| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.9765925925925925, | |
| "eval_steps": 500, | |
| "global_step": 315, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.047407407407407405, | |
| "grad_norm": 1010.7771606445312, | |
| "learning_rate": 7.8125e-06, | |
| "log_odds_chosen": -1.0667380094528198, | |
| "log_odds_ratio": -11.012346267700195, | |
| "logps/chosen": -22.715862274169922, | |
| "logps/rejected": -21.64935302734375, | |
| "loss": 313.2559, | |
| "nll_loss": 9.84270191192627, | |
| "rewards/accuracies": 0.4781250059604645, | |
| "rewards/chosen": -2.2715861797332764, | |
| "rewards/margins": -0.10665090382099152, | |
| "rewards/rejected": -2.164935350418091, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.09481481481481481, | |
| "grad_norm": 434.8633728027344, | |
| "learning_rate": 1.5625e-05, | |
| "log_odds_chosen": -3.1370198726654053, | |
| "log_odds_ratio": -11.084940910339355, | |
| "logps/chosen": -22.05541229248047, | |
| "logps/rejected": -18.918724060058594, | |
| "loss": 322.0073, | |
| "nll_loss": 8.76218032836914, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -2.2055413722991943, | |
| "rewards/margins": -0.31366899609565735, | |
| "rewards/rejected": -1.8918724060058594, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.14222222222222222, | |
| "grad_norm": 371.8564758300781, | |
| "learning_rate": 2.34375e-05, | |
| "log_odds_chosen": -1.3131778240203857, | |
| "log_odds_ratio": -11.433721542358398, | |
| "logps/chosen": -22.153400421142578, | |
| "logps/rejected": -20.838821411132812, | |
| "loss": 280.4119, | |
| "nll_loss": 8.03809928894043, | |
| "rewards/accuracies": 0.515625, | |
| "rewards/chosen": -2.2153401374816895, | |
| "rewards/margins": -0.13145777583122253, | |
| "rewards/rejected": -2.0838820934295654, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.18962962962962962, | |
| "grad_norm": 2397.505126953125, | |
| "learning_rate": 3.125e-05, | |
| "log_odds_chosen": -4.679871559143066, | |
| "log_odds_ratio": -10.199138641357422, | |
| "logps/chosen": -18.253995895385742, | |
| "logps/rejected": -13.57618236541748, | |
| "loss": 271.0691, | |
| "nll_loss": 6.453103542327881, | |
| "rewards/accuracies": 0.44062501192092896, | |
| "rewards/chosen": -1.82539963722229, | |
| "rewards/margins": -0.4677814841270447, | |
| "rewards/rejected": -1.3576180934906006, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.23703703703703705, | |
| "grad_norm": 314.60101318359375, | |
| "learning_rate": 3.90625e-05, | |
| "log_odds_chosen": -0.22799472510814667, | |
| "log_odds_ratio": -2.3246471881866455, | |
| "logps/chosen": -4.8733930587768555, | |
| "logps/rejected": -4.6329240798950195, | |
| "loss": 100.4913, | |
| "nll_loss": 2.7564172744750977, | |
| "rewards/accuracies": 0.5406249761581421, | |
| "rewards/chosen": -0.4873393177986145, | |
| "rewards/margins": -0.024046871811151505, | |
| "rewards/rejected": -0.4632924199104309, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.28444444444444444, | |
| "grad_norm": 294.3478698730469, | |
| "learning_rate": 4.6875e-05, | |
| "log_odds_chosen": 0.029871661216020584, | |
| "log_odds_ratio": -0.9212434887886047, | |
| "logps/chosen": -1.8298267126083374, | |
| "logps/rejected": -1.8607829809188843, | |
| "loss": 63.4308, | |
| "nll_loss": 1.981960654258728, | |
| "rewards/accuracies": 0.5218750238418579, | |
| "rewards/chosen": -0.1829826682806015, | |
| "rewards/margins": 0.0030956335831433535, | |
| "rewards/rejected": -0.186078280210495, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.33185185185185184, | |
| "grad_norm": 122.69192504882812, | |
| "learning_rate": 4.998613757348784e-05, | |
| "log_odds_chosen": 0.18753251433372498, | |
| "log_odds_ratio": -0.846284031867981, | |
| "logps/chosen": -1.7096046209335327, | |
| "logps/rejected": -1.8797508478164673, | |
| "loss": 59.8376, | |
| "nll_loss": 1.8704582452774048, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.1709604412317276, | |
| "rewards/margins": 0.017014626413583755, | |
| "rewards/rejected": -0.18797507882118225, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.37925925925925924, | |
| "grad_norm": 133.71852111816406, | |
| "learning_rate": 4.990147841143462e-05, | |
| "log_odds_chosen": 0.25348493456840515, | |
| "log_odds_ratio": -0.7433185577392578, | |
| "logps/chosen": -1.4737732410430908, | |
| "logps/rejected": -1.6838853359222412, | |
| "loss": 56.603, | |
| "nll_loss": 1.7691009044647217, | |
| "rewards/accuracies": 0.559374988079071, | |
| "rewards/chosen": -0.14737732708454132, | |
| "rewards/margins": 0.021011201664805412, | |
| "rewards/rejected": -0.1683885157108307, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.4266666666666667, | |
| "grad_norm": 96.86552429199219, | |
| "learning_rate": 4.97401218720448e-05, | |
| "log_odds_chosen": 0.18790796399116516, | |
| "log_odds_ratio": -0.7494683265686035, | |
| "logps/chosen": -1.3950669765472412, | |
| "logps/rejected": -1.559629201889038, | |
| "loss": 53.609, | |
| "nll_loss": 1.6753915548324585, | |
| "rewards/accuracies": 0.559374988079071, | |
| "rewards/chosen": -0.13950669765472412, | |
| "rewards/margins": 0.01645624078810215, | |
| "rewards/rejected": -0.15596294403076172, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.4740740740740741, | |
| "grad_norm": 97.8518295288086, | |
| "learning_rate": 4.9502564938797946e-05, | |
| "log_odds_chosen": 0.23386704921722412, | |
| "log_odds_ratio": -0.7078075408935547, | |
| "logps/chosen": -1.2506229877471924, | |
| "logps/rejected": -1.4465948343276978, | |
| "loss": 49.4392, | |
| "nll_loss": 1.545029878616333, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.12506231665611267, | |
| "rewards/margins": 0.01959715224802494, | |
| "rewards/rejected": -0.14465947449207306, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.5214814814814814, | |
| "grad_norm": 117.51167297363281, | |
| "learning_rate": 4.918953929490768e-05, | |
| "log_odds_chosen": 0.20879492163658142, | |
| "log_odds_ratio": -0.7273464798927307, | |
| "logps/chosen": -1.258131742477417, | |
| "logps/rejected": -1.4245622158050537, | |
| "loss": 50.0452, | |
| "nll_loss": 1.5639737844467163, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": -0.12581318616867065, | |
| "rewards/margins": 0.016643041744828224, | |
| "rewards/rejected": -0.14245623350143433, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.5688888888888889, | |
| "grad_norm": 170.34327697753906, | |
| "learning_rate": 4.88020090697132e-05, | |
| "log_odds_chosen": 0.16209658980369568, | |
| "log_odds_ratio": -0.7264987826347351, | |
| "logps/chosen": -1.2688095569610596, | |
| "logps/rejected": -1.406677484512329, | |
| "loss": 50.7873, | |
| "nll_loss": 1.587153434753418, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.1268809586763382, | |
| "rewards/margins": 0.013786802999675274, | |
| "rewards/rejected": -0.14066775143146515, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.6162962962962963, | |
| "grad_norm": 142.08338928222656, | |
| "learning_rate": 4.834116786912897e-05, | |
| "log_odds_chosen": 0.20509567856788635, | |
| "log_odds_ratio": -0.7091192603111267, | |
| "logps/chosen": -1.2300090789794922, | |
| "logps/rejected": -1.384817123413086, | |
| "loss": 48.5215, | |
| "nll_loss": 1.5163531303405762, | |
| "rewards/accuracies": 0.5843750238418579, | |
| "rewards/chosen": -0.12300090491771698, | |
| "rewards/margins": 0.015480798669159412, | |
| "rewards/rejected": -0.1384817212820053, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.6637037037037037, | |
| "grad_norm": 77.49330139160156, | |
| "learning_rate": 4.7808435099299045e-05, | |
| "log_odds_chosen": 0.29166898131370544, | |
| "log_odds_ratio": -0.6636098623275757, | |
| "logps/chosen": -1.182460069656372, | |
| "logps/rejected": -1.4143471717834473, | |
| "loss": 48.004, | |
| "nll_loss": 1.5001914501190186, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.11824599653482437, | |
| "rewards/margins": 0.023188717663288116, | |
| "rewards/rejected": -0.1414347141981125, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.7111111111111111, | |
| "grad_norm": 133.53330993652344, | |
| "learning_rate": 4.720545159477922e-05, | |
| "log_odds_chosen": 0.28170067071914673, | |
| "log_odds_ratio": -0.6955637335777283, | |
| "logps/chosen": -1.1593291759490967, | |
| "logps/rejected": -1.3801120519638062, | |
| "loss": 47.0242, | |
| "nll_loss": 1.4699668884277344, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.11593292653560638, | |
| "rewards/margins": 0.02207827940583229, | |
| "rewards/rejected": -0.13801121711730957, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.7585185185185185, | |
| "grad_norm": 61.732521057128906, | |
| "learning_rate": 4.653407456471222e-05, | |
| "log_odds_chosen": 0.17284110188484192, | |
| "log_odds_ratio": -0.7073065638542175, | |
| "logps/chosen": -1.1204986572265625, | |
| "logps/rejected": -1.2521705627441406, | |
| "loss": 45.2491, | |
| "nll_loss": 1.4140598773956299, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.11204985529184341, | |
| "rewards/margins": 0.013167202472686768, | |
| "rewards/rejected": -0.12521706521511078, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.8059259259259259, | |
| "grad_norm": 134.90655517578125, | |
| "learning_rate": 4.579637187256222e-05, | |
| "log_odds_chosen": 0.17245015501976013, | |
| "log_odds_ratio": -0.6946243047714233, | |
| "logps/chosen": -1.1173996925354004, | |
| "logps/rejected": -1.2512614727020264, | |
| "loss": 46.4933, | |
| "nll_loss": 1.4529409408569336, | |
| "rewards/accuracies": 0.5843750238418579, | |
| "rewards/chosen": -0.11173997074365616, | |
| "rewards/margins": 0.013386172242462635, | |
| "rewards/rejected": -0.12512615323066711, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.8533333333333334, | |
| "grad_norm": 62.5573844909668, | |
| "learning_rate": 4.499461566702685e-05, | |
| "log_odds_chosen": 0.2704593539237976, | |
| "log_odds_ratio": -0.6617124080657959, | |
| "logps/chosen": -1.0328925848007202, | |
| "logps/rejected": -1.224131464958191, | |
| "loss": 44.1876, | |
| "nll_loss": 1.3808879852294922, | |
| "rewards/accuracies": 0.6156250238418579, | |
| "rewards/chosen": -0.10328926891088486, | |
| "rewards/margins": 0.019123883917927742, | |
| "rewards/rejected": -0.12241314351558685, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.9007407407407407, | |
| "grad_norm": 166.74710083007812, | |
| "learning_rate": 4.413127538374411e-05, | |
| "log_odds_chosen": 0.1248466745018959, | |
| "log_odds_ratio": -0.7291629910469055, | |
| "logps/chosen": -1.1157370805740356, | |
| "logps/rejected": -1.213266372680664, | |
| "loss": 45.8715, | |
| "nll_loss": 1.4335052967071533, | |
| "rewards/accuracies": 0.534375011920929, | |
| "rewards/chosen": -0.1115737110376358, | |
| "rewards/margins": 0.009752912446856499, | |
| "rewards/rejected": -0.12132661044597626, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.9481481481481482, | |
| "grad_norm": 94.61317443847656, | |
| "learning_rate": 4.320901013934887e-05, | |
| "log_odds_chosen": 0.21559691429138184, | |
| "log_odds_ratio": -0.6755672693252563, | |
| "logps/chosen": -1.0831605195999146, | |
| "logps/rejected": -1.2379772663116455, | |
| "loss": 44.4355, | |
| "nll_loss": 1.3886381387710571, | |
| "rewards/accuracies": 0.6031249761581421, | |
| "rewards/chosen": -0.10831604897975922, | |
| "rewards/margins": 0.015481671318411827, | |
| "rewards/rejected": -0.12379772961139679, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.9955555555555555, | |
| "grad_norm": 63.372169494628906, | |
| "learning_rate": 4.223066054130568e-05, | |
| "log_odds_chosen": 0.25206008553504944, | |
| "log_odds_ratio": -0.6876682043075562, | |
| "logps/chosen": -1.0718275308609009, | |
| "logps/rejected": -1.2811346054077148, | |
| "loss": 45.085, | |
| "nll_loss": 1.4090317487716675, | |
| "rewards/accuracies": 0.578125, | |
| "rewards/chosen": -0.10718274116516113, | |
| "rewards/margins": 0.02093072049319744, | |
| "rewards/rejected": -0.12811346352100372, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.037925925925926, | |
| "grad_norm": 81.14254760742188, | |
| "learning_rate": 4.1199239938743797e-05, | |
| "log_odds_chosen": 0.49455782771110535, | |
| "log_odds_ratio": -0.5730230212211609, | |
| "logps/chosen": -0.9436227083206177, | |
| "logps/rejected": -1.257029414176941, | |
| "loss": 34.4497, | |
| "nll_loss": 1.2045941352844238, | |
| "rewards/accuracies": 0.6853147149085999, | |
| "rewards/chosen": -0.0943622887134552, | |
| "rewards/margins": 0.031340666115283966, | |
| "rewards/rejected": -0.12570294737815857, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.0853333333333333, | |
| "grad_norm": 47.50578308105469, | |
| "learning_rate": 4.0117925141242174e-05, | |
| "log_odds_chosen": 0.6659696102142334, | |
| "log_odds_ratio": -0.5298361778259277, | |
| "logps/chosen": -0.8568960428237915, | |
| "logps/rejected": -1.2688976526260376, | |
| "loss": 36.7976, | |
| "nll_loss": 1.150004267692566, | |
| "rewards/accuracies": 0.721875011920929, | |
| "rewards/chosen": -0.08568960428237915, | |
| "rewards/margins": 0.04120016470551491, | |
| "rewards/rejected": -0.12688976526260376, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.1327407407407408, | |
| "grad_norm": 51.826683044433594, | |
| "learning_rate": 3.899004663415084e-05, | |
| "log_odds_chosen": 0.6182760000228882, | |
| "log_odds_ratio": -0.5470128655433655, | |
| "logps/chosen": -0.8712148666381836, | |
| "logps/rejected": -1.2713781595230103, | |
| "loss": 37.6544, | |
| "nll_loss": 1.1768953800201416, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.08712148666381836, | |
| "rewards/margins": 0.040016330778598785, | |
| "rewards/rejected": -0.12713780999183655, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.1801481481481482, | |
| "grad_norm": 73.44404602050781, | |
| "learning_rate": 3.781907832058587e-05, | |
| "log_odds_chosen": 0.6142861247062683, | |
| "log_odds_ratio": -0.5300159454345703, | |
| "logps/chosen": -0.9014692306518555, | |
| "logps/rejected": -1.2914025783538818, | |
| "loss": 36.6759, | |
| "nll_loss": 1.1462795734405518, | |
| "rewards/accuracies": 0.7406250238418579, | |
| "rewards/chosen": -0.09014692157506943, | |
| "rewards/margins": 0.03899333253502846, | |
| "rewards/rejected": -0.1291402280330658, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.2275555555555555, | |
| "grad_norm": 88.10698699951172, | |
| "learning_rate": 3.660862682169282e-05, | |
| "log_odds_chosen": 0.6761590242385864, | |
| "log_odds_ratio": -0.5286710858345032, | |
| "logps/chosen": -0.8857125043869019, | |
| "logps/rejected": -1.3080275058746338, | |
| "loss": 37.3742, | |
| "nll_loss": 1.168211817741394, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.08857125788927078, | |
| "rewards/margins": 0.04223150387406349, | |
| "rewards/rejected": -0.13080275058746338, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.274962962962963, | |
| "grad_norm": 61.116153717041016, | |
| "learning_rate": 3.5362420368134356e-05, | |
| "log_odds_chosen": 0.6304064393043518, | |
| "log_odds_ratio": -0.5227637887001038, | |
| "logps/chosen": -0.8533610105514526, | |
| "logps/rejected": -1.233307123184204, | |
| "loss": 36.041, | |
| "nll_loss": 1.1264115571975708, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.0853360965847969, | |
| "rewards/margins": 0.03799460828304291, | |
| "rewards/rejected": -0.12333071231842041, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.3223703703703704, | |
| "grad_norm": 66.3697509765625, | |
| "learning_rate": 3.408429731701635e-05, | |
| "log_odds_chosen": 0.6969801187515259, | |
| "log_odds_ratio": -0.5261715650558472, | |
| "logps/chosen": -0.8764761090278625, | |
| "logps/rejected": -1.3215680122375488, | |
| "loss": 37.4153, | |
| "nll_loss": 1.1695525646209717, | |
| "rewards/accuracies": 0.7281249761581421, | |
| "rewards/chosen": -0.08764760941267014, | |
| "rewards/margins": 0.04450917989015579, | |
| "rewards/rejected": -0.13215678930282593, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.3697777777777778, | |
| "grad_norm": 46.39073181152344, | |
| "learning_rate": 3.2778194329621104e-05, | |
| "log_odds_chosen": 0.6316584944725037, | |
| "log_odds_ratio": -0.5412708520889282, | |
| "logps/chosen": -0.8719720840454102, | |
| "logps/rejected": -1.2730590105056763, | |
| "loss": 37.6538, | |
| "nll_loss": 1.176888108253479, | |
| "rewards/accuracies": 0.7281249761581421, | |
| "rewards/chosen": -0.0871971994638443, | |
| "rewards/margins": 0.040108706802129745, | |
| "rewards/rejected": -0.12730590999126434, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.417185185185185, | |
| "grad_norm": 81.8355941772461, | |
| "learning_rate": 3.144813424636031e-05, | |
| "log_odds_chosen": 0.6246966123580933, | |
| "log_odds_ratio": -0.5357092618942261, | |
| "logps/chosen": -0.8369342684745789, | |
| "logps/rejected": -1.228888750076294, | |
| "loss": 35.6394, | |
| "nll_loss": 1.113890290260315, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.08369342982769012, | |
| "rewards/margins": 0.03919545188546181, | |
| "rewards/rejected": -0.12288890033960342, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.4645925925925927, | |
| "grad_norm": 72.14729309082031, | |
| "learning_rate": 3.0098213696293542e-05, | |
| "log_odds_chosen": 0.6171873211860657, | |
| "log_odds_ratio": -0.541162371635437, | |
| "logps/chosen": -0.8621460795402527, | |
| "logps/rejected": -1.2617683410644531, | |
| "loss": 36.6333, | |
| "nll_loss": 1.144904375076294, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.08621460944414139, | |
| "rewards/margins": 0.03996223211288452, | |
| "rewards/rejected": -0.1261768341064453, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.512, | |
| "grad_norm": 100.7032241821289, | |
| "learning_rate": 2.8732590479375165e-05, | |
| "log_odds_chosen": 0.5847410559654236, | |
| "log_odds_ratio": -0.5625023245811462, | |
| "logps/chosen": -0.8291622400283813, | |
| "logps/rejected": -1.2009413242340088, | |
| "loss": 35.8312, | |
| "nll_loss": 1.1198655366897583, | |
| "rewards/accuracies": 0.671875, | |
| "rewards/chosen": -0.08291621506214142, | |
| "rewards/margins": 0.037177909165620804, | |
| "rewards/rejected": -0.12009412050247192, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.5594074074074074, | |
| "grad_norm": 41.177188873291016, | |
| "learning_rate": 2.7355470760292956e-05, | |
| "log_odds_chosen": 0.6445590853691101, | |
| "log_odds_ratio": -0.5316762924194336, | |
| "logps/chosen": -0.8270977139472961, | |
| "logps/rejected": -1.2208188772201538, | |
| "loss": 36.8138, | |
| "nll_loss": 1.150538682937622, | |
| "rewards/accuracies": 0.721875011920929, | |
| "rewards/chosen": -0.08270977437496185, | |
| "rewards/margins": 0.039372123777866364, | |
| "rewards/rejected": -0.12208189070224762, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.6068148148148147, | |
| "grad_norm": 50.124290466308594, | |
| "learning_rate": 2.597109611334169e-05, | |
| "log_odds_chosen": 0.6327283382415771, | |
| "log_odds_ratio": -0.5253031849861145, | |
| "logps/chosen": -0.8413885831832886, | |
| "logps/rejected": -1.2489277124404907, | |
| "loss": 36.2972, | |
| "nll_loss": 1.1343696117401123, | |
| "rewards/accuracies": 0.7281249761581421, | |
| "rewards/chosen": -0.08413885533809662, | |
| "rewards/margins": 0.04075390845537186, | |
| "rewards/rejected": -0.12489277124404907, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.6542222222222223, | |
| "grad_norm": 45.2021369934082, | |
| "learning_rate": 2.458373045823404e-05, | |
| "log_odds_chosen": 0.7051500082015991, | |
| "log_odds_ratio": -0.5123899579048157, | |
| "logps/chosen": -0.8304961919784546, | |
| "logps/rejected": -1.2771947383880615, | |
| "loss": 35.1566, | |
| "nll_loss": 1.0988842248916626, | |
| "rewards/accuracies": 0.746874988079071, | |
| "rewards/chosen": -0.08304962515830994, | |
| "rewards/margins": 0.04466987028717995, | |
| "rewards/rejected": -0.12771949172019958, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.7016296296296296, | |
| "grad_norm": 56.0422477722168, | |
| "learning_rate": 2.3197646927086697e-05, | |
| "log_odds_chosen": 0.597710371017456, | |
| "log_odds_ratio": -0.5408257842063904, | |
| "logps/chosen": -0.8311992883682251, | |
| "logps/rejected": -1.1982812881469727, | |
| "loss": 35.7218, | |
| "nll_loss": 1.1163814067840576, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.0831199362874031, | |
| "rewards/margins": 0.03670820593833923, | |
| "rewards/rejected": -0.11982814967632294, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.749037037037037, | |
| "grad_norm": 45.33614730834961, | |
| "learning_rate": 2.1817114703032176e-05, | |
| "log_odds_chosen": 0.5593910813331604, | |
| "log_odds_ratio": -0.5753989815711975, | |
| "logps/chosen": -0.8772571682929993, | |
| "logps/rejected": -1.2458473443984985, | |
| "loss": 35.5883, | |
| "nll_loss": 1.1122016906738281, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.08772572129964828, | |
| "rewards/margins": 0.03685901314020157, | |
| "rewards/rejected": -0.12458473443984985, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.7964444444444445, | |
| "grad_norm": 44.086551666259766, | |
| "learning_rate": 2.0446385870993467e-05, | |
| "log_odds_chosen": 0.6592080593109131, | |
| "log_odds_ratio": -0.5308284759521484, | |
| "logps/chosen": -0.839551568031311, | |
| "logps/rejected": -1.2457295656204224, | |
| "loss": 35.969, | |
| "nll_loss": 1.124164342880249, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.08395515382289886, | |
| "rewards/margins": 0.04061780497431755, | |
| "rewards/rejected": -0.12457295507192612, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.8438518518518519, | |
| "grad_norm": 49.89115524291992, | |
| "learning_rate": 1.9089682321121834e-05, | |
| "log_odds_chosen": 0.6262322664260864, | |
| "log_odds_ratio": -0.5255475640296936, | |
| "logps/chosen": -0.8536340594291687, | |
| "logps/rejected": -1.24855375289917, | |
| "loss": 35.548, | |
| "nll_loss": 1.1109614372253418, | |
| "rewards/accuracies": 0.7406250238418579, | |
| "rewards/chosen": -0.08536341041326523, | |
| "rewards/margins": 0.03949197009205818, | |
| "rewards/rejected": -0.12485537678003311, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.8912592592592592, | |
| "grad_norm": 48.73372268676758, | |
| "learning_rate": 1.775118274523545e-05, | |
| "log_odds_chosen": 0.6317826509475708, | |
| "log_odds_ratio": -0.5390036702156067, | |
| "logps/chosen": -0.866468608379364, | |
| "logps/rejected": -1.2655459642410278, | |
| "loss": 36.7251, | |
| "nll_loss": 1.1477723121643066, | |
| "rewards/accuracies": 0.7093750238418579, | |
| "rewards/chosen": -0.08664685487747192, | |
| "rewards/margins": 0.03990774229168892, | |
| "rewards/rejected": -0.12655460834503174, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.9386666666666668, | |
| "grad_norm": 42.03281784057617, | |
| "learning_rate": 1.643500976631037e-05, | |
| "log_odds_chosen": 0.6149218678474426, | |
| "log_odds_ratio": -0.5417319536209106, | |
| "logps/chosen": -0.8332918882369995, | |
| "logps/rejected": -1.207820177078247, | |
| "loss": 35.5065, | |
| "nll_loss": 1.10965096950531, | |
| "rewards/accuracies": 0.6968749761581421, | |
| "rewards/chosen": -0.08332918584346771, | |
| "rewards/margins": 0.03745283931493759, | |
| "rewards/rejected": -0.12078201770782471, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 1.986074074074074, | |
| "grad_norm": 41.75816345214844, | |
| "learning_rate": 1.514521724066537e-05, | |
| "log_odds_chosen": 0.50700443983078, | |
| "log_odds_ratio": -0.5639246702194214, | |
| "logps/chosen": -0.8441019058227539, | |
| "logps/rejected": -1.1364227533340454, | |
| "loss": 36.5811, | |
| "nll_loss": 1.1431875228881836, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.08441020548343658, | |
| "rewards/margins": 0.02923208475112915, | |
| "rewards/rejected": -0.11364229023456573, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 2.0284444444444443, | |
| "grad_norm": 53.94225311279297, | |
| "learning_rate": 1.3885777771950348e-05, | |
| "log_odds_chosen": 0.9303293228149414, | |
| "log_odds_ratio": -0.4375574588775635, | |
| "logps/chosen": -0.6674253344535828, | |
| "logps/rejected": -1.1656166315078735, | |
| "loss": 25.6503, | |
| "nll_loss": 0.8970839381217957, | |
| "rewards/accuracies": 0.7867133021354675, | |
| "rewards/chosen": -0.06674253195524216, | |
| "rewards/margins": 0.04981912672519684, | |
| "rewards/rejected": -0.11656166613101959, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 2.075851851851852, | |
| "grad_norm": 61.88585662841797, | |
| "learning_rate": 1.2660570475395683e-05, | |
| "log_odds_chosen": 1.1436392068862915, | |
| "log_odds_ratio": -0.3958616554737091, | |
| "logps/chosen": -0.6872319579124451, | |
| "logps/rejected": -1.3230359554290771, | |
| "loss": 28.5064, | |
| "nll_loss": 0.8910220861434937, | |
| "rewards/accuracies": 0.8343750238418579, | |
| "rewards/chosen": -0.06872320175170898, | |
| "rewards/margins": 0.06358040124177933, | |
| "rewards/rejected": -0.13230359554290771, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 2.1232592592592594, | |
| "grad_norm": 46.268699645996094, | |
| "learning_rate": 1.1473369030008974e-05, | |
| "log_odds_chosen": 1.2310889959335327, | |
| "log_odds_ratio": -0.3688361942768097, | |
| "logps/chosen": -0.6646565198898315, | |
| "logps/rejected": -1.3415366411209106, | |
| "loss": 27.6811, | |
| "nll_loss": 0.8652756810188293, | |
| "rewards/accuracies": 0.871874988079071, | |
| "rewards/chosen": -0.06646565347909927, | |
| "rewards/margins": 0.06768801063299179, | |
| "rewards/rejected": -0.13415366411209106, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 2.1706666666666665, | |
| "grad_norm": 52.781517028808594, | |
| "learning_rate": 1.0327830055518842e-05, | |
| "log_odds_chosen": 1.2866785526275635, | |
| "log_odds_ratio": -0.36883771419525146, | |
| "logps/chosen": -0.6471323370933533, | |
| "logps/rejected": -1.3661425113677979, | |
| "loss": 27.6196, | |
| "nll_loss": 0.8633133172988892, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.06471323221921921, | |
| "rewards/margins": 0.07190100848674774, | |
| "rewards/rejected": -0.13661423325538635, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 2.218074074074074, | |
| "grad_norm": 48.32188415527344, | |
| "learning_rate": 9.227481849865235e-06, | |
| "log_odds_chosen": 1.2247092723846436, | |
| "log_odds_ratio": -0.3755631446838379, | |
| "logps/chosen": -0.6431256532669067, | |
| "logps/rejected": -1.3039284944534302, | |
| "loss": 28.5384, | |
| "nll_loss": 0.8919950723648071, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.06431256979703903, | |
| "rewards/margins": 0.06608027219772339, | |
| "rewards/rejected": -0.1303928643465042, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 2.2654814814814817, | |
| "grad_norm": 39.736480712890625, | |
| "learning_rate": 8.175713521924978e-06, | |
| "log_odds_chosen": 1.3375396728515625, | |
| "log_odds_ratio": -0.34778350591659546, | |
| "logps/chosen": -0.6190803050994873, | |
| "logps/rejected": -1.3734327554702759, | |
| "loss": 26.7795, | |
| "nll_loss": 0.8374263644218445, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": -0.06190803647041321, | |
| "rewards/margins": 0.07543523609638214, | |
| "rewards/rejected": -0.13734325766563416, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.3128888888888888, | |
| "grad_norm": 44.297664642333984, | |
| "learning_rate": 7.1757645529443665e-06, | |
| "log_odds_chosen": 1.2268487215042114, | |
| "log_odds_ratio": -0.3551596403121948, | |
| "logps/chosen": -0.6334061622619629, | |
| "logps/rejected": -1.3128724098205566, | |
| "loss": 27.5044, | |
| "nll_loss": 0.859671950340271, | |
| "rewards/accuracies": 0.871874988079071, | |
| "rewards/chosen": -0.06334061920642853, | |
| "rewards/margins": 0.06794662773609161, | |
| "rewards/rejected": -0.13128724694252014, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 2.3602962962962963, | |
| "grad_norm": 46.82137680053711, | |
| "learning_rate": 6.230714818829733e-06, | |
| "log_odds_chosen": 1.2968767881393433, | |
| "log_odds_ratio": -0.3475486636161804, | |
| "logps/chosen": -0.6015971302986145, | |
| "logps/rejected": -1.290184736251831, | |
| "loss": 27.7176, | |
| "nll_loss": 0.8663715124130249, | |
| "rewards/accuracies": 0.859375, | |
| "rewards/chosen": -0.06015970557928085, | |
| "rewards/margins": 0.06885877251625061, | |
| "rewards/rejected": -0.12901848554611206, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.407703703703704, | |
| "grad_norm": 41.21742630004883, | |
| "learning_rate": 5.343475104027743e-06, | |
| "log_odds_chosen": 1.367424726486206, | |
| "log_odds_ratio": -0.3471328616142273, | |
| "logps/chosen": -0.6164557337760925, | |
| "logps/rejected": -1.3936684131622314, | |
| "loss": 27.3975, | |
| "nll_loss": 0.8567373156547546, | |
| "rewards/accuracies": 0.903124988079071, | |
| "rewards/chosen": -0.06164557859301567, | |
| "rewards/margins": 0.0777212530374527, | |
| "rewards/rejected": -0.13936683535575867, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 2.455111111111111, | |
| "grad_norm": 36.144256591796875, | |
| "learning_rate": 4.516778136213037e-06, | |
| "log_odds_chosen": 1.2632501125335693, | |
| "log_odds_ratio": -0.3451572358608246, | |
| "logps/chosen": -0.6035677194595337, | |
| "logps/rejected": -1.296513319015503, | |
| "loss": 27.0539, | |
| "nll_loss": 0.8455725908279419, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.06035677716135979, | |
| "rewards/margins": 0.06929457187652588, | |
| "rewards/rejected": -0.12965135276317596, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.5025185185185186, | |
| "grad_norm": 49.33629608154297, | |
| "learning_rate": 3.7531701693965554e-06, | |
| "log_odds_chosen": 1.1750519275665283, | |
| "log_odds_ratio": -0.3728707730770111, | |
| "logps/chosen": -0.6593388915061951, | |
| "logps/rejected": -1.3207858800888062, | |
| "loss": 27.5885, | |
| "nll_loss": 0.862401008605957, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": -0.06593388319015503, | |
| "rewards/margins": 0.06614471226930618, | |
| "rewards/rejected": -0.1320786029100418, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 2.549925925925926, | |
| "grad_norm": 42.38345718383789, | |
| "learning_rate": 3.055003141378948e-06, | |
| "log_odds_chosen": 1.3022594451904297, | |
| "log_odds_ratio": -0.35613012313842773, | |
| "logps/chosen": -0.625469982624054, | |
| "logps/rejected": -1.349654197692871, | |
| "loss": 26.6012, | |
| "nll_loss": 0.8316141366958618, | |
| "rewards/accuracies": 0.878125011920929, | |
| "rewards/chosen": -0.0625469908118248, | |
| "rewards/margins": 0.07241840660572052, | |
| "rewards/rejected": -0.1349654197692871, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 2.5973333333333333, | |
| "grad_norm": 41.206031799316406, | |
| "learning_rate": 2.424427429704365e-06, | |
| "log_odds_chosen": 1.2805674076080322, | |
| "log_odds_ratio": -0.3571448028087616, | |
| "logps/chosen": -0.646803081035614, | |
| "logps/rejected": -1.382863163948059, | |
| "loss": 26.8844, | |
| "nll_loss": 0.8408975601196289, | |
| "rewards/accuracies": 0.890625, | |
| "rewards/chosen": -0.0646803081035614, | |
| "rewards/margins": 0.07360599935054779, | |
| "rewards/rejected": -0.1382863074541092, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 2.644740740740741, | |
| "grad_norm": 41.09498977661133, | |
| "learning_rate": 1.8633852284264508e-06, | |
| "log_odds_chosen": 1.1956632137298584, | |
| "log_odds_ratio": -0.37835055589675903, | |
| "logps/chosen": -0.6253945231437683, | |
| "logps/rejected": -1.2808663845062256, | |
| "loss": 27.366, | |
| "nll_loss": 0.8556842803955078, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": -0.06253945827484131, | |
| "rewards/margins": 0.06554718315601349, | |
| "rewards/rejected": -0.1280866414308548, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.6921481481481484, | |
| "grad_norm": 46.02020263671875, | |
| "learning_rate": 1.3736045660864034e-06, | |
| "log_odds_chosen": 1.1979916095733643, | |
| "log_odds_ratio": -0.39684250950813293, | |
| "logps/chosen": -0.6870399713516235, | |
| "logps/rejected": -1.3697882890701294, | |
| "loss": 28.5888, | |
| "nll_loss": 0.893951416015625, | |
| "rewards/accuracies": 0.831250011920929, | |
| "rewards/chosen": -0.06870399415493011, | |
| "rewards/margins": 0.06827482581138611, | |
| "rewards/rejected": -0.1369788497686386, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 2.7395555555555555, | |
| "grad_norm": 37.892303466796875, | |
| "learning_rate": 9.565939833279192e-07, | |
| "log_odds_chosen": 1.2452318668365479, | |
| "log_odds_ratio": -0.37130576372146606, | |
| "logps/chosen": -0.6673291325569153, | |
| "logps/rejected": -1.3500759601593018, | |
| "loss": 27.5715, | |
| "nll_loss": 0.8617628812789917, | |
| "rewards/accuracies": 0.8531249761581421, | |
| "rewards/chosen": -0.06673291325569153, | |
| "rewards/margins": 0.06827469915151596, | |
| "rewards/rejected": -0.13500761985778809, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 2.786962962962963, | |
| "grad_norm": 39.933834075927734, | |
| "learning_rate": 6.136378865420872e-07, | |
| "log_odds_chosen": 1.1233781576156616, | |
| "log_odds_ratio": -0.394573837518692, | |
| "logps/chosen": -0.6726820468902588, | |
| "logps/rejected": -1.2876471281051636, | |
| "loss": 28.6167, | |
| "nll_loss": 0.8944187164306641, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.06726820766925812, | |
| "rewards/margins": 0.06149650737643242, | |
| "rewards/rejected": -0.12876471877098083, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 2.83437037037037, | |
| "grad_norm": 49.5613899230957, | |
| "learning_rate": 3.45792591853214e-07, | |
| "log_odds_chosen": 1.2906441688537598, | |
| "log_odds_ratio": -0.37283915281295776, | |
| "logps/chosen": -0.6245847344398499, | |
| "logps/rejected": -1.3110054731369019, | |
| "loss": 27.1459, | |
| "nll_loss": 0.8486896753311157, | |
| "rewards/accuracies": 0.840624988079071, | |
| "rewards/chosen": -0.062458474189043045, | |
| "rewards/margins": 0.06864207237958908, | |
| "rewards/rejected": -0.13110056519508362, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.8817777777777778, | |
| "grad_norm": 53.47651290893555, | |
| "learning_rate": 1.538830716302092e-07, | |
| "log_odds_chosen": 1.3686002492904663, | |
| "log_odds_ratio": -0.3520243167877197, | |
| "logps/chosen": -0.5930364727973938, | |
| "logps/rejected": -1.3434317111968994, | |
| "loss": 26.485, | |
| "nll_loss": 0.8286612629890442, | |
| "rewards/accuracies": 0.878125011920929, | |
| "rewards/chosen": -0.0593036524951458, | |
| "rewards/margins": 0.07503952085971832, | |
| "rewards/rejected": -0.13434317708015442, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 2.9291851851851853, | |
| "grad_norm": 41.2106819152832, | |
| "learning_rate": 3.8500413544415025e-08, | |
| "log_odds_chosen": 1.4645073413848877, | |
| "log_odds_ratio": -0.2996174693107605, | |
| "logps/chosen": -0.5810352563858032, | |
| "logps/rejected": -1.3896446228027344, | |
| "loss": 26.7052, | |
| "nll_loss": 0.834790050983429, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": -0.0581035315990448, | |
| "rewards/margins": 0.08086094260215759, | |
| "rewards/rejected": -0.1389644742012024, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 2.9765925925925925, | |
| "grad_norm": 42.9834098815918, | |
| "learning_rate": 0.0, | |
| "log_odds_chosen": 1.2527662515640259, | |
| "log_odds_ratio": -0.354759156703949, | |
| "logps/chosen": -0.5776538848876953, | |
| "logps/rejected": -1.2106915712356567, | |
| "loss": 25.6734, | |
| "nll_loss": 0.8024199604988098, | |
| "rewards/accuracies": 0.8656250238418579, | |
| "rewards/chosen": -0.05776538699865341, | |
| "rewards/margins": 0.06330376863479614, | |
| "rewards/rejected": -0.12106915563344955, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 2.9765925925925925, | |
| "step": 315, | |
| "total_flos": 0.0, | |
| "train_loss": 54.311234392438614, | |
| "train_runtime": 9726.6555, | |
| "train_samples_per_second": 2.082, | |
| "train_steps_per_second": 0.032 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 315, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 100000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |