Invalid JSON:Unexpected token 'N', ..."ad_norm": NaN,
"... is not valid JSON
| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 939, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.016, | |
| "grad_norm": 3.5558087825775146, | |
| "learning_rate": 6.896551724137931e-07, | |
| "logits/chosen": -1.0070860385894775, | |
| "logits/rejected": -0.978116512298584, | |
| "logps/chosen": -211.4395294189453, | |
| "logps/rejected": -259.907958984375, | |
| "loss": 0.6918, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": 0.003007554914802313, | |
| "rewards/margins": 0.00275672459974885, | |
| "rewards/rejected": 0.0002508305769879371, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.032, | |
| "grad_norm": 4.531008243560791, | |
| "learning_rate": 1.5517241379310346e-06, | |
| "logits/chosen": -1.0405652523040771, | |
| "logits/rejected": -1.108757734298706, | |
| "logps/chosen": -225.45095825195312, | |
| "logps/rejected": -228.9770965576172, | |
| "loss": 0.6757, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": 0.029836362227797508, | |
| "rewards/margins": 0.03624327853322029, | |
| "rewards/rejected": -0.00640691677108407, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.048, | |
| "grad_norm": 3.2132256031036377, | |
| "learning_rate": 2.4137931034482762e-06, | |
| "logits/chosen": -0.9458276033401489, | |
| "logits/rejected": -0.8570151329040527, | |
| "logps/chosen": -229.1313934326172, | |
| "logps/rejected": -243.45675659179688, | |
| "loss": 0.6407, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": 0.10421393811702728, | |
| "rewards/margins": 0.11481194198131561, | |
| "rewards/rejected": -0.010597997345030308, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.064, | |
| "grad_norm": 2.8418328762054443, | |
| "learning_rate": 3.2758620689655175e-06, | |
| "logits/chosen": -1.0654828548431396, | |
| "logits/rejected": -1.0304937362670898, | |
| "logps/chosen": -220.5907440185547, | |
| "logps/rejected": -223.35400390625, | |
| "loss": 0.6019, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": 0.20341631770133972, | |
| "rewards/margins": 0.2232368290424347, | |
| "rewards/rejected": -0.01982051134109497, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 3.0401999950408936, | |
| "learning_rate": 4.137931034482759e-06, | |
| "logits/chosen": -1.1209378242492676, | |
| "logits/rejected": -1.0013506412506104, | |
| "logps/chosen": -222.6129150390625, | |
| "logps/rejected": -236.6962127685547, | |
| "loss": 0.5516, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": 0.34540629386901855, | |
| "rewards/margins": 0.3849923014640808, | |
| "rewards/rejected": -0.039586011320352554, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.096, | |
| "grad_norm": 2.5417866706848145, | |
| "learning_rate": 5e-06, | |
| "logits/chosen": -1.1623287200927734, | |
| "logits/rejected": -1.1034960746765137, | |
| "logps/chosen": -210.07327270507812, | |
| "logps/rejected": -253.70626831054688, | |
| "loss": 0.4866, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": 0.51075279712677, | |
| "rewards/margins": 0.6025584936141968, | |
| "rewards/rejected": -0.09180571138858795, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.112, | |
| "grad_norm": 2.4233880043029785, | |
| "learning_rate": 4.999627560102124e-06, | |
| "logits/chosen": -1.158803105354309, | |
| "logits/rejected": -1.154788851737976, | |
| "logps/chosen": -210.58798217773438, | |
| "logps/rejected": -244.96206665039062, | |
| "loss": 0.4686, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": 0.6626904606819153, | |
| "rewards/margins": 0.7746375203132629, | |
| "rewards/rejected": -0.11194711923599243, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.128, | |
| "grad_norm": 2.218043088912964, | |
| "learning_rate": 4.998510351377676e-06, | |
| "logits/chosen": -1.1345102787017822, | |
| "logits/rejected": -1.017688512802124, | |
| "logps/chosen": -211.61611938476562, | |
| "logps/rejected": -234.57315063476562, | |
| "loss": 0.4477, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": 0.7445486187934875, | |
| "rewards/margins": 0.899127185344696, | |
| "rewards/rejected": -0.15457859635353088, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.144, | |
| "grad_norm": 2.5353105068206787, | |
| "learning_rate": 4.996648706701141e-06, | |
| "logits/chosen": -1.1606733798980713, | |
| "logits/rejected": -1.1573020219802856, | |
| "logps/chosen": -192.83447265625, | |
| "logps/rejected": -236.6874542236328, | |
| "loss": 0.4351, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": 0.8413721323013306, | |
| "rewards/margins": 1.077791452407837, | |
| "rewards/rejected": -0.23641936480998993, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 3.485431671142578, | |
| "learning_rate": 4.99404318075312e-06, | |
| "logits/chosen": -1.1803537607192993, | |
| "logits/rejected": -1.0980823040008545, | |
| "logps/chosen": -221.9828643798828, | |
| "logps/rejected": -223.74551391601562, | |
| "loss": 0.4058, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": 1.0138968229293823, | |
| "rewards/margins": 1.2538959980010986, | |
| "rewards/rejected": -0.23999913036823273, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.176, | |
| "grad_norm": 1.676897406578064, | |
| "learning_rate": 4.990694549855066e-06, | |
| "logits/chosen": -1.1700044870376587, | |
| "logits/rejected": -1.1240018606185913, | |
| "logps/chosen": -201.88095092773438, | |
| "logps/rejected": -241.2851104736328, | |
| "loss": 0.3752, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": 1.2212125062942505, | |
| "rewards/margins": 1.552638292312622, | |
| "rewards/rejected": -0.33142566680908203, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.192, | |
| "grad_norm": 2.7544007301330566, | |
| "learning_rate": 4.986603811737982e-06, | |
| "logits/chosen": -1.1804865598678589, | |
| "logits/rejected": -1.1661874055862427, | |
| "logps/chosen": -226.8423309326172, | |
| "logps/rejected": -266.3681945800781, | |
| "loss": 0.3889, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": 0.9815112352371216, | |
| "rewards/margins": 1.3240571022033691, | |
| "rewards/rejected": -0.34254592657089233, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.208, | |
| "grad_norm": 2.14679217338562, | |
| "learning_rate": 4.981772185245135e-06, | |
| "logits/chosen": -1.1490398645401, | |
| "logits/rejected": -1.1126385927200317, | |
| "logps/chosen": -208.37521362304688, | |
| "logps/rejected": -251.33090209960938, | |
| "loss": 0.3573, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": 1.256662130355835, | |
| "rewards/margins": 1.747520089149475, | |
| "rewards/rejected": -0.4908579885959625, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.224, | |
| "grad_norm": 2.0803537368774414, | |
| "learning_rate": 4.976201109968909e-06, | |
| "logits/chosen": -1.2029155492782593, | |
| "logits/rejected": -1.1176910400390625, | |
| "logps/chosen": -207.2776336669922, | |
| "logps/rejected": -223.60400390625, | |
| "loss": 0.3721, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": 1.2059996128082275, | |
| "rewards/margins": 1.7384388446807861, | |
| "rewards/rejected": -0.5324392318725586, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 2.2244112491607666, | |
| "learning_rate": 4.969892245821867e-06, | |
| "logits/chosen": -1.2101242542266846, | |
| "logits/rejected": -1.2214491367340088, | |
| "logps/chosen": -191.7788543701172, | |
| "logps/rejected": -240.3572235107422, | |
| "loss": 0.3254, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": 1.2775731086730957, | |
| "rewards/margins": 1.9345057010650635, | |
| "rewards/rejected": -0.656932532787323, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.256, | |
| "grad_norm": 2.4931509494781494, | |
| "learning_rate": 4.9628474725421845e-06, | |
| "logits/chosen": -1.2070999145507812, | |
| "logits/rejected": -1.125981092453003, | |
| "logps/chosen": -214.82461547851562, | |
| "logps/rejected": -259.69122314453125, | |
| "loss": 0.3627, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": 1.1662296056747437, | |
| "rewards/margins": 1.7516357898712158, | |
| "rewards/rejected": -0.5854062438011169, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.272, | |
| "grad_norm": 2.0334951877593994, | |
| "learning_rate": 4.955068889133577e-06, | |
| "logits/chosen": -1.257777214050293, | |
| "logits/rejected": -1.2115817070007324, | |
| "logps/chosen": -201.50735473632812, | |
| "logps/rejected": -231.436767578125, | |
| "loss": 0.3581, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 1.2795923948287964, | |
| "rewards/margins": 1.8879226446151733, | |
| "rewards/rejected": -0.6083300709724426, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.288, | |
| "grad_norm": NaN, | |
| "learning_rate": 4.946558813239889e-06, | |
| "logits/chosen": -1.26598060131073, | |
| "logits/rejected": -1.211882472038269, | |
| "logps/chosen": -204.62889099121094, | |
| "logps/rejected": -248.060546875, | |
| "loss": 0.3342, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": 1.2854232788085938, | |
| "rewards/margins": 2.1888251304626465, | |
| "rewards/rejected": -0.9034019708633423, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.304, | |
| "grad_norm": 2.3634698390960693, | |
| "learning_rate": 4.937319780454559e-06, | |
| "logits/chosen": -1.350507378578186, | |
| "logits/rejected": -1.2982906103134155, | |
| "logps/chosen": -194.1533966064453, | |
| "logps/rejected": -239.66055297851562, | |
| "loss": 0.3292, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": 1.2391879558563232, | |
| "rewards/margins": 2.0913984775543213, | |
| "rewards/rejected": -0.8522104024887085, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 3.3314573764801025, | |
| "learning_rate": 4.927354543565131e-06, | |
| "logits/chosen": -1.2342275381088257, | |
| "logits/rejected": -1.1977652311325073, | |
| "logps/chosen": -217.9714813232422, | |
| "logps/rejected": -260.415283203125, | |
| "loss": 0.3733, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": 1.1353275775909424, | |
| "rewards/margins": 1.9522864818572998, | |
| "rewards/rejected": -0.816959023475647, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.336, | |
| "grad_norm": 2.4203505516052246, | |
| "learning_rate": 4.9166660717330496e-06, | |
| "logits/chosen": -1.3102691173553467, | |
| "logits/rejected": -1.2377029657363892, | |
| "logps/chosen": -187.3673858642578, | |
| "logps/rejected": -238.6230926513672, | |
| "loss": 0.3051, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": 1.4179824590682983, | |
| "rewards/margins": 2.268038272857666, | |
| "rewards/rejected": -0.8500560522079468, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.352, | |
| "grad_norm": 2.4510674476623535, | |
| "learning_rate": 4.905257549609002e-06, | |
| "logits/chosen": -1.305759072303772, | |
| "logits/rejected": -1.2166777849197388, | |
| "logps/chosen": -207.846435546875, | |
| "logps/rejected": -244.6707000732422, | |
| "loss": 0.281, | |
| "rewards/accuracies": 0.893750011920929, | |
| "rewards/chosen": 1.5557820796966553, | |
| "rewards/margins": 2.547685146331787, | |
| "rewards/rejected": -0.9919027090072632, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.368, | |
| "grad_norm": 2.3315584659576416, | |
| "learning_rate": 4.89313237638404e-06, | |
| "logits/chosen": -1.2491403818130493, | |
| "logits/rejected": -1.2307875156402588, | |
| "logps/chosen": -213.06405639648438, | |
| "logps/rejected": -255.96499633789062, | |
| "loss": 0.3359, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": 1.2078200578689575, | |
| "rewards/margins": 2.0568363666534424, | |
| "rewards/rejected": -0.8490163087844849, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.384, | |
| "grad_norm": 2.4910717010498047, | |
| "learning_rate": 4.880294164776785e-06, | |
| "logits/chosen": -1.3219393491744995, | |
| "logits/rejected": -1.3105016946792603, | |
| "logps/chosen": -225.34326171875, | |
| "logps/rejected": -267.8374938964844, | |
| "loss": 0.2741, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": 1.3729044198989868, | |
| "rewards/margins": 2.491551637649536, | |
| "rewards/rejected": -1.1186472177505493, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 2.8208534717559814, | |
| "learning_rate": 4.866746739957016e-06, | |
| "logits/chosen": -1.2614940404891968, | |
| "logits/rejected": -1.3334258794784546, | |
| "logps/chosen": -209.5437774658203, | |
| "logps/rejected": -268.48028564453125, | |
| "loss": 0.3283, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": 0.9299019575119019, | |
| "rewards/margins": 2.1982016563415527, | |
| "rewards/rejected": -1.2682996988296509, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.416, | |
| "grad_norm": 2.4839680194854736, | |
| "learning_rate": 4.852494138405942e-06, | |
| "logits/chosen": -1.3196954727172852, | |
| "logits/rejected": -1.242071509361267, | |
| "logps/chosen": -222.31631469726562, | |
| "logps/rejected": -253.88211059570312, | |
| "loss": 0.2927, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 1.3626937866210938, | |
| "rewards/margins": 2.622807025909424, | |
| "rewards/rejected": -1.2601131200790405, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.432, | |
| "grad_norm": 2.6777167320251465, | |
| "learning_rate": 4.837540606713538e-06, | |
| "logits/chosen": -1.3383070230484009, | |
| "logits/rejected": -1.2611116170883179, | |
| "logps/chosen": -215.638916015625, | |
| "logps/rejected": -233.745361328125, | |
| "loss": 0.3246, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": 1.1914758682250977, | |
| "rewards/margins": 2.3704941272735596, | |
| "rewards/rejected": -1.179018259048462, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.448, | |
| "grad_norm": 2.0181572437286377, | |
| "learning_rate": 4.821890600313256e-06, | |
| "logits/chosen": -1.2718770503997803, | |
| "logits/rejected": -1.2419809103012085, | |
| "logps/chosen": -207.59140014648438, | |
| "logps/rejected": -246.1212158203125, | |
| "loss": 0.294, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": 1.152263879776001, | |
| "rewards/margins": 2.4711754322052, | |
| "rewards/rejected": -1.3189116716384888, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.464, | |
| "grad_norm": 2.557889461517334, | |
| "learning_rate": 4.805548782154527e-06, | |
| "logits/chosen": -1.3852113485336304, | |
| "logits/rejected": -1.317633867263794, | |
| "logps/chosen": -207.5449676513672, | |
| "logps/rejected": -258.5287780761719, | |
| "loss": 0.2983, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": 1.0099232196807861, | |
| "rewards/margins": 2.554460048675537, | |
| "rewards/rejected": -1.544536828994751, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 2.6911659240722656, | |
| "learning_rate": 4.788520021313416e-06, | |
| "logits/chosen": -1.3613696098327637, | |
| "logits/rejected": -1.3260200023651123, | |
| "logps/chosen": -203.2208251953125, | |
| "logps/rejected": -252.5656280517578, | |
| "loss": 0.2815, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 1.1206295490264893, | |
| "rewards/margins": 2.8727269172668457, | |
| "rewards/rejected": -1.7520973682403564, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.496, | |
| "grad_norm": 2.3627328872680664, | |
| "learning_rate": 4.770809391541885e-06, | |
| "logits/chosen": -1.4218701124191284, | |
| "logits/rejected": -1.4015001058578491, | |
| "logps/chosen": -208.8138427734375, | |
| "logps/rejected": -256.91021728515625, | |
| "loss": 0.3085, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 0.9383912086486816, | |
| "rewards/margins": 2.533010721206665, | |
| "rewards/rejected": -1.5946193933486938, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.512, | |
| "grad_norm": 3.0470287799835205, | |
| "learning_rate": 4.752422169756048e-06, | |
| "logits/chosen": -1.367304801940918, | |
| "logits/rejected": -1.3290746212005615, | |
| "logps/chosen": -195.11813354492188, | |
| "logps/rejected": -259.2236022949219, | |
| "loss": 0.3068, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": 0.8705726861953735, | |
| "rewards/margins": 2.558838367462158, | |
| "rewards/rejected": -1.6882654428482056, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.528, | |
| "grad_norm": 2.524092435836792, | |
| "learning_rate": 4.733363834463908e-06, | |
| "logits/chosen": -1.3467628955841064, | |
| "logits/rejected": -1.3406982421875, | |
| "logps/chosen": -209.3506622314453, | |
| "logps/rejected": -261.6465759277344, | |
| "loss": 0.2852, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": 0.982197105884552, | |
| "rewards/margins": 3.0284159183502197, | |
| "rewards/rejected": -2.0462186336517334, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.544, | |
| "grad_norm": 2.598055601119995, | |
| "learning_rate": 4.7136400641330245e-06, | |
| "logits/chosen": -1.3876451253890991, | |
| "logits/rejected": -1.3981963396072388, | |
| "logps/chosen": -198.97781372070312, | |
| "logps/rejected": -261.360595703125, | |
| "loss": 0.3937, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": 0.6369429230690002, | |
| "rewards/margins": 2.3431198596954346, | |
| "rewards/rejected": -1.706176996231079, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 2.1487221717834473, | |
| "learning_rate": 4.693256735498605e-06, | |
| "logits/chosen": -1.27901291847229, | |
| "logits/rejected": -1.2287042140960693, | |
| "logps/chosen": -207.0474395751953, | |
| "logps/rejected": -225.71542358398438, | |
| "loss": 0.3105, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 1.0815143585205078, | |
| "rewards/margins": 2.551909923553467, | |
| "rewards/rejected": -1.470395565032959, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.576, | |
| "grad_norm": 2.4707322120666504, | |
| "learning_rate": 4.672219921812517e-06, | |
| "logits/chosen": -1.3839061260223389, | |
| "logits/rejected": -1.394359827041626, | |
| "logps/chosen": -202.1638946533203, | |
| "logps/rejected": -260.03961181640625, | |
| "loss": 0.3076, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": 0.9943957328796387, | |
| "rewards/margins": 2.6735429763793945, | |
| "rewards/rejected": -1.6791471242904663, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.592, | |
| "grad_norm": 2.918832540512085, | |
| "learning_rate": 4.650535891033752e-06, | |
| "logits/chosen": -1.425442099571228, | |
| "logits/rejected": -1.400042176246643, | |
| "logps/chosen": -212.57003784179688, | |
| "logps/rejected": -266.183349609375, | |
| "loss": 0.277, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": 1.072869896888733, | |
| "rewards/margins": 2.737441301345825, | |
| "rewards/rejected": -1.6645710468292236, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.608, | |
| "grad_norm": 2.2655458450317383, | |
| "learning_rate": 4.6282111039608786e-06, | |
| "logits/chosen": -1.4577895402908325, | |
| "logits/rejected": -1.412097692489624, | |
| "logps/chosen": -211.5107421875, | |
| "logps/rejected": -260.93280029296875, | |
| "loss": 0.2882, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 1.102726697921753, | |
| "rewards/margins": 2.8348031044006348, | |
| "rewards/rejected": -1.7320764064788818, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.624, | |
| "grad_norm": 2.698307991027832, | |
| "learning_rate": 4.60525221230703e-06, | |
| "logits/chosen": -1.428170919418335, | |
| "logits/rejected": -1.317826509475708, | |
| "logps/chosen": -202.50094604492188, | |
| "logps/rejected": -244.787841796875, | |
| "loss": 0.2431, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 1.3533462285995483, | |
| "rewards/margins": 3.1192307472229004, | |
| "rewards/rejected": -1.7658843994140625, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 4.034365177154541, | |
| "learning_rate": 4.581666056718016e-06, | |
| "logits/chosen": -1.3452891111373901, | |
| "logits/rejected": -1.3130595684051514, | |
| "logps/chosen": -216.2296905517578, | |
| "logps/rejected": -259.84454345703125, | |
| "loss": 0.29, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": 1.1018333435058594, | |
| "rewards/margins": 2.8792738914489746, | |
| "rewards/rejected": -1.7774404287338257, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.656, | |
| "grad_norm": 3.709115982055664, | |
| "learning_rate": 4.5574596647341414e-06, | |
| "logits/chosen": -1.4108755588531494, | |
| "logits/rejected": -1.3489899635314941, | |
| "logps/chosen": -215.44210815429688, | |
| "logps/rejected": -245.09335327148438, | |
| "loss": 0.323, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": 0.8439299464225769, | |
| "rewards/margins": 2.641413688659668, | |
| "rewards/rejected": -1.7974838018417358, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.672, | |
| "grad_norm": 2.070817470550537, | |
| "learning_rate": 4.532640248696331e-06, | |
| "logits/chosen": -1.4538488388061523, | |
| "logits/rejected": -1.37591552734375, | |
| "logps/chosen": -207.18453979492188, | |
| "logps/rejected": -269.21502685546875, | |
| "loss": 0.2839, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": 1.192795991897583, | |
| "rewards/margins": 3.0717904567718506, | |
| "rewards/rejected": -1.878994345664978, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.688, | |
| "grad_norm": 2.577613592147827, | |
| "learning_rate": 4.5072152035972065e-06, | |
| "logits/chosen": -1.371187448501587, | |
| "logits/rejected": -1.313145399093628, | |
| "logps/chosen": -216.05545043945312, | |
| "logps/rejected": -257.0478820800781, | |
| "loss": 0.3076, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": 1.1174415349960327, | |
| "rewards/margins": 2.8477675914764404, | |
| "rewards/rejected": -1.7303262948989868, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.704, | |
| "grad_norm": 2.084120750427246, | |
| "learning_rate": 4.481192104877727e-06, | |
| "logits/chosen": -1.4179775714874268, | |
| "logits/rejected": -1.4000885486602783, | |
| "logps/chosen": -199.14834594726562, | |
| "logps/rejected": -259.9361267089844, | |
| "loss": 0.2761, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 1.2094013690948486, | |
| "rewards/margins": 2.9881272315979004, | |
| "rewards/rejected": -1.7787256240844727, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 2.672084331512451, | |
| "learning_rate": 4.454578706170075e-06, | |
| "logits/chosen": -1.4160033464431763, | |
| "logits/rejected": -1.381691336631775, | |
| "logps/chosen": -212.1450653076172, | |
| "logps/rejected": -264.46978759765625, | |
| "loss": 0.2776, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 1.1448103189468384, | |
| "rewards/margins": 3.152707099914551, | |
| "rewards/rejected": -2.007896661758423, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.736, | |
| "grad_norm": 2.2846269607543945, | |
| "learning_rate": 4.427382936987449e-06, | |
| "logits/chosen": -1.4172203540802002, | |
| "logits/rejected": -1.3477474451065063, | |
| "logps/chosen": -201.10195922851562, | |
| "logps/rejected": -253.28067016601562, | |
| "loss": 0.2495, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": 1.2827614545822144, | |
| "rewards/margins": 3.226905345916748, | |
| "rewards/rejected": -1.9441440105438232, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.752, | |
| "grad_norm": 2.592393398284912, | |
| "learning_rate": 4.3996129003614476e-06, | |
| "logits/chosen": -1.3589997291564941, | |
| "logits/rejected": -1.329886794090271, | |
| "logps/chosen": -269.73968505859375, | |
| "logps/rejected": -273.5852355957031, | |
| "loss": 0.234, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 1.4174926280975342, | |
| "rewards/margins": 3.680424213409424, | |
| "rewards/rejected": -2.2629313468933105, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.768, | |
