301818847656, "loss": 0.6677, "rewards/accuracies": 0.987500011920929, "rewards/chosen": 0.04915199428796768, "rewards/margins": 0.050243012607097626, "rewards/margins_max": 0.07468355447053909, "rewards/margins_min": 0.025802483782172203, "rewards/margins_std": 0.03456413000822067, "rewards/rejected": -0.0010910223936662078, "step": 330 }, { "epoch": 0.25, "grad_norm": 0.451171875, "learning_rate": 4.664528152865105e-07, "logits/chosen": 0.015779225155711174, "logits/rejected": 0.6966221332550049, "logps/chosen": -291.6715087890625, "logps/rejected": -260.85784912109375, "loss": 0.6643, "rewards/accuracies": 0.9375, "rewards/chosen": 0.06338638812303543, "rewards/margins": 0.06281675398349762, "rewards/margins_max": 0.09652182459831238, "rewards/margins_min": 0.029111674055457115, "rewards/margins_std": 0.04766617342829704, "rewards/rejected": 0.0005696387379430234, "step": 340 }, { "epoch": 0.26, "grad_norm": 0.56640625, "learning_rate": 4.6316842923059816e-07, "logits/chosen": -0.15278920531272888, "logits/rejected": 0.6084717512130737, "logps/chosen": -247.9635009765625, "logps/rejected": -236.7021484375, "loss": 0.6618, "rewards/accuracies": 0.987500011920929, "rewards/chosen": 0.07176478952169418, "rewards/margins": 0.07197294384241104, "rewards/margins_max": 0.09642891585826874, "rewards/margins_min": 0.047516968101263046, "rewards/margins_std": 0.03458596393465996, "rewards/rejected": -0.0002081508719129488, "step": 350 }, { "epoch": 0.26, "grad_norm": 0.45703125, "learning_rate": 4.5974339119950334e-07, "logits/chosen": -0.0013063043588772416, "logits/rejected": 0.6344822645187378, "logps/chosen": -263.31610107421875, "logps/rejected": -204.03660583496094, "loss": 0.6657, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": 0.052458666265010834, "rewards/margins": 0.04967236518859863, "rewards/margins_max": 0.07377265393733978, "rewards/margins_min": 0.025572070851922035, "rewards/margins_std": 0.034082964062690735, "rewards/rejected": 0.0027863027062267065, "step": 360 }, { "epoch": 0.27, "grad_norm": 0.451171875, "learning_rate": 4.5617996108867997e-07, "logits/chosen": -0.09778765588998795, "logits/rejected": 0.8527771234512329, "logps/chosen": -272.1876525878906, "logps/rejected": -224.5320281982422, "loss": 0.6615, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.06909666955471039, "rewards/margins": 0.06635682284832001, "rewards/margins_max": 0.09376729279756546, "rewards/margins_min": 0.03894636407494545, "rewards/margins_std": 0.03876424580812454, "rewards/rejected": 0.0027398415841162205, "step": 370 }, { "epoch": 0.28, "grad_norm": 0.39453125, "learning_rate": 4.5248049010691304e-07, "logits/chosen": -0.04781482741236687, "logits/rejected": 0.6669384241104126, "logps/chosen": -270.28253173828125, "logps/rejected": -231.1532440185547, "loss": 0.6605, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.06150456517934799, "rewards/margins": 0.0636376217007637, "rewards/margins_max": 0.09591880440711975, "rewards/margins_min": 0.03135643154382706, "rewards/margins_std": 0.04565249755978584, "rewards/rejected": -0.0021330583840608597, "step": 380 }, { "epoch": 0.29, "grad_norm": 0.5546875, "learning_rate": 4.486474192249533e-07, "logits/chosen": 0.034309376031160355, "logits/rejected": 0.7420383095741272, "logps/chosen": -246.19668579101562, "logps/rejected": -215.16323852539062, "loss": 0.6629, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.056318044662475586, "rewards/margins": 0.06057443469762802, "rewards/margins_max": 0.08743326365947723, "rewards/margins_min": 0.03371559828519821, "rewards/margins_std": 0.037984125316143036, "rewards/rejected": -0.0042563872411847115, "step": 390 }, { "epoch": 0.29, "grad_norm": 0.5390625, "learning_rate": 4.4468327756492504e-07, "logits/chosen": 0.09597498923540115, "logits/rejected": 0.7928622961044312, "logps/chosen": -275.2891540527344, "logps/rejected": -246.2664794921875, "loss": 0.6631, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.06533468514680862, "rewards/margins": 0.06880095601081848, "rewards/margins_max": 0.1037551760673523, "rewards/margins_min": 0.03384672850370407, "rewards/margins_std": 0.04943273216485977, "rewards/rejected": -0.0034662713296711445, "step": 400 }, { "epoch": 0.3, "grad_norm": 0.4765625, "learning_rate": 4.405906807315705e-07, "logits/chosen": -0.021347884088754654, "logits/rejected": 0.6999279260635376, "logps/chosen": -227.08447265625, "logps/rejected": -222.6078643798828, "loss": 0.6565, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.0646035447716713, "rewards/margins": 0.0676642507314682, "rewards/margins_max": 0.09568590670824051, "rewards/margins_min": 0.039642591029405594, "rewards/margins_std": 0.039628610014915466, "rewards/rejected": -0.003060706425458193, "step": 410 }, { "epoch": 0.31, "grad_norm": 0.41796875, "learning_rate": 4.363723290864314e-07, "logits/chosen": 0.11118383705615997, "logits/rejected": 0.9948932528495789, "logps/chosen": -271.27496337890625, "logps/rejected": -222.7627410888672, "loss": 0.6589, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.07119729369878769, "rewards/margins": 0.07638658583164215, "rewards/margins_max": 0.120518758893013, "rewards/margins_min": 0.032254405319690704, "rewards/margins_std": 0.0624123141169548, "rewards/rejected": -0.005189282353967428, "step": 420 }, { "epoch": 0.32, "grad_norm": 0.486328125, "learning_rate": 4.3203100596610723e-07, "logits/chosen": -0.016591979190707207, "logits/rejected": 0.813875675201416, "logps/chosen": -300.334716796875, "logps/rejected": -259.6979064941406, "loss": 0.656, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.07723168283700943, "rewards/margins": 0.0816957950592041, "rewards/margins_max": 0.1194806843996048, "rewards/margins_min": 0.043910913169384, "rewards/margins_std": 0.05343589186668396, "rewards/rejected": -0.004464106634259224, "step": 430 }, { "epoch": 0.32, "grad_norm": 0.384765625, "learning_rate": 4.2756957584576436e-07, "logits/chosen": 0.051903001964092255, "logits/rejected": 0.7332680821418762, "logps/chosen": -291.6619873046875, "logps/rejected": -264.76763916015625, "loss": 0.6551, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.07283146679401398, "rewards/margins": 0.07750260829925537, "rewards/margins_max": 0.11894078552722931, "rewards/margins_min": 0.03606444224715233, "rewards/margins_std": 0.05860241502523422, "rewards/rejected": -0.004671142902225256, "step": 440 }, { "epoch": 0.33, "grad_norm": 0.38671875, "learning_rate": 4.22990982449109e-07, "logits/chosen": 0.00915027130395174, "logits/rejected": 0.7316643595695496, "logps/chosen": -236.2955780029297, "logps/rejected": -220.3606414794922, "loss": 0.6569, "rewards/accuracies": 0.9375, "rewards/chosen": 0.06984328478574753, "rewards/margins": 0.07647494226694107, "rewards/margins_max": 0.11472654342651367, "rewards/margins_min": 0.038223348557949066, "rewards/margins_std": 0.05409592390060425, "rewards/rejected": -0.006631660275161266, "step": 450 }, { "epoch": 0.34, "grad_norm": 0.412109375, "learning_rate": 4.1829824680607104e-07, "logits/chosen": -0.07657341659069061, "logits/rejected": 0.6622528433799744, "logps/chosen": -279.7924499511719, "logps/rejected": -269.6399230957031, "loss": 