| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 2.3792731903222646, |
| "eval_steps": 35, |
| "global_step": 1520, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.023508668821627974, |
| "grad_norm": 66.55298614501953, |
| "learning_rate": 9.333333333333334e-08, |
| "logits/chosen": -0.8181892037391663, |
| "logits/rejected": -0.6109388470649719, |
| "logps/chosen": -360.2866516113281, |
| "logps/rejected": -290.2643127441406, |
| "loss": 0.6916, |
| "rewards/accuracies": 0.4333333373069763, |
| "rewards/chosen": 0.0018199360929429531, |
| "rewards/margins": 0.004990327637642622, |
| "rewards/rejected": -0.0031703924760222435, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.04701733764325595, |
| "grad_norm": 80.19271850585938, |
| "learning_rate": 1.9333333333333332e-07, |
| "logits/chosen": -0.9239326119422913, |
| "logits/rejected": -0.6111201643943787, |
| "logps/chosen": -361.14398193359375, |
| "logps/rejected": -279.4661865234375, |
| "loss": 0.6957, |
| "rewards/accuracies": 0.4791666567325592, |
| "rewards/chosen": -0.0014068834716454148, |
| "rewards/margins": -0.002790238242596388, |
| "rewards/rejected": 0.0013833552366122603, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.05485356058379861, |
| "eval_logits/chosen": -0.9112520217895508, |
| "eval_logits/rejected": -0.6521649360656738, |
| "eval_logps/chosen": -352.1986999511719, |
| "eval_logps/rejected": -285.4880065917969, |
| "eval_loss": 0.6858627796173096, |
| "eval_rewards/accuracies": 0.5549542307853699, |
| "eval_rewards/chosen": 0.011523518711328506, |
| "eval_rewards/margins": 0.017005058005452156, |
| "eval_rewards/rejected": -0.005481537897139788, |
| "eval_runtime": 181.0795, |
| "eval_samples_per_second": 13.265, |
| "eval_steps_per_second": 6.632, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.07052600646488393, |
| "grad_norm": 61.10358428955078, |
| "learning_rate": 2.933333333333333e-07, |
| "logits/chosen": -0.983363926410675, |
| "logits/rejected": -0.6932616829872131, |
| "logps/chosen": -343.0465393066406, |
| "logps/rejected": -297.50042724609375, |
| "loss": 0.6847, |
| "rewards/accuracies": 0.5562499761581421, |
| "rewards/chosen": 0.012528100982308388, |
| "rewards/margins": 0.019641580060124397, |
| "rewards/rejected": -0.0071134804747998714, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.0940346752865119, |
| "grad_norm": 62.27146911621094, |
| "learning_rate": 3.933333333333333e-07, |
| "logits/chosen": -0.923626720905304, |
| "logits/rejected": -0.6753237247467041, |
| "logps/chosen": -379.0172119140625, |
| "logps/rejected": -298.9611511230469, |
| "loss": 0.6765, |
| "rewards/accuracies": 0.6041666865348816, |
| "rewards/chosen": 0.024433977901935577, |
| "rewards/margins": 0.038544632494449615, |
| "rewards/rejected": -0.014110651798546314, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.10970712116759722, |
| "eval_logits/chosen": -0.8566208481788635, |
| "eval_logits/rejected": -0.5961222648620605, |
| "eval_logps/chosen": -351.8851318359375, |
| "eval_logps/rejected": -285.7581481933594, |
| "eval_loss": 0.6613612771034241, |
| "eval_rewards/accuracies": 0.6353039145469666, |
| "eval_rewards/chosen": 0.04288659617304802, |
| "eval_rewards/margins": 0.07538340985774994, |
| "eval_rewards/rejected": -0.03249681368470192, |
| "eval_runtime": 181.1266, |
| "eval_samples_per_second": 13.261, |
| "eval_steps_per_second": 6.631, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.11754334410813988, |
| "grad_norm": 60.890411376953125, |
| "learning_rate": 4.933333333333333e-07, |
| "logits/chosen": -0.8443899154663086, |
| "logits/rejected": -0.5239131450653076, |
| "logps/chosen": -340.6385803222656, |
| "logps/rejected": -286.01763916015625, |
| "loss": 0.6611, |
| "rewards/accuracies": 0.6520833373069763, |
| "rewards/chosen": 0.03822815418243408, |
| "rewards/margins": 0.07544167339801788, |
| "rewards/rejected": -0.0372135154902935, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.14105201292976785, |
| "grad_norm": 61.06071853637695, |
| "learning_rate": 5.933333333333334e-07, |
| "logits/chosen": -0.8115103840827942, |
| "logits/rejected": -0.5216272473335266, |
| "logps/chosen": -354.4888916015625, |
| "logps/rejected": -267.555419921875, |
| "loss": 0.637, |
| "rewards/accuracies": 0.6895833611488342, |
| "rewards/chosen": 0.05075030028820038, |
| "rewards/margins": 0.13897046446800232, |
| "rewards/rejected": -0.08822017163038254, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.16456068175139582, |
| "grad_norm": 56.0710563659668, |
| "learning_rate": 6.933333333333333e-07, |
| "logits/chosen": -0.8203279972076416, |
| "logits/rejected": -0.4818764925003052, |
| "logps/chosen": -363.0785827636719, |
| "logps/rejected": -297.43756103515625, |
| "loss": 0.6295, |
| "rewards/accuracies": 0.6729166507720947, |
| "rewards/chosen": 0.053048841655254364, |
| "rewards/margins": 0.17447903752326965, |
| "rewards/rejected": -0.12143018841743469, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.16456068175139582, |
| "eval_logits/chosen": -0.8201783895492554, |
| "eval_logits/rejected": -0.5542652606964111, |
| "eval_logps/chosen": -351.9151916503906, |
| "eval_logps/rejected": -286.86541748046875, |
| "eval_loss": 0.6280434131622314, |
| "eval_rewards/accuracies": 0.6615320444107056, |
| "eval_rewards/chosen": 0.0398729108273983, |
| "eval_rewards/margins": 0.18310010433197021, |
| "eval_rewards/rejected": -0.1432272046804428, |
| "eval_runtime": 181.1366, |
| "eval_samples_per_second": 13.261, |
| "eval_steps_per_second": 6.63, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.1880693505730238, |
| "grad_norm": 54.890419006347656, |
| "learning_rate": 7.933333333333333e-07, |
| "logits/chosen": -0.8400412201881409, |
| "logits/rejected": -0.5213409662246704, |
| "logps/chosen": -365.0088195800781, |
| "logps/rejected": -308.81427001953125, |
| "loss": 0.6043, |
| "rewards/accuracies": 0.7104166746139526, |
| "rewards/chosen": 0.053323231637477875, |
| "rewards/margins": 0.24515444040298462, |
| "rewards/rejected": -0.19183121621608734, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.2115780193946518, |
| "grad_norm": 60.136470794677734, |
| "learning_rate": 8.933333333333333e-07, |
| "logits/chosen": -0.8502958416938782, |
| "logits/rejected": -0.5787886381149292, |
| "logps/chosen": -350.45751953125, |
| "logps/rejected": -284.59967041015625, |
| "loss": 0.6126, |
| "rewards/accuracies": 0.6854166388511658, |
| "rewards/chosen": 0.02512473240494728, |
| "rewards/margins": 0.2510988712310791, |
| "rewards/rejected": -0.22597412765026093, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.21941424233519444, |
| "eval_logits/chosen": -0.8695166707038879, |
| "eval_logits/rejected": -0.6029744148254395, |
| "eval_logps/chosen": -351.90997314453125, |
| "eval_logps/rejected": -287.88140869140625, |
| "eval_loss": 0.5991479158401489, |
| "eval_rewards/accuracies": 0.6965029239654541, |
| "eval_rewards/chosen": 0.0403980053961277, |
| "eval_rewards/margins": 0.28522104024887085, |
| "eval_rewards/rejected": -0.24482303857803345, |
| "eval_runtime": 181.2258, |
| "eval_samples_per_second": 13.254, |
| "eval_steps_per_second": 6.627, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.23508668821627976, |
| "grad_norm": 64.90687561035156, |
| "learning_rate": 9.933333333333333e-07, |
| "logits/chosen": -0.8242729306221008, |
| "logits/rejected": -0.5118594765663147, |
| "logps/chosen": -360.62066650390625, |
| "logps/rejected": -290.8011779785156, |
| "loss": 0.6232, |
| "rewards/accuracies": 0.6604166626930237, |
| "rewards/chosen": 0.005629782099276781, |
| "rewards/margins": 0.2206929475069046, |
| "rewards/rejected": -0.2150631695985794, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.2585953570379077, |
| "grad_norm": 55.48398208618164, |
| "learning_rate": 9.998451182298019e-07, |
| "logits/chosen": -0.903767466545105, |
| "logits/rejected": -0.5809265375137329, |
| "logps/chosen": -338.2618713378906, |
| "logps/rejected": -285.73797607421875, |
| "loss": 0.6014, |
| "rewards/accuracies": 0.6770833134651184, |
| "rewards/chosen": 0.0597819946706295, |
| "rewards/margins": 0.28081849217414856, |
| "rewards/rejected": -0.22103647887706757, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.27426780291899305, |
| "eval_logits/chosen": -0.9308104515075684, |
| "eval_logits/rejected": -0.6681538224220276, |
| "eval_logps/chosen": -351.89520263671875, |
| "eval_logps/rejected": -288.6672058105469, |
| "eval_loss": 0.5756543874740601, |
| "eval_rewards/accuracies": 0.7173188924789429, |
| "eval_rewards/chosen": 0.04187745228409767, |
| "eval_rewards/margins": 0.36528098583221436, |
| "eval_rewards/rejected": -0.32340356707572937, |
| "eval_runtime": 181.4158, |
| "eval_samples_per_second": 13.24, |
| "eval_steps_per_second": 6.62, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.2821040258595357, |
| "grad_norm": 53.91843032836914, |
| "learning_rate": 9.993355436786068e-07, |
| "logits/chosen": -0.9650304317474365, |
| "logits/rejected": -0.5752300024032593, |
| "logps/chosen": -352.0422668457031, |
| "logps/rejected": -305.9342041015625, |
| "loss": 0.5756, |
| "rewards/accuracies": 0.7145833373069763, |
| "rewards/chosen": 0.0450352244079113, |
| "rewards/margins": 0.34943825006484985, |
| "rewards/rejected": -0.3044029772281647, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.3056126946811637, |
| "grad_norm": 58.17976379394531, |
| "learning_rate": 9.98470848495185e-07, |
| "logits/chosen": -0.9314310550689697, |
| "logits/rejected": -0.6725074648857117, |
| "logps/chosen": -367.3880310058594, |