| "grad_norm": 3.361455202102661, | |
| "learning_rate": 4.3712768704277535e-06, | |
| "logits/chosen": -1.3936150074005127, | |
| "logits/rejected": -1.396785855293274, | |
| "logps/chosen": -199.318115234375, | |
| "logps/rejected": -256.87396240234375, | |
| "loss": 0.245, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": 1.1769810914993286, | |
| "rewards/margins": 3.183269739151001, | |
| "rewards/rejected": -2.006288766860962, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.784, | |
| "grad_norm": 2.744699239730835, | |
| "learning_rate": 4.342383289960842e-06, | |
| "logits/chosen": -1.4649932384490967, | |
| "logits/rejected": -1.3714121580123901, | |
| "logps/chosen": -190.34593200683594, | |
| "logps/rejected": -262.5522766113281, | |
| "loss": 0.2459, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": 1.1242296695709229, | |
| "rewards/margins": 3.3215816020965576, | |
| "rewards/rejected": -2.1973519325256348, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 2.5331287384033203, | |
| "learning_rate": 4.312940767858442e-06, | |
| "logits/chosen": -1.5104731321334839, | |
| "logits/rejected": -1.4501935243606567, | |
| "logps/chosen": -218.8745880126953, | |
| "logps/rejected": -274.6693115234375, | |
| "loss": 0.2521, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 1.0992300510406494, | |
| "rewards/margins": 3.235896348953247, | |
| "rewards/rejected": -2.1366658210754395, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.816, | |
| "grad_norm": 2.387657880783081, | |
| "learning_rate": 4.282958076576491e-06, | |
| "logits/chosen": -1.495585560798645, | |
| "logits/rejected": -1.4296038150787354, | |
| "logps/chosen": -249.4803466796875, | |
| "logps/rejected": -273.93994140625, | |
| "loss": 0.25, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": 1.0271832942962646, | |
| "rewards/margins": 3.5675711631774902, | |
| "rewards/rejected": -2.5403876304626465, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.832, | |
| "grad_norm": 2.605201244354248, | |
| "learning_rate": 4.252444149515374e-06, | |
| "logits/chosen": -1.4731948375701904, | |
| "logits/rejected": -1.4050095081329346, | |
| "logps/chosen": -223.21548461914062, | |
| "logps/rejected": -277.0063171386719, | |
| "loss": 0.2489, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": 1.0619561672210693, | |
| "rewards/margins": 3.671313524246216, | |
| "rewards/rejected": -2.6093573570251465, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.848, | |
| "grad_norm": 2.5914604663848877, | |
| "learning_rate": 4.221408078358193e-06, | |
| "logits/chosen": -1.582438349723816, | |
| "logits/rejected": -1.4810370206832886, | |
| "logps/chosen": -196.49365234375, | |
| "logps/rejected": -268.8774108886719, | |
| "loss": 0.229, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 1.138505220413208, | |
| "rewards/margins": 3.519007921218872, | |
| "rewards/rejected": -2.380502462387085, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.864, | |
| "grad_norm": 2.9997308254241943, | |
| "learning_rate": 4.189859110361886e-06, | |
| "logits/chosen": -1.4258604049682617, | |
| "logits/rejected": -1.2593319416046143, | |
| "logps/chosen": -233.5980224609375, | |
| "logps/rejected": -268.10211181640625, | |
| "loss": 0.2899, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 1.0289585590362549, | |
| "rewards/margins": 3.284339189529419, | |
| "rewards/rejected": -2.255380868911743, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 3.4195799827575684, | |
| "learning_rate": 4.1578066456019885e-06, | |
| "logits/chosen": -1.4348300695419312, | |
| "logits/rejected": -1.3643486499786377, | |
| "logps/chosen": -201.9141082763672, | |
| "logps/rejected": -248.4167022705078, | |
| "loss": 0.2646, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": 0.7493404150009155, | |
| "rewards/margins": 3.0913987159729004, | |
| "rewards/rejected": -2.3420581817626953, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.896, | |
| "grad_norm": 2.695443868637085, | |
| "learning_rate": 4.125260234171861e-06, | |
| "logits/chosen": -1.4055596590042114, | |
| "logits/rejected": -1.4040924310684204, | |
| "logps/chosen": -191.24880981445312, | |
| "logps/rejected": -268.87591552734375, | |
| "loss": 0.2082, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": 1.337165355682373, | |
| "rewards/margins": 4.025852203369141, | |
| "rewards/rejected": -2.6886870861053467, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.912, | |
| "grad_norm": 2.7110860347747803, | |
| "learning_rate": 4.092229573337223e-06, | |
| "logits/chosen": -1.5467432737350464, | |
| "logits/rejected": -1.4956064224243164, | |
| "logps/chosen": -217.4794158935547, | |
| "logps/rejected": -271.25726318359375, | |
| "loss": 0.2636, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 0.9303010106086731, | |
| "rewards/margins": 3.6637215614318848, | |
| "rewards/rejected": -2.7334208488464355, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.928, | |
| "grad_norm": 3.4274234771728516, | |
| "learning_rate": 4.058724504646834e-06, | |
| "logits/chosen": -1.4762035608291626, | |
| "logits/rejected": -1.4678940773010254, | |
| "logps/chosen": -215.96139526367188, | |
| "logps/rejected": -281.8223876953125, | |
| "loss": 0.2417, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 0.948248565196991, | |
| "rewards/margins": 3.616924285888672, | |
| "rewards/rejected": -2.668675661087036, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.944, | |
| "grad_norm": 2.92130708694458, | |
| "learning_rate": 4.0247550110001825e-06, | |
| "logits/chosen": -1.5313862562179565, | |
| "logits/rejected": -1.5171687602996826, | |
| "logps/chosen": -214.7728271484375, | |
| "logps/rejected": -268.3896179199219, | |
| "loss": 0.2797, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": 0.846409797668457, | |
| "rewards/margins": 3.2781710624694824, | |
| "rewards/rejected": -2.4317612648010254, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 3.7328531742095947, | |
| "learning_rate": 3.990331213673064e-06, | |
| "logits/chosen": -1.4543683528900146, | |
| "logits/rejected": -1.4041635990142822, | |
| "logps/chosen": -221.00619506835938, | |
| "logps/rejected": -261.167236328125, | |
| "loss": 0.328, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": 0.697174072265625, | |
| "rewards/margins": 3.078673839569092, | |
| "rewards/rejected": -2.381499767303467, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.976, | |
| "grad_norm": 3.6233444213867188, | |
| "learning_rate": 3.9554633693019275e-06, | |
| "logits/chosen": -1.4626250267028809, | |
| "logits/rejected": -1.4502376317977905, | |
| "logps/chosen": -224.6941375732422, | |
| "logps/rejected": -265.57562255859375, | |
| "loss": 0.2784, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": 0.9059398770332336, | |
| "rewards/margins": 3.199237823486328, | |
| "rewards/rejected": -2.29329776763916, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.992, | |
| "grad_norm": 3.0609471797943115, | |
| "learning_rate": 3.92016186682789e-06, | |
| "logits/chosen": -1.4352532625198364, | |
| "logits/rejected": -1.3892724514007568, | |
| "logps/chosen": -232.20510864257812, | |
| "logps/rejected": -249.70175170898438, | |
| "loss": 0.3045, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 1.0156091451644897, | |
| "rewards/margins": 2.6612770557403564, | |
| "rewards/rejected": -1.6456680297851562, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.0064, | |
| "grad_norm": 2.452713966369629, | |
| "learning_rate": 3.884437224401333e-06, | |
| "logits/chosen": -1.4115742444992065, | |
| "logits/rejected": -1.425600528717041, | |
| "logps/chosen": -197.7564697265625, | |
| "logps/rejected": -261.188232421875, | |
| "loss": 0.2462, | |
| "rewards/accuracies": 0.9097222089767456, | |
| "rewards/chosen": 0.8398535251617432, | |
| "rewards/margins": 3.0637478828430176, | |
| "rewards/rejected": -2.223893880844116, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.0224, | |
| "grad_norm": 2.764373779296875, | |
| "learning_rate": 3.848300086247998e-06, | |
| "logits/chosen": -1.480515718460083, | |
| "logits/rejected": -1.4369785785675049, | |
| "logps/chosen": -227.99496459960938, | |
| "logps/rejected": -269.6391296386719, | |
| "loss": 0.2132, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 1.1980942487716675, | |
| "rewards/margins": 3.635913133621216, | |
| "rewards/rejected": -2.4378185272216797, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.0384, | |
| "grad_norm": 2.349464178085327, | |
| "learning_rate": 3.81176121949752e-06, | |
| "logits/chosen": -1.439501166343689, | |
| "logits/rejected": -1.4518448114395142, | |
| "logps/chosen": -221.046630859375, | |
| "logps/rejected": -268.6219482421875, | |
| "loss": 0.1815, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": 1.5582630634307861, | |
| "rewards/margins": 3.5910651683807373, | |
| "rewards/rejected": -2.0328023433685303, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.0544, | |
| "grad_norm": 2.8748230934143066, | |
| "learning_rate": 3.7748315109753402e-06, | |
| "logits/chosen": -1.473581075668335, | |
| "logits/rejected": -1.3876584768295288, | |
| "logps/chosen": -233.5598907470703, | |
| "logps/rejected": -265.6020812988281, | |
| "loss": 0.1887, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 1.539259433746338, | |
| "rewards/margins": 3.6530539989471436, | |
| "rewards/rejected": -2.1137945652008057, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.0704, | |
| "grad_norm": 2.4874422550201416, | |
| "learning_rate": 3.737521963958954e-06, | |
| "logits/chosen": -1.4513788223266602, | |
| "logits/rejected": -1.4336411952972412, | |
| "logps/chosen": -223.05093383789062, | |
| "logps/rejected": -278.9050598144531, | |
| "loss": 0.1933, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": 1.3904491662979126, | |
| "rewards/margins": 4.047357082366943, | |
| "rewards/rejected": -2.6569082736968994, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.0864, | |
| "grad_norm": 1.6010459661483765, | |
| "learning_rate": 3.6998436948994664e-06, | |
| "logits/chosen": -1.4964760541915894, | |
| "logits/rejected": -1.4747968912124634, | |
| "logps/chosen": -206.62960815429688, | |
| "logps/rejected": -263.10955810546875, | |
| "loss": 0.1859, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 1.4117496013641357, | |
| "rewards/margins": 4.006656169891357, | |
| "rewards/rejected": -2.5949065685272217, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.1024, | |
| "grad_norm": 1.9789769649505615, | |
| "learning_rate": 3.661807930109422e-06, | |
| "logits/chosen": -1.485519528388977, | |
| "logits/rejected": -1.4273617267608643, | |
| "logps/chosen": -211.47769165039062, | |
| "logps/rejected": -276.42559814453125, | |
| "loss": 0.206, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 1.1727440357208252, | |
| "rewards/margins": 3.8037447929382324, | |
| "rewards/rejected": -2.6310009956359863, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.1184, | |
| "grad_norm": 2.5497982501983643, | |
| "learning_rate": 3.6234260024179036e-06, | |
| "logits/chosen": -1.4089610576629639, | |
| "logits/rejected": -1.302335500717163, | |
| "logps/chosen": -213.62350463867188, | |
| "logps/rejected": -266.78814697265625, | |
| "loss": 0.2387, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": 1.048295021057129, | |
| "rewards/margins": 3.6779956817626953, | |
| "rewards/rejected": -2.6297006607055664, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.1344, | |
| "grad_norm": 1.9107615947723389, | |
| "learning_rate": 3.5847093477938955e-06, | |
| "logits/chosen": -1.5314279794692993, | |
| "logits/rejected": -1.4192278385162354, | |
| "logps/chosen": -193.95803833007812, | |