0.6549, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.07170796394348145, "rewards/margins": 0.0784129947423935, "rewards/margins_max": 0.11032243072986603, "rewards/margins_min": 0.046503566205501556, "rewards/margins_std": 0.045126743614673615, "rewards/rejected": -0.0067050280049443245, "step": 460 }, { "epoch": 0.35, "grad_norm": 0.453125, "learning_rate": 4.134944652594794e-07, "logits/chosen": 0.0630350112915039, "logits/rejected": 0.6474081873893738, "logps/chosen": -261.22418212890625, "logps/rejected": -239.87588500976562, "loss": 0.6581, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.05761648342013359, "rewards/margins": 0.06358405947685242, "rewards/margins_max": 0.09130324423313141, "rewards/margins_min": 0.03586486726999283, "rewards/margins_std": 0.03920084983110428, "rewards/rejected": -0.0059675718657672405, "step": 470 }, { "epoch": 0.35, "grad_norm": 0.392578125, "learning_rate": 4.085828074220451e-07, "logits/chosen": -0.19398275017738342, "logits/rejected": 0.5935221910476685, "logps/chosen": -270.58892822265625, "logps/rejected": -245.28076171875, "loss": 0.6561, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.06964448094367981, "rewards/margins": 0.08161848783493042, "rewards/margins_max": 0.12005958706140518, "rewards/margins_min": 0.04317738488316536, "rewards/margins_std": 0.0543639250099659, "rewards/rejected": -0.011974002234637737, "step": 480 }, { "epoch": 0.36, "grad_norm": 0.373046875, "learning_rate": 4.035665140849994e-07, "logits/chosen": -0.14574551582336426, "logits/rejected": 0.5638002157211304, "logps/chosen": -274.3948669433594, "logps/rejected": -228.3678436279297, "loss": 0.6602, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.06451380997896194, "rewards/margins": 0.07035050541162491, "rewards/margins_max": 0.10534896701574326, "rewards/margins_min": 0.035352036356925964, "rewards/margins_std": 0.04949530214071274, "rewards/rejected": -0.005836695898324251, "step": 490 }, { "epoch": 0.37, "grad_norm": 0.470703125, "learning_rate": 3.984488950797678e-07, "logits/chosen": -0.040041010826826096, "logits/rejected": 0.7338499426841736, "logps/chosen": -265.7203063964844, "logps/rejected": -230.03170776367188, "loss": 0.6557, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.07825618237257004, "rewards/margins": 0.07796762883663177, "rewards/margins_max": 0.11045694351196289, "rewards/margins_min": 0.04547831788659096, "rewards/margins_std": 0.04594682529568672, "rewards/rejected": 0.00028855842538177967, "step": 500 }, { "epoch": 0.38, "grad_norm": 0.46484375, "learning_rate": 3.9323332709408904e-07, "logits/chosen": 0.02443014457821846, "logits/rejected": 0.604122519493103, "logps/chosen": -254.78701782226562, "logps/rejected": -233.02084350585938, "loss": 0.6547, "rewards/accuracies": 0.887499988079071, "rewards/chosen": 0.060421328991651535, "rewards/margins": 0.07197172939777374, "rewards/margins_max": 0.11490993201732635, "rewards/margins_min": 0.029033523052930832, "rewards/margins_std": 0.060723792761564255, "rewards/rejected": -0.01155039481818676, "step": 510 }, { "epoch": 0.38, "grad_norm": 0.40234375, "learning_rate": 3.879232514440227e-07, "logits/chosen": 0.04711779206991196, "logits/rejected": 0.9458179473876953, "logps/chosen": -280.1850280761719, "logps/rejected": -221.70785522460938, "loss": 0.6533, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.07137273252010345, "rewards/margins": 0.0817800909280777, "rewards/margins_max": 0.12524864077568054, "rewards/margins_min": 0.038311541080474854, "rewards/margins_std": 0.06147381663322449, "rewards/rejected": -0.01040735375136137, "step": 520 }, { "epoch": 0.39, "grad_norm": 0.51171875, "learning_rate": 3.825221718033129e-07, "logits/chosen": 0.09109713137149811, "logits/rejected": 0.7228142023086548, "logps/chosen": -256.89202880859375, "logps/rejected": -236.75460815429688, "loss": 0.6552, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.06750547885894775, "rewards/margins": 0.0867571011185646, "rewards/margins_max": 0.12318144738674164, "rewards/margins_min": 0.050332747399806976, "rewards/margins_std": 0.05151181295514107, "rewards/rejected": -0.019251624122262, "step": 530 }, { "epoch": 0.4, "grad_norm": 0.375, "learning_rate": 3.7703365189160746e-07, "logits/chosen": -0.09472191333770752, "logits/rejected": 0.6297842264175415, "logps/chosen": -261.3214416503906, "logps/rejected": -244.3570556640625, "loss": 0.6562, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.07096521556377411, "rewards/margins": 0.07419043034315109, "rewards/margins_max": 0.1066773533821106, "rewards/margins_min": 0.041703514754772186, "rewards/margins_std": 0.04594343155622482, "rewards/rejected": -0.0032252087257802486, "step": 540 }, { "epoch": 0.4, "grad_norm": 0.396484375, "learning_rate": 3.714613131230587e-07, "logits/chosen": -0.002119688782840967, "logits/rejected": 0.6269344091415405, "logps/chosen": -249.4778594970703, "logps/rejected": -223.02603149414062, "loss": 0.6572, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.06649978458881378, "rewards/margins": 0.07572688907384872, "rewards/margins_max": 0.10936542600393295, "rewards/margins_min": 0.0420883484184742, "rewards/margins_std": 0.04757208377122879, "rewards/rejected": -0.009227104485034943, "step": 550 }, { "epoch": 0.41, "grad_norm": 0.427734375, "learning_rate": 3.6580883221685533e-07, "logits/chosen": 0.05928317829966545, "logits/rejected": 0.6542156338691711, "logps/chosen": -249.701171875, "logps/rejected": -235.28256225585938, "loss": 0.6546, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.06236664578318596, "rewards/margins": 0.07323845475912094, "rewards/margins_max": 0.11050041764974594, "rewards/margins_min": 0.03597649186849594, "rewards/margins_std": 0.05269637703895569, "rewards/rejected": -0.010871809907257557, "step": 560 }, { "epoch": 0.42, "grad_norm": 0.39453125, "learning_rate": 3.6007993877126386e-07, "logits/chosen": -0.15580764412879944, "logits/rejected": 0.7158504724502563, "logps/chosen": -291.7669372558594, "logps/rejected": -232.95767211914062, "loss": 0.6561, "rewards/accuracies": 0.887499988079071, "rewards/chosen": 0.06417657434940338, "rewards/margins": 0.07455410063266754, "rewards/margins_max": 0.11633304506540298, "rewards/margins_min": 0.032775141298770905, "rewards/margins_std": 0.05908435583114624, "rewards/rejected": -0.010377511382102966, "step": 570 }, { "epoch": 0.43, "grad_norm": 0.490234375, "learning_rate": 3.5427841280277937e-07, "logits/chosen": 0.07802899181842804, "logits/rejected": 0.6800590753555298, "logps/chosen": -253.85107421875, "logps/rejected": -234.7869110107422, "loss": 0.6586, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": 0.058535121381282806, "rewards/margins": 0.06843312084674835, "rewards/margins_max": 0.10552020370960236, "rewards/margins_min": 0.03134603053331375, "rewards/margins_std": 0.0524490587413311, "rewards/rejected": -0.0098979901522398, "step": 580 }, { "epoch": 0.43, "grad_norm": 0.482421875, "learning_rate": 3.484080822520096e-07, "logits/chosen": -0.049510225653648376, "logits/rejected": 0.7604032754898071, "logps/chosen": -268.8267822265625, "logps/rejected": -224.325439453125, "loss": 0.654, "rewards/accuracies": 0.9375, "rewards/chosen": 0.07341635227203369, "rewards/margins": 0.07777970284223557, "rewards/margins_max": 0.11911604553461075, "rewards/margins_min": 0.03644336014986038, "rewards/margins_std": 0.058458417654037476, "rewards/rejected": -0.004363355692476034, "step": 590 }, { "epoch": 0.44, "grad_norm": 0.400390625, "learning_rate": 3.4247282045793797e-07, "logits/chosen": -0.19325688481330872, "logits/rejected": 0.6270472407341003, "logps/chosen": -282.7298889160156, "logps/rejected": -252.6767578125, "loss": 0.6482, "rewards/accuracies": 0.987500011920929, "rewards/chosen": 0.07601656764745712, "rewards/margins": 0.09246112406253815, "rewards/margins_max": 0.13728852570056915, "rewards/margins_min": 0.04763372614979744, "rewards/margins_std": 0.06339551508426666, "rewards/rejected": -0.016444550827145576, "step": 600 }, { "epoch": 0.45, "grad_norm": 0.458984375, "learning_rate": 3.3647654360223144e-07, "logits/chosen": -0.05807774141430855, "logits/rejected": 0.6809174418449402, "logps/chosen": -299.19000244140625, "logps/rejected": -262.0146179199219, "loss": 0.6554, "rewards/accuracies": 0.9375, "rewards/chosen": 0.06890231370925903, "rewards/margins": 0.07707327604293823, "rewards/margins_max": 0.11320241540670395, "rewards/margins_min": 0.04094412922859192, "rewards/margins_std": 0.05109432339668274, "rewards/rejected": -0.0081709586083889, "step": 610 }, { "epoch": 0.46, "grad_norm": 0.4609375, "learning_rate": 3.30423208125281e-07, "logits/chosen": -0.15423426032066345, "logits/rejected": 0.6746962070465088, "logps/chosen": -361.3031921386719, "logps/rejected": -271.8617858886719, "loss": 0.6527, "rewards/accuracies": 0.987500011920929, "rewards/chosen": 0.08182433992624283, "rewards/margins": 0.08147110044956207, "rewards/margins_max": 0.1197957769036293, "rewards/margins_min": 0.04314642399549484, "rewards/margins_std": 0.05419927090406418, "rewards/rejected": 0.00035324436612427235, "step": 620 }, { "epoch": 0.46, "grad_norm": 0.53515625, "learning_rate": 3.2431680811567833e-07, "logits/chosen": -0.18638429045677185, "logits/rejected": 0.5596809387207031, "logps/chosen": -265.38189697265625, "logps/rejected": -216.4676513671875, "loss": 0.6498, "rewards/accuracies": 0.9375, "rewards/chosen": 0.07413068413734436, "rewards/margins": 0.08364250510931015, "rewards/margins_max": 0.11661893129348755, "rewards/margins_min": 0.05066608637571335, "rewards/margins_std": 0.046635713428258896, "rewards/rejected": -0.009511834010481834, "step": 630 }, { "epoch": 0.47, "grad_norm": 0.4765625, "learning_rate": 3.1816137267485136e-07, "logits/chosen": -0.01266663521528244, "logits/rejected": 0.7700913548469543, "logps/chosen": -270.8904113769531, "logps/rejected": -227.817138671875, "loss": 0.6499, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.08034569025039673, "rewards/margins": 0.09530878067016602, "rewards/margins_max": 0.13598743081092834, "rewards/margins_min": 0.05463013797998428, "rewards/margins_std": 0.0575282946228981, "rewards/rejected": -0.014963088557124138, "step": 640 }, { "epoch": 0.48, "grad_norm": 0.458984375, "learning_rate": 3.1196096325859815e-07, "logits/chosen": -0.09016970545053482, "logits/rejected": 0.5758002400398254, "logps/chosen": -248.17721557617188, "logps/rejected": -231.0363311767578, "loss": 0.6539, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.06288131326436996, "rewards/margins": 0.08216816931962967, "rewards/margins_max": 0.12130717188119888, "rewards/margins_min": 0.04302916303277016, "rewards/margins_std": 0.05535091087222099, "rewards/rejected": -0.019286850467324257, "step": 650 }, { "epoch": 0.49, "grad_norm": 0.357421875, "learning_rate": 3.057196709972727e-07, "logits/chosen": -0.06700615584850311, "logits/rejected": 0.7238128781318665, "logps/chosen": -273.21044921875, "logps/rejected": -222.5702667236328, "loss": 0.6541, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.07693549245595932, "rewards/margins": 0.08857451379299164, "rewards/margins_max": 0.13141193985939026, "rewards/margins_min": 0.045737091451883316, "rewards/margins_std": 0.0605812594294548, "rewards/rejected": -0.011639016680419445, "step": 660 }, { "epoch": 0.49, "grad_norm": 0.384765625, "learning_rate": 2.9944161399639086e-07, "logits/chosen": -0.17069509625434875, "logits/rejected": 0.5526743531227112, "logps/chosen": -274.2879943847656, "logps/rejected": -238.9512176513672, "loss": 0.6554, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.07316674292087555, "rewards/margins": 0.0783548504114151, "rewards/margins_max": 0.12053277343511581, "rewards/margins_min": 0.03617694973945618, "rewards/margins_std": 0.059648580849170685, "rewards/rejected": -0.0051881177350878716, "step": 670 }, { "epoch": 0.5, "grad_norm": 0.4453125, "learning_rate": 2.9313093461943824e-07, "logits/chosen": -0.16692019999027252, "logits/rejected": 0.6132463216781616, "logps/chosen": -266.98297119140625, "logps/rejected": -240.8970489501953, "loss": 0.6514, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.0738702192902565, "rewards/margins": 0.08596795052289963, "rewards/margins_max": 0.12484346330165863, "rewards/margins_min": 0.04709243029356003, "rewards/margins_std": 0.05497828125953674, "rewards/rejected": -0.012097733095288277, "step": 680 }, { "epoch": 0.51, "grad_norm": 0.404296875, "learning_rate": 2.8679179675467104e-07, "logits/chosen": -0.028652098029851913, "logits/rejected": 0.7112724781036377, "logps/chosen": -263.5270080566406, "logps/rejected": -232.5438232421875, "loss": 0.6505, "rewards/accuracies": 0.9375, "rewards/chosen": 0.07920118421316147, "rewards/margins": 0.08954565227031708, "rewards/margins_max": 0.13445377349853516, "rewards/margins_min": 0.04463755339384079, "rewards/margins_std": 0.06350965052843094, "rewards/rejected": -0.010344477370381355, "step": 690 }, { "epoch": 0.52, "grad_norm": 0.44140625, "learning_rate": 2.80428383067716e-07, "logits/chosen": 0.04036666825413704, "logits/rejected": 0.5802733302116394, "logps/chosen": -276.1773681640625, "logps/rejected": -284.2413024902344, "loss": 0.6492, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.07053406536579132, "rewards/margins": 0.09340489655733109, "rewards/margins_max": 0.13880935311317444, "rewards/margins_min": 0.048000432550907135, "rewards/margins_std": 0.06421160697937012, "rewards/rejected": -0.022870829328894615, "step": 700 }, { "epoch": 0.52, "grad_norm": 0.484375, "learning_rate": 2.7404489224177973e-07, "logits/chosen": -0.10626095533370972, "logits/rejected": 0.6656783819198608, "logps/chosen": -273.10369873046875, "logps/rejected": -240.2863006591797, "loss": 0.6555, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.06227077171206474, "rewards/margins": 0.07790277898311615, "rewards/margins_max": 0.12013592571020126, "rewards/margins_min": 0.03566960617899895, "rewards/margins_std": 0.05972672253847122, "rewards/rejected": -0.015631992369890213, "step": 710 }, { "epoch": 0.53, "grad_norm": 0.392578125, "learning_rate": 2.676455362072894e-07, "logits/chosen": -0.09553918987512589, "logits/rejected": 0.7383521795272827, "logps/chosen": -302.948974609375, "logps/rejected": -269.911376953125, "loss": 0.6506, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.08486850559711456, "rewards/margins": 0.10140830278396606, "rewards/margins_max": 0.14273163676261902, "rewards/margins_min": 0.06008495017886162, "rewards/margins_std": 0.05844002962112427, "rewards/rejected": -0.016539786010980606, "step": 720 }, { "epoch": 0.54, "grad_norm": 0.37890625, "learning_rate": 2.612345373627937e-07, "logits/chosen": 0.11825668811798096, "logits/rejected": 0.7459693551063538, "logps/chosen": -275.9749450683594, "logps/rejected": -238.9975128173828, "loss": 0.6551, "rewards/accuracies": 0.9375, "rewards/chosen": 0.06423775851726532, "rewards/margins": 0.08155209571123123, "rewards/margins_max": 0.12171584367752075, "rewards/margins_min": 0.04138834401965141, "rewards/margins_std": 0.05680011585354805, "rewards/rejected": -0.017314324155449867, "step": 730 }, { "epoch": 0.54, "grad_norm": 0.40234375, "learning_rate": 2.54816125788955e-07, "logits/chosen": -0.04539821669459343, "logits/rejected": 0.8194332122802734, "logps/chosen": -290.15557861328125, "logps/rejected": -230.5435333251953, "loss": 0.6515, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.074457548558712, "rewards/margins": 0.08972884714603424, "rewards/margins_max": 0.13167297840118408, "rewards/margins_min": 0.04778473079204559, "rewards/margins_std": 0.059317946434020996, "rewards/rejected": -0.015271296724677086, "step": 740 }, { "epoch": 0.55, "grad_norm": 0.38671875, "learning_rate": 2.4839453645747467e-07, "logits/chosen": 0.04359906539320946, "logits/rejected": 0.622580349445343, "logps/chosen": -276.761474609375, "logps/rejected": -252.15261840820312, "loss": 0.6543, "rewards/accuracies": 0.987500011920929, "rewards/chosen": 0.07819774001836777, "rewards/margins": 0.0873616486787796, "rewards/margins_max": 0.14198613166809082, "rewards/margins_min": 0.03273715078830719, "rewards/margins_std": 0.07725068926811218, "rewards/rejected": -0.009163893759250641, "step": 750 }, { "epoch": 0.56, "grad_norm": 0.33984375, "learning_rate": 2.4197400643678987e-07, "logits/chosen": 0.03581676632165909, "logits/rejected": 0.8077651858329773, "logps/chosen": -280.2861328125, "logps/rejected": -233.9840545654297, "loss": 0.6528, "rewards/accuracies": 0.987500011920929, "rewards/chosen": 0.07635350525379181, "rewards/margins": 0.09026426076889038, "rewards/margins_max": 0.1272718459367752, "rewards/margins_min": 0.05325668305158615, "rewards/margins_std": 0.05233662202954292, "rewards/rejected": -0.013910764828324318, "step": 760 }, { "epoch": 0.57, "grad_norm": 0.435546875, "learning_rate": 2.3555877209638726e-07, "logits/chosen": -0.12947851419448853, "logits/rejected": 0.7489079833030701, "logps/chosen": -290.2663879394531, "logps/rejected": -254.322021484375, "loss": 0.6506, "rewards/accuracies": 0.9375, "rewards/chosen": 0.0825168713927269, "rewards/margins": 0.08864180743694305, "rewards/margins_max": 0.12674041092395782, "rewards/margins_min": 0.050543200224637985, "rewards/margins_std": 0.05387955904006958, "rewards/rejected": -0.006124935112893581, "step": 770 }, { "epoch": 0.57, "grad_norm": 0.390625, "learning_rate": 2.2915306631157817e-07, "logits/chosen": -0.04638870805501938, "logits/rejected": 0.6960999965667725, "logps/chosen": -238.04745483398438, "logps/rejected": -215.9326629638672, "loss": 0.6497, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.06742905080318451, "rewards/margins": 0.08723022788763046, "rewards/margins_max": 0.13323740661144257, "rewards/margins_min": 0.04122306779026985, "rewards/margins_std": 0.0650639608502388, "rewards/rejected": -0.019801190122961998, "step": 780 }, { "epoch": 0.58, "grad_norm": 0.451171875, "learning_rate": 2.2276111567057887e-07, "logits/chosen": 0.19736722111701965, "logits/rejected": 0.8112510442733765, "logps/chosen": -263.0642395019531, "logps/rejected": -247.5417022705078, "loss": 0.6481, "rewards/accuracies": 0.9375, "rewards/chosen": 0.07030580937862396, "rewards/margins": 0.08302759379148483, "rewards/margins_max": 0.12399895489215851, "rewards/margins_min": 0.042056240141391754, "rewards/margins_std": 0.05794224888086319, "rewards/rejected": -0.01272178627550602, "step": 790 }, { "epoch": 0.59, "grad_norm": 0.474609375, "learning_rate": 2.1638713768573936e-07, "logits/chosen": 0.0460067018866539, "logits/rejected": 0.6669069528579712, "logps/chosen": -227.3536834716797, "logps/rejected": -220.83984375, "loss": 0.6476, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.0669163167476654, "rewards/margins": 0.07873805612325668, "rewards/margins_max": 0.12541484832763672, "rewards/margins_min": 0.03206125646829605, "rewards/margins_std": 0.0660109594464302, "rewards/rejected": -0.011821739375591278, "step": 800 }, { "epoch": 0.6, "grad_norm": 0.484375, "learning_rate": 2.100353380107609e-07, "logits/chosen": -0.007617829833179712, "logits/rejected": 0.7070167660713196, "logps/chosen": -271.02691650390625, "logps/rejected": -219.64944458007812, "loss": 0.655, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.05975126102566719, "rewards/margins": 0.07805275171995163, "rewards/margins_max": 0.11311063915491104, "rewards/margins_min": 0.04299486428499222, "rewards/margins_std": 0.04957933351397514, "rewards/rejected": -0.01830149069428444, "step": 810 }, { "epoch": 0.6, "grad_norm": 0.400390625, "learning_rate": 2.0370990766573698e-07, "logits/chosen": 0.11678284406661987, "logits/rejected": 0.8694367408752441, "logps/chosen": -319.4522705078125, "logps/rejected": -231.9202423095703, "loss": 0.653, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.060270898044109344, "rewards/margins": 0.08039253205060959, "rewards/margins_max": 0.11875177919864655, "rewards/margins_min": 0.042033277451992035, "rewards/margins_std": 0.054248176515102386, "rewards/rejected": -0.020121637731790543, "step": 820 }, { "epoch": 0.61, "grad_norm": 0.357421875, "learning_rate": 1.974150202718513e-07, "logits/chosen": 0.037704307585954666, "logits/rejected": 0.6765593886375427, "logps/chosen": -252.72128295898438, "logps/rejected": -251.72232055664062, "loss": 0.6551, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.06962188333272934, "rewards/margins": 0.07926660031080246, "rewards/margins_max": 0.11337349563837051, "rewards/margins_min": 0.0451597236096859, "rewards/margins_std": 0.04823441430926323, "rewards/rejected": -0.009644721634685993, "step": 830 }, { "epoch": 0.62, "grad_norm": 0.4296875, "learning_rate": 1.9115482929755445e-07, "logits/chosen": 0.21195778250694275, "logits/rejected": 0.8628422021865845, "logps/chosen": -255.17141723632812, "logps/rejected": -230.54837036132812, "loss": 0.6542, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.05505312606692314, "rewards/margins": 0.08028857409954071, "rewards/margins_max": 0.12079598009586334, "rewards/margins_min": 0.03978116437792778, "rewards/margins_std": 0.05728612467646599, "rewards/rejected": -0.025235449895262718, "step": 840 }, { "epoch": 0.63, "grad_norm": 0.451171875, "learning_rate": 1.8493346531803887e-07, "logits/chosen": -0.0924936980009079, "logits/rejected": 0.6660225987434387, "logps/chosen": -253.78762817382812, "logps/rejected": -230.6340789794922, "loss": 0.654, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.07504816353321075, "rewards/margins": 0.08388366550207138, "rewards/margins_max": 0.13147205114364624, "rewards/margins_min": 0.03629531338810921, "rewards/margins_std": 0.06730011850595474, "rewards/rejected": -0.00883551873266697, "step": 850 }, { "epoch": 0.63, "grad_norm": 0.62890625, "learning_rate": 1.7875503328981807e-07, "logits/chosen": 0.03313580900430679, "logits/rejected": 0.7846869826316833, "logps/chosen": -289.10919189453125, "logps/rejected": -260.46209716796875, "loss": 0.651, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.07691006362438202, "rewards/margins": 0.09373887628316879, "rewards/margins_max": 0.14286434650421143, "rewards/margins_min": 0.04461339861154556, "rewards/margins_std": 0.06947391480207443, "rewards/rejected": -0.01682879962027073, "step": 860 }, { "epoch": 0.64, "grad_norm": 0.4765625, "learning_rate": 1.7262360984221006e-07, "logits/chosen": 0.07170330733060837, "logits/rejected": 0.585077166557312, "logps/chosen": -232.71224975585938, "logps/rejected": -241.5399932861328, "loss": 0.649, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.06494490802288055, "rewards/margins": 0.082070492208004, "rewards/margins_max": 0.11704879999160767, "rewards/margins_min": 0.04709218442440033, "rewards/margins_std": 0.049466803669929504, "rewards/rejected": -0.017125584185123444, "step": 870 }, { "epoch": 0.65, "grad_norm": 0.46875, "learning_rate": 1.6654324058751175e-07, "logits/chosen": -0.03041740693151951, "logits/rejected": 0.655995786190033, "logps/chosen": -307.1095275878906, "logps/rejected": -223.6023712158203, "loss": 0.6511, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.05577283352613449, "rewards/margins": 0.0676901713013649, "rewards/margins_max": 0.10265400260686874, "rewards/margins_min": 0.03272632881999016, "rewards/margins_std": 0.04944632947444916, "rewards/rejected": -0.011917328462004662, "step": 880 }, { "epoch": 0.65, "grad_norm": 0.439453125, "learning_rate": 1.6051793745163812e-07, "logits/chosen": -0.03592555597424507, "logits/rejected": 0.5853763222694397, "logps/chosen": -272.7641296386719, "logps/rejected": -243.679443359375, "loss": 0.6543, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.07209287583827972, "rewards/margins": 0.0827474445104599, "rewards/margins_max": 0.11978437751531601, "rewards/margins_min": 0.045710500329732895, "rewards/margins_std": 0.05237815901637077, "rewards/rejected": -0.010654577054083347, "step": 890 }, { "epoch": 0.66, "grad_norm": 0.435546875, "learning_rate": 1.5455167602698915e-07, "logits/chosen": 0.10257561504840851, "logits/rejected": 0.7078714370727539, "logps/chosen": -239.56201171875, "logps/rejected": -240.7050323486328, "loss": 0.6508, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.06524372100830078, "rewards/margins": 0.08618183434009552, "rewards/margins_max": 0.12894438207149506, "rewards/margins_min": 0.043419282883405685, "rewards/margins_std": 0.060475386679172516, "rewards/rejected": -0.020938115194439888, "step": 900 }, { "epoch": 0.67, "grad_norm": 0.5, "learning_rate": 1.4864839294928924e-07, "logits/chosen": 0.03240091726183891, "logits/rejected": 0.7821232080459595, "logps/chosen": -283.28338623046875, "logps/rejected": -228.65771484375, "loss": 0.6537, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.06496749818325043, "rewards/margins": 0.08208582550287247, "rewards/margins_max": 0.12210579216480255, "rewards/margins_min": 0.04206586629152298, "rewards/margins_std": 0.056596774607896805, "rewards/rejected": -0.01711832545697689, "step": 910 }, { "epoch": 0.68, "grad_norm": 0.515625, "learning_rate": 1.428119833001315e-07, "logits/chosen": -0.23715606331825256, "logits/rejected": 0.6829497218132019, "logps/chosen": -352.04071044921875, "logps/rejected": -288.7782287597656, "loss": 0.6512, "rewards/accuracies": 0.987500011920929, "rewards/chosen": 0.07856696844100952, "rewards/margins": 0.08786121755838394, "rewards/margins_max": 0.1261076182126999, "rewards/margins_min": 0.049614839255809784, "rewards/margins_std": 0.05408855527639389, "rewards/rejected": -0.009294259361922741, "step": 920 }, { "epoch": 0.68, "grad_norm": 0.41796875, "learning_rate": 1.370462980369401e-07, "logits/chosen": 0.029560232535004616, "logits/rejected": 0.729602038860321, "logps/chosen": -257.4678955078125, "logps/rejected": -246.0548858642578, "loss": 0.6487, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.0831000953912735, "rewards/margins": 0.09744415432214737, "rewards/margins_max": 0.15836606919765472, "rewards/margins_min": 0.03652223199605942, "rewards/margins_std": 0.08615661412477493, "rewards/rejected": -0.014344056136906147, "step": 930 }, { "epoch": 0.69, "grad_norm": 0.4453125, "learning_rate": 1.3135514145204606e-07, "logits/chosen": 0.046013571321964264, "logits/rejected": 0.5954084396362305, "logps/chosen": -269.13629150390625, "logps/rejected": -249.99008178710938, "loss": 0.6511, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.06792394816875458, "rewards/margins": 0.08472023159265518, "rewards/margins_max": 0.12758344411849976, "rewards/margins_min": 0.0418570414185524, "rewards/margins_std": 0.06061771512031555, "rewards/rejected": -0.016796285286545753, "step": 940 }, { "epoch": 0.7, "grad_norm": 0.515625, "learning_rate": 1.257422686625539e-07, "logits/chosen": -0.12281881272792816, "logits/rejected": 0.7430060505867004, "logps/chosen": -328.2705383300781, "logps/rejected": -239.69070434570312, "loss": 0.6552, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.06796546280384064, "rewards/margins": 0.08045408874750137, "rewards/margins_max": 0.1251390427350998, "rewards/margins_min": 0.03576914221048355, "rewards/margins_std": 0.06319407373666763, "rewards/rejected": -0.01248864270746708, "step": 950 }, { "epoch": 0.71, "grad_norm": 0.49609375, "learning_rate": 1.2021138313265444e-07, "logits/chosen": -0.05693725496530533, "logits/rejected": 0.5191583633422852, "logps/chosen": -267.0467224121094, "logps/rejected": -253.3563995361328, "loss": 0.6547, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.06574182212352753, "rewards/margins": 0.0813496932387352, "rewards/margins_max": 0.11537833511829376, "rewards/margins_min": 0.04732104018330574, "rewards/margins_std": 0.048123784363269806, "rewards/rejected": -0.015607871115207672, "step": 960 }, { "epoch": 0.71, "grad_norm": 0.421875, "learning_rate": 1.1476613423001974e-07, "logits/chosen": 0.01074630580842495, "logits/rejected": 0.7058780789375305, "logps/chosen": -268.10162353515625, "logps/rejected": -246.88095092773438, "loss": 0.6503, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.06480704247951508, "rewards/margins": 0.07971958816051483, "rewards/margins_max": 0.12575621902942657, "rewards/margins_min": 0.03368296101689339, "rewards/margins_std": 0.0651056244969368, "rewards/rejected": -0.014912550337612629, "step": 970 }, { "epoch": 0.72, "grad_norm": 0.447265625, "learning_rate": 1.0941011481789042e-07, "logits/chosen": 0.16233515739440918, "logits/rejected": 0.6619356274604797, "logps/chosen": -268.57537841796875, "logps/rejected": -248.046142578125, "loss": 0.6547, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.061273593455553055, "rewards/margins": 0.07566945254802704, "rewards/margins_max": 0.1124287024140358, "rewards/margins_min": 0.03891021013259888, "rewards/margins_std": 0.051985424011945724, "rewards/rejected": -0.014395867474377155, "step": 980 }, { "epoch": 0.73, "grad_norm": 0.4296875, "learning_rate": 1.041468588844476e-07, "logits/chosen": 0.021765699610114098, "logits/rejected": 0.5723878145217896, "logps/chosen": -252.99365234375, "logps/rejected": -237.05233764648438, "loss": 0.6506, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.06569187343120575, "rewards/margins": 0.08357974141836166, "rewards/margins_max": 0.12318392097949982, "rewards/margins_min": 0.043975573033094406, "rewards/margins_std": 0.056008756160736084, "rewards/rejected": -0.017887871712446213, "step": 990 }, { "epoch": 0.74, "grad_norm": 0.474609375, "learning_rate": 9.897983921102954e-08, "logits/chosen": -0.08432263880968094, "logits/rejected": 0.5865111351013184, "logps/chosen": -265.72735595703125, "logps/rejected": -234.95028686523438, "loss": 0.6572, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.060374390333890915, "rewards/margins": 0.070238396525383, "rewards/margins_max": 0.10305947065353394, "rewards/margins_min": 0.037417326122522354, "rewards/margins_std": 0.04641600325703621, "rewards/rejected": -0.009864002466201782, "step": 1000 }, { "epoch": 0.74, "grad_norm": 0.42578125, "learning_rate": 9.391246508073433e-08, "logits/chosen": -0.17238160967826843, "logits/rejected": 0.6868064999580383, "logps/chosen": -321.6485595703125, "logps/rejected": -263.62188720703125, "loss": 0.6505, "rewards/accuracies": 0.9375, "rewards/chosen": 0.07812955230474472, "rewards/margins": 0.09409363567829132, "rewards/margins_max": 0.13336913287639618, "rewards/margins_min": 0.054818131029605865, "rewards/margins_std": 0.05554394796490669, "rewards/rejected": -0.015964074060320854, "step": 1010 }, { "epoch": 0.75, "grad_norm": 0.408203125, "learning_rate": 8.894808002892037e-08, "logits/chosen": -0.048636335879564285, "logits/rejected": 0.6718664765357971, "logps/chosen": -286.724365234375, "logps/rejected": -256.17767333984375, "loss": 0.6521, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.07918758690357208, "rewards/margins": 0.09503775835037231, "rewards/margins_max": 0.1390329748392105, "rewards/margins_min": 0.05104254558682442, "rewards/margins_std": 0.06221862882375717, "rewards/rejected": -0.015850163996219635, "step": 1020 }, { "epoch": 0.76, "grad_norm": 0.423828125, "learning_rate": 8.408995963708756e-08, "logits/chosen": -0.03906983509659767, "logits/rejected": 0.7512571215629578, "logps/chosen": -289.33660888671875, "logps/rejected": -260.1453857421875, "loss": 0.6546, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.07737907022237778, "rewards/margins": 0.08742411434650421, "rewards/margins_max": 0.12180934101343155, "rewards/margins_min": 0.053038887679576874, "rewards/margins_std": 0.048628050833940506, "rewards/rejected": -0.010045038536190987, "step": 1030 }, { "epoch": 0.77, "grad_norm": 0.484375, "learning_rate": 7.934130937159508e-08, "logits/chosen": 0.0004897325998172164, "logits/rejected": 0.7238715887069702, "logps/chosen": -279.21624755859375, "logps/rejected": -248.09957885742188, "loss": 0.6519, "rewards/accuracies": 0.987500011920929, "rewards/chosen": 0.07424648851156235, "rewards/margins": 0.08449728786945343, "rewards/margins_max": 0.1293594241142273, "rewards/margins_min": 0.039635173976421356, "rewards/margins_std": 0.06344461441040039, "rewards/rejected": -0.010250799357891083, "step": 1040 }, { "epoch": 0.77, "grad_norm": 0.455078125, "learning_rate": 7.470526246864364e-08, "logits/chosen": -0.028243955224752426, "logits/rejected": 0.6930115818977356, "logps/chosen": -301.51458740234375, "logps/rejected": -251.8220672607422, "loss": 0.6559, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.06157411262392998, "rewards/margins": 0.07462558895349503, "rewards/margins_max": 0.10729455947875977, "rewards/margins_min": 0.04195661470293999, "rewards/margins_std": 0.04620090872049332, "rewards/rejected": -0.01305147260427475, "step": 1050 }, { "epoch": 0.78, "grad_norm": 0.494140625, "learning_rate": 7.018487786691512e-08, "logits/chosen": -0.18514272570610046, "logits/rejected": 0.8259115219116211, "logps/chosen": -284.2102355957031, "logps/rejected": -236.52206420898438, "loss": 0.6474, "rewards/accuracies": 0.987500011920929, "rewards/chosen": 0.09142851829528809, "rewards/margins": 0.10867790132761002, "rewards/margins_max": 0.16026386618614197, "rewards/margins_min": 0.05709192156791687, "rewards/margins_std": 0.07295358926057816, "rewards/rejected": -0.01724938116967678, "step": 1060 }, { "epoch": 0.79, "grad_norm": 0.4296875, "learning_rate": 6.578313818923559e-08, "logits/chosen": 0.12218723446130753, "logits/rejected": 0.6379785537719727, "logps/chosen": -238.366943359375, "logps/rejected": -233.61026000976562, "loss": 0.6535, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.055001966655254364, "rewards/margins": 0.06915679574012756, "rewards/margins_max": 0.09845878183841705, "rewards/margins_min": 0.039854805916547775, "rewards/margins_std": 0.04143926501274109, "rewards/rejected": -0.014154826290905476, "step": 1070 }, { "epoch": 0.79, "grad_norm": 0.40234375, "learning_rate": 6.15029477745925e-08, "logits/chosen": -0.1947169005870819, "logits/rejected": 0.7179273366928101, "logps/chosen": -295.55096435546875, "logps/rejected": -222.1858367919922, "loss": 0.6512, "rewards/accuracies": 0.9375, "rewards/chosen": 0.07841251790523529, "rewards/margins": 0.08169189840555191, "rewards/margins_max": 0.11758895218372345, "rewards/margins_min": 0.045794855803251266, "rewards/margins_std": 0.05076608806848526, "rewards/rejected": -0.003279387950897217, "step": 1080 }, { "epoch": 0.8, "grad_norm": 0.44921875, "learning_rate": 5.734713076180486e-08, "logits/chosen": 0.006981179118156433, "logits/rejected": 0.7452491521835327, "logps/chosen": -260.1736145019531, "logps/rejected": -224.355224609375, "loss": 0.6557, "rewards/accuracies": 0.987500011920929, "rewards/chosen": 0.06287099421024323, "rewards/margins": 0.07665695995092392, "rewards/margins_max": 0.11319824308156967, "rewards/margins_min": 0.04011568799614906, "rewards/margins_std": 0.05167717486619949, "rewards/rejected": -0.013785977847874165, "step": 1090 }, { "epoch": 0.81, "grad_norm": 0.48046875, "learning_rate": 5.3318429226110875e-08, "logits/chosen": 0.015549438074231148, "logits/rejected": 0.6923533082008362, "logps/chosen": -269.09747314453125, "logps/rejected": -236.7939910888672, "loss": 0.6512, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.06426068395376205, "rewards/margins": 0.08864103257656097, "rewards/margins_max": 0.12798473238945007, "rewards/margins_min": 0.04929735139012337, "rewards/margins_std": 0.05564037710428238, "rewards/rejected": -0.024380357936024666, "step": 1100 }, { "epoch": 0.82, "grad_norm": 0.419921875, "learning_rate": 4.9419501369902026e-08, "logits/chosen": -0.029879633337259293, "logits/rejected": 0.7326146960258484, "logps/chosen": -260.98968505859375, "logps/rejected": -220.9407958984375, "loss": 0.6499, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.07826592028141022, "rewards/margins": 0.09226210415363312, "rewards/margins_max": 0.1359224021434784, "rewards/margins_min": 0.04860180616378784, "rewards/margins_std": 0.06174498051404953, "rewards/rejected": -0.0139961838722229, "step": 1110 }, { "epoch": 0.82, "grad_norm": 0.42578125, "learning_rate": 4.5652919768798896e-08, "logits/chosen": -0.1553410142660141, "logits/rejected": 0.6449904441833496, "logps/chosen": -280.3263244628906, "logps/rejected": -242.4776153564453, "loss": 0.6512, "rewards/accuracies": 0.987500011920929, "rewards/chosen": 0.074976846575737, "rewards/margins": 0.08818279206752777, "rewards/margins_max": 0.12042693048715591, "rewards/margins_min": 0.05593867227435112, "rewards/margins_std": 0.04560008645057678, "rewards/rejected": -0.013205950148403645, "step": 1120 }, { "epoch": 0.83, "grad_norm": 0.474609375, "learning_rate": 4.2021169674223536e-08, "logits/chosen": -0.12581007182598114, "logits/rejected": 0.6627474427223206, "logps/chosen": -274.0786437988281, "logps/rejected": -250.4473114013672, "loss": 0.6483, "rewards/accuracies": 0.987500011920929, "rewards/chosen": 0.0676179900765419, "rewards/margins": 0.08669795840978622, "rewards/margins_max": 0.12913702428340912, "rewards/margins_min": 0.04425888508558273, "rewards/margins_std": 0.060017913579940796, "rewards/rejected": -0.01907997578382492, "step": 1130 }, { "epoch": 0.84, "grad_norm": 0.369140625, "learning_rate": 3.852664737359046e-08, "logits/chosen": 0.08468042314052582, "logits/rejected": 0.7667611241340637, "logps/chosen": -249.0133819580078, "logps/rejected": -232.16561889648438, "loss": 0.6497, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.07480134069919586, "rewards/margins": 0.09523622691631317, "rewards/margins_max": 0.15331092476844788, "rewards/margins_min": 0.037161558866500854, "rewards/margins_std": 0.08212999999523163, "rewards/rejected": -0.020434897392988205, "step": 1140 }, { "epoch": 0.85, "grad_norm": 0.373046875, "learning_rate": 3.5171658609197824e-08, "logits/chosen": -0.03485478088259697, "logits/rejected": 0.5982181429862976, "logps/chosen": -252.3005828857422, "logps/rejected": -240.3730926513672, "loss": 0.6515, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.06789231300354004, "rewards/margins": 0.0819520652294159, "rewards/margins_max": 0.11909864842891693, "rewards/margins_min": 0.04480547830462456, "rewards/margins_std": 0.05253320187330246, "rewards/rejected": -0.014059755019843578, "step": 1150 }, { "epoch": 0.85, "grad_norm": 0.435546875, "learning_rate": 3.195841705686139e-08, "logits/chosen": 0.001758894301019609, "logits/rejected": 0.6828774213790894, "logps/chosen": -259.75640869140625, "logps/rejected": -243.09970092773438, "loss": 0.6517, "rewards/accuracies": 0.987500011920929, "rewards/chosen": 0.06515143811702728, "rewards/margins": 0.07973488420248032, "rewards/margins_max": 0.11704728752374649, "rewards/margins_min": 0.042422495782375336, "rewards/margins_std": 0.05276770144701004, "rewards/rejected": -0.014583444222807884, "step": 1160 }, { "epoch": 0.86, "grad_norm": 0.4375, "learning_rate": 2.8889042865294837e-08, "logits/chosen": 0.0009852380026131868, "logits/rejected": 0.7770395278930664, "logps/chosen": -252.7977752685547, "logps/rejected": -211.03494262695312, "loss": 0.652, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.06879086047410965, "rewards/margins": 0.0823785737156868, "rewards/margins_max": 0.126399964094162, "rewards/margins_min": 0.038357168436050415, "rewards/margins_std": 0.06225566938519478, "rewards/rejected": -0.013587711378932, "step": 1170 }, { "epoch": 0.87, "grad_norm": 0.5234375, "learning_rate": 2.5965561257202036e-08, "logits/chosen": -0.0783575028181076, "logits/rejected": 0.7545520067214966, "logps/chosen": -272.24041748046875, "logps/rejected": -247.2455291748047, "loss": 0.6473, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.09078168123960495, "rewards/margins": 0.10268386453390121, "rewards/margins_max": 0.14412382245063782, "rewards/margins_min": 0.06124391406774521, "rewards/margins_std": 0.05860493704676628, "rewards/rejected": -0.011902175843715668, "step": 1180 }, { "epoch": 0.88, "grad_norm": 0.80078125, "learning_rate": 2.318990119300218e-08, "logits/chosen": 0.05255628377199173, "logits/rejected": 0.6947780847549438, "logps/chosen": -243.50509643554688, "logps/rejected": -212.8872528076172, "loss": 0.6546, "rewards/accuracies": 0.9375, "rewards/chosen": 0.05941227823495865, "rewards/margins": 0.07317686825990677, "rewards/margins_max": 0.11058169603347778, "rewards/margins_min": 0.03577204793691635, "rewards/margins_std": 0.052898406982421875, "rewards/rejected": -0.013764592818915844, "step": 1190 }, { "epoch": 0.88, "grad_norm": 0.4296875, "learning_rate": 2.0563894098070216e-08, "logits/chosen": -0.0692586898803711, "logits/rejected": 0.7870053052902222, "logps/chosen": -309.2232666015625, "logps/rejected": -266.5531921386719, "loss": 0.6476, "rewards/accuracies": 0.925000011920929, "rewards/chosen": 0.08217885345220566, "rewards/margins": 0.09460541605949402, "rewards/margins_max": 0.13726098835468292, "rewards/margins_min": 0.05194983631372452, "rewards/margins_std": 0.06032410264015198, "rewards/rejected": -0.012426558881998062, "step": 1200 }, { "epoch": 0.89, "grad_norm": 0.34375, "learning_rate": 1.8089272654333353e-08, "logits/chosen": -0.046720363199710846, "logits/rejected": 0.6504586338996887, "logps/chosen": -261.49298095703125, "logps/rejected": -223.7759552001953, "loss": 0.6526, "rewards/accuracies": 0.9375, "rewards/chosen": 0.07228533923625946, "rewards/margins": 0.07905267179012299, "rewards/margins_max": 0.12283986806869507, "rewards/margins_min": 0.035265471786260605, "rewards/margins_std": 0.06192445755004883, "rewards/rejected": -0.006767333950847387, "step": 1210 }, { "epoch": 0.9, "grad_norm": 0.4140625, "learning_rate": 1.5767669657019005e-08, "logits/chosen": -0.1003817468881607, "logits/rejected": 0.7918148636817932, "logps/chosen": -284.4376525878906, "logps/rejected": -226.739990234375, "loss": 0.6514, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.07430224120616913, "rewards/margins": 0.08967499434947968, "rewards/margins_max": 0.13941441476345062, "rewards/margins_min": 0.03993559628725052, "rewards/margins_std": 0.07034213840961456, "rewards/rejected": -0.015372763387858868, "step": 1220 }, { "epoch": 0.91, "grad_norm": 0.42578125, "learning_rate": 1.3600616937310267e-08, "logits/chosen": -0.13235566020011902, "logits/rejected": 0.602156400680542, "logps/chosen": -261.45050048828125, "logps/rejected": -231.56277465820312, "loss": 0.647, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.0806741714477539, "rewards/margins": 0.10508724302053452, "rewards/margins_max": 0.1610412895679474, "rewards/margins_min": 0.049133218824863434, "rewards/margins_std": 0.07913094758987427, "rewards/rejected": -0.024413082748651505, "step": 1230 }, { "epoch": 0.91, "grad_norm": 0.478515625, "learning_rate": 1.1589544351619047e-08, "logits/chosen": 0.02703939750790596, "logits/rejected": 0.8025895357131958, "logps/chosen": -334.37176513671875, "logps/rejected": -263.4154968261719, "loss": 0.6488, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.07323239743709564, "rewards/margins": 0.08194652944803238, "rewards/margins_max": 0.12296142429113388, "rewards/margins_min": 0.04093163460493088, "rewards/margins_std": 0.05800382047891617, "rewards/rejected": -0.008714130148291588, "step": 1240 }, { "epoch": 0.92, "grad_norm": 0.359375, "learning_rate": 9.735778838143749e-09, "logits/chosen": -0.0286453515291214, "logits/rejected": 0.7510533332824707, "logps/chosen": -311.01959228515625, "logps/rejected": -240.82870483398438, "loss": 0.6492, "rewards/accuracies": 0.987500011920929, "rewards/chosen": 0.07665716856718063, "rewards/margins": 0.09343422949314117, "rewards/margins_max": 0.13625505566596985, "rewards/margins_min": 0.0506133958697319, "rewards/margins_std": 0.06055780500173569, "rewards/rejected": -0.01677705906331539, "step": 1250 }, { "epoch": 0.93, "grad_norm": 0.44140625, "learning_rate": 8.040543541333655e-09, "logits/chosen": 0.05244514346122742, "logits/rejected": 0.7751520276069641, "logps/chosen": -310.7460021972656, "logps/rejected": -286.03826904296875, "loss": 0.6489, "rewards/accuracies": 0.987500011920929, "rewards/chosen": 0.06985687464475632, "rewards/margins": 0.0900772362947464, "rewards/margins_max": 0.12786877155303955, "rewards/margins_min": 0.052285682410001755, "rewards/margins_std": 0.053445328027009964, "rewards/rejected": -0.020220357924699783, "step": 1260 }, { "epoch": 0.93, "grad_norm": 0.5, "learning_rate": 6.504957004838746e-09, "logits/chosen": -0.03165416046977043, "logits/rejected": 0.822761058807373, "logps/chosen": -310.04400634765625, "logps/rejected": -237.48782348632812, "loss": 0.6523, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.06924634426832199, "rewards/margins": 0.08929922431707382, "rewards/margins_max": 0.14400045573711395, "rewards/margins_min": 0.034597981721162796, "rewards/margins_std": 0.07735923677682877, "rewards/rejected": -0.020052876323461533, "step": 1270 }, { "epoch": 0.94, "grad_norm": 0.4765625, "learning_rate": 5.130032433476483e-09, "logits/chosen": -0.011189639568328857, "logits/rejected": 0.691806972026825, "logps/chosen": -271.3976135253906, "logps/rejected": -269.19390869140625, "loss": 0.6545, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.08424172550439835, "rewards/margins": 0.09692950546741486, "rewards/margins_max": 0.14971934258937836, "rewards/margins_min": 0.044139645993709564, "rewards/margins_std": 0.07465612888336182, "rewards/rejected": -0.012687774375081062, "step": 1280 }, { "epoch": 0.95, "grad_norm": 0.43359375, "learning_rate": 3.916677024702858e-09, "logits/chosen": 0.009526084177196026, "logits/rejected": 0.7200338244438171, "logps/chosen": -255.1676788330078, "logps/rejected": -234.27432250976562, "loss": 0.6534, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.06782687455415726, "rewards/margins": 0.08498277515172958, "rewards/margins_max": 0.1279430389404297, "rewards/margins_min": 0.04202253371477127, "rewards/margins_std": 0.06075497344136238, "rewards/rejected": -0.017155908048152924, "step": 1290 }, { "epoch": 0.96, "grad_norm": 0.51953125, "learning_rate": 2.865691370028761e-09, "logits/chosen": -0.11016593128442764, "logits/rejected": 0.7786885499954224, "logps/chosen": -310.4071960449219, "logps/rejected": -242.39505004882812, "loss": 0.6487, "rewards/accuracies": 1.0, "rewards/chosen": 0.0797882154583931, "rewards/margins": 0.09522275626659393, "rewards/margins_max": 0.13675549626350403, "rewards/margins_min": 0.05369000881910324, "rewards/margins_std": 0.0587361678481102, "rewards/rejected": -0.015434542670845985, "step": 1300 }, { "epoch": 0.96, "grad_norm": 0.482421875, "learning_rate": 1.977768926776896e-09, "logits/chosen": 0.07193834334611893, "logits/rejected": 0.7082736492156982, "logps/chosen": -269.1424255371094, "logps/rejected": -258.5067443847656, "loss": 0.6549, "rewards/accuracies": 0.949999988079071, "rewards/chosen": 0.07826878130435944, "rewards/margins": 0.0938304141163826, "rewards/margins_max": 0.13868080079555511, "rewards/margins_min": 0.04898003116250038, "rewards/margins_std": 0.06342802941799164, "rewards/rejected": -0.01556164026260376, "step": 1310 }, { "epoch": 0.97, "grad_norm": 0.474609375, "learning_rate": 1.2534955605274233e-09, "logits/chosen": -0.014053803868591785, "logits/rejected": 0.777262806892395, "logps/chosen": -242.13796997070312, "logps/rejected": -218.4778594970703, "loss": 0.6505, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.07983841001987457, "rewards/margins": 0.09085123986005783, "rewards/margins_max": 0.1351737529039383, "rewards/margins_min": 0.046528737992048264, "rewards/margins_std": 0.06268148869276047, "rewards/rejected": -0.011012842878699303, "step": 1320 }, { "epoch": 0.98, "grad_norm": 0.48828125, "learning_rate": 6.933491585542351e-10, "logits/chosen": 0.03265099227428436, "logits/rejected": 0.7704384326934814, "logps/chosen": -293.1032409667969, "logps/rejected": -263.2946472167969, "loss": 0.6471, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.08791947364807129, "rewards/margins": 0.10431689023971558, "rewards/margins_max": 0.15510039031505585, "rewards/margins_min": 0.05353338643908501, "rewards/margins_std": 0.07181872427463531, "rewards/rejected": -0.016397414728999138, "step": 1330 }, { "epoch": 0.99, "grad_norm": 0.451171875, "learning_rate": 2.9769931450737694e-10, "logits/chosen": 0.03406764939427376, "logits/rejected": 0.7593650221824646, "logps/chosen": -301.3545227050781, "logps/rejected": -253.1154022216797, "loss": 0.6492, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": 0.06837470084428787, "rewards/margins": 0.08594924956560135, "rewards/margins_max": 0.1240011602640152, "rewards/margins_min": 0.0478973314166069, "rewards/margins_std": 0.05381352826952934, "rewards/rejected": -0.017574548721313477, "step": 1340 }, { "epoch": 0.99, "grad_norm": 0.412109375, "learning_rate": 6.680708454906425e-11, "logits/chosen": -0.05887039750814438, "logits/rejected": 0.7168238759040833, "logps/chosen": -274.7958679199219, "logps/rejected": -234.88424682617188, "loss": 0.6506, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": 0.06663821637630463, "rewards/margins": 0.08757266402244568, "rewards/margins_max": 0.1257934719324112, "rewards/margins_min": 0.049351852387189865, "rewards/margins_std": 0.05405237525701523, "rewards/rejected": -0.020934434607625008, "step": 1350 }, { "epoch": 1.0, "eval_logits/chosen": 0.744699239730835, "eval_logits/rejected": 0.8899587988853455, "eval_logps/chosen": -330.823974609375, "eval_logps/rejected": -316.1527404785156, "eval_loss": 0.6946464776992798, "eval_rewards/accuracies": 0.5195000171661377, "eval_rewards/chosen": 0.03155522421002388, "eval_rewards/margins": -0.0017884434200823307, "eval_rewards/margins_max": 0.09517286717891693, "eval_rewards/margins_min": -0.10408221185207367, "eval_rewards/margins_std": 0.06461790204048157, "eval_rewards/rejected": 0.03334366902709007, "eval_runtime": 837.0887, "eval_samples_per_second": 4.778, "eval_steps_per_second": 0.299, "step": 1359 }, { "epoch": 1.0, "step": 1359, "total_flos": 0.0, "train_loss": 0.6604200788417925, "train_runtime": 10358.1715, "train_samples_per_second": 2.099, "train_steps_per_second": 0.131 } ], "logging_steps": 10, "max_steps": 1359, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 100, "total_flos": 0.0, "train_batch_size": 2, "trial_name": null, "trial_params": null }