| "logps/rejected": -280.9293518066406, |
| "loss": 0.5662, |
| "rewards/accuracies": 0.71875, |
| "rewards/chosen": 0.03602181747555733, |
| "rewards/margins": 0.40341871976852417, |
| "rewards/rejected": -0.3673969507217407, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.32912136350279164, |
| "grad_norm": 52.23281478881836, |
| "learning_rate": 9.972516476389642e-07, |
| "logits/chosen": -0.9290924668312073, |
| "logits/rejected": -0.6681386232376099, |
| "logps/chosen": -382.5225830078125, |
| "logps/rejected": -288.5783386230469, |
| "loss": 0.5534, |
| "rewards/accuracies": 0.7354166507720947, |
| "rewards/chosen": 0.009489117190241814, |
| "rewards/margins": 0.4669964611530304, |
| "rewards/rejected": -0.45750725269317627, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.32912136350279164, |
| "eval_logits/chosen": -0.9316383004188538, |
| "eval_logits/rejected": -0.6768124103546143, |
| "eval_logps/chosen": -352.13800048828125, |
| "eval_logps/rejected": -289.91552734375, |
| "eval_loss": 0.5547987222671509, |
| "eval_rewards/accuracies": 0.7285595536231995, |
| "eval_rewards/chosen": 0.01759806089103222, |
| "eval_rewards/margins": 0.4658346474170685, |
| "eval_rewards/rejected": -0.4482365548610687, |
| "eval_runtime": 180.7226, |
| "eval_samples_per_second": 13.291, |
| "eval_steps_per_second": 6.646, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.35263003232441964, |
| "grad_norm": 51.5745849609375, |
| "learning_rate": 9.956788081889744e-07, |
| "logits/chosen": -1.0272151231765747, |
| "logits/rejected": -0.7074090242385864, |
| "logps/chosen": -327.7598876953125, |
| "logps/rejected": -291.8259582519531, |
| "loss": 0.5717, |
| "rewards/accuracies": 0.7083333134651184, |
| "rewards/chosen": -0.03899319842457771, |
| "rewards/margins": 0.4320748746395111, |
| "rewards/rejected": -0.47106799483299255, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.3761387011460476, |
| "grad_norm": 52.133583068847656, |
| "learning_rate": 9.93753448727193e-07, |
| "logits/chosen": -0.9769286513328552, |
| "logits/rejected": -0.6823646426200867, |
| "logps/chosen": -356.5062255859375, |
| "logps/rejected": -284.9337463378906, |
| "loss": 0.5371, |
| "rewards/accuracies": 0.7583333253860474, |
| "rewards/chosen": 0.034630779176950455, |
| "rewards/margins": 0.5447677373886108, |
| "rewards/rejected": -0.5101370215415955, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.3839749240865903, |
| "eval_logits/chosen": -0.9463596940040588, |
| "eval_logits/rejected": -0.6955367922782898, |
| "eval_logps/chosen": -352.4129943847656, |
| "eval_logps/rejected": -291.015380859375, |
| "eval_loss": 0.5392794013023376, |
| "eval_rewards/accuracies": 0.729808509349823, |
| "eval_rewards/chosen": -0.009902803227305412, |
| "eval_rewards/margins": 0.5483154058456421, |
| "eval_rewards/rejected": -0.5582181811332703, |
| "eval_runtime": 181.8777, |
| "eval_samples_per_second": 13.207, |
| "eval_steps_per_second": 6.603, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.3996473699676756, |
| "grad_norm": 43.88309097290039, |
| "learning_rate": 9.914769385430247e-07, |
| "logits/chosen": -0.9463251233100891, |
| "logits/rejected": -0.6262016296386719, |
| "logps/chosen": -357.5391845703125, |
| "logps/rejected": -307.8315124511719, |
| "loss": 0.5629, |
| "rewards/accuracies": 0.7124999761581421, |
| "rewards/chosen": -0.028146782889962196, |
| "rewards/margins": 0.5159488320350647, |
| "rewards/rejected": -0.5440956354141235, |
| "step": 255 |
| }, |
| { |
| "epoch": 0.4231560387893036, |
| "grad_norm": 64.59564971923828, |
| "learning_rate": 9.888508966594814e-07, |
| "logits/chosen": -0.9834640026092529, |
| "logits/rejected": -0.6821132898330688, |
| "logps/chosen": -361.8191833496094, |
| "logps/rejected": -296.5501708984375, |
| "loss": 0.5284, |
| "rewards/accuracies": 0.7479166388511658, |
| "rewards/chosen": -0.042962126433849335, |
| "rewards/margins": 0.56870037317276, |
| "rewards/rejected": -0.6116625070571899, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.4388284846703889, |
| "eval_logits/chosen": -0.9437891244888306, |
| "eval_logits/rejected": -0.6993773579597473, |
| "eval_logps/chosen": -353.16070556640625, |
| "eval_logps/rejected": -292.8016662597656, |
| "eval_loss": 0.5247384905815125, |
| "eval_rewards/accuracies": 0.7348043322563171, |
| "eval_rewards/chosen": -0.0846758484840393, |
| "eval_rewards/margins": 0.6521764397621155, |
| "eval_rewards/rejected": -0.7368523478507996, |
| "eval_runtime": 181.0518, |
| "eval_samples_per_second": 13.267, |
| "eval_steps_per_second": 6.633, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.4466647076109315, |
| "grad_norm": 68.69232177734375, |
| "learning_rate": 9.85877190681755e-07, |
| "logits/chosen": -0.9795821309089661, |
| "logits/rejected": -0.7652584314346313, |
| "logps/chosen": -367.8410339355469, |
| "logps/rejected": -297.42730712890625, |
| "loss": 0.5191, |
| "rewards/accuracies": 0.7708333134651184, |
| "rewards/chosen": -0.10684531182050705, |
| "rewards/margins": 0.6454100012779236, |
| "rewards/rejected": -0.752255380153656, |
| "step": 285 |
| }, |
| { |
| "epoch": 0.4701733764325595, |
| "grad_norm": 44.80474853515625, |
| "learning_rate": 9.825579354690029e-07, |
| "logits/chosen": -1.0019594430923462, |
| "logits/rejected": -0.7404360175132751, |
| "logps/chosen": -349.2023620605469, |
| "logps/rejected": -286.4942932128906, |
| "loss": 0.5028, |
| "rewards/accuracies": 0.7604166865348816, |
| "rewards/chosen": -0.07686860114336014, |
| "rewards/margins": 0.6980764269828796, |
| "rewards/rejected": -0.774945080280304, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.49368204525418746, |
| "grad_norm": 64.48184967041016, |
| "learning_rate": 9.788954916302867e-07, |
| "logits/chosen": -1.033690333366394, |
| "logits/rejected": -0.7505324482917786, |
| "logps/chosen": -353.74407958984375, |
| "logps/rejected": -301.9696350097656, |
| "loss": 0.509, |
| "rewards/accuracies": 0.737500011920929, |
| "rewards/chosen": -0.19922642409801483, |
| "rewards/margins": 0.7380853891372681, |
| "rewards/rejected": -0.9373118877410889, |
| "step": 315 |
| }, |
| { |
| "epoch": 0.49368204525418746, |
| "eval_logits/chosen": -1.0560635328292847, |
| "eval_logits/rejected": -0.8106775879859924, |
| "eval_logps/chosen": -354.1845397949219, |
| "eval_logps/rejected": -294.6258239746094, |
| "eval_loss": 0.5152602195739746, |
| "eval_rewards/accuracies": 0.7435470223426819, |
| "eval_rewards/chosen": -0.18705680966377258, |
| "eval_rewards/margins": 0.7322102189064026, |
| "eval_rewards/rejected": -0.9192669987678528, |
| "eval_runtime": 181.0466, |
| "eval_samples_per_second": 13.267, |
| "eval_steps_per_second": 6.634, |
| "step": 315 |
| }, |
| { |
| "epoch": 0.5171907140758154, |
| "grad_norm": 60.448970794677734, |
| "learning_rate": 9.748924638457425e-07, |
| "logits/chosen": -1.1234601736068726, |
| "logits/rejected": -0.7872902750968933, |
| "logps/chosen": -363.0025329589844, |
| "logps/rejected": -309.54241943359375, |
| "loss": 0.4751, |
| "rewards/accuracies": 0.7562500238418579, |
| "rewards/chosen": -0.08163168281316757, |
| "rewards/margins": 0.8535270094871521, |
| "rewards/rejected": -0.9351587295532227, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.5406993828974435, |
| "grad_norm": 59.353973388671875, |
| "learning_rate": 9.705516990141652e-07, |
| "logits/chosen": -1.035413384437561, |
| "logits/rejected": -0.7837415337562561, |
| "logps/chosen": -374.60638427734375, |
| "logps/rejected": -337.9807434082031, |
| "loss": 0.5057, |
| "rewards/accuracies": 0.7395833134651184, |
| "rewards/chosen": -0.18711723387241364, |
| "rewards/margins": 0.7444573640823364, |
| "rewards/rejected": -0.9315746426582336, |
| "step": 345 |
| }, |
| { |
| "epoch": 0.5485356058379861, |
| "eval_logits/chosen": -1.0895341634750366, |
| "eval_logits/rejected": -0.8445096611976624, |
| "eval_logps/chosen": -354.7453918457031, |
| "eval_logps/rejected": -295.85137939453125, |
| "eval_loss": 0.5081992745399475, |
| "eval_rewards/accuracies": 0.7435470223426819, |
| "eval_rewards/chosen": -0.2431420534849167, |
| "eval_rewards/margins": 0.7986794114112854, |
| "eval_rewards/rejected": -1.0418214797973633, |
| "eval_runtime": 182.0988, |
| "eval_samples_per_second": 13.191, |
| "eval_steps_per_second": 6.595, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.5642080517190714, |
| "grad_norm": 65.68885040283203, |
| "learning_rate": 9.658762842283341e-07, |
| "logits/chosen": -1.07207453250885, |
| "logits/rejected": -0.7635911107063293, |
| "logps/chosen": -380.82977294921875, |
| "logps/rejected": -317.8447570800781, |
| "loss": 0.4982, |
| "rewards/accuracies": 0.75, |
| "rewards/chosen": -0.2277773916721344, |
| "rewards/margins": 0.8403520584106445, |
| "rewards/rejected": -1.0681294202804565, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.5877167205406993, |
| "grad_norm": 58.057342529296875, |
| "learning_rate": 9.608695445795146e-07, |
| "logits/chosen": -1.134562373161316, |
| "logits/rejected": -0.8592749238014221, |
| "logps/chosen": -372.07171630859375, |
| "logps/rejected": -310.8824768066406, |
| "loss": 0.4783, |
| "rewards/accuracies": 0.7645833492279053, |
| "rewards/chosen": -0.17556972801685333, |
| "rewards/margins": 0.9482249021530151, |
| "rewards/rejected": -1.1237945556640625, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.6033891664217848, |
| "eval_logits/chosen": -1.0469961166381836, |
| "eval_logits/rejected": -0.8057866096496582, |
| "eval_logps/chosen": -354.4659118652344, |
| "eval_logps/rejected": -296.0675354003906, |
| "eval_loss": 0.5002036690711975, |
| "eval_rewards/accuracies": 0.7522897720336914, |
| "eval_rewards/chosen": -0.21519356966018677, |
| "eval_rewards/margins": 0.8482457399368286, |
| "eval_rewards/rejected": -1.063439130783081, |
| "eval_runtime": 181.9108, |
| "eval_samples_per_second": 13.204, |
| "eval_steps_per_second": 6.602, |
| "step": 385 |
| }, |
| { |
| "epoch": 0.6112253893623274, |
| "grad_norm": 59.8174934387207, |
| "learning_rate": 9.555350407926977e-07, |
| "logits/chosen": -1.0428589582443237, |
| "logits/rejected": -0.7815055847167969, |
| "logps/chosen": -363.3075866699219, |
| "logps/rejected": -304.4619140625, |
| "loss": 0.5423, |
| "rewards/accuracies": 0.7104166746139526, |
| "rewards/chosen": -0.26339924335479736, |
| "rewards/margins": 0.7616356015205383, |
| "rewards/rejected": -1.0250346660614014, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.6347340581839553, |
| "grad_norm": 57.40476608276367, |
| "learning_rate": 9.498765666942621e-07, |
| "logits/chosen": -1.1355029344558716, |
| "logits/rejected": -0.8117865324020386, |
| "logps/chosen": -338.7142028808594, |
| "logps/rejected": -303.7315673828125, |
| "loss": 0.479, |
| "rewards/accuracies": 0.768750011920929, |
| "rewards/chosen": -0.238851860165596, |
| "rewards/margins": 0.918969988822937, |
| "rewards/rejected": -1.1578218936920166, |
| "step": 405 |
| }, |
| { |
| "epoch": 0.6582427270055833, |
| "grad_norm": 58.45095443725586, |
| "learning_rate": 9.438981465138564e-07, |
| "logits/chosen": -1.0798269510269165, |
| "logits/rejected": -0.860160231590271, |
| "logps/chosen": -368.1922607421875, |
| "logps/rejected": -300.15185546875, |
| "loss": 0.4952, |
| "rewards/accuracies": 0.75, |
| "rewards/chosen": -0.2948712706565857, |
| "rewards/margins": 0.9247942566871643, |
| "rewards/rejected": -1.2196654081344604, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.6582427270055833, |
| "eval_logits/chosen": -1.1024389266967773, |
| "eval_logits/rejected": -0.8648662567138672, |
| "eval_logps/chosen": -355.2312316894531, |
| "eval_logps/rejected": -297.4045104980469, |
| "eval_loss": 0.4926217198371887, |
| "eval_rewards/accuracies": 0.7572855949401855, |
| "eval_rewards/chosen": -0.29172664880752563, |
| "eval_rewards/margins": 0.9054065942764282, |
| "eval_rewards/rejected": -1.197133183479309, |
| "eval_runtime": 181.474, |
| "eval_samples_per_second": 13.236, |
| "eval_steps_per_second": 6.618, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.6817513958272113, |
| "grad_norm": 67.23454284667969, |
| "learning_rate": 9.376040320224207e-07, |
| "logits/chosen": -1.1256804466247559, |
| "logits/rejected": -0.778734028339386, |
| "logps/chosen": -353.2846374511719, |
| "logps/rejected": -294.6732482910156, |
| "loss": 0.5274, |
| "rewards/accuracies": 0.737500011920929, |
| "rewards/chosen": -0.3149377405643463, |
| "rewards/margins": 0.8212650418281555, |
| "rewards/rejected": -1.1362026929855347, |
| "step": 435 |
| }, |
| { |
| "epoch": 0.7052600646488393, |
| "grad_norm": 43.976097106933594, |
| "learning_rate": 9.309986995083865e-07, |
| "logits/chosen": -1.0163235664367676, |
| "logits/rejected": -0.7996082305908203, |
| "logps/chosen": -347.673828125, |
| "logps/rejected": -304.9941101074219, |
| "loss": 0.487, |
| "rewards/accuracies": 0.7520833611488342, |
| "rewards/chosen": -0.27800673246383667, |
| "rewards/margins": 0.9257082343101501, |
| "rewards/rejected": -1.2037149667739868, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.7130962875893819, |
| "eval_logits/chosen": -1.0526485443115234, |
| "eval_logits/rejected": -0.819446325302124, |
| "eval_logps/chosen": -355.0964050292969, |
| "eval_logps/rejected": -297.8984069824219, |
| "eval_loss": 0.4861847162246704, |
| "eval_rewards/accuracies": 0.7643630504608154, |
| "eval_rewards/chosen": -0.2782430350780487, |
| "eval_rewards/margins": 0.9682838916778564, |
| "eval_rewards/rejected": -1.246526837348938, |
| "eval_runtime": 181.8521, |
| "eval_samples_per_second": 13.209, |
| "eval_steps_per_second": 6.604, |
| "step": 455 |
| }, |
| { |
| "epoch": 0.7287687334704672, |
| "grad_norm": 44.1576042175293, |
| "learning_rate": 9.240868465942004e-07, |
| "logits/chosen": -1.0843127965927124, |
| "logits/rejected": -0.8464950323104858, |
| "logps/chosen": -368.0324401855469, |
| "logps/rejected": -298.4980773925781, |
| "loss": 0.443, |
| "rewards/accuracies": 0.7895833253860474, |
| "rewards/chosen": -0.20106664299964905, |
| "rewards/margins": 1.0733890533447266, |
| "rewards/rejected": -1.2744557857513428, |
| "step": 465 |
| }, |
| { |
| "epoch": 0.7522774022920952, |
| "grad_norm": 75.02023315429688, |
| "learning_rate": 9.168733888954398e-07, |
| "logits/chosen": -0.9654378294944763, |
| "logits/rejected": -0.766428530216217, |
| "logps/chosen": -368.84527587890625, |
| "logps/rejected": -310.3114013671875, |
| "loss": 0.4913, |
| "rewards/accuracies": 0.7729166746139526, |
| "rewards/chosen": -0.31580713391304016, |
| "rewards/margins": 0.9536004662513733, |
| "rewards/rejected": -1.2694075107574463, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.7679498481731806, |
| "eval_logits/chosen": -1.018790364265442, |
| "eval_logits/rejected": -0.7895683646202087, |
| "eval_logps/chosen": -355.1246032714844, |
| "eval_logps/rejected": -298.2669677734375, |
| "eval_loss": 0.4843537211418152, |
| "eval_rewards/accuracies": 0.7597835063934326, |
| "eval_rewards/chosen": -0.2810628116130829, |
| "eval_rewards/margins": 1.002318024635315, |
| "eval_rewards/rejected": -1.2833808660507202, |
| "eval_runtime": 181.6862, |
| "eval_samples_per_second": 13.221, |
| "eval_steps_per_second": 6.61, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.7757860711137232, |
| "grad_norm": 57.9295768737793, |
| "learning_rate": 9.093634565248934e-07, |
| "logits/chosen": -1.0683478116989136, |
| "logits/rejected": -0.7814379930496216, |
| "logps/chosen": -346.7208251953125, |
| "logps/rejected": -305.7705383300781, |
| "loss": 0.4662, |
| "rewards/accuracies": 0.762499988079071, |
| "rewards/chosen": -0.22331449389457703, |
| "rewards/margins": 1.0663999319076538, |
| "rewards/rejected": -1.2897144556045532, |
| "step": 495 |
| }, |
| { |
| "epoch": 0.7992947399353512, |
| "grad_norm": 64.89127349853516, |
| "learning_rate": 9.015623904440955e-07, |
| "logits/chosen": -1.0165103673934937, |
| "logits/rejected": -0.7501361966133118, |
| "logps/chosen": -360.36444091796875, |
| "logps/rejected": -311.2584533691406, |
| "loss": 0.4763, |
| "rewards/accuracies": 0.7729166746139526, |
| "rewards/chosen": -0.3126125931739807, |
| "rewards/margins": 1.0347987413406372, |
| "rewards/rejected": -1.3474112749099731, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.8228034087569791, |
| "grad_norm": 57.808197021484375, |
| "learning_rate": 8.934757386649061e-07, |
| "logits/chosen": -1.0641908645629883, |
| "logits/rejected": -0.7976428866386414, |
| "logps/chosen": -331.990234375, |
| "logps/rejected": -290.6216125488281, |
| "loss": 0.4478, |
| "rewards/accuracies": 0.7854166626930237, |
| "rewards/chosen": -0.32272768020629883, |
| "rewards/margins": 1.1051963567733765, |
| "rewards/rejected": -1.4279239177703857, |
| "step": 525 |
| }, |
| { |
| "epoch": 0.8228034087569791, |
| "eval_logits/chosen": -1.0868600606918335, |
| "eval_logits/rejected": -0.8659496307373047, |
| "eval_logps/chosen": -355.5617370605469, |
| "eval_logps/rejected": -299.2372741699219, |
| "eval_loss": 0.4808862507343292, |
| "eval_rewards/accuracies": 0.7585345506668091, |
| "eval_rewards/chosen": -0.32477837800979614, |
| "eval_rewards/margins": 1.0556353330612183, |
| "eval_rewards/rejected": -1.3804137706756592, |
| "eval_runtime": 182.9841, |
| "eval_samples_per_second": 13.127, |
| "eval_steps_per_second": 6.563, |
| "step": 525 |
| }, |
| { |
| "epoch": 0.8463120775786072, |
| "grad_norm": 53.98827362060547, |
| "learning_rate": 8.851092523038417e-07, |
| "logits/chosen": -1.118863821029663, |
| "logits/rejected": -0.8719570636749268, |
| "logps/chosen": -363.63299560546875, |
| "logps/rejected": -302.99853515625, |
| "loss": 0.4774, |
| "rewards/accuracies": 0.7770833373069763, |
| "rewards/chosen": -0.3193167746067047, |
| "rewards/margins": 1.0706781148910522, |
| "rewards/rejected": -1.3899948596954346, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.8698207464002351, |
| "grad_norm": 53.19316482543945, |
| "learning_rate": 8.764688814919589e-07, |
| "logits/chosen": -1.0061967372894287, |
| "logits/rejected": -0.8136522173881531, |
| "logps/chosen": -387.64447021484375, |
| "logps/rejected": -306.2342529296875, |
| "loss": 0.452, |
| "rewards/accuracies": 0.7541666626930237, |
| "rewards/chosen": -0.3774346113204956, |
| "rewards/margins": 1.1483807563781738, |
| "rewards/rejected": -1.525815486907959, |
| "step": 555 |
| }, |
| { |
| "epoch": 0.8776569693407777, |
| "eval_logits/chosen": -1.0304757356643677, |
| "eval_logits/rejected": -0.8115791082382202, |
| "eval_logps/chosen": -355.95196533203125, |
| "eval_logps/rejected": -300.0523681640625, |
| "eval_loss": 0.47852277755737305, |
| "eval_rewards/accuracies": 0.7635303735733032, |
| "eval_rewards/chosen": -0.36379873752593994, |
| "eval_rewards/margins": 1.09812331199646, |
| "eval_rewards/rejected": -1.4619219303131104, |
| "eval_runtime": 182.2848, |
| "eval_samples_per_second": 13.177, |
| "eval_steps_per_second": 6.589, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.893329415221863, |
| "grad_norm": 52.40211486816406, |
| "learning_rate": 8.675607711432023e-07, |
| "logits/chosen": -1.012203574180603, |
| "logits/rejected": -0.8512925505638123, |
| "logps/chosen": -353.43341064453125, |
| "logps/rejected": -299.34576416015625, |
| "loss": 0.4917, |
| "rewards/accuracies": 0.7583333253860474, |
| "rewards/chosen": -0.38326555490493774, |
| "rewards/margins": 1.1050751209259033, |
| "rewards/rejected": -1.4883407354354858, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.916838084043491, |
| "grad_norm": 77.06608581542969, |
| "learning_rate": 8.583912565842256e-07, |
| "logits/chosen": -1.0286543369293213, |
| "logits/rejected": -0.8386385440826416, |
| "logps/chosen": -363.56439208984375, |
| "logps/rejected": -330.7981872558594, |
| "loss": 0.4654, |
| "rewards/accuracies": 0.7833333611488342, |
| "rewards/chosen": -0.44688889384269714, |
| "rewards/margins": 1.2291133403778076, |
| "rewards/rejected": -1.6760022640228271, |
| "step": 585 |
| }, |
| { |
| "epoch": 0.9325105299245764, |
| "eval_logits/chosen": -1.075386643409729, |
| "eval_logits/rejected": -0.8566927909851074, |
| "eval_logps/chosen": -355.85455322265625, |
| "eval_logps/rejected": -299.8857727050781, |
| "eval_loss": 0.4771224856376648, |
| "eval_rewards/accuracies": 0.7572855949401855, |
| "eval_rewards/chosen": -0.3540586531162262, |
| "eval_rewards/margins": 1.091202735900879, |
| "eval_rewards/rejected": -1.4452612400054932, |
| "eval_runtime": 181.9619, |
| "eval_samples_per_second": 13.201, |
| "eval_steps_per_second": 6.6, |
| "step": 595 |
| }, |
| { |
| "epoch": 0.940346752865119, |
| "grad_norm": 62.413719177246094, |
| "learning_rate": 8.489668590487934e-07, |
| "logits/chosen": -1.0846012830734253, |
| "logits/rejected": -0.7979000210762024, |
| "logps/chosen": -384.29083251953125, |
| "logps/rejected": -316.1002197265625, |
| "loss": 0.4709, |
| "rewards/accuracies": 0.7708333134651184, |
| "rewards/chosen": -0.3900051712989807, |
| "rewards/margins": 1.184963345527649, |
| "rewards/rejected": -1.574968695640564, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.963855421686747, |
| "grad_norm": 53.96653366088867, |
| "learning_rate": 8.392942810399692e-07, |
| "logits/chosen": -1.0858737230300903, |
| "logits/rejected": -0.7765557765960693, |
| "logps/chosen": -354.53985595703125, |
| "logps/rejected": -297.6081848144531, |
| "loss": 0.4754, |
| "rewards/accuracies": 0.7708333134651184, |
| "rewards/chosen": -0.3535262644290924, |
| "rewards/margins": 1.0648608207702637, |
| "rewards/rejected": -1.4183870553970337, |
| "step": 615 |
| }, |
| { |
| "epoch": 0.9873640905083749, |
| "grad_norm": 52.692047119140625, |
| "learning_rate": 8.293804015633861e-07, |
| "logits/chosen": -0.9902106523513794, |
| "logits/rejected": -0.8622989654541016, |
| "logps/chosen": -364.4175109863281, |
| "logps/rejected": -307.4686279296875, |
| "loss": 0.4829, |
| "rewards/accuracies": 0.7791666388511658, |
| "rewards/chosen": -0.3149837851524353, |
| "rewards/margins": 1.1350115537643433, |
| "rewards/rejected": -1.4499951601028442, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.9873640905083749, |
| "eval_logits/chosen": -1.1149685382843018, |
| "eval_logits/rejected": -0.8937057256698608, |
| "eval_logps/chosen": -355.9053649902344, |
| "eval_logps/rejected": -300.0727233886719, |
| "eval_loss": 0.4741358458995819, |
| "eval_rewards/accuracies": 0.7643630504608154, |
| "eval_rewards/chosen": -0.35914117097854614, |
| "eval_rewards/margins": 1.10481595993042, |
| "eval_rewards/rejected": -1.4639569520950317, |
| "eval_runtime": 181.7303, |
| "eval_samples_per_second": 13.217, |
| "eval_steps_per_second": 6.609, |
| "step": 630 |
| }, |
| { |
| "epoch": 1.0094034675286512, |
| "grad_norm": 31.40093421936035, |
| "learning_rate": 8.192322712349917e-07, |
| "logits/chosen": -1.1683887243270874, |
| "logits/rejected": -0.944835364818573, |
| "logps/chosen": -368.8553466796875, |
| "logps/rejected": -316.08819580078125, |
| "loss": 0.381, |
| "rewards/accuracies": 0.8333333134651184, |
| "rewards/chosen": -0.2540862560272217, |
| "rewards/margins": 1.389145016670227, |
| "rewards/rejected": -1.6432311534881592, |
| "step": 645 |
| }, |
| { |
| "epoch": 1.0329121363502791, |
| "grad_norm": 35.006961822509766, |
| "learning_rate": 8.088571072667467e-07, |
| "logits/chosen": -1.1861752271652222, |
| "logits/rejected": -0.9679210782051086, |
| "logps/chosen": -358.0697937011719, |
| "logps/rejected": -311.8282470703125, |
| "loss": 0.3093, |
| "rewards/accuracies": 0.8979166746139526, |
| "rewards/chosen": -0.12442357838153839, |
| "rewards/margins": 1.7039713859558105, |
| "rewards/rejected": -1.828395128250122, |
| "step": 660 |
| }, |
| { |
| "epoch": 1.0407483592908218, |
| "eval_logits/chosen": -1.1818084716796875, |
| "eval_logits/rejected": -0.9633015394210815, |
| "eval_logps/chosen": -356.53192138671875, |
| "eval_logps/rejected": -301.01409912109375, |
| "eval_loss": 0.4742245674133301, |
| "eval_rewards/accuracies": 0.7726894021034241, |
| "eval_rewards/chosen": -0.4217943549156189, |
| "eval_rewards/margins": 1.1362972259521484, |
| "eval_rewards/rejected": -1.558091640472412, |
| "eval_runtime": 182.4015, |
| "eval_samples_per_second": 13.169, |
| "eval_steps_per_second": 6.584, |
| "step": 665 |
| }, |
| { |
| "epoch": 1.056420805171907, |
| "grad_norm": 38.69508361816406, |
| "learning_rate": 7.982622883338412e-07, |
| "logits/chosen": -1.3052853345870972, |
| "logits/rejected": -0.9795024991035461, |
| "logps/chosen": -354.4381103515625, |
| "logps/rejected": -300.4755859375, |
| "loss": 0.2806, |
| "rewards/accuracies": 0.9145833253860474, |
| "rewards/chosen": -0.05176251009106636, |
| "rewards/margins": 1.8534504175186157, |
| "rewards/rejected": -1.9052128791809082, |
| "step": 675 |
| }, |
| { |
| "epoch": 1.079929473993535, |
| "grad_norm": 38.487979888916016, |
| "learning_rate": 7.87455349327083e-07, |
| "logits/chosen": -1.1824771165847778, |
| "logits/rejected": -1.0179836750030518, |
| "logps/chosen": -359.6732482910156, |
| "logps/rejected": -310.7752685546875, |
| "loss": 0.3073, |
| "rewards/accuracies": 0.9145833253860474, |
| "rewards/chosen": -0.17797540128231049, |
| "rewards/margins": 1.6661934852600098, |
| "rewards/rejected": -1.8441689014434814, |
| "step": 690 |
| }, |
| { |
| "epoch": 1.0956019198746205, |
| "eval_logits/chosen": -1.2035633325576782, |
| "eval_logits/rejected": -0.988406777381897, |
| "eval_logps/chosen": -357.3915100097656, |
| "eval_logps/rejected": -302.4732666015625, |
| "eval_loss": 0.47595611214637756, |
| "eval_rewards/accuracies": 0.765612006187439, |
| "eval_rewards/chosen": -0.5077541470527649, |
| "eval_rewards/margins": 1.1962535381317139, |
| "eval_rewards/rejected": -1.7040073871612549, |
| "eval_runtime": 182.096, |
| "eval_samples_per_second": 13.191, |
| "eval_steps_per_second": 6.595, |
| "step": 700 |
| }, |
| { |
| "epoch": 1.1034381428151632, |
| "grad_norm": 36.37774658203125, |
| "learning_rate": 7.76443975994184e-07, |
| "logits/chosen": -1.2869796752929688, |
| "logits/rejected": -0.9491466879844666, |
| "logps/chosen": -367.9986877441406, |
| "logps/rejected": -300.4508972167969, |
| "loss": 0.2897, |
| "rewards/accuracies": 0.893750011920929, |
| "rewards/chosen": -0.16376915574073792, |
| "rewards/margins": 1.7783530950546265, |
| "rewards/rejected": -1.942122220993042, |
| "step": 705 |
| }, |
| { |
| "epoch": 1.1269468116367911, |
| "grad_norm": 46.2029914855957, |
| "learning_rate": 7.652359994737627e-07, |
| "logits/chosen": -1.236419916152954, |
| "logits/rejected": -1.0354564189910889, |
| "logps/chosen": -379.1826171875, |
| "logps/rejected": -316.03631591796875, |
| "loss": 0.2747, |
| "rewards/accuracies": 0.9083333611488342, |
| "rewards/chosen": -0.18529005348682404, |
| "rewards/margins": 1.8945099115371704, |
| "rewards/rejected": -2.0797998905181885, |
| "step": 720 |
| }, |
| { |
| "epoch": 1.150455480458419, |
| "grad_norm": 30.26227569580078, |
| "learning_rate": 7.538393907259448e-07, |
| "logits/chosen": -1.1049808263778687, |
| "logits/rejected": -0.9194839596748352, |
| "logps/chosen": -378.98443603515625, |
| "logps/rejected": -327.0231628417969, |
| "loss": 0.2608, |
| "rewards/accuracies": 0.9270833134651184, |
| "rewards/chosen": -0.19015556573867798, |
| "rewards/margins": 1.9543516635894775, |
| "rewards/rejected": -2.14450740814209, |
| "step": 735 |
| }, |
| { |
| "epoch": 1.150455480458419, |
| "eval_logits/chosen": -1.1943178176879883, |
| "eval_logits/rejected": -0.9850893020629883, |
| "eval_logps/chosen": -358.00390625, |
| "eval_logps/rejected": -303.1407775878906, |
| "eval_loss": 0.4748242497444153, |
| "eval_rewards/accuracies": 0.7668609619140625, |
| "eval_rewards/chosen": -0.568992018699646, |
| "eval_rewards/margins": 1.2017687559127808, |
| "eval_rewards/rejected": -1.7707608938217163, |
| "eval_runtime": 182.4368, |
| "eval_samples_per_second": 13.166, |
| "eval_steps_per_second": 6.583, |
| "step": 735 |
| }, |
| { |
| "epoch": 1.173964149280047, |
| "grad_norm": 37.521915435791016, |
| "learning_rate": 7.422622548635244e-07, |
| "logits/chosen": -1.2676866054534912, |
| "logits/rejected": -1.0028654336929321, |
| "logps/chosen": -362.83648681640625, |
| "logps/rejected": -321.3612060546875, |
| "loss": 0.2729, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -0.2053557187318802, |
| "rewards/margins": 1.8843697309494019, |
| "rewards/rejected": -2.0897254943847656, |
| "step": 750 |
| }, |
| { |
| "epoch": 1.197472818101675, |
| "grad_norm": 33.50147247314453, |
| "learning_rate": 7.305128253877195e-07, |
| "logits/chosen": -1.2824881076812744, |
| "logits/rejected": -0.964570939540863, |
| "logps/chosen": -340.29644775390625, |
| "logps/rejected": -311.6012268066406, |
| "loss": 0.3037, |
| "rewards/accuracies": 0.8999999761581421, |
| "rewards/chosen": -0.22407926619052887, |
| "rewards/margins": 1.7219882011413574, |
| "rewards/rejected": -1.9460675716400146, |
| "step": 765 |
| }, |
| { |
| "epoch": 1.2053090410422176, |
| "eval_logits/chosen": -1.2377439737319946, |
| "eval_logits/rejected": -1.0323067903518677, |
| "eval_logps/chosen": -357.77703857421875, |
| "eval_logps/rejected": -302.7647399902344, |
| "eval_loss": 0.4716549813747406, |
| "eval_rewards/accuracies": 0.7697752118110657, |
| "eval_rewards/chosen": -0.5463067293167114, |
| "eval_rewards/margins": 1.1868516206741333, |
| "eval_rewards/rejected": -1.7331583499908447, |
| "eval_runtime": 181.9712, |
| "eval_samples_per_second": 13.2, |
| "eval_steps_per_second": 6.6, |
| "step": 770 |
| }, |
| { |
| "epoch": 1.2209814869233029, |
| "grad_norm": 37.49959182739258, |
| "learning_rate": 7.185994583326165e-07, |
| "logits/chosen": -1.1782593727111816, |
| "logits/rejected": -1.0032784938812256, |
| "logps/chosen": -380.5348205566406, |
| "logps/rejected": -333.3002624511719, |
| "loss": 0.2849, |
| "rewards/accuracies": 0.9104166626930237, |
| "rewards/chosen": -0.21467427909374237, |
| "rewards/margins": 1.8065526485443115, |
| "rewards/rejected": -2.0212271213531494, |
| "step": 780 |
| }, |
| { |
| "epoch": 1.244490155744931, |
| "grad_norm": 35.8784294128418, |
| "learning_rate": 7.065306263224742e-07, |
| "logits/chosen": -1.148066759109497, |
| "logits/rejected": -0.9383103847503662, |
| "logps/chosen": -367.6258239746094, |
| "logps/rejected": -306.7835388183594, |
| "loss": 0.2828, |
| "rewards/accuracies": 0.9020833373069763, |
| "rewards/chosen": -0.21600770950317383, |
| "rewards/margins": 1.806153416633606, |
| "rewards/rejected": -2.0221610069274902, |
| "step": 795 |
| }, |
| { |
| "epoch": 1.2601626016260163, |
| "eval_logits/chosen": -1.2068735361099243, |
| "eval_logits/rejected": -1.0000766515731812, |
| "eval_logps/chosen": -358.01068115234375, |
| "eval_logps/rejected": -303.16278076171875, |
| "eval_loss": 0.47309279441833496, |
| "eval_rewards/accuracies": 0.7714404463768005, |
| "eval_rewards/chosen": -0.569669246673584, |
| "eval_rewards/margins": 1.203292965888977, |
| "eval_rewards/rejected": -1.772962212562561, |
| "eval_runtime": 182.4089, |
| "eval_samples_per_second": 13.168, |
| "eval_steps_per_second": 6.584, |
| "step": 805 |
| }, |
| { |
| "epoch": 1.267998824566559, |
| "grad_norm": 24.38459014892578, |
| "learning_rate": 6.94314912546108e-07, |
| "logits/chosen": -1.2680124044418335, |
| "logits/rejected": -0.9894136786460876, |
| "logps/chosen": -350.9635009765625, |
| "logps/rejected": -310.0843811035156, |
| "loss": 0.2591, |
| "rewards/accuracies": 0.9041666388511658, |
| "rewards/chosen": -0.1324070394039154, |
| "rewards/margins": 1.9815576076507568, |
| "rewards/rejected": -2.113964796066284, |
| "step": 810 |
| }, |
| { |
| "epoch": 1.291507493388187, |
| "grad_norm": 39.18712615966797, |
| "learning_rate": 6.819610046526436e-07, |
| "logits/chosen": -1.2195533514022827, |
| "logits/rejected": -1.0207655429840088, |
| "logps/chosen": -365.35595703125, |
| "logps/rejected": -308.2247009277344, |
| "loss": 0.29, |
| "rewards/accuracies": 0.8916666507720947, |
| "rewards/chosen": -0.2736612856388092, |
| "rewards/margins": 1.791273593902588, |
| "rewards/rejected": -2.0649349689483643, |
| "step": 825 |
| }, |
| { |
| "epoch": 1.3150161622098149, |
| "grad_norm": 32.96370315551758, |
| "learning_rate": 6.694776885729787e-07, |
| "logits/chosen": -1.1510590314865112, |
| "logits/rejected": -0.9895613193511963, |
| "logps/chosen": -346.2775573730469, |
| "logps/rejected": -303.6460876464844, |
| "loss": 0.3013, |
| "rewards/accuracies": 0.8895833492279053, |
| "rewards/chosen": -0.276450514793396, |
| "rewards/margins": 1.8475831747055054, |
| "rewards/rejected": -2.1240336894989014, |
| "step": 840 |
| }, |
| { |
| "epoch": 1.3150161622098149, |
| "eval_logits/chosen": -1.2594456672668457, |
| "eval_logits/rejected": -1.056649923324585, |
| "eval_logps/chosen": -358.96270751953125, |
| "eval_logps/rejected": -304.4568176269531, |
| "eval_loss": 0.47552865743637085, |
| "eval_rewards/accuracies": 0.770191490650177, |
| "eval_rewards/chosen": -0.6648746728897095, |
| "eval_rewards/margins": 1.2374926805496216, |
| "eval_rewards/rejected": -1.9023675918579102, |
| "eval_runtime": 182.3073, |
| "eval_samples_per_second": 13.176, |
| "eval_steps_per_second": 6.588, |
| "step": 840 |
| }, |
| { |
| "epoch": 1.3385248310314428, |
| "grad_norm": 43.61619567871094, |
| "learning_rate": 6.568738422713492e-07, |
| "logits/chosen": -1.3479042053222656, |
| "logits/rejected": -1.1390700340270996, |
| "logps/chosen": -356.8882751464844, |
| "logps/rejected": -316.7053527832031, |
| "loss": 0.253, |
| "rewards/accuracies": 0.9125000238418579, |
| "rewards/chosen": -0.34459179639816284, |
| "rewards/margins": 2.0109331607818604, |
| "rewards/rejected": -2.355525016784668, |
| "step": 855 |
| }, |
| { |
| "epoch": 1.3620334998530708, |
| "grad_norm": 37.48102569580078, |
| "learning_rate": 6.441584294314419e-07, |
| "logits/chosen": -1.2495485544204712, |
| "logits/rejected": -0.9884099364280701, |
| "logps/chosen": -361.2688293457031, |
| "logps/rejected": -326.4720458984375, |
| "loss": 0.2693, |
| "rewards/accuracies": 0.9020833373069763, |
| "rewards/chosen": -0.33273938298225403, |
| "rewards/margins": 2.039118528366089, |
| "rewards/rejected": -2.3718576431274414, |
| "step": 870 |
| }, |
| { |
| "epoch": 1.3698697227936134, |
| "eval_logits/chosen": -1.2140861749649048, |
| "eval_logits/rejected": -1.0111212730407715, |
| "eval_logps/chosen": -359.7378845214844, |
| "eval_logps/rejected": -305.63800048828125, |
| "eval_loss": 0.47619202733039856, |
| "eval_rewards/accuracies": 0.765612006187439, |
| "eval_rewards/chosen": -0.7423911094665527, |
| "eval_rewards/margins": 1.2780914306640625, |
| "eval_rewards/rejected": -2.0204825401306152, |
| "eval_runtime": 182.596, |
| "eval_samples_per_second": 13.155, |
| "eval_steps_per_second": 6.577, |
| "step": 875 |
| }, |
| { |
| "epoch": 1.3855421686746987, |
| "grad_norm": 49.734066009521484, |
| "learning_rate": 6.313404930815452e-07, |
| "logits/chosen": -1.163791537284851, |
| "logits/rejected": -0.9558045864105225, |
| "logps/chosen": -359.283203125, |
| "logps/rejected": -306.8938293457031, |
| "loss": 0.2767, |
| "rewards/accuracies": 0.887499988079071, |
| "rewards/chosen": -0.4080369770526886, |
| "rewards/margins": 1.97806978225708, |
| "rewards/rejected": -2.3861069679260254, |
| "step": 885 |
| }, |
| { |
| "epoch": 1.4090508374963266, |
| "grad_norm": 30.948827743530273, |
| "learning_rate": 6.184291491632694e-07, |
| "logits/chosen": -1.1903069019317627, |
| "logits/rejected": -0.9102679491043091, |
| "logps/chosen": -341.7632141113281, |
| "logps/rejected": -303.9745178222656, |
| "loss": 0.2875, |
| "rewards/accuracies": 0.8979166746139526, |
| "rewards/chosen": -0.32621437311172485, |
| "rewards/margins": 1.8810745477676392, |
| "rewards/rejected": -2.207289218902588, |
| "step": 900 |
| }, |
| { |
| "epoch": 1.4247232833774122, |
| "eval_logits/chosen": -1.2810581922531128, |
| "eval_logits/rejected": -1.083061933517456, |
| "eval_logps/chosen": -359.7813720703125, |
| "eval_logps/rejected": -305.91448974609375, |
| "eval_loss": 0.4744855761528015, |
| "eval_rewards/accuracies": 0.768109917640686, |
| "eval_rewards/chosen": -0.7467413544654846, |
| "eval_rewards/margins": 1.3013904094696045, |
| "eval_rewards/rejected": -2.0481317043304443, |
| "eval_runtime": 181.9732, |
| "eval_samples_per_second": 13.2, |
| "eval_steps_per_second": 6.6, |
| "step": 910 |
| }, |
| { |
| "epoch": 1.4325595063179548, |
| "grad_norm": 39.57698440551758, |
| "learning_rate": 6.054335800484152e-07, |
| "logits/chosen": -1.3292855024337769, |
| "logits/rejected": -0.9953449368476868, |
| "logps/chosen": -373.2196044921875, |
| "logps/rejected": -335.6629943847656, |
| "loss": 0.2296, |
| "rewards/accuracies": 0.9291666746139526, |
| "rewards/chosen": -0.20129725337028503, |
| "rewards/margins": 2.1831979751586914, |
| "rewards/rejected": -2.384495258331299, |
| "step": 915 |
| }, |
| { |
| "epoch": 1.4560681751395828, |
| "grad_norm": 43.79376983642578, |
| "learning_rate": 5.923630280085947e-07, |
| "logits/chosen": -1.2759658098220825, |
| "logits/rejected": -1.0600124597549438, |
| "logps/chosen": -356.8258056640625, |
| "logps/rejected": -317.5994567871094, |
| "loss": 0.271, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -0.3783254027366638, |
| "rewards/margins": 2.0156712532043457, |
| "rewards/rejected": -2.3939969539642334, |
| "step": 930 |
| }, |
| { |
| "epoch": 1.4795768439612107, |
| "grad_norm": 36.227500915527344, |
| "learning_rate": 5.792267886422537e-07, |
| "logits/chosen": -1.1942765712738037, |
| "logits/rejected": -0.9782482385635376, |
| "logps/chosen": -368.8127136230469, |
| "logps/rejected": -322.2507629394531, |
| "loss": 0.2462, |
| "rewards/accuracies": 0.9145833253860474, |
| "rewards/chosen": -0.2551910877227783, |
| "rewards/margins": 2.0672876834869385, |
| "rewards/rejected": -2.322478771209717, |
| "step": 945 |
| }, |
| { |
| "epoch": 1.4795768439612107, |
| "eval_logits/chosen": -1.2686865329742432, |
| "eval_logits/rejected": -1.0674816370010376, |
| "eval_logps/chosen": -359.4958801269531, |
| "eval_logps/rejected": -305.484130859375, |
| "eval_loss": 0.473550945520401, |
| "eval_rewards/accuracies": 0.7722731232643127, |
| "eval_rewards/chosen": -0.7181934118270874, |
| "eval_rewards/margins": 1.286901593208313, |
| "eval_rewards/rejected": -2.0050950050354004, |
| "eval_runtime": 182.2235, |
| "eval_samples_per_second": 13.182, |
| "eval_steps_per_second": 6.591, |
| "step": 945 |
| }, |
| { |
| "epoch": 1.5030855127828386, |
| "grad_norm": 35.31517028808594, |
| "learning_rate": 5.660342042637701e-07, |
| "logits/chosen": -1.278725504875183, |
| "logits/rejected": -0.9818956255912781, |
| "logps/chosen": -354.6424560546875, |
| "logps/rejected": -320.46343994140625, |
| "loss": 0.266, |
| "rewards/accuracies": 0.9041666388511658, |
| "rewards/chosen": -0.37783387303352356, |
| "rewards/margins": 2.0141122341156006, |
| "rewards/rejected": -2.3919460773468018, |
| "step": 960 |
| }, |
| { |
| "epoch": 1.5265941816044668, |
| "grad_norm": 43.222660064697266, |
| "learning_rate": 5.527946572593254e-07, |
| "logits/chosen": -1.3914339542388916, |
| "logits/rejected": -1.0797264575958252, |
| "logps/chosen": -360.6291198730469, |
| "logps/rejected": -320.04779052734375, |
| "loss": 0.2524, |
| "rewards/accuracies": 0.9145833253860474, |
| "rewards/chosen": -0.3379184603691101, |
| "rewards/margins": 2.10139536857605, |
| "rewards/rejected": -2.4393134117126465, |
| "step": 975 |
| }, |
| { |
| "epoch": 1.5344304045450095, |
| "eval_logits/chosen": -1.2813137769699097, |
| "eval_logits/rejected": -1.0804452896118164, |
| "eval_logps/chosen": -359.6990661621094, |
| "eval_logps/rejected": -306.07623291015625, |
| "eval_loss": 0.47156351804733276, |
| "eval_rewards/accuracies": 0.770191490650177, |
| "eval_rewards/chosen": -0.7385069131851196, |
| "eval_rewards/margins": 1.3257992267608643, |
| "eval_rewards/rejected": -2.0643060207366943, |
| "eval_runtime": 182.7063, |
| "eval_samples_per_second": 13.147, |
| "eval_steps_per_second": 6.573, |
| "step": 980 |
| }, |
| { |
| "epoch": 1.5501028504260947, |
| "grad_norm": 26.045185089111328, |
| "learning_rate": 5.395175634142814e-07, |
| "logits/chosen": -1.3037978410720825, |
| "logits/rejected": -1.1191579103469849, |
| "logps/chosen": -376.4510498046875, |
| "logps/rejected": -316.0899658203125, |
| "loss": 0.2476, |
| "rewards/accuracies": 0.9208333492279053, |
| "rewards/chosen": -0.27149498462677, |
| "rewards/margins": 2.1316168308258057, |
| "rewards/rejected": -2.4031119346618652, |
| "step": 990 |
| }, |
| { |
| "epoch": 1.5736115192477227, |
| "grad_norm": 32.610233306884766, |
| "learning_rate": 5.262123652168005e-07, |
| "logits/chosen": -1.2782458066940308, |
| "logits/rejected": -0.9592511653900146, |
| "logps/chosen": -351.68524169921875, |
| "logps/rejected": -310.4751281738281, |
| "loss": 0.2421, |
| "rewards/accuracies": 0.925000011920929, |
| "rewards/chosen": -0.37502792477607727, |
| "rewards/margins": 2.187502145767212, |
| "rewards/rejected": -2.562530040740967, |
| "step": 1005 |
| }, |
| { |
| "epoch": 1.589283965128808, |
| "eval_logits/chosen": -1.2604247331619263, |
| "eval_logits/rejected": -1.0631768703460693, |
| "eval_logps/chosen": -360.0327453613281, |
| "eval_logps/rejected": -306.8144226074219, |
| "eval_loss": 0.4726716876029968, |
| "eval_rewards/accuracies": 0.7743546962738037, |
| "eval_rewards/chosen": -0.7718815803527832, |
| "eval_rewards/margins": 1.366243839263916, |
| "eval_rewards/rejected": -2.138125419616699, |
| "eval_runtime": 182.6755, |
| "eval_samples_per_second": 13.149, |
| "eval_steps_per_second": 6.574, |
| "step": 1015 |
| }, |
| { |
| "epoch": 1.5971201880693506, |
| "grad_norm": 39.39152908325195, |
| "learning_rate": 5.128885251424781e-07, |
| "logits/chosen": -1.367023229598999, |
| "logits/rejected": -1.0569322109222412, |
| "logps/chosen": -360.5495910644531, |
| "logps/rejected": -323.3975524902344, |
| "loss": 0.2304, |
| "rewards/accuracies": 0.9208333492279053, |
| "rewards/chosen": -0.40084362030029297, |
| "rewards/margins": 2.2848780155181885, |
| "rewards/rejected": -2.6857216358184814, |
| "step": 1020 |
| }, |
| { |
| "epoch": 1.6206288568909786, |
| "grad_norm": 27.369558334350586, |
| "learning_rate": 4.995555189247576e-07, |
| "logits/chosen": -1.2406485080718994, |
| "logits/rejected": -0.989262044429779, |
| "logps/chosen": -357.1865234375, |
| "logps/rejected": -297.4217529296875, |
| "loss": 0.2357, |
| "rewards/accuracies": 0.9354166388511658, |
| "rewards/chosen": -0.38933005928993225, |
| "rewards/margins": 2.214608907699585, |
| "rewards/rejected": -2.6039390563964844, |
| "step": 1035 |
| }, |
| { |
| "epoch": 1.6441375257126065, |
| "grad_norm": 35.30168533325195, |
| "learning_rate": 4.86222828815919e-07, |
| "logits/chosen": -1.3355810642242432, |
| "logits/rejected": -1.0357699394226074, |
| "logps/chosen": -360.48468017578125, |
| "logps/rejected": -318.0264892578125, |
| "loss": 0.2545, |
| "rewards/accuracies": 0.893750011920929, |
| "rewards/chosen": -0.3565267324447632, |
| "rewards/margins": 2.160374641418457, |
| "rewards/rejected": -2.5169014930725098, |
| "step": 1050 |
| }, |
| { |
| "epoch": 1.6441375257126065, |
| "eval_logits/chosen": -1.3108830451965332, |
| "eval_logits/rejected": -1.1151154041290283, |
| "eval_logps/chosen": -360.46136474609375, |
| "eval_logps/rejected": -307.2978820800781, |
| "eval_loss": 0.47149062156677246, |
| "eval_rewards/accuracies": 0.7706078290939331, |
| "eval_rewards/chosen": -0.8147412538528442, |
| "eval_rewards/margins": 1.3717304468154907, |
| "eval_rewards/rejected": -2.186471700668335, |
| "eval_runtime": 182.8736, |
| "eval_samples_per_second": 13.135, |
| "eval_steps_per_second": 6.567, |
| "step": 1050 |
| }, |
| { |
| "epoch": 1.6676461945342345, |
| "grad_norm": 23.27956199645996, |
| "learning_rate": 4.728999368434301e-07, |
| "logits/chosen": -1.3302825689315796, |
| "logits/rejected": -1.044356107711792, |
| "logps/chosen": -348.8330078125, |
| "logps/rejected": -303.2927551269531, |
| "loss": 0.2709, |
| "rewards/accuracies": 0.8958333134651184, |
| "rewards/chosen": -0.4363751709461212, |
| "rewards/margins": 2.0858278274536133, |
| "rewards/rejected": -2.522202968597412, |
| "step": 1065 |
| }, |
| { |
| "epoch": 1.6911548633558624, |
| "grad_norm": 30.386754989624023, |
| "learning_rate": 4.595963180664576e-07, |
| "logits/chosen": -1.2675700187683105, |
| "logits/rejected": -1.0812220573425293, |
| "logps/chosen": -342.9148864746094, |
| "logps/rejected": -290.79949951171875, |
| "loss": 0.253, |
| "rewards/accuracies": 0.9145833253860474, |
| "rewards/chosen": -0.4274654984474182, |
| "rewards/margins": 2.0667319297790527, |
| "rewards/rejected": -2.494197368621826, |
| "step": 1080 |
| }, |
| { |
| "epoch": 1.698991086296405, |
| "eval_logits/chosen": -1.3290081024169922, |
| "eval_logits/rejected": -1.1328283548355103, |
| "eval_logps/chosen": -359.980224609375, |
| "eval_logps/rejected": -306.26348876953125, |
| "eval_loss": 0.4710884392261505, |
| "eval_rewards/accuracies": 0.7726894021034241, |
| "eval_rewards/chosen": -0.7666258215904236, |
| "eval_rewards/margins": 1.3164042234420776, |
| "eval_rewards/rejected": -2.0830299854278564, |
| "eval_runtime": 182.8851, |
| "eval_samples_per_second": 13.134, |
| "eval_steps_per_second": 6.567, |
| "step": 1085 |
| }, |
| { |
| "epoch": 1.7146635321774903, |
| "grad_norm": 41.24742126464844, |
| "learning_rate": 4.4632143383733394e-07, |
| "logits/chosen": -1.3585143089294434, |
| "logits/rejected": -1.1536871194839478, |
| "logps/chosen": -355.0205993652344, |
| "logps/rejected": -318.7183532714844, |
| "loss": 0.2484, |
| "rewards/accuracies": 0.925000011920929, |
| "rewards/chosen": -0.43984732031822205, |
| "rewards/margins": 2.0352494716644287, |
| "rewards/rejected": -2.4750969409942627, |
| "step": 1095 |
| }, |
| { |
| "epoch": 1.7381722009991183, |
| "grad_norm": 26.021894454956055, |
| "learning_rate": 4.330847250727732e-07, |
| "logits/chosen": -1.3933275938034058, |
| "logits/rejected": -1.1843206882476807, |
| "logps/chosen": -370.0199279785156, |
| "logps/rejected": -322.68731689453125, |
| "loss": 0.2263, |
| "rewards/accuracies": 0.9395833611488342, |
| "rewards/chosen": -0.4698244035243988, |
| "rewards/margins": 2.1898274421691895, |
| "rewards/rejected": -2.6596519947052, |
| "step": 1110 |
| }, |
| { |
| "epoch": 1.7538446468802036, |
| "eval_logits/chosen": -1.3134483098983765, |
| "eval_logits/rejected": -1.1194194555282593, |
| "eval_logps/chosen": -360.1936340332031, |
| "eval_logps/rejected": -306.6624755859375, |
| "eval_loss": 0.46780040860176086, |
| "eval_rewards/accuracies": 0.7756036520004272, |
| "eval_rewards/chosen": -0.7879676818847656, |
| "eval_rewards/margins": 1.3349637985229492, |
| "eval_rewards/rejected": -2.122931480407715, |
| "eval_runtime": 182.6424, |
| "eval_samples_per_second": 13.151, |
| "eval_steps_per_second": 6.576, |
| "step": 1120 |
| }, |
| { |
| "epoch": 1.7616808698207465, |
| "grad_norm": 32.77103042602539, |
| "learning_rate": 4.198956055396194e-07, |
| "logits/chosen": -1.3157938718795776, |
| "logits/rejected": -1.1473113298416138, |
| "logps/chosen": -370.41925048828125, |
| "logps/rejected": -322.4455261230469, |
| "loss": 0.2544, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -0.35879072546958923, |
| "rewards/margins": 2.2119410037994385, |
| "rewards/rejected": -2.5707318782806396, |
| "step": 1125 |
| }, |
| { |
| "epoch": 1.7851895386423744, |
| "grad_norm": 33.03691101074219, |
| "learning_rate": 4.0676345515990384e-07, |
| "logits/chosen": -1.2871196269989014, |
| "logits/rejected": -1.11886465549469, |
| "logps/chosen": -371.6841735839844, |
| "logps/rejected": -308.32806396484375, |
| "loss": 0.2314, |
| "rewards/accuracies": 0.9166666865348816, |
| "rewards/chosen": -0.3601396977901459, |
| "rewards/margins": 2.2169458866119385, |
| "rewards/rejected": -2.577085256576538, |
| "step": 1140 |
| }, |
| { |
| "epoch": 1.8086982074640023, |
| "grad_norm": 35.31661605834961, |
| "learning_rate": 3.9369761333997193e-07, |
| "logits/chosen": -1.3224281072616577, |
| "logits/rejected": -1.102236270904541, |
| "logps/chosen": -355.3537292480469, |
| "logps/rejected": -324.50274658203125, |
| "loss": 0.2507, |
| "rewards/accuracies": 0.9125000238418579, |
| "rewards/chosen": -0.41861438751220703, |
| "rewards/margins": 2.1263999938964844, |
| "rewards/rejected": -2.5450143814086914, |
| "step": 1155 |
| }, |
| { |
| "epoch": 1.8086982074640023, |
| "eval_logits/chosen": -1.2986286878585815, |
| "eval_logits/rejected": -1.1063010692596436, |
| "eval_logps/chosen": -360.1191101074219, |
| "eval_logps/rejected": -306.6952209472656, |
| "eval_loss": 0.4674054980278015, |
| "eval_rewards/accuracies": 0.7781015634536743, |
| "eval_rewards/chosen": -0.7805167436599731, |
| "eval_rewards/margins": 1.3456897735595703, |
| "eval_rewards/rejected": -2.126206636428833, |
| "eval_runtime": 182.9858, |
| "eval_samples_per_second": 13.127, |
| "eval_steps_per_second": 6.563, |
| "step": 1155 |
| }, |
| { |
| "epoch": 1.8322068762856303, |
| "grad_norm": 29.87853240966797, |
| "learning_rate": 3.8070737232842614e-07, |
| "logits/chosen": -1.3065639734268188, |
| "logits/rejected": -1.0679255723953247, |
| "logps/chosen": -374.07391357421875, |
| "logps/rejected": -335.5907287597656, |
| "loss": 0.2381, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -0.30956602096557617, |
| "rewards/margins": 2.238311290740967, |
| "rewards/rejected": -2.547877311706543, |
| "step": 1170 |
| }, |
| { |
| "epoch": 1.8557155451072584, |
| "grad_norm": 39.42518997192383, |
| "learning_rate": 3.678019706076039e-07, |
| "logits/chosen": -1.3437250852584839, |
| "logits/rejected": -1.1545518636703491, |
| "logps/chosen": -377.6987609863281, |
| "logps/rejected": -333.1975402832031, |
| "loss": 0.2465, |
| "rewards/accuracies": 0.9208333492279053, |
| "rewards/chosen": -0.3938901126384735, |
| "rewards/margins": 2.213219165802002, |
| "rewards/rejected": -2.607109308242798, |
| "step": 1185 |
| }, |
| { |
| "epoch": 1.863551768047801, |
| "eval_logits/chosen": -1.3109138011932373, |
| "eval_logits/rejected": -1.1203300952911377, |
| "eval_logps/chosen": -360.5334167480469, |
| "eval_logps/rejected": -307.3110656738281, |
| "eval_loss": 0.46806973218917847, |
| "eval_rewards/accuracies": 0.7764363288879395, |
| "eval_rewards/chosen": -0.8219457864761353, |
| "eval_rewards/margins": 1.3658442497253418, |
| "eval_rewards/rejected": -2.1877899169921875, |
| "eval_runtime": 183.179, |
| "eval_samples_per_second": 13.113, |
| "eval_steps_per_second": 6.556, |
| "step": 1190 |
| }, |
| { |
| "epoch": 1.8792242139288864, |
| "grad_norm": 35.16496276855469, |
| "learning_rate": 3.5499058632329536e-07, |
| "logits/chosen": -1.3551256656646729, |
| "logits/rejected": -1.003768801689148, |
| "logps/chosen": -343.90545654296875, |
| "logps/rejected": -318.3465576171875, |
| "loss": 0.2508, |
| "rewards/accuracies": 0.9104166626930237, |
| "rewards/chosen": -0.38343942165374756, |
| "rewards/margins": 2.2458739280700684, |
| "rewards/rejected": -2.6293132305145264, |
| "step": 1200 |
| }, |
| { |
| "epoch": 1.9027328827505143, |
| "grad_norm": 40.798072814941406, |
| "learning_rate": 3.4228233075737223e-07, |
| "logits/chosen": -1.3036648035049438, |
| "logits/rejected": -1.1327162981033325, |
| "logps/chosen": -375.63385009765625, |
| "logps/rejected": -311.95782470703125, |
| "loss": 0.2338, |
| "rewards/accuracies": 0.9125000238418579, |
| "rewards/chosen": -0.48579010367393494, |
| "rewards/margins": 2.2717065811157227, |
| "rewards/rejected": -2.7574965953826904, |
| "step": 1215 |
| }, |
| { |
| "epoch": 1.9184053286315996, |
| "eval_logits/chosen": -1.2938474416732788, |
| "eval_logits/rejected": -1.1041306257247925, |
| "eval_logps/chosen": -360.88861083984375, |
| "eval_logps/rejected": -307.92333984375, |
| "eval_loss": 0.4688262343406677, |
| "eval_rewards/accuracies": 0.777685284614563, |
| "eval_rewards/chosen": -0.8574661016464233, |
| "eval_rewards/margins": 1.391554355621338, |
| "eval_rewards/rejected": -2.2490200996398926, |
| "eval_runtime": 183.0197, |
| "eval_samples_per_second": 13.124, |
| "eval_steps_per_second": 6.562, |
| "step": 1225 |
| }, |
| { |
| "epoch": 1.9262415515721423, |
| "grad_norm": 25.399438858032227, |
| "learning_rate": 3.2968624184796654e-07, |
| "logits/chosen": -1.2950618267059326, |
| "logits/rejected": -1.0804049968719482, |
| "logps/chosen": -351.4801025390625, |
| "logps/rejected": -326.7780456542969, |
| "loss": 0.2601, |
| "rewards/accuracies": 0.8895833492279053, |
| "rewards/chosen": -0.4700445830821991, |
| "rewards/margins": 2.1944639682769775, |
| "rewards/rejected": -2.664508581161499, |
| "step": 1230 |
| }, |
| { |
| "epoch": 1.9497502203937702, |
| "grad_norm": 25.898283004760742, |
| "learning_rate": 3.1721127776181364e-07, |
| "logits/chosen": -1.2864807844161987, |
| "logits/rejected": -0.9478799104690552, |
| "logps/chosen": -376.6116027832031, |
| "logps/rejected": -320.6463623046875, |
| "loss": 0.2545, |
| "rewards/accuracies": 0.9166666865348816, |
| "rewards/chosen": -0.3715561330318451, |
| "rewards/margins": 2.254369020462036, |
| "rewards/rejected": -2.6259255409240723, |
| "step": 1245 |
| }, |
| { |
| "epoch": 1.9732588892153982, |
| "grad_norm": 29.050125122070312, |
| "learning_rate": 3.0486631052332534e-07, |
| "logits/chosen": -1.2400730848312378, |
| "logits/rejected": -1.0113612413406372, |
| "logps/chosen": -350.2702331542969, |
| "logps/rejected": -303.7456359863281, |
| "loss": 0.2545, |
| "rewards/accuracies": 0.8916666507720947, |
| "rewards/chosen": -0.4329583942890167, |
| "rewards/margins": 2.1629526615142822, |
| "rewards/rejected": -2.5959107875823975, |
| "step": 1260 |
| }, |
| { |
| "epoch": 1.9732588892153982, |
| "eval_logits/chosen": -1.279727816581726, |
| "eval_logits/rejected": -1.0888946056365967, |
| "eval_logps/chosen": -360.5728454589844, |
| "eval_logps/rejected": -307.5393981933594, |
| "eval_loss": 0.4675270617008209, |
| "eval_rewards/accuracies": 0.7735220789909363, |
| "eval_rewards/chosen": -0.8258867859840393, |
| "eval_rewards/margins": 1.3847379684448242, |
| "eval_rewards/rejected": -2.210624933242798, |
| "eval_runtime": 182.6913, |
| "eval_samples_per_second": 13.148, |
| "eval_steps_per_second": 6.574, |
| "step": 1260 |
| }, |
| { |
| "epoch": 1.996767558037026, |
| "grad_norm": 34.406803131103516, |
| "learning_rate": 2.9266011970492775e-07, |
| "logits/chosen": -1.347962737083435, |
| "logits/rejected": -1.1148425340652466, |
| "logps/chosen": -360.68511962890625, |
| "logps/rejected": -309.3987731933594, |
| "loss": 0.2562, |
| "rewards/accuracies": 0.9104166626930237, |
| "rewards/chosen": -0.32166415452957153, |
| "rewards/margins": 2.071373224258423, |
| "rewards/rejected": -2.3930373191833496, |
| "step": 1275 |
| }, |
| { |
| "epoch": 2.0188069350573024, |
| "grad_norm": 16.827268600463867, |
| "learning_rate": 2.806013861831484e-07, |
| "logits/chosen": -1.38746178150177, |
| "logits/rejected": -1.0861637592315674, |
| "logps/chosen": -369.17529296875, |
| "logps/rejected": -314.560302734375, |
| "loss": 0.1532, |
| "rewards/accuracies": 0.9577777981758118, |
| "rewards/chosen": -0.18039265275001526, |
| "rewards/margins": 2.5118191242218018, |
| "rewards/rejected": -2.692211627960205, |
| "step": 1290 |
| }, |
| { |
| "epoch": 2.026643157997845, |
| "eval_logits/chosen": -1.3171147108078003, |
| "eval_logits/rejected": -1.1266576051712036, |
| "eval_logps/chosen": -361.1548767089844, |
| "eval_logps/rejected": -308.2939758300781, |
| "eval_loss": 0.4683936536312103, |
| "eval_rewards/accuracies": 0.7735220789909363, |
| "eval_rewards/chosen": -0.8840944170951843, |
| "eval_rewards/margins": 1.4019862413406372, |
| "eval_rewards/rejected": -2.286080837249756, |
| "eval_runtime": 183.0594, |
| "eval_samples_per_second": 13.121, |
| "eval_steps_per_second": 6.561, |
| "step": 1295 |
| }, |
| { |
| "epoch": 2.0423156038789303, |
| "grad_norm": 20.37907600402832, |
| "learning_rate": 2.686986859648953e-07, |
| "logits/chosen": -1.3132938146591187, |
| "logits/rejected": -1.0580551624298096, |
| "logps/chosen": -353.71502685546875, |
| "logps/rejected": -316.21600341796875, |
| "loss": 0.1542, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -0.22374172508716583, |
| "rewards/margins": 2.5709550380706787, |
| "rewards/rejected": -2.794696807861328, |
| "step": 1305 |
| }, |
| { |
| "epoch": 2.0658242727005582, |
| "grad_norm": 31.620746612548828, |
| "learning_rate": 2.5696048408831833e-07, |
| "logits/chosen": -1.3634592294692993, |
| "logits/rejected": -1.1027321815490723, |
| "logps/chosen": -355.3858947753906, |
| "logps/rejected": -324.46295166015625, |
| "loss": 0.159, |
| "rewards/accuracies": 0.9729166626930237, |
| "rewards/chosen": -0.32911619544029236, |
| "rewards/margins": 2.6501662731170654, |
| "rewards/rejected": -2.979282855987549, |
| "step": 1320 |
| }, |
| { |
| "epoch": 2.0814967185816435, |
| "eval_logits/chosen": -1.3730605840682983, |
| "eval_logits/rejected": -1.1844755411148071, |
| "eval_logps/chosen": -362.084716796875, |
| "eval_logps/rejected": -309.5851745605469, |
| "eval_loss": 0.4718104600906372, |
| "eval_rewards/accuracies": 0.7722731232643127, |
| "eval_rewards/chosen": -0.9770740866661072, |
| "eval_rewards/margins": 1.438130497932434, |
| "eval_rewards/rejected": -2.4152042865753174, |
| "eval_runtime": 184.2931, |
| "eval_samples_per_second": 13.034, |
| "eval_steps_per_second": 6.517, |
| "step": 1330 |
| }, |
| { |
| "epoch": 2.089332941522186, |
| "grad_norm": 24.57793426513672, |
| "learning_rate": 2.4539512860258874e-07, |
| "logits/chosen": -1.40749990940094, |
| "logits/rejected": -1.219661831855774, |
| "logps/chosen": -375.73541259765625, |
| "logps/rejected": -332.43707275390625, |
| "loss": 0.1548, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -0.33681097626686096, |
| "rewards/margins": 2.658933401107788, |
| "rewards/rejected": -2.9957449436187744, |
| "step": 1335 |
| }, |
| { |
| "epoch": 2.112841610343814, |
| "grad_norm": 24.045860290527344, |
| "learning_rate": 2.3401084463088095e-07, |
| "logits/chosen": -1.3906124830245972, |
| "logits/rejected": -1.10317862033844, |
| "logps/chosen": -353.56524658203125, |
| "logps/rejected": -316.1902160644531, |
| "loss": 0.1594, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -0.5038239359855652, |
| "rewards/margins": 2.6673552989959717, |
| "rewards/rejected": -3.1711785793304443, |
| "step": 1350 |
| }, |
| { |
| "epoch": 2.136350279165442, |
| "grad_norm": 15.747769355773926, |
| "learning_rate": 2.2281572852077602e-07, |
| "logits/chosen": -1.3541206121444702, |
| "logits/rejected": -1.1997876167297363, |
| "logps/chosen": -352.85357666015625, |
| "logps/rejected": -318.2190246582031, |
| "loss": 0.1644, |
| "rewards/accuracies": 0.9520833492279053, |
| "rewards/chosen": -0.3332318961620331, |
| "rewards/margins": 2.7003839015960693, |
| "rewards/rejected": -3.033615827560425, |
| "step": 1365 |
| }, |
| { |
| "epoch": 2.136350279165442, |
| "eval_logits/chosen": -1.3989009857177734, |
| "eval_logits/rejected": -1.2132837772369385, |
| "eval_logps/chosen": -362.2392272949219, |
| "eval_logps/rejected": -309.7401428222656, |
| "eval_loss": 0.47149989008903503, |
| "eval_rewards/accuracies": 0.7739383578300476, |
| "eval_rewards/chosen": -0.9925247430801392, |
| "eval_rewards/margins": 1.438172459602356, |
| "eval_rewards/rejected": -2.430696964263916, |
| "eval_runtime": 183.7972, |
| "eval_samples_per_second": 13.069, |
| "eval_steps_per_second": 6.534, |
| "step": 1365 |
| }, |
| { |
| "epoch": 2.15985894798707, |
| "grad_norm": 19.07766342163086, |
| "learning_rate": 2.1181774208625169e-07, |
| "logits/chosen": -1.4445375204086304, |
| "logits/rejected": -1.2832697629928589, |
| "logps/chosen": -351.9157409667969, |
| "logps/rejected": -310.8402404785156, |
| "loss": 0.1382, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -0.36597153544425964, |
| "rewards/margins": 2.7161738872528076, |
| "rewards/rejected": -3.0821452140808105, |
| "step": 1380 |
| }, |
| { |
| "epoch": 2.1833676168086984, |
| "grad_norm": 26.579313278198242, |
| "learning_rate": 2.010247069453462e-07, |
| "logits/chosen": -1.3750087022781372, |
| "logits/rejected": -1.1288143396377563, |
| "logps/chosen": -363.9015197753906, |
| "logps/rejected": -326.6436767578125, |
| "loss": 0.1696, |
| "rewards/accuracies": 0.9479166865348816, |
| "rewards/chosen": -0.4505316913127899, |
| "rewards/margins": 2.662269115447998, |
| "rewards/rejected": -3.1128008365631104, |
| "step": 1395 |
| }, |
| { |
| "epoch": 2.191203839749241, |
| "eval_logits/chosen": -1.4125560522079468, |
| "eval_logits/rejected": -1.2285879850387573, |
| "eval_logps/chosen": -362.7055358886719, |
| "eval_logps/rejected": -310.5049743652344, |
| "eval_loss": 0.4731534719467163, |
| "eval_rewards/accuracies": 0.7718567848205566, |
| "eval_rewards/chosen": -1.0391572713851929, |
| "eval_rewards/margins": 1.4680242538452148, |
| "eval_rewards/rejected": -2.5071816444396973, |
| "eval_runtime": 183.2882, |
| "eval_samples_per_second": 13.105, |
| "eval_steps_per_second": 6.553, |
| "step": 1400 |
| }, |
| { |
| "epoch": 2.2068762856303263, |
| "grad_norm": 31.265209197998047, |
| "learning_rate": 1.904442989575324e-07, |
| "logits/chosen": -1.4215842485427856, |
| "logits/rejected": -1.2600691318511963, |
| "logps/chosen": -371.39471435546875, |
| "logps/rejected": -328.9493408203125, |
| "loss": 0.1491, |
| "rewards/accuracies": 0.9666666388511658, |
| "rewards/chosen": -0.3961075246334076, |
| "rewards/margins": 2.718886375427246, |
| "rewards/rejected": -3.1149940490722656, |
| "step": 1410 |
| }, |
| { |
| "epoch": 2.2303849544519543, |
| "grad_norm": 19.422006607055664, |
| "learning_rate": 1.800840427647492e-07, |
| "logits/chosen": -1.4670078754425049, |
| "logits/rejected": -1.1978638172149658, |
| "logps/chosen": -366.31884765625, |
| "logps/rejected": -310.2874450683594, |
| "loss": 0.1714, |
| "rewards/accuracies": 0.9604166746139526, |
| "rewards/chosen": -0.40646833181381226, |
| "rewards/margins": 2.6268420219421387, |
| "rewards/rejected": -3.0333104133605957, |
| "step": 1425 |
| }, |
| { |
| "epoch": 2.2460574003330396, |
| "eval_logits/chosen": -1.4236711263656616, |
| "eval_logits/rejected": -1.2401769161224365, |
| "eval_logps/chosen": -362.6986389160156, |
| "eval_logps/rejected": -310.4892272949219, |
| "eval_loss": 0.47280353307724, |
| "eval_rewards/accuracies": 0.7726894021034241, |
| "eval_rewards/chosen": -1.0384666919708252, |
| "eval_rewards/margins": 1.4671398401260376, |
| "eval_rewards/rejected": -2.5056064128875732, |
| "eval_runtime": 183.4182, |
| "eval_samples_per_second": 13.096, |
| "eval_steps_per_second": 6.548, |
| "step": 1435 |
| }, |
| { |
| "epoch": 2.2538936232735822, |
| "grad_norm": 22.030029296875, |
| "learning_rate": 1.699513064399812e-07, |
| "logits/chosen": -1.4837629795074463, |
| "logits/rejected": -1.2376011610031128, |
| "logps/chosen": -347.2449035644531, |
| "logps/rejected": -323.3914489746094, |
| "loss": 0.1497, |
| "rewards/accuracies": 0.9708333611488342, |
| "rewards/chosen": -0.4454355239868164, |
| "rewards/margins": 2.712066650390625, |
| "rewards/rejected": -3.1575021743774414, |
| "step": 1440 |
| }, |
| { |
| "epoch": 2.27740229209521, |
| "grad_norm": 22.95709991455078, |
| "learning_rate": 1.6005329624718272e-07, |
| "logits/chosen": -1.3890153169631958, |
| "logits/rejected": -1.2241790294647217, |
| "logps/chosen": -359.70880126953125, |
| "logps/rejected": -325.8527526855469, |
| "loss": 0.1543, |
| "rewards/accuracies": 0.9645833373069763, |
| "rewards/chosen": -0.3865669369697571, |
| "rewards/margins": 2.7711870670318604, |
| "rewards/rejected": -3.1577537059783936, |
| "step": 1455 |
| }, |
| { |
| "epoch": 2.300910960916838, |
| "grad_norm": 16.600971221923828, |
| "learning_rate": 1.503970515162834e-07, |
| "logits/chosen": -1.4796699285507202, |
| "logits/rejected": -1.283089518547058, |
| "logps/chosen": -346.07806396484375, |
| "logps/rejected": -313.4685974121094, |
| "loss": 0.161, |
| "rewards/accuracies": 0.9666666388511658, |
| "rewards/chosen": -0.3793748915195465, |
| "rewards/margins": 2.6482481956481934, |
| "rewards/rejected": -3.027622938156128, |
| "step": 1470 |
| }, |
| { |
| "epoch": 2.300910960916838, |
| "eval_logits/chosen": -1.4251654148101807, |
| "eval_logits/rejected": -1.2433199882507324, |
| "eval_logps/chosen": -362.67913818359375, |
| "eval_logps/rejected": -310.4259338378906, |
| "eval_loss": 0.4735627770423889, |
| "eval_rewards/accuracies": 0.7739383578300476, |
| "eval_rewards/chosen": -1.0365179777145386, |
| "eval_rewards/margins": 1.4627617597579956, |
| "eval_rewards/rejected": -2.499279499053955, |
| "eval_runtime": 183.8286, |
| "eval_samples_per_second": 13.067, |
| "eval_steps_per_second": 6.533, |
| "step": 1470 |
| }, |
| { |
| "epoch": 2.324419629738466, |
| "grad_norm": 21.0710506439209, |
| "learning_rate": 1.4098943963691178e-07, |
| "logits/chosen": -1.469247817993164, |
| "logits/rejected": -1.2050501108169556, |
| "logps/chosen": -346.5850830078125, |
| "logps/rejected": -338.1049499511719, |
| "loss": 0.1465, |
| "rewards/accuracies": 0.9729166626930237, |
| "rewards/chosen": -0.35343825817108154, |
| "rewards/margins": 2.7935383319854736, |
| "rewards/rejected": -3.1469767093658447, |
| "step": 1485 |
| }, |
| { |
| "epoch": 2.347928298560094, |
| "grad_norm": 22.436262130737305, |
| "learning_rate": 1.3183715117440142e-07, |
| "logits/chosen": -1.5117619037628174, |
| "logits/rejected": -1.2061889171600342, |
| "logps/chosen": -347.35565185546875, |
| "logps/rejected": -311.8982238769531, |
| "loss": 0.1468, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -0.4175235331058502, |
| "rewards/margins": 2.734891653060913, |
| "rewards/rejected": -3.1524152755737305, |
| "step": 1500 |
| }, |
| { |
| "epoch": 2.3557645215006366, |
| "eval_logits/chosen": -1.4360556602478027, |
| "eval_logits/rejected": -1.2548272609710693, |
| "eval_logps/chosen": -363.044921875, |
| "eval_logps/rejected": -311.045166015625, |
| "eval_loss": 0.4749804139137268, |
| "eval_rewards/accuracies": 0.7764363288879395, |
| "eval_rewards/chosen": -1.0730968713760376, |
| "eval_rewards/margins": 1.4881056547164917, |
| "eval_rewards/rejected": -2.5612027645111084, |
| "eval_runtime": 183.375, |
| "eval_samples_per_second": 13.099, |
| "eval_steps_per_second": 6.549, |
| "step": 1505 |
| }, |
| { |
| "epoch": 2.371436967381722, |
| "grad_norm": 12.820830345153809, |
| "learning_rate": 1.229466951115519e-07, |
| "logits/chosen": -1.4124829769134521, |
| "logits/rejected": -1.3270103931427002, |
| "logps/chosen": -363.55633544921875, |
| "logps/rejected": -315.1530456542969, |
| "loss": 0.138, |
| "rewards/accuracies": 0.9708333611488342, |
| "rewards/chosen": -0.4280143082141876, |
| "rewards/margins": 2.7927613258361816, |
| "rewards/rejected": -3.220775604248047, |
| "step": 1515 |
| } |
| ], |
| "logging_steps": 15, |
| "max_steps": 1917, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 40, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 0.0, |
| "train_batch_size": 2, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