| "logps/rejected": -266.6246643066406, | |
| "loss": 0.1881, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 1.1405754089355469, | |
| "rewards/margins": 3.900803804397583, | |
| "rewards/rejected": -2.7602286338806152, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.1504, | |
| "grad_norm": 2.248903274536133, | |
| "learning_rate": 3.545669501938913e-06, | |
| "logits/chosen": -1.3808566331863403, | |
| "logits/rejected": -1.3538700342178345, | |
| "logps/chosen": -229.12655639648438, | |
| "logps/rejected": -274.583251953125, | |
| "loss": 0.1907, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 1.035677433013916, | |
| "rewards/margins": 3.7907638549804688, | |
| "rewards/rejected": -2.7550864219665527, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.1663999999999999, | |
| "grad_norm": 2.1139893531799316, | |
| "learning_rate": 3.506318096849919e-06, | |
| "logits/chosen": -1.5006558895111084, | |
| "logits/rejected": -1.4343509674072266, | |
| "logps/chosen": -210.866943359375, | |
| "logps/rejected": -255.4267120361328, | |
| "loss": 0.2048, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": 0.8776493072509766, | |
| "rewards/margins": 3.7163491249084473, | |
| "rewards/rejected": -2.8386993408203125, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.1824, | |
| "grad_norm": 2.827563524246216, | |
| "learning_rate": 3.466666857353547e-06, | |
| "logits/chosen": -1.4484174251556396, | |
| "logits/rejected": -1.3797426223754883, | |
| "logps/chosen": -228.63720703125, | |
| "logps/rejected": -262.3194274902344, | |
| "loss": 0.24, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": 1.26779043674469, | |
| "rewards/margins": 3.7388503551483154, | |
| "rewards/rejected": -2.471060276031494, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.1984, | |
| "grad_norm": 5.360053539276123, | |
| "learning_rate": 3.4267275976126697e-06, | |
| "logits/chosen": -1.5434255599975586, | |
| "logits/rejected": -1.4696922302246094, | |
| "logps/chosen": -205.205322265625, | |
| "logps/rejected": -279.88714599609375, | |
| "loss": 0.1568, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": 1.2206922769546509, | |
| "rewards/margins": 4.399322032928467, | |
| "rewards/rejected": -3.1786296367645264, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.2144, | |
| "grad_norm": 2.5186727046966553, | |
| "learning_rate": 3.386512217606339e-06, | |
| "logits/chosen": -1.481487512588501, | |
| "logits/rejected": -1.4067240953445435, | |
| "logps/chosen": -199.1824951171875, | |
| "logps/rejected": -262.4269714355469, | |
| "loss": 0.1983, | |
| "rewards/accuracies": 0.893750011920929, | |
| "rewards/chosen": 1.1041431427001953, | |
| "rewards/margins": 3.9722588062286377, | |
| "rewards/rejected": -2.8681156635284424, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.2304, | |
| "grad_norm": 2.019970417022705, | |
| "learning_rate": 3.346032699584176e-06, | |
| "logits/chosen": -1.5157972574234009, | |
| "logits/rejected": -1.4852614402770996, | |
| "logps/chosen": -197.70455932617188, | |
| "logps/rejected": -257.8968505859375, | |
| "loss": 0.2178, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": 1.0223619937896729, | |
| "rewards/margins": 3.5193912982940674, | |
| "rewards/rejected": -2.4970290660858154, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.2464, | |
| "grad_norm": 2.2139105796813965, | |
| "learning_rate": 3.3053011044962268e-06, | |
| "logits/chosen": -1.564191222190857, | |
| "logits/rejected": -1.5597954988479614, | |
| "logps/chosen": -203.78013610839844, | |
| "logps/rejected": -282.21136474609375, | |
| "loss": 0.1759, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 1.0344898700714111, | |
| "rewards/margins": 4.304172992706299, | |
| "rewards/rejected": -3.2696831226348877, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.2624, | |
| "grad_norm": 3.2327651977539062, | |
| "learning_rate": 3.2643295683993827e-06, | |
| "logits/chosen": -1.504197120666504, | |
| "logits/rejected": -1.4315654039382935, | |
| "logps/chosen": -217.15869140625, | |
| "logps/rejected": -265.8768310546875, | |
| "loss": 0.2149, | |
| "rewards/accuracies": 0.893750011920929, | |
| "rewards/chosen": 0.929955780506134, | |
| "rewards/margins": 4.235381126403809, | |
| "rewards/rejected": -3.3054251670837402, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.2784, | |
| "grad_norm": 3.1434683799743652, | |
| "learning_rate": 3.2231302988414198e-06, | |
| "logits/chosen": -1.4585120677947998, | |
| "logits/rejected": -1.4169162511825562, | |
| "logps/chosen": -210.6068115234375, | |
| "logps/rejected": -273.0541687011719, | |
| "loss": 0.2233, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": 0.8234063982963562, | |
| "rewards/margins": 4.173564910888672, | |
| "rewards/rejected": -3.35015869140625, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.2944, | |
| "grad_norm": 1.7649414539337158, | |
| "learning_rate": 3.1817155712237347e-06, | |
| "logits/chosen": -1.4741700887680054, | |
| "logits/rejected": -1.3722097873687744, | |
| "logps/chosen": -226.2657928466797, | |
| "logps/rejected": -277.5989990234375, | |
| "loss": 0.221, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.9392329454421997, | |
| "rewards/margins": 3.8948147296905518, | |
| "rewards/rejected": -2.9555816650390625, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.3104, | |
| "grad_norm": 3.173729658126831, | |
| "learning_rate": 3.140097725143868e-06, | |
| "logits/chosen": -1.4934290647506714, | |
| "logits/rejected": -1.412099838256836, | |
| "logps/chosen": -197.9171905517578, | |
| "logps/rejected": -267.22479248046875, | |
| "loss": 0.1667, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 0.9158923029899597, | |
| "rewards/margins": 4.115560531616211, | |
| "rewards/rejected": -3.1996684074401855, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.3264, | |
| "grad_norm": 3.198397159576416, | |
| "learning_rate": 3.0982891607188948e-06, | |
| "logits/chosen": -1.490950345993042, | |
| "logits/rejected": -1.4556682109832764, | |
| "logps/chosen": -238.0470428466797, | |
| "logps/rejected": -283.5078430175781, | |
| "loss": 0.2348, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": 0.9115694165229797, | |
| "rewards/margins": 3.748671054840088, | |
| "rewards/rejected": -2.837101697921753, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.3424, | |
| "grad_norm": 2.510286331176758, | |
| "learning_rate": 3.056302334890786e-06, | |
| "logits/chosen": -1.4701006412506104, | |
| "logits/rejected": -1.441331386566162, | |
| "logps/chosen": -260.3006591796875, | |
| "logps/rejected": -261.59100341796875, | |
| "loss": 0.202, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 1.0139939785003662, | |
| "rewards/margins": 3.927891254425049, | |
| "rewards/rejected": -2.913897752761841, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.3584, | |
| "grad_norm": 3.634739637374878, | |
| "learning_rate": 3.0141497577148416e-06, | |
| "logits/chosen": -1.5961452722549438, | |
| "logits/rejected": -1.57537841796875, | |
| "logps/chosen": -205.42941284179688, | |
| "logps/rejected": -282.03997802734375, | |
| "loss": 0.2036, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 0.551750659942627, | |
| "rewards/margins": 4.165204048156738, | |
| "rewards/rejected": -3.6134536266326904, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.3744, | |
| "grad_norm": 3.698240041732788, | |
| "learning_rate": 2.971843988632292e-06, | |
| "logits/chosen": -1.4715049266815186, | |
| "logits/rejected": -1.3983992338180542, | |
| "logps/chosen": -231.0693359375, | |
| "logps/rejected": -257.81280517578125, | |
| "loss": 0.2249, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": 0.7772096991539001, | |
| "rewards/margins": 3.659933567047119, | |
| "rewards/rejected": -2.882723808288574, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.3904, | |
| "grad_norm": 3.1096959114074707, | |
| "learning_rate": 2.929397632728191e-06, | |
| "logits/chosen": -1.5016343593597412, | |
| "logits/rejected": -1.4842969179153442, | |
| "logps/chosen": -207.78262329101562, | |
| "logps/rejected": -275.2723388671875, | |
| "loss": 0.1894, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": 1.0016006231307983, | |
| "rewards/margins": 4.241255760192871, | |
| "rewards/rejected": -3.2396552562713623, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.4064, | |
| "grad_norm": 3.0866386890411377, | |
| "learning_rate": 2.886823336975703e-06, | |
| "logits/chosen": -1.5346386432647705, | |
| "logits/rejected": -1.437782645225525, | |
| "logps/chosen": -202.8625030517578, | |
| "logps/rejected": -287.2112731933594, | |
| "loss": 0.2577, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": 0.8549686670303345, | |
| "rewards/margins": 3.8706657886505127, | |
| "rewards/rejected": -3.0156967639923096, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.4224, | |
| "grad_norm": 4.283688545227051, | |
| "learning_rate": 2.8441337864679213e-06, | |
| "logits/chosen": -1.4911736249923706, | |
| "logits/rejected": -1.426809310913086, | |
| "logps/chosen": -198.9265899658203, | |
| "logps/rejected": -265.6153564453125, | |
| "loss": 0.1915, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": 1.185701608657837, | |
| "rewards/margins": 4.455019950866699, | |
| "rewards/rejected": -3.2693183422088623, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.4384000000000001, | |
| "grad_norm": 2.7477753162384033, | |
| "learning_rate": 2.8013417006383078e-06, | |
| "logits/chosen": -1.4503509998321533, | |
| "logits/rejected": -1.4662247896194458, | |
| "logps/chosen": -197.8849334716797, | |
| "logps/rejected": -275.83831787109375, | |
| "loss": 0.1803, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.975446879863739, | |
| "rewards/margins": 4.213322162628174, | |
| "rewards/rejected": -3.237874984741211, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.4544000000000001, | |
| "grad_norm": 2.467759132385254, | |
| "learning_rate": 2.758459829470925e-06, | |
| "logits/chosen": -1.5110101699829102, | |
| "logits/rejected": -1.5128719806671143, | |
| "logps/chosen": -222.1479949951172, | |
| "logps/rejected": -272.8663330078125, | |
| "loss": 0.1873, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.9241386651992798, | |
| "rewards/margins": 4.218895435333252, | |
| "rewards/rejected": -3.2947566509246826, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.4704, | |
| "grad_norm": 2.5419845581054688, | |
| "learning_rate": 2.7155009497015487e-06, | |
| "logits/chosen": -1.549300193786621, | |
| "logits/rejected": -1.4836896657943726, | |
| "logps/chosen": -195.63414001464844, | |
| "logps/rejected": -276.05230712890625, | |
| "loss": 0.1573, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": 1.0924745798110962, | |
| "rewards/margins": 4.838970184326172, | |
| "rewards/rejected": -3.7464957237243652, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.4864, | |
| "grad_norm": 2.3716790676116943, | |
| "learning_rate": 2.6724778610108123e-06, | |
| "logits/chosen": -1.5963730812072754, | |
| "logits/rejected": -1.545782446861267, | |
| "logps/chosen": -204.3538818359375, | |
| "logps/rejected": -261.65625, | |
| "loss": 0.1937, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 1.0204435586929321, | |
| "rewards/margins": 4.1371049880981445, | |
| "rewards/rejected": -3.116661787033081, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.5024, | |
| "grad_norm": 3.689908981323242, | |
| "learning_rate": 2.629403382210524e-06, | |
| "logits/chosen": -1.5731112957000732, | |
| "logits/rejected": -1.458322525024414, | |
| "logps/chosen": -189.004638671875, | |
| "logps/rejected": -253.52145385742188, | |
| "loss": 0.2311, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 0.8893483281135559, | |
| "rewards/margins": 3.971355438232422, | |
| "rewards/rejected": -3.08200740814209, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.5184, | |
| "grad_norm": 2.376858949661255, | |
| "learning_rate": 2.5862903474242707e-06, | |
| "logits/chosen": -1.4432743787765503, | |
| "logits/rejected": -1.369016408920288, | |
| "logps/chosen": -203.9170684814453, | |
| "logps/rejected": -268.00201416015625, | |
| "loss": 0.1701, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": 1.1208674907684326, | |
| "rewards/margins": 4.389878273010254, | |
| "rewards/rejected": -3.269010543823242, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.5344, | |
| "grad_norm": 3.032092332839966, | |
| "learning_rate": 2.5431516022634718e-06, | |
| "logits/chosen": -1.5671919584274292, | |
| "logits/rejected": -1.4809482097625732, | |
| "logps/chosen": -200.5327606201172, | |
| "logps/rejected": -268.4273986816406, | |
| "loss": 0.1361, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": 1.168104887008667, | |
| "rewards/margins": 4.882475852966309, | |
| "rewards/rejected": -3.7143707275390625, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.5504, | |
| "grad_norm": 3.788256883621216, | |
| "learning_rate": 2.5e-06, | |
| "logits/chosen": -1.5944483280181885, | |
| "logits/rejected": -1.4670864343643188, | |
| "logps/chosen": -216.2355194091797, | |
| "logps/rejected": -270.3284606933594, | |
| "loss": 0.1597, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 0.7749621272087097, | |
| "rewards/margins": 4.562745094299316, | |
| "rewards/rejected": -3.787782669067383, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.5664, | |
| "grad_norm": 2.4552183151245117, | |
| "learning_rate": 2.456848397736529e-06, | |
| "logits/chosen": -1.586390733718872, | |
| "logits/rejected": -1.5117604732513428, | |
| "logps/chosen": -226.70364379882812, | |
| "logps/rejected": -263.7825012207031, | |
| "loss": 0.221, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 0.6633533239364624, | |
| "rewards/margins": 4.1091156005859375, | |
| "rewards/rejected": -3.4457621574401855, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.5824, | |
| "grad_norm": 2.8182895183563232, | |
| "learning_rate": 2.4137096525757306e-06, | |
| "logits/chosen": -1.445420503616333, | |
| "logits/rejected": -1.5098474025726318, | |
| "logps/chosen": -211.97216796875, | |
| "logps/rejected": -244.12893676757812, | |
| "loss": 0.1896, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": 0.7570455074310303, | |
| "rewards/margins": 4.359705924987793, | |
| "rewards/rejected": -3.6026604175567627, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.5984, | |
| "grad_norm": 2.062126636505127, | |
| "learning_rate": 2.3705966177894763e-06, | |
| "logits/chosen": -1.5797353982925415, | |
| "logits/rejected": -1.5158863067626953, | |
| "logps/chosen": -191.18060302734375, | |
| "logps/rejected": -275.715576171875, | |
| "loss": 0.1436, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": 0.751899778842926, | |
| "rewards/margins": 4.745448112487793, | |
| "rewards/rejected": -3.9935481548309326, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.6143999999999998, | |
| "grad_norm": 2.1525497436523438, | |
| "learning_rate": 2.3275221389891877e-06, | |
| "logits/chosen": -1.4490973949432373, | |
| "logits/rejected": -1.4142483472824097, | |
| "logps/chosen": -223.93295288085938, | |
| "logps/rejected": -267.38995361328125, | |
| "loss": 0.1898, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 0.8560250401496887, | |
| "rewards/margins": 4.311391353607178, | |
| "rewards/rejected": -3.4553661346435547, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.6303999999999998, | |
| "grad_norm": 3.303349018096924, | |
| "learning_rate": 2.2844990502984517e-06, | |
| "logits/chosen": -1.5563932657241821, | |
| "logits/rejected": -1.5370930433273315, | |
| "logps/chosen": -207.7513885498047, | |
| "logps/rejected": -284.32928466796875, | |
| "loss": 0.2041, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 0.6124199628829956, | |
| "rewards/margins": 4.572120666503906, | |
| "rewards/rejected": -3.9597008228302, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.6463999999999999, | |
| "grad_norm": 3.3144888877868652, | |
| "learning_rate": 2.2415401705290753e-06, | |
| "logits/chosen": -1.5163837671279907, | |
| "logits/rejected": -1.477477788925171, | |
| "logps/chosen": -215.9023895263672, | |
| "logps/rejected": -288.0118713378906, | |
| "loss": 0.2097, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 0.6554079055786133, | |
| "rewards/margins": 4.482608795166016, | |
| "rewards/rejected": -3.8272006511688232, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.6623999999999999, | |
| "grad_norm": 3.2930357456207275, | |
| "learning_rate": 2.1986582993616926e-06, | |
| "logits/chosen": -1.4894039630889893, | |
| "logits/rejected": -1.4551079273223877, | |
| "logps/chosen": -201.579833984375, | |
| "logps/rejected": -288.08392333984375, | |
| "loss": 0.179, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": 0.6922101974487305, | |
| "rewards/margins": 4.913764953613281, | |
| "rewards/rejected": -4.221554756164551, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.6784, | |
| "grad_norm": 3.814143657684326, | |
| "learning_rate": 2.15586621353208e-06, | |
| "logits/chosen": -1.499711513519287, | |
| "logits/rejected": -1.4773409366607666, | |
| "logps/chosen": -211.90798950195312, | |
| "logps/rejected": -274.95892333984375, | |
| "loss": 0.1965, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 0.6713994741439819, | |
| "rewards/margins": 4.400740623474121, | |
| "rewards/rejected": -3.7293410301208496, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.6944, | |
| "grad_norm": 2.245388984680176, | |
| "learning_rate": 2.113176663024297e-06, | |
| "logits/chosen": -1.5954089164733887, | |
| "logits/rejected": -1.5338319540023804, | |
| "logps/chosen": -255.9158172607422, | |
| "logps/rejected": -277.632080078125, | |
| "loss": 0.2137, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.6438032388687134, | |
| "rewards/margins": 4.144532203674316, | |
| "rewards/rejected": -3.5007293224334717, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.7104, | |
| "grad_norm": 3.450652837753296, | |
| "learning_rate": 2.07060236727181e-06, | |
| "logits/chosen": -1.555140733718872, | |
| "logits/rejected": -1.5839719772338867, | |
| "logps/chosen": -215.4705810546875, | |
| "logps/rejected": -282.3954162597656, | |
| "loss": 0.1855, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.5197333097457886, | |
| "rewards/margins": 4.1995954513549805, | |
| "rewards/rejected": -3.6798622608184814, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.7264, | |
| "grad_norm": 3.3166344165802, | |
| "learning_rate": 2.0281560113677085e-06, | |
| "logits/chosen": -1.4750263690948486, | |
| "logits/rejected": -1.4056106805801392, | |
| "logps/chosen": -227.0777587890625, | |
| "logps/rejected": -278.08203125, | |
| "loss": 0.1951, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 0.8925558924674988, | |
| "rewards/margins": 4.614056587219238, | |
| "rewards/rejected": -3.7215003967285156, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.7424, | |
| "grad_norm": 2.740474224090576, | |
| "learning_rate": 1.9858502422851588e-06, | |
| "logits/chosen": -1.4764506816864014, | |
| "logits/rejected": -1.5012338161468506, | |
| "logps/chosen": -214.189453125, | |
| "logps/rejected": -290.9407653808594, | |
| "loss": 0.1765, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 0.7208338379859924, | |
| "rewards/margins": 4.320445537567139, | |
| "rewards/rejected": -3.599611759185791, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.7584, | |
| "grad_norm": 1.636268138885498, | |
| "learning_rate": 1.9436976651092143e-06, | |
| "logits/chosen": -1.4227992296218872, | |
| "logits/rejected": -1.5083410739898682, | |
| "logps/chosen": -190.32284545898438, | |
| "logps/rejected": -269.3763427734375, | |
| "loss": 0.1326, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 1.1259499788284302, | |
| "rewards/margins": 4.734012603759766, | |
| "rewards/rejected": -3.608062267303467, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.7744, | |
| "grad_norm": 3.211097240447998, | |
| "learning_rate": 1.9017108392811065e-06, | |
| "logits/chosen": -1.6095863580703735, | |
| "logits/rejected": -1.6652412414550781, | |
| "logps/chosen": -198.27940368652344, | |
| "logps/rejected": -273.74053955078125, | |
| "loss": 0.1919, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": 0.6028407216072083, | |
| "rewards/margins": 4.351764678955078, | |
| "rewards/rejected": -3.7489237785339355, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.7904, | |
| "grad_norm": 3.039146900177002, | |
| "learning_rate": 1.8599022748561324e-06, | |
| "logits/chosen": -1.6256183385849, | |
| "logits/rejected": -1.5564262866973877, | |
| "logps/chosen": -207.5953826904297, | |
| "logps/rejected": -292.1823425292969, | |
| "loss": 0.192, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": 0.6595524549484253, | |
| "rewards/margins": 4.907757759094238, | |
| "rewards/rejected": -4.248204708099365, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.8064, | |
| "grad_norm": 2.752476215362549, | |
| "learning_rate": 1.8182844287762657e-06, | |
| "logits/chosen": -1.540544867515564, | |
| "logits/rejected": -1.529164433479309, | |
| "logps/chosen": -222.51266479492188, | |
| "logps/rejected": -307.46258544921875, | |
| "loss": 0.1604, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 0.8695205450057983, | |
| "rewards/margins": 5.046850204467773, | |
| "rewards/rejected": -4.177329063415527, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.8224, | |
| "grad_norm": 2.99002742767334, | |
| "learning_rate": 1.776869701158581e-06, | |
| "logits/chosen": -1.5254290103912354, | |
| "logits/rejected": -1.4722168445587158, | |
| "logps/chosen": -221.71719360351562, | |
| "logps/rejected": -269.9870910644531, | |
| "loss": 0.1846, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.9478768110275269, | |
| "rewards/margins": 4.497414588928223, | |
| "rewards/rejected": -3.5495376586914062, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.8384, | |
| "grad_norm": 4.398141384124756, | |
| "learning_rate": 1.7356704316006185e-06, | |
| "logits/chosen": -1.501982569694519, | |
| "logits/rejected": -1.4923641681671143, | |
| "logps/chosen": -239.463623046875, | |
| "logps/rejected": -276.9585876464844, | |
| "loss": 0.2674, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 0.11826063692569733, | |
| "rewards/margins": 3.5606513023376465, | |
| "rewards/rejected": -3.4423904418945312, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.8544, | |
| "grad_norm": 3.2502195835113525, | |
| "learning_rate": 1.694698895503774e-06, | |
| "logits/chosen": -1.521781325340271, | |
| "logits/rejected": -1.445136308670044, | |
| "logps/chosen": -222.3773651123047, | |
| "logps/rejected": -268.7160339355469, | |
| "loss": 0.1946, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.7389503717422485, | |
| "rewards/margins": 4.303567886352539, | |
| "rewards/rejected": -3.56461763381958, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.8704, | |
| "grad_norm": 3.165996551513672, | |
| "learning_rate": 1.653967300415824e-06, | |
| "logits/chosen": -1.4852544069290161, | |
| "logits/rejected": -1.4845850467681885, | |
| "logps/chosen": -214.6300811767578, | |
| "logps/rejected": -295.7364501953125, | |
| "loss": 0.1623, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": 0.7326163053512573, | |
| "rewards/margins": 5.156301975250244, | |
| "rewards/rejected": -4.4236860275268555, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.8864, | |
| "grad_norm": 4.204633712768555, | |
| "learning_rate": 1.613487782393661e-06, | |
| "logits/chosen": -1.598530650138855, | |
| "logits/rejected": -1.5577715635299683, | |
| "logps/chosen": -200.7525634765625, | |
| "logps/rejected": -280.6080017089844, | |
| "loss": 0.2124, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 0.6972262263298035, | |
| "rewards/margins": 4.553775787353516, | |
| "rewards/rejected": -3.8565497398376465, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.9024, | |
| "grad_norm": 2.8563499450683594, | |
| "learning_rate": 1.5732724023873313e-06, | |
| "logits/chosen": -1.4955016374588013, | |
| "logits/rejected": -1.526990294456482, | |
| "logps/chosen": -195.61587524414062, | |
| "logps/rejected": -277.1825256347656, | |
| "loss": 0.1821, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": 0.7116910219192505, | |
| "rewards/margins": 4.586084842681885, | |
| "rewards/rejected": -3.874393939971924, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.9184, | |
| "grad_norm": 2.4262192249298096, | |
| "learning_rate": 1.5333331426464532e-06, | |
| "logits/chosen": -1.5559725761413574, | |
| "logits/rejected": -1.4919229745864868, | |
| "logps/chosen": -226.164306640625, | |
| "logps/rejected": -290.63018798828125, | |
| "loss": 0.1933, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.4328826069831848, | |
| "rewards/margins": 4.367493629455566, | |
| "rewards/rejected": -3.9346108436584473, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.9344000000000001, | |
| "grad_norm": 4.227684497833252, | |
| "learning_rate": 1.4936819031500816e-06, | |
| "logits/chosen": -1.45127272605896, | |
| "logits/rejected": -1.4866652488708496, | |
| "logps/chosen": -222.46701049804688, | |
| "logps/rejected": -293.39520263671875, | |
| "loss": 0.1848, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.8863641619682312, | |
| "rewards/margins": 4.865546226501465, | |
| "rewards/rejected": -3.979182004928589, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.9504000000000001, | |
| "grad_norm": 3.343823194503784, | |
| "learning_rate": 1.4543304980610879e-06, | |
| "logits/chosen": -1.5442382097244263, | |
| "logits/rejected": -1.5331324338912964, | |
| "logps/chosen": -207.97799682617188, | |
| "logps/rejected": -280.4158935546875, | |
| "loss": 0.2311, | |
| "rewards/accuracies": 0.893750011920929, | |
| "rewards/chosen": 0.5620196461677551, | |
| "rewards/margins": 4.68914270401001, | |
| "rewards/rejected": -4.1271233558654785, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.9664000000000001, | |
| "grad_norm": 2.6413931846618652, | |
| "learning_rate": 1.415290652206105e-06, | |
| "logits/chosen": -1.5758134126663208, | |
| "logits/rejected": -1.5755757093429565, | |
| "logps/chosen": -194.215576171875, | |
| "logps/rejected": -279.52703857421875, | |
| "loss": 0.191, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": 0.5947729349136353, | |
| "rewards/margins": 4.675285339355469, | |
| "rewards/rejected": -4.080512046813965, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.9824000000000002, | |
| "grad_norm": 4.83159065246582, | |
| "learning_rate": 1.3765739975820964e-06, | |
| "logits/chosen": -1.5410311222076416, | |
| "logits/rejected": -1.46727454662323, | |
| "logps/chosen": -205.7766876220703, | |
| "logps/rejected": -287.05706787109375, | |
| "loss": 0.1981, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 0.6785244345664978, | |
| "rewards/margins": 4.577788829803467, | |
| "rewards/rejected": -3.8992645740509033, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.9984, | |
| "grad_norm": 3.0916264057159424, | |
| "learning_rate": 1.3381920698905788e-06, | |
| "logits/chosen": -1.5568665266036987, | |
| "logits/rejected": -1.487683892250061, | |
| "logps/chosen": -197.45993041992188, | |
| "logps/rejected": -270.3889465332031, | |
| "loss": 0.1795, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 0.41252049803733826, | |
| "rewards/margins": 4.434720039367676, | |
| "rewards/rejected": -4.0221991539001465, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.0128, | |
| "grad_norm": 3.156958818435669, | |
| "learning_rate": 1.3001563051005348e-06, | |
| "logits/chosen": -1.5213212966918945, | |
| "logits/rejected": -1.4491430521011353, | |
| "logps/chosen": -214.64614868164062, | |
| "logps/rejected": -271.3830871582031, | |
| "loss": 0.1604, | |
| "rewards/accuracies": 0.9444444179534912, | |
| "rewards/chosen": 0.5151737928390503, | |
| "rewards/margins": 4.692739963531494, | |
| "rewards/rejected": -4.177565574645996, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 2.0288, | |
| "grad_norm": 3.899384021759033, | |
| "learning_rate": 1.2624780360410465e-06, | |
| "logits/chosen": -1.5627011060714722, | |
| "logits/rejected": -1.5318704843521118, | |
| "logps/chosen": -258.4732971191406, | |
| "logps/rejected": -268.80194091796875, | |
| "loss": 0.2228, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": 0.5114110112190247, | |
| "rewards/margins": 4.497435569763184, | |
| "rewards/rejected": -3.9860243797302246, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 2.0448, | |
| "grad_norm": 1.9564834833145142, | |
| "learning_rate": 1.225168489024661e-06, | |
| "logits/chosen": -1.570343255996704, | |
| "logits/rejected": -1.539205551147461, | |
| "logps/chosen": -221.6617889404297, | |
| "logps/rejected": -303.56744384765625, | |
| "loss": 0.1121, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.5438711047172546, | |
| "rewards/margins": 5.046863079071045, | |
| "rewards/rejected": -4.502991676330566, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 2.0608, | |
| "grad_norm": 3.1204640865325928, | |
| "learning_rate": 1.18823878050248e-06, | |
| "logits/chosen": -1.4727885723114014, | |
| "logits/rejected": -1.416222333908081, | |
| "logps/chosen": -217.26416015625, | |
| "logps/rejected": -290.2035827636719, | |
| "loss": 0.1569, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 0.9948328733444214, | |
| "rewards/margins": 5.736484527587891, | |
| "rewards/rejected": -4.741652011871338, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 2.0768, | |
| "grad_norm": 3.6491470336914062, | |
| "learning_rate": 1.1516999137520023e-06, | |
| "logits/chosen": -1.4982631206512451, | |
| "logits/rejected": -1.4064592123031616, | |
| "logps/chosen": -214.3899383544922, | |
| "logps/rejected": -274.42340087890625, | |
| "loss": 0.164, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.9111789464950562, | |
| "rewards/margins": 4.822937965393066, | |
| "rewards/rejected": -3.9117591381073, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.0928, | |
| "grad_norm": 2.44218373298645, | |
| "learning_rate": 1.1155627755986673e-06, | |
| "logits/chosen": -1.4995273351669312, | |
| "logits/rejected": -1.399436354637146, | |
| "logps/chosen": -200.63796997070312, | |
| "logps/rejected": -254.8256072998047, | |
| "loss": 0.1737, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": 0.758064866065979, | |
| "rewards/margins": 4.460127353668213, | |
| "rewards/rejected": -3.7020626068115234, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 2.1088, | |
| "grad_norm": 2.5277323722839355, | |
| "learning_rate": 1.079838133172111e-06, | |
| "logits/chosen": -1.5891015529632568, | |
| "logits/rejected": -1.5126923322677612, | |
| "logps/chosen": -214.89492797851562, | |
| "logps/rejected": -271.93463134765625, | |
| "loss": 0.1561, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 0.9002867937088013, | |
| "rewards/margins": 4.9788923263549805, | |
| "rewards/rejected": -4.0786051750183105, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.1248, | |
| "grad_norm": 3.2694408893585205, | |
| "learning_rate": 1.044536630698074e-06, | |
| "logits/chosen": -1.4245771169662476, | |
| "logits/rejected": -1.4857099056243896, | |
| "logps/chosen": -227.65475463867188, | |
| "logps/rejected": -280.5589904785156, | |
| "loss": 0.1571, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": 0.8741976618766785, | |
| "rewards/margins": 4.782471656799316, | |
| "rewards/rejected": -3.9082744121551514, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 2.1408, | |
| "grad_norm": 3.1822643280029297, | |
| "learning_rate": 1.0096687863269368e-06, | |
| "logits/chosen": -1.6040668487548828, | |
| "logits/rejected": -1.5315070152282715, | |
| "logps/chosen": -216.56002807617188, | |
| "logps/rejected": -272.2256774902344, | |
| "loss": 0.1276, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 1.028387188911438, | |
| "rewards/margins": 5.059295654296875, | |
| "rewards/rejected": -4.030908584594727, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.1568, | |
| "grad_norm": 1.985778570175171, | |
| "learning_rate": 9.752449889998177e-07, | |
| "logits/chosen": -1.5215928554534912, | |
| "logits/rejected": -1.4317477941513062, | |
| "logps/chosen": -220.37283325195312, | |
| "logps/rejected": -282.06463623046875, | |
| "loss": 0.1527, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 0.8296896815299988, | |
| "rewards/margins": 4.47981595993042, | |
| "rewards/rejected": -3.6501266956329346, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.1728, | |
| "grad_norm": 2.4375979900360107, | |
| "learning_rate": 9.412754953531664e-07, | |
| "logits/chosen": -1.5252634286880493, | |
| "logits/rejected": -1.471828818321228, | |
| "logps/chosen": -214.5895538330078, | |
| "logps/rejected": -291.8826904296875, | |
| "loss": 0.126, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": 0.6852453947067261, | |
| "rewards/margins": 5.221624851226807, | |
| "rewards/rejected": -4.536379814147949, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 2.1888, | |
| "grad_norm": 2.8848726749420166, | |
| "learning_rate": 9.077704266627777e-07, | |
| "logits/chosen": -1.5290806293487549, | |
| "logits/rejected": -1.5034786462783813, | |
| "logps/chosen": -207.0309600830078, | |
| "logps/rejected": -277.1182556152344, | |
| "loss": 0.1651, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 0.6543574333190918, | |
| "rewards/margins": 4.433718681335449, | |
| "rewards/rejected": -3.7793612480163574, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 2.2048, | |
| "grad_norm": 2.8718600273132324, | |
| "learning_rate": 8.747397658281396e-07, | |
| "logits/chosen": -1.5480024814605713, | |
| "logits/rejected": -1.5038466453552246, | |
| "logps/chosen": -229.7807159423828, | |
| "logps/rejected": -297.35284423828125, | |
| "loss": 0.1557, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.940424919128418, | |
| "rewards/margins": 5.296424388885498, | |
| "rewards/rejected": -4.355999946594238, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.2208, | |
| "grad_norm": 1.3786566257476807, | |
| "learning_rate": 8.421933543980126e-07, | |
| "logits/chosen": -1.4576784372329712, | |
| "logits/rejected": -1.5554091930389404, | |
| "logps/chosen": -215.29415893554688, | |
| "logps/rejected": -277.7773132324219, | |
| "loss": 0.1193, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.7863073348999023, | |
| "rewards/margins": 5.157085418701172, | |
| "rewards/rejected": -4.3707780838012695, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 2.2368, | |
| "grad_norm": 4.430466651916504, | |
| "learning_rate": 8.101408896381141e-07, | |
| "logits/chosen": -1.5009676218032837, | |
| "logits/rejected": -1.4523426294326782, | |
| "logps/chosen": -229.9296875, | |
| "logps/rejected": -281.8008728027344, | |
| "loss": 0.1953, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 0.5093691349029541, | |
| "rewards/margins": 5.013900279998779, | |
| "rewards/rejected": -4.504530906677246, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.2528, | |
| "grad_norm": 3.049445152282715, | |
| "learning_rate": 7.78591921641807e-07, | |
| "logits/chosen": -1.5671838521957397, | |
| "logits/rejected": -1.508034348487854, | |
| "logps/chosen": -223.31234741210938, | |
| "logps/rejected": -278.34185791015625, | |
| "loss": 0.1751, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": 0.33358827233314514, | |
| "rewards/margins": 4.723074436187744, | |
| "rewards/rejected": -4.389485836029053, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 2.2688, | |
| "grad_norm": 2.8345718383789062, | |
| "learning_rate": 7.475558504846264e-07, | |
| "logits/chosen": -1.5353517532348633, | |
| "logits/rejected": -1.5015804767608643, | |
| "logps/chosen": -213.0556182861328, | |
| "logps/rejected": -278.02197265625, | |
| "loss": 0.1505, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": 0.681841254234314, | |
| "rewards/margins": 5.173250675201416, | |
| "rewards/rejected": -4.4914093017578125, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.2848, | |
| "grad_norm": 3.8184285163879395, | |
| "learning_rate": 7.170419234235099e-07, | |
| "logits/chosen": -1.5723527669906616, | |
| "logits/rejected": -1.5454447269439697, | |
| "logps/chosen": -220.0974578857422, | |
| "logps/rejected": -285.02471923828125, | |
| "loss": 0.1645, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 0.32882851362228394, | |
| "rewards/margins": 4.826219081878662, | |
| "rewards/rejected": -4.4973907470703125, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 2.3008, | |
| "grad_norm": 1.8742855787277222, | |
| "learning_rate": 6.870592321415595e-07, | |
| "logits/chosen": -1.5036770105361938, | |
| "logits/rejected": -1.5060780048370361, | |
| "logps/chosen": -258.08514404296875, | |
| "logps/rejected": -288.4046630859375, | |
| "loss": 0.1302, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.4841408133506775, | |
| "rewards/margins": 4.661401271820068, | |
| "rewards/rejected": -4.177260398864746, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.3168, | |
| "grad_norm": 2.858125686645508, | |
| "learning_rate": 6.576167100391583e-07, | |
| "logits/chosen": -1.5316962003707886, | |
| "logits/rejected": -1.5005123615264893, | |
| "logps/chosen": -200.00645446777344, | |
| "logps/rejected": -274.73797607421875, | |
| "loss": 0.1702, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 0.2455310821533203, | |
| "rewards/margins": 4.6962361335754395, | |
| "rewards/rejected": -4.450705051422119, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 2.3327999999999998, | |
| "grad_norm": 2.4755117893218994, | |
| "learning_rate": 6.28723129572247e-07, | |
| "logits/chosen": -1.4955435991287231, | |
| "logits/rejected": -1.450901985168457, | |
| "logps/chosen": -215.6560821533203, | |
| "logps/rejected": -282.6337890625, | |
| "loss": 0.1465, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": 0.7359136343002319, | |
| "rewards/margins": 5.175809860229492, | |
| "rewards/rejected": -4.4398956298828125, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.3487999999999998, | |
| "grad_norm": 2.082693338394165, | |
| "learning_rate": 6.003870996385533e-07, | |
| "logits/chosen": -1.5163285732269287, | |
| "logits/rejected": -1.4916880130767822, | |
| "logps/chosen": -223.2830810546875, | |
| "logps/rejected": -304.0013122558594, | |
| "loss": 0.1236, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.4678799510002136, | |
| "rewards/margins": 5.262962341308594, | |
| "rewards/rejected": -4.7950825691223145, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 2.3648, | |
| "grad_norm": 3.079440116882324, | |
| "learning_rate": 5.72617063012551e-07, | |
| "logits/chosen": -1.4600715637207031, | |
| "logits/rejected": -1.4372762441635132, | |
| "logps/chosen": -200.4602813720703, | |
| "logps/rejected": -298.87261962890625, | |
| "loss": 0.1165, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 0.6318634748458862, | |
| "rewards/margins": 5.3328537940979, | |
| "rewards/rejected": -4.700989723205566, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 2.3808, | |
| "grad_norm": 2.735065221786499, | |
| "learning_rate": 5.454212938299256e-07, | |
| "logits/chosen": -1.5163583755493164, | |
| "logits/rejected": -1.4097801446914673, | |
| "logps/chosen": -222.529052734375, | |
| "logps/rejected": -293.6036682128906, | |
| "loss": 0.1375, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.6007111072540283, | |
| "rewards/margins": 5.02976655960083, | |
| "rewards/rejected": -4.429055213928223, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 2.3968, | |
| "grad_norm": 3.09622859954834, | |
| "learning_rate": 5.188078951222745e-07, | |
| "logits/chosen": -1.5613389015197754, | |
| "logits/rejected": -1.421422004699707, | |
| "logps/chosen": -225.4494171142578, | |
| "logps/rejected": -287.4031982421875, | |
| "loss": 0.1347, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": 0.6756893992424011, | |
| "rewards/margins": 5.113866329193115, | |
| "rewards/rejected": -4.438177585601807, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.4128, | |
| "grad_norm": 4.569268703460693, | |
| "learning_rate": 4.92784796402794e-07, | |
| "logits/chosen": -1.5446864366531372, | |
| "logits/rejected": -1.5669465065002441, | |
| "logps/chosen": -213.2777862548828, | |
| "logps/rejected": -303.1000671386719, | |
| "loss": 0.1699, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": 0.12444484233856201, | |
| "rewards/margins": 4.8673295974731445, | |
| "rewards/rejected": -4.742884635925293, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 2.4288, | |
| "grad_norm": 1.9897822141647339, | |
| "learning_rate": 4.673597513036684e-07, | |
| "logits/chosen": -1.5755380392074585, | |
| "logits/rejected": -1.5424786806106567, | |
| "logps/chosen": -217.6551971435547, | |
| "logps/rejected": -267.38604736328125, | |
| "loss": 0.1802, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": 0.19554877281188965, | |
| "rewards/margins": 4.060165882110596, | |
| "rewards/rejected": -3.864617109298706, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.4448, | |
| "grad_norm": 1.4262510538101196, | |
| "learning_rate": 4.4254033526585917e-07, | |
| "logits/chosen": -1.4375520944595337, | |
| "logits/rejected": -1.4742629528045654, | |
| "logps/chosen": -217.225341796875, | |
| "logps/rejected": -281.43182373046875, | |
| "loss": 0.1281, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": 0.7436832189559937, | |
| "rewards/margins": 5.335335731506348, | |
| "rewards/rejected": -4.591652870178223, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 2.4608, | |
| "grad_norm": 3.0517566204071045, | |
| "learning_rate": 4.183339432819844e-07, | |
| "logits/chosen": -1.5090440511703491, | |
| "logits/rejected": -1.4726752042770386, | |
| "logps/chosen": -222.68173217773438, | |
| "logps/rejected": -276.00872802734375, | |
| "loss": 0.1476, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": 0.494690477848053, | |
| "rewards/margins": 4.942327976226807, | |
| "rewards/rejected": -4.447638034820557, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 2.4768, | |
| "grad_norm": 2.9095730781555176, | |
| "learning_rate": 3.947477876929706e-07, | |
| "logits/chosen": -1.5281938314437866, | |
| "logits/rejected": -1.4848145246505737, | |
| "logps/chosen": -214.83279418945312, | |
| "logps/rejected": -294.42193603515625, | |
| "loss": 0.1613, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 0.24510130286216736, | |
| "rewards/margins": 5.043450355529785, | |
| "rewards/rejected": -4.798348903656006, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 2.4928, | |
| "grad_norm": 1.4994947910308838, | |
| "learning_rate": 3.717888960391222e-07, | |
| "logits/chosen": -1.5372960567474365, | |
| "logits/rejected": -1.5170810222625732, | |
| "logps/chosen": -230.2967987060547, | |
| "logps/rejected": -311.89129638671875, | |
| "loss": 0.1262, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": 0.27188095450401306, | |
| "rewards/margins": 5.6029887199401855, | |
| "rewards/rejected": -5.331108570098877, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.5088, | |
| "grad_norm": 3.1163201332092285, | |
| "learning_rate": 3.4946410896624816e-07, | |
| "logits/chosen": -1.5684281587600708, | |
| "logits/rejected": -1.5009348392486572, | |
| "logps/chosen": -204.79660034179688, | |
| "logps/rejected": -296.203857421875, | |
| "loss": 0.1305, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 0.44786787033081055, | |
| "rewards/margins": 5.553648948669434, | |
| "rewards/rejected": -5.105780601501465, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 2.5248, | |
| "grad_norm": 4.7359795570373535, | |
| "learning_rate": 3.2778007818748376e-07, | |
| "logits/chosen": -1.495949149131775, | |
| "logits/rejected": -1.5144822597503662, | |
| "logps/chosen": -235.7005615234375, | |
| "logps/rejected": -299.01824951171875, | |
| "loss": 0.2294, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": 0.26778358221054077, | |
| "rewards/margins": 4.562989234924316, | |
| "rewards/rejected": -4.295205593109131, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 2.5408, | |
| "grad_norm": 3.072680711746216, | |
| "learning_rate": 3.0674326450139516e-07, | |
| "logits/chosen": -1.386596918106079, | |
| "logits/rejected": -1.4416801929473877, | |
| "logps/chosen": -222.2554168701172, | |
| "logps/rejected": -300.201171875, | |
| "loss": 0.1592, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": 0.29924696683883667, | |
| "rewards/margins": 5.083845615386963, | |
| "rewards/rejected": -4.784598350524902, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 2.5568, | |
| "grad_norm": 2.27390193939209, | |
| "learning_rate": 2.8635993586697555e-07, | |
| "logits/chosen": -1.5572984218597412, | |
| "logits/rejected": -1.4875266551971436, | |
| "logps/chosen": -209.93283081054688, | |
| "logps/rejected": -285.22491455078125, | |
| "loss": 0.1291, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": 0.31827136874198914, | |
| "rewards/margins": 5.4527387619018555, | |
| "rewards/rejected": -5.134467124938965, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.5728, | |
| "grad_norm": 3.9169278144836426, | |
| "learning_rate": 2.6663616553609273e-07, | |
| "logits/chosen": -1.5840399265289307, | |
| "logits/rejected": -1.533908486366272, | |
| "logps/chosen": -218.97805786132812, | |
| "logps/rejected": -276.19989013671875, | |
| "loss": 0.169, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 0.1213858500123024, | |
| "rewards/margins": 4.739739418029785, | |
| "rewards/rejected": -4.618352890014648, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 2.5888, | |
| "grad_norm": 2.456167697906494, | |
| "learning_rate": 2.4757783024395244e-07, | |
| "logits/chosen": -1.463147521018982, | |
| "logits/rejected": -1.3830325603485107, | |
| "logps/chosen": -219.8505401611328, | |
| "logps/rejected": -282.1579284667969, | |
| "loss": 0.1611, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": 0.33140283823013306, | |
| "rewards/margins": 4.928020477294922, | |
| "rewards/rejected": -4.596617698669434, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.6048, | |
| "grad_norm": 1.365941047668457, | |
| "learning_rate": 2.2919060845811498e-07, | |
| "logits/chosen": -1.585924744606018, | |
| "logits/rejected": -1.5624620914459229, | |
| "logps/chosen": -207.739013671875, | |
| "logps/rejected": -302.1373291015625, | |
| "loss": 0.1271, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 0.21847054362297058, | |
| "rewards/margins": 5.681158542633057, | |
| "rewards/rejected": -5.4626874923706055, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 2.6208, | |
| "grad_norm": 3.1357429027557373, | |
| "learning_rate": 2.1147997868658427e-07, | |
| "logits/chosen": -1.5514700412750244, | |
| "logits/rejected": -1.6299651861190796, | |
| "logps/chosen": -210.6826934814453, | |
| "logps/rejected": -283.8253173828125, | |
| "loss": 0.174, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": 0.06875920295715332, | |
| "rewards/margins": 5.225507736206055, | |
| "rewards/rejected": -5.1567487716674805, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.6368, | |
| "grad_norm": 2.4194045066833496, | |
| "learning_rate": 1.9445121784547437e-07, | |
| "logits/chosen": -1.5748323202133179, | |
| "logits/rejected": -1.5742114782333374, | |
| "logps/chosen": -209.5052947998047, | |
| "logps/rejected": -294.8690185546875, | |
| "loss": 0.1372, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 0.09001219272613525, | |
| "rewards/margins": 4.978179454803467, | |
| "rewards/rejected": -4.888167858123779, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 2.6528, | |
| "grad_norm": 1.028726577758789, | |
| "learning_rate": 1.7810939968674418e-07, | |
| "logits/chosen": -1.6239761114120483, | |
| "logits/rejected": -1.5699875354766846, | |
| "logps/chosen": -211.88369750976562, | |
| "logps/rejected": -284.1302795410156, | |
| "loss": 0.0973, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": 0.3777560591697693, | |
| "rewards/margins": 5.902177333831787, | |
| "rewards/rejected": -5.524421691894531, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 2.6688, | |
| "grad_norm": 2.883683681488037, | |
| "learning_rate": 1.6245939328646322e-07, | |
| "logits/chosen": -1.4824936389923096, | |
| "logits/rejected": -1.535343050956726, | |
| "logps/chosen": -221.0633087158203, | |
| "logps/rejected": -297.4151611328125, | |
| "loss": 0.1735, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.1900520920753479, | |
| "rewards/margins": 5.30252742767334, | |
| "rewards/rejected": -5.112476348876953, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 2.6848, | |
| "grad_norm": 3.589721918106079, | |
| "learning_rate": 1.4750586159405917e-07, | |
| "logits/chosen": -1.5421957969665527, | |
| "logits/rejected": -1.5106103420257568, | |
| "logps/chosen": -210.3826904296875, | |
| "logps/rejected": -274.41046142578125, | |
| "loss": 0.2073, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.28717076778411865, | |
| "rewards/margins": 4.679255485534668, | |
| "rewards/rejected": -4.392085075378418, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.7008, | |
| "grad_norm": 3.69374418258667, | |
| "learning_rate": 1.3325326004298533e-07, | |
| "logits/chosen": -1.558068037033081, | |
| "logits/rejected": -1.5273410081863403, | |
| "logps/chosen": -231.698974609375, | |
| "logps/rejected": -311.16204833984375, | |
| "loss": 0.2213, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": -0.017372488975524902, | |
| "rewards/margins": 4.810871124267578, | |
| "rewards/rejected": -4.828243732452393, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 2.7168, | |
| "grad_norm": 3.5620641708374023, | |
| "learning_rate": 1.197058352232147e-07, | |
| "logits/chosen": -1.561061143875122, | |
| "logits/rejected": -1.5442357063293457, | |
| "logps/chosen": -220.67630004882812, | |
| "logps/rejected": -293.6253356933594, | |
| "loss": 0.1862, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 0.19432000815868378, | |
| "rewards/margins": 5.36147928237915, | |
| "rewards/rejected": -5.167159080505371, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.7328, | |
| "grad_norm": 1.8132295608520508, | |
| "learning_rate": 1.0686762361596075e-07, | |
| "logits/chosen": -1.6041780710220337, | |
| "logits/rejected": -1.501832365989685, | |
| "logps/chosen": -213.18447875976562, | |
| "logps/rejected": -277.6732177734375, | |
| "loss": 0.136, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": 0.015193851664662361, | |
| "rewards/margins": 4.558278560638428, | |
| "rewards/rejected": -4.54308557510376, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 2.7488, | |
| "grad_norm": 4.747691631317139, | |
| "learning_rate": 9.474245039099883e-08, | |
| "logits/chosen": -1.4521749019622803, | |
| "logits/rejected": -1.3812751770019531, | |
| "logps/chosen": -219.65438842773438, | |
| "logps/rejected": -263.45977783203125, | |
| "loss": 0.1762, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.5863351821899414, | |
| "rewards/margins": 4.733705997467041, | |
| "rewards/rejected": -4.147371292114258, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 2.7648, | |
| "grad_norm": 2.4241857528686523, | |
| "learning_rate": 8.333392826695125e-08, | |
| "logits/chosen": -1.6109880208969116, | |
| "logits/rejected": -1.5573093891143799, | |
| "logps/chosen": -207.94308471679688, | |
| "logps/rejected": -267.8423767089844, | |
| "loss": 0.1597, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": 0.2518244683742523, | |
| "rewards/margins": 4.8867506980896, | |
| "rewards/rejected": -4.6349263191223145, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 2.7808, | |
| "grad_norm": 3.291642189025879, | |
| "learning_rate": 7.264545643486997e-08, | |
| "logits/chosen": -1.5958778858184814, | |
| "logits/rejected": -1.5443265438079834, | |
| "logps/chosen": -229.17300415039062, | |
| "logps/rejected": -307.0359802246094, | |
| "loss": 0.1593, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 0.4986957013607025, | |
| "rewards/margins": 5.3819475173950195, | |
| "rewards/rejected": -4.883251667022705, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 2.7968, | |
| "grad_norm": 3.9328432083129883, | |
| "learning_rate": 6.268021954544095e-08, | |
| "logits/chosen": -1.5294109582901, | |
| "logits/rejected": -1.5232346057891846, | |
| "logps/chosen": -215.351318359375, | |
| "logps/rejected": -281.74481201171875, | |
| "loss": 0.1916, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": 0.16991707682609558, | |
| "rewards/margins": 4.9698686599731445, | |
| "rewards/rejected": -4.799951076507568, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 2.8128, | |
| "grad_norm": 1.6685928106307983, | |
| "learning_rate": 5.344118676011173e-08, | |
| "logits/chosen": -1.517317295074463, | |
| "logits/rejected": -1.5692014694213867, | |
| "logps/chosen": -202.7231903076172, | |
| "logps/rejected": -289.62152099609375, | |
| "loss": 0.1346, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": 0.6345641016960144, | |
| "rewards/margins": 5.580275058746338, | |
| "rewards/rejected": -4.945711135864258, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 2.8288, | |
| "grad_norm": 2.229893207550049, | |
| "learning_rate": 4.493111086642438e-08, | |
| "logits/chosen": -1.568162202835083, | |
| "logits/rejected": -1.5379860401153564, | |
| "logps/chosen": -197.50967407226562, | |
| "logps/rejected": -292.4415283203125, | |
| "loss": 0.1472, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 0.33922216296195984, | |
| "rewards/margins": 5.462419509887695, | |
| "rewards/rejected": -5.123197555541992, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 2.8448, | |
| "grad_norm": 2.519378900527954, | |
| "learning_rate": 3.7152527457815504e-08, | |
| "logits/chosen": -1.5041329860687256, | |
| "logits/rejected": -1.5115489959716797, | |
| "logps/chosen": -207.9475555419922, | |
| "logps/rejected": -274.80438232421875, | |
| "loss": 0.1982, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": 0.23130455613136292, | |
| "rewards/margins": 4.7877349853515625, | |
| "rewards/rejected": -4.556429862976074, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 2.8608000000000002, | |
| "grad_norm": 3.1473171710968018, | |
| "learning_rate": 3.010775417813388e-08, | |
| "logits/chosen": -1.522871971130371, | |
| "logits/rejected": -1.5445659160614014, | |
| "logps/chosen": -196.432861328125, | |
| "logps/rejected": -287.74359130859375, | |
| "loss": 0.2051, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -0.02917574718594551, | |
| "rewards/margins": 4.636673927307129, | |
| "rewards/rejected": -4.665849685668945, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 2.8768000000000002, | |
| "grad_norm": 3.1271698474884033, | |
| "learning_rate": 2.3798890031092037e-08, | |
| "logits/chosen": -1.5359059572219849, | |
| "logits/rejected": -1.4754010438919067, | |
| "logps/chosen": -223.46484375, | |
| "logps/rejected": -294.39752197265625, | |
| "loss": 0.1698, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": 0.7118966579437256, | |
| "rewards/margins": 5.0428667068481445, | |
| "rewards/rejected": -4.33096981048584, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.8928000000000003, | |
| "grad_norm": 1.7171223163604736, | |
| "learning_rate": 1.822781475486507e-08, | |
| "logits/chosen": -1.5836378335952759, | |
| "logits/rejected": -1.5234249830245972, | |
| "logps/chosen": -209.51748657226562, | |
| "logps/rejected": -297.83343505859375, | |
| "loss": 0.1182, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.35548344254493713, | |
| "rewards/margins": 5.583395481109619, | |
| "rewards/rejected": -5.227911949157715, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 2.9088000000000003, | |
| "grad_norm": 2.023606538772583, | |
| "learning_rate": 1.3396188262018439e-08, | |
| "logits/chosen": -1.486228346824646, | |
| "logits/rejected": -1.506572961807251, | |
| "logps/chosen": -211.44338989257812, | |
| "logps/rejected": -291.0889892578125, | |
| "loss": 0.1161, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": 0.8655111193656921, | |
| "rewards/margins": 5.486937522888184, | |
| "rewards/rejected": -4.621427059173584, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.9248, | |
| "grad_norm": 2.3930633068084717, | |
| "learning_rate": 9.305450144933836e-09, | |
| "logits/chosen": -1.4977796077728271, | |
| "logits/rejected": -1.5438463687896729, | |
| "logps/chosen": -212.6026611328125, | |
| "logps/rejected": -279.10528564453125, | |
| "loss": 0.1812, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 0.12152069807052612, | |
| "rewards/margins": 4.457780838012695, | |
| "rewards/rejected": -4.336259841918945, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 2.9408, | |
| "grad_norm": 2.2170960903167725, | |
| "learning_rate": 5.9568192468811844e-09, | |
| "logits/chosen": -1.6552884578704834, | |
| "logits/rejected": -1.611583948135376, | |
| "logps/chosen": -212.38027954101562, | |
| "logps/rejected": -305.21142578125, | |
| "loss": 0.1603, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": 0.3693106472492218, | |
| "rewards/margins": 5.326704978942871, | |
| "rewards/rejected": -4.957394599914551, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.9568, | |
| "grad_norm": 3.3993589878082275, | |
| "learning_rate": 3.3512932988596924e-09, | |
| "logits/chosen": -1.5539171695709229, | |
| "logits/rejected": -1.5044705867767334, | |
| "logps/chosen": -212.821044921875, | |
| "logps/rejected": -275.41632080078125, | |
| "loss": 0.1584, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": 0.17114581167697906, | |
| "rewards/margins": 4.8249664306640625, | |
| "rewards/rejected": -4.653820991516113, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.9728, | |
| "grad_norm": 2.7821054458618164, | |
| "learning_rate": 1.4896486223239802e-09, | |
| "logits/chosen": -1.6075645685195923, | |
| "logits/rejected": -1.5205880403518677, | |
| "logps/chosen": -211.8487091064453, | |
| "logps/rejected": -275.73980712890625, | |
| "loss": 0.1672, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": 0.3071829676628113, | |
| "rewards/margins": 4.825101375579834, | |
| "rewards/rejected": -4.517918586730957, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.9888, | |
| "grad_norm": 3.1003592014312744, | |
| "learning_rate": 3.7243989787633106e-10, | |
| "logits/chosen": -1.535521149635315, | |
| "logits/rejected": -1.5245225429534912, | |
| "logps/chosen": -230.11026000976562, | |
| "logps/rejected": -291.1710510253906, | |
| "loss": 0.1983, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": 0.3992445170879364, | |
| "rewards/margins": 4.833321571350098, | |
| "rewards/rejected": -4.43407678604126, | |
| "step": 935 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 939, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |