7112 lines
240 KiB
JSON
7112 lines
240 KiB
JSON
{
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.9981298423724285,
|
|
"eval_steps": 100,
|
|
"global_step": 467,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.0021373230029388193,
|
|
"grad_norm": 5.9665925980173595,
|
|
"learning_rate": 1.7021276595744678e-08,
|
|
"logits/chosen": -1.1223822832107544,
|
|
"logits/rejected": -1.1192574501037598,
|
|
"logps/chosen": -257.6567687988281,
|
|
"logps/rejected": -268.0864562988281,
|
|
"loss": 0.6931,
|
|
"rewards/accuracies": 0.0,
|
|
"rewards/chosen": 0.0,
|
|
"rewards/margins": 0.0,
|
|
"rewards/rejected": 0.0,
|
|
"step": 1
|
|
},
|
|
{
|
|
"epoch": 0.004274646005877639,
|
|
"grad_norm": 5.827849982093965,
|
|
"learning_rate": 3.4042553191489356e-08,
|
|
"logits/chosen": -1.1125223636627197,
|
|
"logits/rejected": -1.0834215879440308,
|
|
"logps/chosen": -333.40960693359375,
|
|
"logps/rejected": -285.8963928222656,
|
|
"loss": 0.6931,
|
|
"rewards/accuracies": 0.0,
|
|
"rewards/chosen": 0.0,
|
|
"rewards/margins": 0.0,
|
|
"rewards/rejected": 0.0,
|
|
"step": 2
|
|
},
|
|
{
|
|
"epoch": 0.006411969008816457,
|
|
"grad_norm": 6.028572669421686,
|
|
"learning_rate": 5.106382978723404e-08,
|
|
"logits/chosen": -0.9786796569824219,
|
|
"logits/rejected": -0.9414944052696228,
|
|
"logps/chosen": -242.9634552001953,
|
|
"logps/rejected": -243.12557983398438,
|
|
"loss": 0.6936,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": 0.00023853778839111328,
|
|
"rewards/margins": 7.291347719728947e-05,
|
|
"rewards/rejected": 0.0001656246604397893,
|
|
"step": 3
|
|
},
|
|
{
|
|
"epoch": 0.008549292011755277,
|
|
"grad_norm": 5.496288011485378,
|
|
"learning_rate": 6.808510638297871e-08,
|
|
"logits/chosen": -1.1184200048446655,
|
|
"logits/rejected": -1.1394802331924438,
|
|
"logps/chosen": -210.64151000976562,
|
|
"logps/rejected": -217.98028564453125,
|
|
"loss": 0.6934,
|
|
"rewards/accuracies": 0.3125,
|
|
"rewards/chosen": -0.001957583473995328,
|
|
"rewards/margins": -0.0010110187577083707,
|
|
"rewards/rejected": -0.0009465646580792964,
|
|
"step": 4
|
|
},
|
|
{
|
|
"epoch": 0.010686615014694095,
|
|
"grad_norm": 6.1423037995058385,
|
|
"learning_rate": 8.51063829787234e-08,
|
|
"logits/chosen": -1.1924649477005005,
|
|
"logits/rejected": -1.2157042026519775,
|
|
"logps/chosen": -229.38040161132812,
|
|
"logps/rejected": -228.48341369628906,
|
|
"loss": 0.6924,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.0004083870444446802,
|
|
"rewards/margins": 0.002144141122698784,
|
|
"rewards/rejected": -0.002552528167143464,
|
|
"step": 5
|
|
},
|
|
{
|
|
"epoch": 0.012823938017632914,
|
|
"grad_norm": 6.659651187507062,
|
|
"learning_rate": 1.0212765957446807e-07,
|
|
"logits/chosen": -1.152483344078064,
|
|
"logits/rejected": -1.0924268960952759,
|
|
"logps/chosen": -317.67449951171875,
|
|
"logps/rejected": -295.94219970703125,
|
|
"loss": 0.693,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.001339879003353417,
|
|
"rewards/margins": -0.00023622988373972476,
|
|
"rewards/rejected": -0.0011036490323022008,
|
|
"step": 6
|
|
},
|
|
{
|
|
"epoch": 0.014961261020571734,
|
|
"grad_norm": 7.17635843844568,
|
|
"learning_rate": 1.1914893617021275e-07,
|
|
"logits/chosen": -1.2364102602005005,
|
|
"logits/rejected": -1.1785768270492554,
|
|
"logps/chosen": -242.46392822265625,
|
|
"logps/rejected": -235.65455627441406,
|
|
"loss": 0.6926,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.0010340880835428834,
|
|
"rewards/margins": 0.0029830788262188435,
|
|
"rewards/rejected": -0.004017166793346405,
|
|
"step": 7
|
|
},
|
|
{
|
|
"epoch": 0.017098584023510555,
|
|
"grad_norm": 5.718849974150984,
|
|
"learning_rate": 1.3617021276595742e-07,
|
|
"logits/chosen": -1.1600620746612549,
|
|
"logits/rejected": -1.1021112203598022,
|
|
"logps/chosen": -253.3247528076172,
|
|
"logps/rejected": -266.04168701171875,
|
|
"loss": 0.6926,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": 0.0006916046258993447,
|
|
"rewards/margins": 0.0022089341655373573,
|
|
"rewards/rejected": -0.0015173291321843863,
|
|
"step": 8
|
|
},
|
|
{
|
|
"epoch": 0.01923590702644937,
|
|
"grad_norm": 5.084147578805341,
|
|
"learning_rate": 1.531914893617021e-07,
|
|
"logits/chosen": -1.079232096672058,
|
|
"logits/rejected": -1.0460667610168457,
|
|
"logps/chosen": -286.5476379394531,
|
|
"logps/rejected": -294.2388610839844,
|
|
"loss": 0.6935,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": 0.000238885753788054,
|
|
"rewards/margins": 0.0006698797806166112,
|
|
"rewards/rejected": -0.00043099402682855725,
|
|
"step": 9
|
|
},
|
|
{
|
|
"epoch": 0.02137323002938819,
|
|
"grad_norm": 6.119066349537701,
|
|
"learning_rate": 1.702127659574468e-07,
|
|
"logits/chosen": -1.3416539430618286,
|
|
"logits/rejected": -1.3090091943740845,
|
|
"logps/chosen": -269.2237548828125,
|
|
"logps/rejected": -290.6747741699219,
|
|
"loss": 0.6933,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.0022063730284571648,
|
|
"rewards/margins": -0.00022077548783272505,
|
|
"rewards/rejected": -0.001985597424209118,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.02351055303232701,
|
|
"grad_norm": 6.296522722355036,
|
|
"learning_rate": 1.872340425531915e-07,
|
|
"logits/chosen": -1.1501847505569458,
|
|
"logits/rejected": -1.1223665475845337,
|
|
"logps/chosen": -268.7527160644531,
|
|
"logps/rejected": -282.72100830078125,
|
|
"loss": 0.6927,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": 0.0022443102207034826,
|
|
"rewards/margins": 0.0018837449606508017,
|
|
"rewards/rejected": 0.0003605652309488505,
|
|
"step": 11
|
|
},
|
|
{
|
|
"epoch": 0.02564787603526583,
|
|
"grad_norm": 4.946921694352282,
|
|
"learning_rate": 2.0425531914893615e-07,
|
|
"logits/chosen": -1.0480014085769653,
|
|
"logits/rejected": -1.0393519401550293,
|
|
"logps/chosen": -236.5433349609375,
|
|
"logps/rejected": -217.593017578125,
|
|
"loss": 0.693,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.0006607104442082345,
|
|
"rewards/margins": -0.000438790419138968,
|
|
"rewards/rejected": -0.00022191996686160564,
|
|
"step": 12
|
|
},
|
|
{
|
|
"epoch": 0.027785199038204648,
|
|
"grad_norm": 5.73018010615819,
|
|
"learning_rate": 2.2127659574468084e-07,
|
|
"logits/chosen": -1.0228816270828247,
|
|
"logits/rejected": -0.9232650995254517,
|
|
"logps/chosen": -265.681884765625,
|
|
"logps/rejected": -292.8697509765625,
|
|
"loss": 0.6932,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.004240307491272688,
|
|
"rewards/margins": -0.001113958191126585,
|
|
"rewards/rejected": -0.003126349300146103,
|
|
"step": 13
|
|
},
|
|
{
|
|
"epoch": 0.029922522041143467,
|
|
"grad_norm": 6.486456105963774,
|
|
"learning_rate": 2.382978723404255e-07,
|
|
"logits/chosen": -1.1559597253799438,
|
|
"logits/rejected": -1.204541563987732,
|
|
"logps/chosen": -302.4142150878906,
|
|
"logps/rejected": -285.40521240234375,
|
|
"loss": 0.6927,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.0032664297614246607,
|
|
"rewards/margins": 0.0017348529072478414,
|
|
"rewards/rejected": -0.0050012823194265366,
|
|
"step": 14
|
|
},
|
|
{
|
|
"epoch": 0.03205984504408229,
|
|
"grad_norm": 5.03221979114363,
|
|
"learning_rate": 2.553191489361702e-07,
|
|
"logits/chosen": -1.3137496709823608,
|
|
"logits/rejected": -1.2676270008087158,
|
|
"logps/chosen": -271.43218994140625,
|
|
"logps/rejected": -278.537109375,
|
|
"loss": 0.6928,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.0035740709863603115,
|
|
"rewards/margins": 0.0014838932547718287,
|
|
"rewards/rejected": -0.0050579640083014965,
|
|
"step": 15
|
|
},
|
|
{
|
|
"epoch": 0.03419716804702111,
|
|
"grad_norm": 5.805233594511206,
|
|
"learning_rate": 2.7234042553191485e-07,
|
|
"logits/chosen": -1.2339807748794556,
|
|
"logits/rejected": -1.1998343467712402,
|
|
"logps/chosen": -278.9430236816406,
|
|
"logps/rejected": -287.36444091796875,
|
|
"loss": 0.6929,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.005366215482354164,
|
|
"rewards/margins": -0.0015738580841571093,
|
|
"rewards/rejected": -0.003792357398197055,
|
|
"step": 16
|
|
},
|
|
{
|
|
"epoch": 0.03633449104995993,
|
|
"grad_norm": 5.939188331286007,
|
|
"learning_rate": 2.893617021276596e-07,
|
|
"logits/chosen": -1.267714500427246,
|
|
"logits/rejected": -1.2154172658920288,
|
|
"logps/chosen": -243.30702209472656,
|
|
"logps/rejected": -243.65267944335938,
|
|
"loss": 0.6924,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -0.005914730951189995,
|
|
"rewards/margins": -0.003907041624188423,
|
|
"rewards/rejected": -0.0020076895598322153,
|
|
"step": 17
|
|
},
|
|
{
|
|
"epoch": 0.03847181405289874,
|
|
"grad_norm": 5.736074036017852,
|
|
"learning_rate": 3.063829787234042e-07,
|
|
"logits/chosen": -1.211553692817688,
|
|
"logits/rejected": -1.218402624130249,
|
|
"logps/chosen": -246.43824768066406,
|
|
"logps/rejected": -262.7088928222656,
|
|
"loss": 0.6922,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.0035472153685986996,
|
|
"rewards/margins": -0.00027114871772937477,
|
|
"rewards/rejected": -0.0032760666217654943,
|
|
"step": 18
|
|
},
|
|
{
|
|
"epoch": 0.04060913705583756,
|
|
"grad_norm": 8.028547080597,
|
|
"learning_rate": 3.234042553191489e-07,
|
|
"logits/chosen": -1.2778618335723877,
|
|
"logits/rejected": -1.151484727859497,
|
|
"logps/chosen": -263.84149169921875,
|
|
"logps/rejected": -285.849365234375,
|
|
"loss": 0.6917,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.004957227967679501,
|
|
"rewards/margins": 0.004752359353005886,
|
|
"rewards/rejected": -0.009709586389362812,
|
|
"step": 19
|
|
},
|
|
{
|
|
"epoch": 0.04274646005877638,
|
|
"grad_norm": 5.812341074260827,
|
|
"learning_rate": 3.404255319148936e-07,
|
|
"logits/chosen": -1.2942702770233154,
|
|
"logits/rejected": -1.2191565036773682,
|
|
"logps/chosen": -309.5222473144531,
|
|
"logps/rejected": -300.4433898925781,
|
|
"loss": 0.6914,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.014837831258773804,
|
|
"rewards/margins": 0.004754701629281044,
|
|
"rewards/rejected": -0.019592532888054848,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.0448837830617152,
|
|
"grad_norm": 5.722892787713038,
|
|
"learning_rate": 3.574468085106383e-07,
|
|
"logits/chosen": -1.2925513982772827,
|
|
"logits/rejected": -1.2902947664260864,
|
|
"logps/chosen": -201.62168884277344,
|
|
"logps/rejected": -207.10025024414062,
|
|
"loss": 0.6911,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.005997480824589729,
|
|
"rewards/margins": 0.002215881133452058,
|
|
"rewards/rejected": -0.008213362656533718,
|
|
"step": 21
|
|
},
|
|
{
|
|
"epoch": 0.04702110606465402,
|
|
"grad_norm": 5.066561101041897,
|
|
"learning_rate": 3.74468085106383e-07,
|
|
"logits/chosen": -1.2172021865844727,
|
|
"logits/rejected": -1.2375967502593994,
|
|
"logps/chosen": -213.11264038085938,
|
|
"logps/rejected": -208.974609375,
|
|
"loss": 0.6908,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.011541333049535751,
|
|
"rewards/margins": 0.0011011171154677868,
|
|
"rewards/rejected": -0.012642450630664825,
|
|
"step": 22
|
|
},
|
|
{
|
|
"epoch": 0.04915842906759284,
|
|
"grad_norm": 5.8862642088746355,
|
|
"learning_rate": 3.914893617021276e-07,
|
|
"logits/chosen": -1.0798557996749878,
|
|
"logits/rejected": -0.9454042911529541,
|
|
"logps/chosen": -252.95816040039062,
|
|
"logps/rejected": -280.6626892089844,
|
|
"loss": 0.6905,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.014871005900204182,
|
|
"rewards/margins": 0.007070617750287056,
|
|
"rewards/rejected": -0.021941622719168663,
|
|
"step": 23
|
|
},
|
|
{
|
|
"epoch": 0.05129575207053166,
|
|
"grad_norm": 9.358799637903934,
|
|
"learning_rate": 4.085106382978723e-07,
|
|
"logits/chosen": -1.229708194732666,
|
|
"logits/rejected": -1.1125763654708862,
|
|
"logps/chosen": -205.67190551757812,
|
|
"logps/rejected": -238.33160400390625,
|
|
"loss": 0.6902,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.012925582006573677,
|
|
"rewards/margins": 0.0010348652722314,
|
|
"rewards/rejected": -0.01396044623106718,
|
|
"step": 24
|
|
},
|
|
{
|
|
"epoch": 0.053433075073470476,
|
|
"grad_norm": 5.386776465639343,
|
|
"learning_rate": 4.25531914893617e-07,
|
|
"logits/chosen": -1.0455987453460693,
|
|
"logits/rejected": -1.0133016109466553,
|
|
"logps/chosen": -286.6942138671875,
|
|
"logps/rejected": -276.3359069824219,
|
|
"loss": 0.6897,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.017084330320358276,
|
|
"rewards/margins": 0.006912917364388704,
|
|
"rewards/rejected": -0.023997247219085693,
|
|
"step": 25
|
|
},
|
|
{
|
|
"epoch": 0.055570398076409296,
|
|
"grad_norm": 5.50227897442877,
|
|
"learning_rate": 4.425531914893617e-07,
|
|
"logits/chosen": -1.0649359226226807,
|
|
"logits/rejected": -1.0070443153381348,
|
|
"logps/chosen": -246.0877685546875,
|
|
"logps/rejected": -261.14007568359375,
|
|
"loss": 0.689,
|
|
"rewards/accuracies": 0.375,
|
|
"rewards/chosen": -0.020686931908130646,
|
|
"rewards/margins": 0.0015359835233539343,
|
|
"rewards/rejected": -0.022222913801670074,
|
|
"step": 26
|
|
},
|
|
{
|
|
"epoch": 0.057707721079348115,
|
|
"grad_norm": 5.926262302660941,
|
|
"learning_rate": 4.595744680851064e-07,
|
|
"logits/chosen": -1.1902655363082886,
|
|
"logits/rejected": -1.2254102230072021,
|
|
"logps/chosen": -257.3299255371094,
|
|
"logps/rejected": -252.44285583496094,
|
|
"loss": 0.6881,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.015050854533910751,
|
|
"rewards/margins": 0.007191185839474201,
|
|
"rewards/rejected": -0.022242041304707527,
|
|
"step": 27
|
|
},
|
|
{
|
|
"epoch": 0.059845044082286934,
|
|
"grad_norm": 5.816673583594112,
|
|
"learning_rate": 4.76595744680851e-07,
|
|
"logits/chosen": -1.2353150844573975,
|
|
"logits/rejected": -1.2537351846694946,
|
|
"logps/chosen": -239.78131103515625,
|
|
"logps/rejected": -237.89956665039062,
|
|
"loss": 0.6871,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.0323203019797802,
|
|
"rewards/margins": 0.00978852715343237,
|
|
"rewards/rejected": -0.04210883006453514,
|
|
"step": 28
|
|
},
|
|
{
|
|
"epoch": 0.061982367085225754,
|
|
"grad_norm": 6.282142408603975,
|
|
"learning_rate": 4.936170212765957e-07,
|
|
"logits/chosen": -1.1677157878875732,
|
|
"logits/rejected": -1.1619088649749756,
|
|
"logps/chosen": -272.3318176269531,
|
|
"logps/rejected": -272.49407958984375,
|
|
"loss": 0.687,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.03659503534436226,
|
|
"rewards/margins": 0.010677139274775982,
|
|
"rewards/rejected": -0.047272179275751114,
|
|
"step": 29
|
|
},
|
|
{
|
|
"epoch": 0.06411969008816458,
|
|
"grad_norm": 6.059359505202537,
|
|
"learning_rate": 5.106382978723404e-07,
|
|
"logits/chosen": -1.1942038536071777,
|
|
"logits/rejected": -1.2899129390716553,
|
|
"logps/chosen": -215.77291870117188,
|
|
"logps/rejected": -200.72760009765625,
|
|
"loss": 0.6857,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.02227736823260784,
|
|
"rewards/margins": 0.0018395374063402414,
|
|
"rewards/rejected": -0.02411690726876259,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.06625701309110339,
|
|
"grad_norm": 5.482721306670962,
|
|
"learning_rate": 5.276595744680851e-07,
|
|
"logits/chosen": -1.0668655633926392,
|
|
"logits/rejected": -0.9902481436729431,
|
|
"logps/chosen": -231.6899871826172,
|
|
"logps/rejected": -255.43142700195312,
|
|
"loss": 0.6856,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.03268000856041908,
|
|
"rewards/margins": 0.016983937472105026,
|
|
"rewards/rejected": -0.04966394975781441,
|
|
"step": 31
|
|
},
|
|
{
|
|
"epoch": 0.06839433609404222,
|
|
"grad_norm": 6.948193767681259,
|
|
"learning_rate": 5.446808510638297e-07,
|
|
"logits/chosen": -1.193525791168213,
|
|
"logits/rejected": -1.2220956087112427,
|
|
"logps/chosen": -300.5246887207031,
|
|
"logps/rejected": -285.05584716796875,
|
|
"loss": 0.6797,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.06849750876426697,
|
|
"rewards/margins": 0.023764295503497124,
|
|
"rewards/rejected": -0.09226180613040924,
|
|
"step": 32
|
|
},
|
|
{
|
|
"epoch": 0.07053165909698103,
|
|
"grad_norm": 6.841964931394082,
|
|
"learning_rate": 5.617021276595744e-07,
|
|
"logits/chosen": -1.037040114402771,
|
|
"logits/rejected": -1.0746084451675415,
|
|
"logps/chosen": -326.0146484375,
|
|
"logps/rejected": -331.83868408203125,
|
|
"loss": 0.6816,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.07449866831302643,
|
|
"rewards/margins": 0.029566621407866478,
|
|
"rewards/rejected": -0.10406528413295746,
|
|
"step": 33
|
|
},
|
|
{
|
|
"epoch": 0.07266898209991986,
|
|
"grad_norm": 6.505679088181679,
|
|
"learning_rate": 5.787234042553192e-07,
|
|
"logits/chosen": -1.2144380807876587,
|
|
"logits/rejected": -1.172149658203125,
|
|
"logps/chosen": -285.81732177734375,
|
|
"logps/rejected": -277.4313659667969,
|
|
"loss": 0.6811,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.06557968258857727,
|
|
"rewards/margins": 0.01972799003124237,
|
|
"rewards/rejected": -0.08530767261981964,
|
|
"step": 34
|
|
},
|
|
{
|
|
"epoch": 0.07480630510285867,
|
|
"grad_norm": 5.874795982379736,
|
|
"learning_rate": 5.957446808510638e-07,
|
|
"logits/chosen": -1.2059704065322876,
|
|
"logits/rejected": -1.1337679624557495,
|
|
"logps/chosen": -316.4410400390625,
|
|
"logps/rejected": -358.4386901855469,
|
|
"loss": 0.6786,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.08414824306964874,
|
|
"rewards/margins": 0.04845026880502701,
|
|
"rewards/rejected": -0.13259851932525635,
|
|
"step": 35
|
|
},
|
|
{
|
|
"epoch": 0.07694362810579748,
|
|
"grad_norm": 6.990449192598191,
|
|
"learning_rate": 6.127659574468084e-07,
|
|
"logits/chosen": -1.2996914386749268,
|
|
"logits/rejected": -1.1766685247421265,
|
|
"logps/chosen": -225.32235717773438,
|
|
"logps/rejected": -242.12782287597656,
|
|
"loss": 0.6773,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.0663793534040451,
|
|
"rewards/margins": 0.00477500818669796,
|
|
"rewards/rejected": -0.07115435600280762,
|
|
"step": 36
|
|
},
|
|
{
|
|
"epoch": 0.07908095110873631,
|
|
"grad_norm": 6.539356394704236,
|
|
"learning_rate": 6.297872340425532e-07,
|
|
"logits/chosen": -1.3612180948257446,
|
|
"logits/rejected": -1.365153431892395,
|
|
"logps/chosen": -247.11227416992188,
|
|
"logps/rejected": -245.68914794921875,
|
|
"loss": 0.6767,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.10457728803157806,
|
|
"rewards/margins": 0.018666967749595642,
|
|
"rewards/rejected": -0.12324424088001251,
|
|
"step": 37
|
|
},
|
|
{
|
|
"epoch": 0.08121827411167512,
|
|
"grad_norm": 7.001992517945579,
|
|
"learning_rate": 6.468085106382978e-07,
|
|
"logits/chosen": -1.206650733947754,
|
|
"logits/rejected": -1.1853901147842407,
|
|
"logps/chosen": -290.7265319824219,
|
|
"logps/rejected": -278.7143249511719,
|
|
"loss": 0.6706,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.08214958757162094,
|
|
"rewards/margins": 0.03963039815425873,
|
|
"rewards/rejected": -0.12177997827529907,
|
|
"step": 38
|
|
},
|
|
{
|
|
"epoch": 0.08335559711461395,
|
|
"grad_norm": 6.9572266440707455,
|
|
"learning_rate": 6.638297872340425e-07,
|
|
"logits/chosen": -1.2570126056671143,
|
|
"logits/rejected": -1.209625005722046,
|
|
"logps/chosen": -220.4678955078125,
|
|
"logps/rejected": -248.70547485351562,
|
|
"loss": 0.6708,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.10038035362958908,
|
|
"rewards/margins": 0.0464475080370903,
|
|
"rewards/rejected": -0.14682786166667938,
|
|
"step": 39
|
|
},
|
|
{
|
|
"epoch": 0.08549292011755276,
|
|
"grad_norm": 6.646103732875039,
|
|
"learning_rate": 6.808510638297872e-07,
|
|
"logits/chosen": -1.2964568138122559,
|
|
"logits/rejected": -1.2421560287475586,
|
|
"logps/chosen": -268.4561767578125,
|
|
"logps/rejected": -277.7322998046875,
|
|
"loss": 0.6691,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.12192462384700775,
|
|
"rewards/margins": 0.024914879351854324,
|
|
"rewards/rejected": -0.14683951437473297,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.08763024312049159,
|
|
"grad_norm": 6.963601763048083,
|
|
"learning_rate": 6.978723404255319e-07,
|
|
"logits/chosen": -1.2182767391204834,
|
|
"logits/rejected": -1.144841194152832,
|
|
"logps/chosen": -208.9151611328125,
|
|
"logps/rejected": -223.7606201171875,
|
|
"loss": 0.6683,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.10518630594015121,
|
|
"rewards/margins": 0.0583706796169281,
|
|
"rewards/rejected": -0.1635569930076599,
|
|
"step": 41
|
|
},
|
|
{
|
|
"epoch": 0.0897675661234304,
|
|
"grad_norm": 6.059193610856394,
|
|
"learning_rate": 7.148936170212766e-07,
|
|
"logits/chosen": -1.2251960039138794,
|
|
"logits/rejected": -1.1969135999679565,
|
|
"logps/chosen": -275.060791015625,
|
|
"logps/rejected": -310.41168212890625,
|
|
"loss": 0.6666,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -0.17626452445983887,
|
|
"rewards/margins": 0.06843668222427368,
|
|
"rewards/rejected": -0.24470120668411255,
|
|
"step": 42
|
|
},
|
|
{
|
|
"epoch": 0.09190488912636922,
|
|
"grad_norm": 6.5707188570559385,
|
|
"learning_rate": 7.319148936170212e-07,
|
|
"logits/chosen": -1.1565831899642944,
|
|
"logits/rejected": -1.101088285446167,
|
|
"logps/chosen": -232.02914428710938,
|
|
"logps/rejected": -240.09799194335938,
|
|
"loss": 0.6632,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.10670860856771469,
|
|
"rewards/margins": 0.05728934705257416,
|
|
"rewards/rejected": -0.16399797797203064,
|
|
"step": 43
|
|
},
|
|
{
|
|
"epoch": 0.09404221212930804,
|
|
"grad_norm": 6.11669964043902,
|
|
"learning_rate": 7.48936170212766e-07,
|
|
"logits/chosen": -1.042961835861206,
|
|
"logits/rejected": -1.1094971895217896,
|
|
"logps/chosen": -259.85546875,
|
|
"logps/rejected": -273.7874755859375,
|
|
"loss": 0.6633,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.1373087465763092,
|
|
"rewards/margins": 0.02613016963005066,
|
|
"rewards/rejected": -0.16343891620635986,
|
|
"step": 44
|
|
},
|
|
{
|
|
"epoch": 0.09617953513224686,
|
|
"grad_norm": 6.39410867008386,
|
|
"learning_rate": 7.659574468085107e-07,
|
|
"logits/chosen": -1.2064584493637085,
|
|
"logits/rejected": -1.1550904512405396,
|
|
"logps/chosen": -342.9076232910156,
|
|
"logps/rejected": -351.82830810546875,
|
|
"loss": 0.6587,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.20879533886909485,
|
|
"rewards/margins": 0.049167849123477936,
|
|
"rewards/rejected": -0.2579631805419922,
|
|
"step": 45
|
|
},
|
|
{
|
|
"epoch": 0.09831685813518568,
|
|
"grad_norm": 5.9662902412289816,
|
|
"learning_rate": 7.829787234042552e-07,
|
|
"logits/chosen": -1.2047083377838135,
|
|
"logits/rejected": -1.1663964986801147,
|
|
"logps/chosen": -279.71356201171875,
|
|
"logps/rejected": -309.82489013671875,
|
|
"loss": 0.6615,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.21875953674316406,
|
|
"rewards/margins": 0.14003317058086395,
|
|
"rewards/rejected": -0.3587927222251892,
|
|
"step": 46
|
|
},
|
|
{
|
|
"epoch": 0.1004541811381245,
|
|
"grad_norm": 6.848260346060141,
|
|
"learning_rate": 8e-07,
|
|
"logits/chosen": -1.2341694831848145,
|
|
"logits/rejected": -1.198740839958191,
|
|
"logps/chosen": -273.4669189453125,
|
|
"logps/rejected": -289.4903564453125,
|
|
"loss": 0.6422,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -0.2072269469499588,
|
|
"rewards/margins": 0.15279962122440338,
|
|
"rewards/rejected": -0.3600265383720398,
|
|
"step": 47
|
|
},
|
|
{
|
|
"epoch": 0.10259150414106331,
|
|
"grad_norm": 6.627952291422947,
|
|
"learning_rate": 7.999888100245079e-07,
|
|
"logits/chosen": -1.1258509159088135,
|
|
"logits/rejected": -1.0670114755630493,
|
|
"logps/chosen": -218.09140014648438,
|
|
"logps/rejected": -236.913818359375,
|
|
"loss": 0.6483,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -0.10230693221092224,
|
|
"rewards/margins": 0.11233773827552795,
|
|
"rewards/rejected": -0.2146446704864502,
|
|
"step": 48
|
|
},
|
|
{
|
|
"epoch": 0.10472882714400214,
|
|
"grad_norm": 7.206747256622627,
|
|
"learning_rate": 7.999552407241093e-07,
|
|
"logits/chosen": -1.1644402742385864,
|
|
"logits/rejected": -1.1633307933807373,
|
|
"logps/chosen": -222.858642578125,
|
|
"logps/rejected": -237.24234008789062,
|
|
"loss": 0.6281,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.17772327363491058,
|
|
"rewards/margins": 0.06820763647556305,
|
|
"rewards/rejected": -0.24593091011047363,
|
|
"step": 49
|
|
},
|
|
{
|
|
"epoch": 0.10686615014694095,
|
|
"grad_norm": 6.2831933314047435,
|
|
"learning_rate": 7.998992939770026e-07,
|
|
"logits/chosen": -1.2497477531433105,
|
|
"logits/rejected": -1.1775400638580322,
|
|
"logps/chosen": -223.41224670410156,
|
|
"logps/rejected": -229.77484130859375,
|
|
"loss": 0.6441,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.176996111869812,
|
|
"rewards/margins": 0.10983872413635254,
|
|
"rewards/rejected": -0.28683483600616455,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.10900347314987978,
|
|
"grad_norm": 6.77971339252681,
|
|
"learning_rate": 7.998209729134013e-07,
|
|
"logits/chosen": -1.2855724096298218,
|
|
"logits/rejected": -1.1933931112289429,
|
|
"logps/chosen": -291.30792236328125,
|
|
"logps/rejected": -304.0755310058594,
|
|
"loss": 0.6383,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.43583106994628906,
|
|
"rewards/margins": 0.13815590739250183,
|
|
"rewards/rejected": -0.5739869475364685,
|
|
"step": 51
|
|
},
|
|
{
|
|
"epoch": 0.11114079615281859,
|
|
"grad_norm": 7.218967742515957,
|
|
"learning_rate": 7.997202819153594e-07,
|
|
"logits/chosen": -1.1941241025924683,
|
|
"logits/rejected": -1.2352594137191772,
|
|
"logps/chosen": -296.0772399902344,
|
|
"logps/rejected": -312.9319152832031,
|
|
"loss": 0.6107,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.3638754189014435,
|
|
"rewards/margins": 0.1599450707435608,
|
|
"rewards/rejected": -0.5238205194473267,
|
|
"step": 52
|
|
},
|
|
{
|
|
"epoch": 0.11327811915575742,
|
|
"grad_norm": 6.701000620809773,
|
|
"learning_rate": 7.995972266165259e-07,
|
|
"logits/chosen": -1.2163739204406738,
|
|
"logits/rejected": -1.1790796518325806,
|
|
"logps/chosen": -342.8826599121094,
|
|
"logps/rejected": -385.2430419921875,
|
|
"loss": 0.6203,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.6190586686134338,
|
|
"rewards/margins": 0.20686545968055725,
|
|
"rewards/rejected": -0.8259240984916687,
|
|
"step": 53
|
|
},
|
|
{
|
|
"epoch": 0.11541544215869623,
|
|
"grad_norm": 7.252043517716344,
|
|
"learning_rate": 7.994518139018295e-07,
|
|
"logits/chosen": -1.2206201553344727,
|
|
"logits/rejected": -1.1347448825836182,
|
|
"logps/chosen": -285.89044189453125,
|
|
"logps/rejected": -303.69091796875,
|
|
"loss": 0.6351,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -0.45011359453201294,
|
|
"rewards/margins": 0.11390627920627594,
|
|
"rewards/rejected": -0.5640199184417725,
|
|
"step": 54
|
|
},
|
|
{
|
|
"epoch": 0.11755276516163506,
|
|
"grad_norm": 9.977900301785272,
|
|
"learning_rate": 7.99284051907094e-07,
|
|
"logits/chosen": -1.0799576044082642,
|
|
"logits/rejected": -1.00977623462677,
|
|
"logps/chosen": -271.39154052734375,
|
|
"logps/rejected": -310.511474609375,
|
|
"loss": 0.5842,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.5156710147857666,
|
|
"rewards/margins": 0.30485641956329346,
|
|
"rewards/rejected": -0.8205274343490601,
|
|
"step": 55
|
|
},
|
|
{
|
|
"epoch": 0.11969008816457387,
|
|
"grad_norm": 6.400760110377084,
|
|
"learning_rate": 7.990939500185824e-07,
|
|
"logits/chosen": -1.1088342666625977,
|
|
"logits/rejected": -1.1030653715133667,
|
|
"logps/chosen": -364.3983459472656,
|
|
"logps/rejected": -385.44000244140625,
|
|
"loss": 0.618,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.8900222182273865,
|
|
"rewards/margins": 0.1941274106502533,
|
|
"rewards/rejected": -1.0841495990753174,
|
|
"step": 56
|
|
},
|
|
{
|
|
"epoch": 0.1218274111675127,
|
|
"grad_norm": 6.9939690893971695,
|
|
"learning_rate": 7.98881518872472e-07,
|
|
"logits/chosen": -1.2345085144042969,
|
|
"logits/rejected": -1.148017406463623,
|
|
"logps/chosen": -302.2966003417969,
|
|
"logps/rejected": -330.4090576171875,
|
|
"loss": 0.5864,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -0.7428092360496521,
|
|
"rewards/margins": 0.30247360467910767,
|
|
"rewards/rejected": -1.0452828407287598,
|
|
"step": 57
|
|
},
|
|
{
|
|
"epoch": 0.12396473417045151,
|
|
"grad_norm": 7.089640899087372,
|
|
"learning_rate": 7.986467703542595e-07,
|
|
"logits/chosen": -1.2412291765213013,
|
|
"logits/rejected": -1.2027617692947388,
|
|
"logps/chosen": -309.7779235839844,
|
|
"logps/rejected": -313.6658935546875,
|
|
"loss": 0.6085,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.9433287978172302,
|
|
"rewards/margins": 0.06926216185092926,
|
|
"rewards/rejected": -1.0125908851623535,
|
|
"step": 58
|
|
},
|
|
{
|
|
"epoch": 0.12610205717339032,
|
|
"grad_norm": 8.578352434954352,
|
|
"learning_rate": 7.983897175980956e-07,
|
|
"logits/chosen": -1.187761902809143,
|
|
"logits/rejected": -1.1473008394241333,
|
|
"logps/chosen": -522.899169921875,
|
|
"logps/rejected": -563.5382080078125,
|
|
"loss": 0.5888,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -1.4695754051208496,
|
|
"rewards/margins": 0.39022523164749146,
|
|
"rewards/rejected": -1.8598008155822754,
|
|
"step": 59
|
|
},
|
|
{
|
|
"epoch": 0.12823938017632916,
|
|
"grad_norm": 7.479375095677279,
|
|
"learning_rate": 7.981103749860505e-07,
|
|
"logits/chosen": -1.1445815563201904,
|
|
"logits/rejected": -1.1645028591156006,
|
|
"logps/chosen": -345.0823974609375,
|
|
"logps/rejected": -342.9745788574219,
|
|
"loss": 0.5963,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -0.995075523853302,
|
|
"rewards/margins": 0.0031197145581245422,
|
|
"rewards/rejected": -0.9981951713562012,
|
|
"step": 60
|
|
},
|
|
{
|
|
"epoch": 0.13037670317926797,
|
|
"grad_norm": 9.767907382058546,
|
|
"learning_rate": 7.978087581473093e-07,
|
|
"logits/chosen": -1.1243810653686523,
|
|
"logits/rejected": -1.064988374710083,
|
|
"logps/chosen": -456.0157470703125,
|
|
"logps/rejected": -530.0255737304688,
|
|
"loss": 0.5985,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -1.5085893869400024,
|
|
"rewards/margins": 0.4878966808319092,
|
|
"rewards/rejected": -1.996485948562622,
|
|
"step": 61
|
|
},
|
|
{
|
|
"epoch": 0.13251402618220678,
|
|
"grad_norm": 8.307089325438865,
|
|
"learning_rate": 7.97484883957297e-07,
|
|
"logits/chosen": -1.2906808853149414,
|
|
"logits/rejected": -1.232366919517517,
|
|
"logps/chosen": -459.9493103027344,
|
|
"logps/rejected": -529.822021484375,
|
|
"loss": 0.5535,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.7710328102111816,
|
|
"rewards/margins": 0.7507337331771851,
|
|
"rewards/rejected": -2.5217666625976562,
|
|
"step": 62
|
|
},
|
|
{
|
|
"epoch": 0.1346513491851456,
|
|
"grad_norm": 9.937869344696164,
|
|
"learning_rate": 7.971387705367349e-07,
|
|
"logits/chosen": -1.0343647003173828,
|
|
"logits/rejected": -1.009024739265442,
|
|
"logps/chosen": -344.3237609863281,
|
|
"logps/rejected": -414.5412292480469,
|
|
"loss": 0.5707,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.04449462890625,
|
|
"rewards/margins": 0.5895835757255554,
|
|
"rewards/rejected": -1.6340781450271606,
|
|
"step": 63
|
|
},
|
|
{
|
|
"epoch": 0.13678867218808444,
|
|
"grad_norm": 8.102078008830782,
|
|
"learning_rate": 7.967704372506265e-07,
|
|
"logits/chosen": -1.136821985244751,
|
|
"logits/rejected": -1.145324468612671,
|
|
"logps/chosen": -454.86968994140625,
|
|
"logps/rejected": -522.869140625,
|
|
"loss": 0.5878,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.6470240354537964,
|
|
"rewards/margins": 0.73206627368927,
|
|
"rewards/rejected": -2.3790900707244873,
|
|
"step": 64
|
|
},
|
|
{
|
|
"epoch": 0.13892599519102325,
|
|
"grad_norm": 9.062627671259957,
|
|
"learning_rate": 7.963799047071739e-07,
|
|
"logits/chosen": -1.1571276187896729,
|
|
"logits/rejected": -1.1334136724472046,
|
|
"logps/chosen": -454.8720397949219,
|
|
"logps/rejected": -572.2128295898438,
|
|
"loss": 0.5829,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.8390952348709106,
|
|
"rewards/margins": 1.0353538990020752,
|
|
"rewards/rejected": -2.8744490146636963,
|
|
"step": 65
|
|
},
|
|
{
|
|
"epoch": 0.14106331819396206,
|
|
"grad_norm": 9.107794878722252,
|
|
"learning_rate": 7.959671947566251e-07,
|
|
"logits/chosen": -1.3037822246551514,
|
|
"logits/rejected": -1.254806637763977,
|
|
"logps/chosen": -514.9037475585938,
|
|
"logps/rejected": -532.9312133789062,
|
|
"loss": 0.6234,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -2.43453049659729,
|
|
"rewards/margins": 0.19514000415802002,
|
|
"rewards/rejected": -2.6296703815460205,
|
|
"step": 66
|
|
},
|
|
{
|
|
"epoch": 0.14320064119690087,
|
|
"grad_norm": 8.171092519186132,
|
|
"learning_rate": 7.955323304900513e-07,
|
|
"logits/chosen": -1.2573140859603882,
|
|
"logits/rejected": -1.2275471687316895,
|
|
"logps/chosen": -609.5222778320312,
|
|
"logps/rejected": -656.034912109375,
|
|
"loss": 0.5297,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -2.932328224182129,
|
|
"rewards/margins": 0.5183333158493042,
|
|
"rewards/rejected": -3.4506616592407227,
|
|
"step": 67
|
|
},
|
|
{
|
|
"epoch": 0.14533796419983971,
|
|
"grad_norm": 8.998419071484543,
|
|
"learning_rate": 7.95075336238055e-07,
|
|
"logits/chosen": -1.0774922370910645,
|
|
"logits/rejected": -1.0322424173355103,
|
|
"logps/chosen": -358.639892578125,
|
|
"logps/rejected": -422.7215576171875,
|
|
"loss": 0.5631,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -1.590600609779358,
|
|
"rewards/margins": 0.5412168502807617,
|
|
"rewards/rejected": -2.13181734085083,
|
|
"step": 68
|
|
},
|
|
{
|
|
"epoch": 0.14747528720277853,
|
|
"grad_norm": 9.57704334804126,
|
|
"learning_rate": 7.945962375694085e-07,
|
|
"logits/chosen": -1.1822316646575928,
|
|
"logits/rejected": -1.061968207359314,
|
|
"logps/chosen": -442.8787536621094,
|
|
"logps/rejected": -535.091552734375,
|
|
"loss": 0.6019,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.9403026103973389,
|
|
"rewards/margins": 0.7558570504188538,
|
|
"rewards/rejected": -2.696159601211548,
|
|
"step": 69
|
|
},
|
|
{
|
|
"epoch": 0.14961261020571734,
|
|
"grad_norm": 8.335354984340409,
|
|
"learning_rate": 7.940950612896237e-07,
|
|
"logits/chosen": -1.1393486261367798,
|
|
"logits/rejected": -1.148930311203003,
|
|
"logps/chosen": -565.1220703125,
|
|
"logps/rejected": -670.720458984375,
|
|
"loss": 0.562,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -2.9813404083251953,
|
|
"rewards/margins": 0.9398919343948364,
|
|
"rewards/rejected": -3.921232223510742,
|
|
"step": 70
|
|
},
|
|
{
|
|
"epoch": 0.15174993320865615,
|
|
"grad_norm": 9.614295261199452,
|
|
"learning_rate": 7.935718354394518e-07,
|
|
"logits/chosen": -1.2340784072875977,
|
|
"logits/rejected": -1.2443894147872925,
|
|
"logps/chosen": -579.1088256835938,
|
|
"logps/rejected": -634.750732421875,
|
|
"loss": 0.5595,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -2.7751572132110596,
|
|
"rewards/margins": 0.6060917973518372,
|
|
"rewards/rejected": -3.381248950958252,
|
|
"step": 71
|
|
},
|
|
{
|
|
"epoch": 0.15388725621159496,
|
|
"grad_norm": 12.368431705912641,
|
|
"learning_rate": 7.930265892933153e-07,
|
|
"logits/chosen": -1.2153799533843994,
|
|
"logits/rejected": -1.24309241771698,
|
|
"logps/chosen": -538.3513793945312,
|
|
"logps/rejected": -561.787353515625,
|
|
"loss": 0.574,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -2.4809069633483887,
|
|
"rewards/margins": 0.3710068166255951,
|
|
"rewards/rejected": -2.8519139289855957,
|
|
"step": 72
|
|
},
|
|
{
|
|
"epoch": 0.1560245792145338,
|
|
"grad_norm": 11.318087514647862,
|
|
"learning_rate": 7.92459353357669e-07,
|
|
"logits/chosen": -1.1397067308425903,
|
|
"logits/rejected": -1.1050128936767578,
|
|
"logps/chosen": -494.8123779296875,
|
|
"logps/rejected": -600.887451171875,
|
|
"loss": 0.607,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.3441858291625977,
|
|
"rewards/margins": 0.8333718776702881,
|
|
"rewards/rejected": -3.177557945251465,
|
|
"step": 73
|
|
},
|
|
{
|
|
"epoch": 0.15816190221747262,
|
|
"grad_norm": 12.66693529075148,
|
|
"learning_rate": 7.918701593692942e-07,
|
|
"logits/chosen": -1.1999236345291138,
|
|
"logits/rejected": -1.1343518495559692,
|
|
"logps/chosen": -537.6787719726562,
|
|
"logps/rejected": -661.78271484375,
|
|
"loss": 0.6214,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.2442550659179688,
|
|
"rewards/margins": 0.8728527426719666,
|
|
"rewards/rejected": -3.117107629776001,
|
|
"step": 74
|
|
},
|
|
{
|
|
"epoch": 0.16029922522041143,
|
|
"grad_norm": 18.744343302034743,
|
|
"learning_rate": 7.912590402935222e-07,
|
|
"logits/chosen": -1.210806965827942,
|
|
"logits/rejected": -1.2487571239471436,
|
|
"logps/chosen": -598.8636474609375,
|
|
"logps/rejected": -607.9540405273438,
|
|
"loss": 0.6434,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -3.0624806880950928,
|
|
"rewards/margins": 0.30628111958503723,
|
|
"rewards/rejected": -3.3687617778778076,
|
|
"step": 75
|
|
},
|
|
{
|
|
"epoch": 0.16243654822335024,
|
|
"grad_norm": 10.387480961194791,
|
|
"learning_rate": 7.906260303223906e-07,
|
|
"logits/chosen": -1.2011109590530396,
|
|
"logits/rejected": -1.1724956035614014,
|
|
"logps/chosen": -523.1990356445312,
|
|
"logps/rejected": -635.5238647460938,
|
|
"loss": 0.5103,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -2.499415636062622,
|
|
"rewards/margins": 0.956606388092041,
|
|
"rewards/rejected": -3.456022024154663,
|
|
"step": 76
|
|
},
|
|
{
|
|
"epoch": 0.16457387122628908,
|
|
"grad_norm": 8.286082639772292,
|
|
"learning_rate": 7.899711648727293e-07,
|
|
"logits/chosen": -1.1222546100616455,
|
|
"logits/rejected": -1.0313266515731812,
|
|
"logps/chosen": -422.1160583496094,
|
|
"logps/rejected": -486.4732971191406,
|
|
"loss": 0.5272,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.8823583126068115,
|
|
"rewards/margins": 0.557906985282898,
|
|
"rewards/rejected": -2.44026517868042,
|
|
"step": 77
|
|
},
|
|
{
|
|
"epoch": 0.1667111942292279,
|
|
"grad_norm": 14.710358539113331,
|
|
"learning_rate": 7.892944805841805e-07,
|
|
"logits/chosen": -1.18240487575531,
|
|
"logits/rejected": -1.1336525678634644,
|
|
"logps/chosen": -457.12652587890625,
|
|
"logps/rejected": -520.3361206054688,
|
|
"loss": 0.5602,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.255852222442627,
|
|
"rewards/margins": 0.5986082553863525,
|
|
"rewards/rejected": -2.8544602394104004,
|
|
"step": 78
|
|
},
|
|
{
|
|
"epoch": 0.1688485172321667,
|
|
"grad_norm": 13.126930480373984,
|
|
"learning_rate": 7.885960153171469e-07,
|
|
"logits/chosen": -1.0567469596862793,
|
|
"logits/rejected": -1.0342040061950684,
|
|
"logps/chosen": -449.7267761230469,
|
|
"logps/rejected": -539.2164306640625,
|
|
"loss": 0.535,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.0118534564971924,
|
|
"rewards/margins": 0.6608402729034424,
|
|
"rewards/rejected": -2.672694206237793,
|
|
"step": 79
|
|
},
|
|
{
|
|
"epoch": 0.17098584023510552,
|
|
"grad_norm": 11.307140941363295,
|
|
"learning_rate": 7.878758081506747e-07,
|
|
"logits/chosen": -1.114970326423645,
|
|
"logits/rejected": -1.1288405656814575,
|
|
"logps/chosen": -453.8138732910156,
|
|
"logps/rejected": -491.1424865722656,
|
|
"loss": 0.5674,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.1642098426818848,
|
|
"rewards/margins": 0.4818393886089325,
|
|
"rewards/rejected": -2.6460490226745605,
|
|
"step": 80
|
|
},
|
|
{
|
|
"epoch": 0.17312316323804436,
|
|
"grad_norm": 11.077741078561617,
|
|
"learning_rate": 7.871338993802666e-07,
|
|
"logits/chosen": -1.127083659172058,
|
|
"logits/rejected": -1.0359755754470825,
|
|
"logps/chosen": -568.3497314453125,
|
|
"logps/rejected": -610.1666259765625,
|
|
"loss": 0.5281,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.4274895191192627,
|
|
"rewards/margins": 0.5119785070419312,
|
|
"rewards/rejected": -2.9394679069519043,
|
|
"step": 81
|
|
},
|
|
{
|
|
"epoch": 0.17526048624098317,
|
|
"grad_norm": 9.584538258710456,
|
|
"learning_rate": 7.863703305156272e-07,
|
|
"logits/chosen": -1.2093491554260254,
|
|
"logits/rejected": -1.2194944620132446,
|
|
"logps/chosen": -477.71746826171875,
|
|
"logps/rejected": -516.8004760742188,
|
|
"loss": 0.5721,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.3491222858428955,
|
|
"rewards/margins": 0.46743592619895935,
|
|
"rewards/rejected": -2.8165578842163086,
|
|
"step": 82
|
|
},
|
|
{
|
|
"epoch": 0.17739780924392198,
|
|
"grad_norm": 9.911683367044649,
|
|
"learning_rate": 7.855851442783412e-07,
|
|
"logits/chosen": -1.185421109199524,
|
|
"logits/rejected": -1.1721405982971191,
|
|
"logps/chosen": -466.0152587890625,
|
|
"logps/rejected": -567.0381469726562,
|
|
"loss": 0.5891,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -1.8479394912719727,
|
|
"rewards/margins": 1.1030150651931763,
|
|
"rewards/rejected": -2.9509544372558594,
|
|
"step": 83
|
|
},
|
|
{
|
|
"epoch": 0.1795351322468608,
|
|
"grad_norm": 9.609331845196555,
|
|
"learning_rate": 7.847783845994824e-07,
|
|
"logits/chosen": -1.2735304832458496,
|
|
"logits/rejected": -1.3312420845031738,
|
|
"logps/chosen": -473.52703857421875,
|
|
"logps/rejected": -493.1807861328125,
|
|
"loss": 0.554,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.2089741230010986,
|
|
"rewards/margins": 0.2816333770751953,
|
|
"rewards/rejected": -2.490607500076294,
|
|
"step": 84
|
|
},
|
|
{
|
|
"epoch": 0.18167245524979964,
|
|
"grad_norm": 9.131968723887285,
|
|
"learning_rate": 7.839500966171556e-07,
|
|
"logits/chosen": -1.2005490064620972,
|
|
"logits/rejected": -1.029706358909607,
|
|
"logps/chosen": -393.1463928222656,
|
|
"logps/rejected": -525.7657470703125,
|
|
"loss": 0.5741,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -1.82315993309021,
|
|
"rewards/margins": 1.128662109375,
|
|
"rewards/rejected": -2.95182204246521,
|
|
"step": 85
|
|
},
|
|
{
|
|
"epoch": 0.18380977825273845,
|
|
"grad_norm": 9.252010710109433,
|
|
"learning_rate": 7.831003266739721e-07,
|
|
"logits/chosen": -1.2110841274261475,
|
|
"logits/rejected": -1.1710755825042725,
|
|
"logps/chosen": -540.1366577148438,
|
|
"logps/rejected": -684.7119140625,
|
|
"loss": 0.5182,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -2.6053380966186523,
|
|
"rewards/margins": 1.2827905416488647,
|
|
"rewards/rejected": -3.8881282806396484,
|
|
"step": 86
|
|
},
|
|
{
|
|
"epoch": 0.18594710125567726,
|
|
"grad_norm": 9.076549815928802,
|
|
"learning_rate": 7.822291223144563e-07,
|
|
"logits/chosen": -1.0235605239868164,
|
|
"logits/rejected": -0.9067903757095337,
|
|
"logps/chosen": -615.156982421875,
|
|
"logps/rejected": -752.937255859375,
|
|
"loss": 0.5167,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.8930633068084717,
|
|
"rewards/margins": 1.34428071975708,
|
|
"rewards/rejected": -4.237343788146973,
|
|
"step": 87
|
|
},
|
|
{
|
|
"epoch": 0.18808442425861607,
|
|
"grad_norm": 11.373953224514699,
|
|
"learning_rate": 7.81336532282385e-07,
|
|
"logits/chosen": -1.1721646785736084,
|
|
"logits/rejected": -1.0930900573730469,
|
|
"logps/chosen": -280.4423828125,
|
|
"logps/rejected": -336.436279296875,
|
|
"loss": 0.597,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -1.236398458480835,
|
|
"rewards/margins": 0.5181585550308228,
|
|
"rewards/rejected": -1.7545570135116577,
|
|
"step": 88
|
|
},
|
|
{
|
|
"epoch": 0.1902217472615549,
|
|
"grad_norm": 17.85017844805296,
|
|
"learning_rate": 7.804226065180615e-07,
|
|
"logits/chosen": -1.1734278202056885,
|
|
"logits/rejected": -1.1101915836334229,
|
|
"logps/chosen": -382.6453552246094,
|
|
"logps/rejected": -466.7173156738281,
|
|
"loss": 0.5902,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.7278035879135132,
|
|
"rewards/margins": 0.6392509937286377,
|
|
"rewards/rejected": -2.3670544624328613,
|
|
"step": 89
|
|
},
|
|
{
|
|
"epoch": 0.19235907026449373,
|
|
"grad_norm": 12.773370813435598,
|
|
"learning_rate": 7.794873961555198e-07,
|
|
"logits/chosen": -1.1437866687774658,
|
|
"logits/rejected": -1.1589542627334595,
|
|
"logps/chosen": -544.008544921875,
|
|
"logps/rejected": -615.4200439453125,
|
|
"loss": 0.5274,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -2.523742198944092,
|
|
"rewards/margins": 0.7184704542160034,
|
|
"rewards/rejected": -3.2422127723693848,
|
|
"step": 90
|
|
},
|
|
{
|
|
"epoch": 0.19449639326743254,
|
|
"grad_norm": 9.908778922785883,
|
|
"learning_rate": 7.785309535196656e-07,
|
|
"logits/chosen": -1.1318367719650269,
|
|
"logits/rejected": -1.089951992034912,
|
|
"logps/chosen": -529.7106323242188,
|
|
"logps/rejected": -568.3753051757812,
|
|
"loss": 0.528,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -2.680912494659424,
|
|
"rewards/margins": 0.3908342123031616,
|
|
"rewards/rejected": -3.071746826171875,
|
|
"step": 91
|
|
},
|
|
{
|
|
"epoch": 0.19663371627037135,
|
|
"grad_norm": 10.509283821989197,
|
|
"learning_rate": 7.77553332123347e-07,
|
|
"logits/chosen": -1.2806425094604492,
|
|
"logits/rejected": -1.2847963571548462,
|
|
"logps/chosen": -492.8918762207031,
|
|
"logps/rejected": -563.6588134765625,
|
|
"loss": 0.537,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.380342483520508,
|
|
"rewards/margins": 0.6923278570175171,
|
|
"rewards/rejected": -3.0726704597473145,
|
|
"step": 92
|
|
},
|
|
{
|
|
"epoch": 0.1987710392733102,
|
|
"grad_norm": 9.899922277845048,
|
|
"learning_rate": 7.765545866643613e-07,
|
|
"logits/chosen": -1.0114997625350952,
|
|
"logits/rejected": -1.0235319137573242,
|
|
"logps/chosen": -381.5235290527344,
|
|
"logps/rejected": -387.6109619140625,
|
|
"loss": 0.5506,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -1.6469519138336182,
|
|
"rewards/margins": 0.23874562978744507,
|
|
"rewards/rejected": -1.8856974840164185,
|
|
"step": 93
|
|
},
|
|
{
|
|
"epoch": 0.200908362276249,
|
|
"grad_norm": 11.043390646196134,
|
|
"learning_rate": 7.755347730223943e-07,
|
|
"logits/chosen": -1.1114813089370728,
|
|
"logits/rejected": -1.1337555646896362,
|
|
"logps/chosen": -550.198486328125,
|
|
"logps/rejected": -619.2704467773438,
|
|
"loss": 0.553,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.6842494010925293,
|
|
"rewards/margins": 0.6715414524078369,
|
|
"rewards/rejected": -3.355790615081787,
|
|
"step": 94
|
|
},
|
|
{
|
|
"epoch": 0.20304568527918782,
|
|
"grad_norm": 9.638994670369192,
|
|
"learning_rate": 7.744939482558948e-07,
|
|
"logits/chosen": -1.1464406251907349,
|
|
"logits/rejected": -1.1081212759017944,
|
|
"logps/chosen": -422.6688537597656,
|
|
"logps/rejected": -483.55584716796875,
|
|
"loss": 0.5182,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -2.2197909355163574,
|
|
"rewards/margins": 0.568549633026123,
|
|
"rewards/rejected": -2.7883403301239014,
|
|
"step": 95
|
|
},
|
|
{
|
|
"epoch": 0.20518300828212663,
|
|
"grad_norm": 10.66227269593484,
|
|
"learning_rate": 7.734321705988807e-07,
|
|
"logits/chosen": -1.2760132551193237,
|
|
"logits/rejected": -1.2601702213287354,
|
|
"logps/chosen": -515.2978515625,
|
|
"logps/rejected": -529.2376098632812,
|
|
"loss": 0.5024,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -1.9085440635681152,
|
|
"rewards/margins": 0.5009690523147583,
|
|
"rewards/rejected": -2.409513235092163,
|
|
"step": 96
|
|
},
|
|
{
|
|
"epoch": 0.20732033128506547,
|
|
"grad_norm": 11.057889862172901,
|
|
"learning_rate": 7.723494994576816e-07,
|
|
"logits/chosen": -1.1963937282562256,
|
|
"logits/rejected": -1.125280499458313,
|
|
"logps/chosen": -494.753662109375,
|
|
"logps/rejected": -545.3011474609375,
|
|
"loss": 0.5381,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.26602840423584,
|
|
"rewards/margins": 0.47012466192245483,
|
|
"rewards/rejected": -2.7361528873443604,
|
|
"step": 97
|
|
},
|
|
{
|
|
"epoch": 0.20945765428800428,
|
|
"grad_norm": 11.127088588048807,
|
|
"learning_rate": 7.712459954076156e-07,
|
|
"logits/chosen": -1.0651479959487915,
|
|
"logits/rejected": -1.1355326175689697,
|
|
"logps/chosen": -647.6420288085938,
|
|
"logps/rejected": -641.8303833007812,
|
|
"loss": 0.5441,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.322498083114624,
|
|
"rewards/margins": 0.05157832056283951,
|
|
"rewards/rejected": -3.3740763664245605,
|
|
"step": 98
|
|
},
|
|
{
|
|
"epoch": 0.2115949772909431,
|
|
"grad_norm": 11.207670370998112,
|
|
"learning_rate": 7.701217201895986e-07,
|
|
"logits/chosen": -1.052258014678955,
|
|
"logits/rejected": -1.1536017656326294,
|
|
"logps/chosen": -514.69287109375,
|
|
"logps/rejected": -530.2621459960938,
|
|
"loss": 0.5424,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -2.5572896003723145,
|
|
"rewards/margins": 0.346367210149765,
|
|
"rewards/rejected": -2.9036567211151123,
|
|
"step": 99
|
|
},
|
|
{
|
|
"epoch": 0.2137323002938819,
|
|
"grad_norm": 10.113988708766959,
|
|
"learning_rate": 7.689767367066917e-07,
|
|
"logits/chosen": -0.8341108560562134,
|
|
"logits/rejected": -0.8912850618362427,
|
|
"logps/chosen": -377.82342529296875,
|
|
"logps/rejected": -454.91064453125,
|
|
"loss": 0.5793,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -1.8863927125930786,
|
|
"rewards/margins": 0.5190551280975342,
|
|
"rewards/rejected": -2.4054477214813232,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.2137323002938819,
|
|
"eval_logits/chosen": -1.2699782848358154,
|
|
"eval_logits/rejected": -1.2844398021697998,
|
|
"eval_logps/chosen": -544.71923828125,
|
|
"eval_logps/rejected": -616.7446899414062,
|
|
"eval_loss": 0.5182811617851257,
|
|
"eval_rewards/accuracies": 0.7459677457809448,
|
|
"eval_rewards/chosen": -2.652132511138916,
|
|
"eval_rewards/margins": 0.7797600030899048,
|
|
"eval_rewards/rejected": -3.4318928718566895,
|
|
"eval_runtime": 142.3104,
|
|
"eval_samples_per_second": 13.78,
|
|
"eval_steps_per_second": 0.436,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.21586962329682075,
|
|
"grad_norm": 16.391709000284067,
|
|
"learning_rate": 7.678111090205803e-07,
|
|
"logits/chosen": -1.2273062467575073,
|
|
"logits/rejected": -1.2417653799057007,
|
|
"logps/chosen": -559.35888671875,
|
|
"logps/rejected": -655.1536865234375,
|
|
"loss": 0.6137,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.6603152751922607,
|
|
"rewards/margins": 0.9901705384254456,
|
|
"rewards/rejected": -3.6504859924316406,
|
|
"step": 101
|
|
},
|
|
{
|
|
"epoch": 0.21800694629975956,
|
|
"grad_norm": 10.524629928454354,
|
|
"learning_rate": 7.666249023479905e-07,
|
|
"logits/chosen": -1.253726601600647,
|
|
"logits/rejected": -1.0882092714309692,
|
|
"logps/chosen": -479.20758056640625,
|
|
"logps/rejected": -595.8485107421875,
|
|
"loss": 0.4448,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.2103652954101562,
|
|
"rewards/margins": 0.9676500558853149,
|
|
"rewards/rejected": -3.1780154705047607,
|
|
"step": 102
|
|
},
|
|
{
|
|
"epoch": 0.22014426930269837,
|
|
"grad_norm": 9.72211474274192,
|
|
"learning_rate": 7.654181830570403e-07,
|
|
"logits/chosen": -1.0320769548416138,
|
|
"logits/rejected": -0.9883528351783752,
|
|
"logps/chosen": -515.2326049804688,
|
|
"logps/rejected": -605.5062255859375,
|
|
"loss": 0.4677,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.22302508354187,
|
|
"rewards/margins": 0.9804313778877258,
|
|
"rewards/rejected": -3.203456401824951,
|
|
"step": 103
|
|
},
|
|
{
|
|
"epoch": 0.22228159230563718,
|
|
"grad_norm": 10.463756107018286,
|
|
"learning_rate": 7.641910186635264e-07,
|
|
"logits/chosen": -1.2872788906097412,
|
|
"logits/rejected": -1.2068657875061035,
|
|
"logps/chosen": -533.40576171875,
|
|
"logps/rejected": -672.7639770507812,
|
|
"loss": 0.4858,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.8758130073547363,
|
|
"rewards/margins": 1.288372278213501,
|
|
"rewards/rejected": -4.164185523986816,
|
|
"step": 104
|
|
},
|
|
{
|
|
"epoch": 0.224418915308576,
|
|
"grad_norm": 9.94692219065209,
|
|
"learning_rate": 7.629434778271459e-07,
|
|
"logits/chosen": -1.2782255411148071,
|
|
"logits/rejected": -1.2008510828018188,
|
|
"logps/chosen": -422.54052734375,
|
|
"logps/rejected": -536.9180297851562,
|
|
"loss": 0.5088,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -2.0897161960601807,
|
|
"rewards/margins": 0.9998084306716919,
|
|
"rewards/rejected": -3.089524269104004,
|
|
"step": 105
|
|
},
|
|
{
|
|
"epoch": 0.22655623831151483,
|
|
"grad_norm": 15.519123908773544,
|
|
"learning_rate": 7.616756303476559e-07,
|
|
"logits/chosen": -1.2410894632339478,
|
|
"logits/rejected": -1.2540771961212158,
|
|
"logps/chosen": -524.3763427734375,
|
|
"logps/rejected": -598.2949829101562,
|
|
"loss": 0.6234,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -2.4764623641967773,
|
|
"rewards/margins": 0.8137304186820984,
|
|
"rewards/rejected": -3.2901928424835205,
|
|
"step": 106
|
|
},
|
|
{
|
|
"epoch": 0.22869356131445365,
|
|
"grad_norm": 10.169089359472466,
|
|
"learning_rate": 7.603875471609676e-07,
|
|
"logits/chosen": -1.0019772052764893,
|
|
"logits/rejected": -0.9863028526306152,
|
|
"logps/chosen": -614.7957763671875,
|
|
"logps/rejected": -719.146728515625,
|
|
"loss": 0.5029,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.299185276031494,
|
|
"rewards/margins": 0.8879890441894531,
|
|
"rewards/rejected": -4.187174320220947,
|
|
"step": 107
|
|
},
|
|
{
|
|
"epoch": 0.23083088431739246,
|
|
"grad_norm": 10.471849086193684,
|
|
"learning_rate": 7.590793003351773e-07,
|
|
"logits/chosen": -1.132460355758667,
|
|
"logits/rejected": -1.115264654159546,
|
|
"logps/chosen": -640.571044921875,
|
|
"logps/rejected": -740.5921020507812,
|
|
"loss": 0.4963,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.463143825531006,
|
|
"rewards/margins": 0.9680495858192444,
|
|
"rewards/rejected": -4.431193828582764,
|
|
"step": 108
|
|
},
|
|
{
|
|
"epoch": 0.23296820732033127,
|
|
"grad_norm": 11.823001541225018,
|
|
"learning_rate": 7.577509630665347e-07,
|
|
"logits/chosen": -1.1503872871398926,
|
|
"logits/rejected": -1.0826492309570312,
|
|
"logps/chosen": -555.7222290039062,
|
|
"logps/rejected": -684.6669311523438,
|
|
"loss": 0.5508,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.875490427017212,
|
|
"rewards/margins": 1.2122423648834229,
|
|
"rewards/rejected": -4.087732315063477,
|
|
"step": 109
|
|
},
|
|
{
|
|
"epoch": 0.2351055303232701,
|
|
"grad_norm": 9.86238586725783,
|
|
"learning_rate": 7.564026096753471e-07,
|
|
"logits/chosen": -1.2278306484222412,
|
|
"logits/rejected": -1.140399694442749,
|
|
"logps/chosen": -543.5636596679688,
|
|
"logps/rejected": -715.5054321289062,
|
|
"loss": 0.4682,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.983625888824463,
|
|
"rewards/margins": 1.4977326393127441,
|
|
"rewards/rejected": -4.481358528137207,
|
|
"step": 110
|
|
},
|
|
{
|
|
"epoch": 0.23724285332620892,
|
|
"grad_norm": 9.440230861402858,
|
|
"learning_rate": 7.550343156018216e-07,
|
|
"logits/chosen": -1.2607563734054565,
|
|
"logits/rejected": -1.253735065460205,
|
|
"logps/chosen": -500.310546875,
|
|
"logps/rejected": -607.3995971679688,
|
|
"loss": 0.5163,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.420441150665283,
|
|
"rewards/margins": 1.0024170875549316,
|
|
"rewards/rejected": -3.422858238220215,
|
|
"step": 111
|
|
},
|
|
{
|
|
"epoch": 0.23938017632914774,
|
|
"grad_norm": 12.215223201725872,
|
|
"learning_rate": 7.536461574018439e-07,
|
|
"logits/chosen": -1.108474850654602,
|
|
"logits/rejected": -1.146998643875122,
|
|
"logps/chosen": -474.21197509765625,
|
|
"logps/rejected": -562.3438110351562,
|
|
"loss": 0.5669,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -2.460186243057251,
|
|
"rewards/margins": 0.8443668484687805,
|
|
"rewards/rejected": -3.3045530319213867,
|
|
"step": 112
|
|
},
|
|
{
|
|
"epoch": 0.24151749933208655,
|
|
"grad_norm": 12.559970980315386,
|
|
"learning_rate": 7.522382127426952e-07,
|
|
"logits/chosen": -1.2318087816238403,
|
|
"logits/rejected": -1.1735649108886719,
|
|
"logps/chosen": -432.732421875,
|
|
"logps/rejected": -507.0394287109375,
|
|
"loss": 0.4479,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -2.251821756362915,
|
|
"rewards/margins": 0.5928990840911865,
|
|
"rewards/rejected": -2.8447208404541016,
|
|
"step": 113
|
|
},
|
|
{
|
|
"epoch": 0.2436548223350254,
|
|
"grad_norm": 13.07315315212208,
|
|
"learning_rate": 7.508105603987066e-07,
|
|
"logits/chosen": -1.0219734907150269,
|
|
"logits/rejected": -0.9402412176132202,
|
|
"logps/chosen": -579.5034790039062,
|
|
"logps/rejected": -664.8965454101562,
|
|
"loss": 0.5591,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -2.9037399291992188,
|
|
"rewards/margins": 0.7655159831047058,
|
|
"rewards/rejected": -3.6692559719085693,
|
|
"step": 114
|
|
},
|
|
{
|
|
"epoch": 0.2457921453379642,
|
|
"grad_norm": 10.646810060304825,
|
|
"learning_rate": 7.493632802468518e-07,
|
|
"logits/chosen": -1.1602777242660522,
|
|
"logits/rejected": -1.0831139087677002,
|
|
"logps/chosen": -389.80419921875,
|
|
"logps/rejected": -458.3884582519531,
|
|
"loss": 0.5419,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -1.8653407096862793,
|
|
"rewards/margins": 0.5108134746551514,
|
|
"rewards/rejected": -2.3761539459228516,
|
|
"step": 115
|
|
},
|
|
{
|
|
"epoch": 0.24792946834090301,
|
|
"grad_norm": 10.71566783907731,
|
|
"learning_rate": 7.478964532622782e-07,
|
|
"logits/chosen": -1.1830374002456665,
|
|
"logits/rejected": -1.1820306777954102,
|
|
"logps/chosen": -591.6640625,
|
|
"logps/rejected": -658.7415771484375,
|
|
"loss": 0.4764,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.90578031539917,
|
|
"rewards/margins": 0.7861341238021851,
|
|
"rewards/rejected": -3.6919145584106445,
|
|
"step": 116
|
|
},
|
|
{
|
|
"epoch": 0.25006679134384185,
|
|
"grad_norm": 11.694350723954798,
|
|
"learning_rate": 7.464101615137755e-07,
|
|
"logits/chosen": -1.0726900100708008,
|
|
"logits/rejected": -1.1676019430160522,
|
|
"logps/chosen": -554.7371215820312,
|
|
"logps/rejected": -629.0760498046875,
|
|
"loss": 0.534,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -2.7668631076812744,
|
|
"rewards/margins": 0.9604626297950745,
|
|
"rewards/rejected": -3.727325916290283,
|
|
"step": 117
|
|
},
|
|
{
|
|
"epoch": 0.25220411434678064,
|
|
"grad_norm": 11.734514848455067,
|
|
"learning_rate": 7.44904488159185e-07,
|
|
"logits/chosen": -1.0949318408966064,
|
|
"logits/rejected": -1.0768948793411255,
|
|
"logps/chosen": -397.58837890625,
|
|
"logps/rejected": -468.90264892578125,
|
|
"loss": 0.4996,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -2.029899835586548,
|
|
"rewards/margins": 0.6923526525497437,
|
|
"rewards/rejected": -2.722252607345581,
|
|
"step": 118
|
|
},
|
|
{
|
|
"epoch": 0.2543414373497195,
|
|
"grad_norm": 10.42794481881483,
|
|
"learning_rate": 7.433795174407464e-07,
|
|
"logits/chosen": -1.266892910003662,
|
|
"logits/rejected": -1.1528326272964478,
|
|
"logps/chosen": -570.2450561523438,
|
|
"logps/rejected": -709.3895263671875,
|
|
"loss": 0.5043,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -2.995210647583008,
|
|
"rewards/margins": 1.1049041748046875,
|
|
"rewards/rejected": -4.100114822387695,
|
|
"step": 119
|
|
},
|
|
{
|
|
"epoch": 0.2564787603526583,
|
|
"grad_norm": 10.929127899973288,
|
|
"learning_rate": 7.418353346803846e-07,
|
|
"logits/chosen": -1.0802345275878906,
|
|
"logits/rejected": -1.1326872110366821,
|
|
"logps/chosen": -526.001708984375,
|
|
"logps/rejected": -621.600341796875,
|
|
"loss": 0.4601,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -2.6054108142852783,
|
|
"rewards/margins": 1.0312682390213013,
|
|
"rewards/rejected": -3.636679172515869,
|
|
"step": 120
|
|
},
|
|
{
|
|
"epoch": 0.2586160833555971,
|
|
"grad_norm": 11.494420054527998,
|
|
"learning_rate": 7.402720262749359e-07,
|
|
"logits/chosen": -1.2067303657531738,
|
|
"logits/rejected": -1.2558921575546265,
|
|
"logps/chosen": -564.7210693359375,
|
|
"logps/rejected": -711.4730224609375,
|
|
"loss": 0.498,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -2.781289577484131,
|
|
"rewards/margins": 1.5313748121261597,
|
|
"rewards/rejected": -4.31266450881958,
|
|
"step": 121
|
|
},
|
|
{
|
|
"epoch": 0.26075340635853594,
|
|
"grad_norm": 11.148239249520412,
|
|
"learning_rate": 7.386896796913136e-07,
|
|
"logits/chosen": -1.1556284427642822,
|
|
"logits/rejected": -1.01291823387146,
|
|
"logps/chosen": -429.47967529296875,
|
|
"logps/rejected": -533.7469482421875,
|
|
"loss": 0.4546,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -2.000453233718872,
|
|
"rewards/margins": 0.8496807813644409,
|
|
"rewards/rejected": -2.8501341342926025,
|
|
"step": 122
|
|
},
|
|
{
|
|
"epoch": 0.26289072936147473,
|
|
"grad_norm": 11.713757273838597,
|
|
"learning_rate": 7.370883834616156e-07,
|
|
"logits/chosen": -1.1531327962875366,
|
|
"logits/rejected": -1.1401838064193726,
|
|
"logps/chosen": -551.8589477539062,
|
|
"logps/rejected": -632.5919799804688,
|
|
"loss": 0.5588,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -2.5737900733947754,
|
|
"rewards/margins": 0.8271616101264954,
|
|
"rewards/rejected": -3.400951623916626,
|
|
"step": 123
|
|
},
|
|
{
|
|
"epoch": 0.26502805236441357,
|
|
"grad_norm": 10.989454375451436,
|
|
"learning_rate": 7.354682271781696e-07,
|
|
"logits/chosen": -1.1755902767181396,
|
|
"logits/rejected": -1.1291191577911377,
|
|
"logps/chosen": -571.854736328125,
|
|
"logps/rejected": -655.4735107421875,
|
|
"loss": 0.5171,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -2.706859827041626,
|
|
"rewards/margins": 0.9184459447860718,
|
|
"rewards/rejected": -3.625305652618408,
|
|
"step": 124
|
|
},
|
|
{
|
|
"epoch": 0.2671653753673524,
|
|
"grad_norm": 9.756298897089394,
|
|
"learning_rate": 7.33829301488521e-07,
|
|
"logits/chosen": -1.1387619972229004,
|
|
"logits/rejected": -1.1437007188796997,
|
|
"logps/chosen": -448.17193603515625,
|
|
"logps/rejected": -520.0236206054688,
|
|
"loss": 0.4732,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.275339126586914,
|
|
"rewards/margins": 0.6221529841423035,
|
|
"rewards/rejected": -2.8974921703338623,
|
|
"step": 125
|
|
},
|
|
{
|
|
"epoch": 0.2693026983702912,
|
|
"grad_norm": 11.552300951274553,
|
|
"learning_rate": 7.321716980903615e-07,
|
|
"logits/chosen": -1.2335395812988281,
|
|
"logits/rejected": -1.2084323167800903,
|
|
"logps/chosen": -494.1284484863281,
|
|
"logps/rejected": -644.2607421875,
|
|
"loss": 0.5453,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.216590404510498,
|
|
"rewards/margins": 1.4140255451202393,
|
|
"rewards/rejected": -3.630615711212158,
|
|
"step": 126
|
|
},
|
|
{
|
|
"epoch": 0.27144002137323003,
|
|
"grad_norm": 14.030573939890216,
|
|
"learning_rate": 7.304955097263979e-07,
|
|
"logits/chosen": -1.0867464542388916,
|
|
"logits/rejected": -1.1234714984893799,
|
|
"logps/chosen": -524.39453125,
|
|
"logps/rejected": -586.8427734375,
|
|
"loss": 0.5351,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -2.543902635574341,
|
|
"rewards/margins": 0.9019243717193604,
|
|
"rewards/rejected": -3.445826530456543,
|
|
"step": 127
|
|
},
|
|
{
|
|
"epoch": 0.2735773443761689,
|
|
"grad_norm": 11.313049836006805,
|
|
"learning_rate": 7.288008301791639e-07,
|
|
"logits/chosen": -1.1349008083343506,
|
|
"logits/rejected": -1.1048234701156616,
|
|
"logps/chosen": -461.23516845703125,
|
|
"logps/rejected": -517.0540771484375,
|
|
"loss": 0.5452,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.255331039428711,
|
|
"rewards/margins": 0.661991536617279,
|
|
"rewards/rejected": -2.9173226356506348,
|
|
"step": 128
|
|
},
|
|
{
|
|
"epoch": 0.27571466737910766,
|
|
"grad_norm": 10.703951432530653,
|
|
"learning_rate": 7.270877542657725e-07,
|
|
"logits/chosen": -1.1094213724136353,
|
|
"logits/rejected": -1.1715359687805176,
|
|
"logps/chosen": -489.816650390625,
|
|
"logps/rejected": -586.4013671875,
|
|
"loss": 0.4797,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.231893301010132,
|
|
"rewards/margins": 1.1094894409179688,
|
|
"rewards/rejected": -3.3413827419281006,
|
|
"step": 129
|
|
},
|
|
{
|
|
"epoch": 0.2778519903820465,
|
|
"grad_norm": 10.775339737914736,
|
|
"learning_rate": 7.253563778326111e-07,
|
|
"logits/chosen": -1.2365562915802002,
|
|
"logits/rejected": -1.1621637344360352,
|
|
"logps/chosen": -537.2908325195312,
|
|
"logps/rejected": -638.2408447265625,
|
|
"loss": 0.5017,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -2.384437322616577,
|
|
"rewards/margins": 1.190860629081726,
|
|
"rewards/rejected": -3.5752978324890137,
|
|
"step": 130
|
|
},
|
|
{
|
|
"epoch": 0.2799893133849853,
|
|
"grad_norm": 11.015612451823644,
|
|
"learning_rate": 7.23606797749979e-07,
|
|
"logits/chosen": -1.0056509971618652,
|
|
"logits/rejected": -1.0789600610733032,
|
|
"logps/chosen": -605.8868408203125,
|
|
"logps/rejected": -681.7349243164062,
|
|
"loss": 0.4895,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.0001766681671143,
|
|
"rewards/margins": 0.875503659248352,
|
|
"rewards/rejected": -3.875680685043335,
|
|
"step": 131
|
|
},
|
|
{
|
|
"epoch": 0.2821266363879241,
|
|
"grad_norm": 14.664837385616957,
|
|
"learning_rate": 7.218391119066673e-07,
|
|
"logits/chosen": -1.0803807973861694,
|
|
"logits/rejected": -1.0349315404891968,
|
|
"logps/chosen": -613.3341064453125,
|
|
"logps/rejected": -670.0115356445312,
|
|
"loss": 0.5508,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -3.3352606296539307,
|
|
"rewards/margins": 0.5032040476799011,
|
|
"rewards/rejected": -3.8384647369384766,
|
|
"step": 132
|
|
},
|
|
{
|
|
"epoch": 0.28426395939086296,
|
|
"grad_norm": 11.315998787147386,
|
|
"learning_rate": 7.200534192044825e-07,
|
|
"logits/chosen": -1.0804975032806396,
|
|
"logits/rejected": -0.9459331035614014,
|
|
"logps/chosen": -410.5482482910156,
|
|
"logps/rejected": -552.6442260742188,
|
|
"loss": 0.5149,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -1.9860866069793701,
|
|
"rewards/margins": 1.2312902212142944,
|
|
"rewards/rejected": -3.217376708984375,
|
|
"step": 133
|
|
},
|
|
{
|
|
"epoch": 0.28640128239380175,
|
|
"grad_norm": 11.906263072694715,
|
|
"learning_rate": 7.182498195527123e-07,
|
|
"logits/chosen": -0.9738950133323669,
|
|
"logits/rejected": -1.0062488317489624,
|
|
"logps/chosen": -404.84100341796875,
|
|
"logps/rejected": -420.71380615234375,
|
|
"loss": 0.4926,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -2.204134464263916,
|
|
"rewards/margins": 0.22545769810676575,
|
|
"rewards/rejected": -2.4295921325683594,
|
|
"step": 134
|
|
},
|
|
{
|
|
"epoch": 0.2885386053967406,
|
|
"grad_norm": 10.372611012110845,
|
|
"learning_rate": 7.164284138625366e-07,
|
|
"logits/chosen": -1.0062649250030518,
|
|
"logits/rejected": -0.8770849704742432,
|
|
"logps/chosen": -423.56182861328125,
|
|
"logps/rejected": -553.545166015625,
|
|
"loss": 0.4668,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.139157295227051,
|
|
"rewards/margins": 1.1258550882339478,
|
|
"rewards/rejected": -3.265012264251709,
|
|
"step": 135
|
|
},
|
|
{
|
|
"epoch": 0.29067592839967943,
|
|
"grad_norm": 12.207304324310757,
|
|
"learning_rate": 7.1458930404138e-07,
|
|
"logits/chosen": -1.084909439086914,
|
|
"logits/rejected": -1.0752151012420654,
|
|
"logps/chosen": -564.194091796875,
|
|
"logps/rejected": -667.5319213867188,
|
|
"loss": 0.4906,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.760105848312378,
|
|
"rewards/margins": 1.005811333656311,
|
|
"rewards/rejected": -3.7659173011779785,
|
|
"step": 136
|
|
},
|
|
{
|
|
"epoch": 0.2928132514026182,
|
|
"grad_norm": 13.61174460411604,
|
|
"learning_rate": 7.12732592987212e-07,
|
|
"logits/chosen": -1.1829478740692139,
|
|
"logits/rejected": -1.0823060274124146,
|
|
"logps/chosen": -590.635009765625,
|
|
"logps/rejected": -802.473388671875,
|
|
"loss": 0.5161,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -2.7893006801605225,
|
|
"rewards/margins": 1.989109754562378,
|
|
"rewards/rejected": -4.7784104347229,
|
|
"step": 137
|
|
},
|
|
{
|
|
"epoch": 0.29495057440555705,
|
|
"grad_norm": 10.128959641227988,
|
|
"learning_rate": 7.108583845827883e-07,
|
|
"logits/chosen": -1.1610594987869263,
|
|
"logits/rejected": -1.1704530715942383,
|
|
"logps/chosen": -387.8962097167969,
|
|
"logps/rejected": -480.23797607421875,
|
|
"loss": 0.4904,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -1.9382271766662598,
|
|
"rewards/margins": 0.8355373740196228,
|
|
"rewards/rejected": -2.7737643718719482,
|
|
"step": 138
|
|
},
|
|
{
|
|
"epoch": 0.29708789740849584,
|
|
"grad_norm": 10.050177230914596,
|
|
"learning_rate": 7.089667836898397e-07,
|
|
"logits/chosen": -1.114013910293579,
|
|
"logits/rejected": -1.110438585281372,
|
|
"logps/chosen": -525.3073120117188,
|
|
"logps/rejected": -606.350830078125,
|
|
"loss": 0.4506,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -2.643073320388794,
|
|
"rewards/margins": 0.7284079790115356,
|
|
"rewards/rejected": -3.371481418609619,
|
|
"step": 139
|
|
},
|
|
{
|
|
"epoch": 0.2992252204114347,
|
|
"grad_norm": 16.443121928276934,
|
|
"learning_rate": 7.070578961432044e-07,
|
|
"logits/chosen": -1.006872296333313,
|
|
"logits/rejected": -0.9997127652168274,
|
|
"logps/chosen": -532.4028930664062,
|
|
"logps/rejected": -579.0374755859375,
|
|
"loss": 0.5552,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -2.788207769393921,
|
|
"rewards/margins": 0.6314295530319214,
|
|
"rewards/rejected": -3.4196369647979736,
|
|
"step": 140
|
|
},
|
|
{
|
|
"epoch": 0.3013625434143735,
|
|
"grad_norm": 10.5536510889646,
|
|
"learning_rate": 7.051318287449066e-07,
|
|
"logits/chosen": -1.0748560428619385,
|
|
"logits/rejected": -1.0380382537841797,
|
|
"logps/chosen": -439.17962646484375,
|
|
"logps/rejected": -555.1817626953125,
|
|
"loss": 0.5109,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.2144086360931396,
|
|
"rewards/margins": 1.0603525638580322,
|
|
"rewards/rejected": -3.274761199951172,
|
|
"step": 141
|
|
},
|
|
{
|
|
"epoch": 0.3034998664173123,
|
|
"grad_norm": 12.492156055956597,
|
|
"learning_rate": 7.031886892581812e-07,
|
|
"logits/chosen": -1.0383498668670654,
|
|
"logits/rejected": -1.008802056312561,
|
|
"logps/chosen": -501.4347229003906,
|
|
"logps/rejected": -582.19189453125,
|
|
"loss": 0.5302,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.4712347984313965,
|
|
"rewards/margins": 0.7897406816482544,
|
|
"rewards/rejected": -3.2609758377075195,
|
|
"step": 142
|
|
},
|
|
{
|
|
"epoch": 0.30563718942025114,
|
|
"grad_norm": 14.048982773810426,
|
|
"learning_rate": 7.012285864014444e-07,
|
|
"logits/chosen": -1.0521856546401978,
|
|
"logits/rejected": -1.0056830644607544,
|
|
"logps/chosen": -521.499755859375,
|
|
"logps/rejected": -605.831787109375,
|
|
"loss": 0.5057,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.7067787647247314,
|
|
"rewards/margins": 0.7501015663146973,
|
|
"rewards/rejected": -3.4568800926208496,
|
|
"step": 143
|
|
},
|
|
{
|
|
"epoch": 0.3077745124231899,
|
|
"grad_norm": 10.420316980175429,
|
|
"learning_rate": 6.992516298422107e-07,
|
|
"logits/chosen": -1.0325193405151367,
|
|
"logits/rejected": -1.072160005569458,
|
|
"logps/chosen": -711.7697143554688,
|
|
"logps/rejected": -853.23681640625,
|
|
"loss": 0.426,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.530585289001465,
|
|
"rewards/margins": 1.588570237159729,
|
|
"rewards/rejected": -5.119154930114746,
|
|
"step": 144
|
|
},
|
|
{
|
|
"epoch": 0.30991183542612877,
|
|
"grad_norm": 20.310072106279815,
|
|
"learning_rate": 6.972579301909577e-07,
|
|
"logits/chosen": -1.07466459274292,
|
|
"logits/rejected": -1.0684446096420288,
|
|
"logps/chosen": -638.5875244140625,
|
|
"logps/rejected": -820.4556884765625,
|
|
"loss": 0.4766,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.3279354572296143,
|
|
"rewards/margins": 1.7255775928497314,
|
|
"rewards/rejected": -5.0535125732421875,
|
|
"step": 145
|
|
},
|
|
{
|
|
"epoch": 0.3120491584290676,
|
|
"grad_norm": 12.2030303667386,
|
|
"learning_rate": 6.952475989949362e-07,
|
|
"logits/chosen": -1.0073144435882568,
|
|
"logits/rejected": -1.0062808990478516,
|
|
"logps/chosen": -554.0878295898438,
|
|
"logps/rejected": -649.923828125,
|
|
"loss": 0.5041,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.752135992050171,
|
|
"rewards/margins": 1.0968319177627563,
|
|
"rewards/rejected": -3.848968029022217,
|
|
"step": 146
|
|
},
|
|
{
|
|
"epoch": 0.3141864814320064,
|
|
"grad_norm": 16.554518168883288,
|
|
"learning_rate": 6.932207487319305e-07,
|
|
"logits/chosen": -1.0791057348251343,
|
|
"logits/rejected": -0.9959053993225098,
|
|
"logps/chosen": -520.7820434570312,
|
|
"logps/rejected": -573.4239501953125,
|
|
"loss": 0.5661,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.844965934753418,
|
|
"rewards/margins": 0.715225338935852,
|
|
"rewards/rejected": -3.5601911544799805,
|
|
"step": 147
|
|
},
|
|
{
|
|
"epoch": 0.31632380443494523,
|
|
"grad_norm": 18.440264006008654,
|
|
"learning_rate": 6.911774928039643e-07,
|
|
"logits/chosen": -1.1887952089309692,
|
|
"logits/rejected": -1.1027898788452148,
|
|
"logps/chosen": -543.8815307617188,
|
|
"logps/rejected": -748.257080078125,
|
|
"loss": 0.524,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.785444498062134,
|
|
"rewards/margins": 1.9438941478729248,
|
|
"rewards/rejected": -4.729338645935059,
|
|
"step": 148
|
|
},
|
|
{
|
|
"epoch": 0.3184611274378841,
|
|
"grad_norm": 15.107186630332485,
|
|
"learning_rate": 6.891179455309565e-07,
|
|
"logits/chosen": -1.1168442964553833,
|
|
"logits/rejected": -1.0895863771438599,
|
|
"logps/chosen": -551.2578125,
|
|
"logps/rejected": -583.7507934570312,
|
|
"loss": 0.4379,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -3.0725977420806885,
|
|
"rewards/margins": 0.5573362112045288,
|
|
"rewards/rejected": -3.6299338340759277,
|
|
"step": 149
|
|
},
|
|
{
|
|
"epoch": 0.32059845044082286,
|
|
"grad_norm": 13.129470938801255,
|
|
"learning_rate": 6.870422221443247e-07,
|
|
"logits/chosen": -1.241693139076233,
|
|
"logits/rejected": -1.2060009241104126,
|
|
"logps/chosen": -577.2138061523438,
|
|
"logps/rejected": -712.0526123046875,
|
|
"loss": 0.5511,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.028244972229004,
|
|
"rewards/margins": 1.214041829109192,
|
|
"rewards/rejected": -4.242286682128906,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 0.3227357734437617,
|
|
"grad_norm": 17.950223217812194,
|
|
"learning_rate": 6.849504387805379e-07,
|
|
"logits/chosen": -0.9408307075500488,
|
|
"logits/rejected": -1.0360286235809326,
|
|
"logps/chosen": -588.658447265625,
|
|
"logps/rejected": -604.2407836914062,
|
|
"loss": 0.5004,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -3.3046035766601562,
|
|
"rewards/margins": 0.3405230641365051,
|
|
"rewards/rejected": -3.645127058029175,
|
|
"step": 151
|
|
},
|
|
{
|
|
"epoch": 0.3248730964467005,
|
|
"grad_norm": 15.115382544003781,
|
|
"learning_rate": 6.828427124746189e-07,
|
|
"logits/chosen": -1.0206228494644165,
|
|
"logits/rejected": -0.9669525027275085,
|
|
"logps/chosen": -562.2371215820312,
|
|
"logps/rejected": -633.0101928710938,
|
|
"loss": 0.5519,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.1128549575805664,
|
|
"rewards/margins": 0.6220429539680481,
|
|
"rewards/rejected": -3.7348976135253906,
|
|
"step": 152
|
|
},
|
|
{
|
|
"epoch": 0.3270104194496393,
|
|
"grad_norm": 12.660384376081153,
|
|
"learning_rate": 6.807191611535966e-07,
|
|
"logits/chosen": -1.1048712730407715,
|
|
"logits/rejected": -1.0781325101852417,
|
|
"logps/chosen": -564.8250732421875,
|
|
"logps/rejected": -678.5778198242188,
|
|
"loss": 0.474,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.999598264694214,
|
|
"rewards/margins": 1.3427116870880127,
|
|
"rewards/rejected": -4.342309474945068,
|
|
"step": 153
|
|
},
|
|
{
|
|
"epoch": 0.32914774245257816,
|
|
"grad_norm": 15.357120824304575,
|
|
"learning_rate": 6.785799036299067e-07,
|
|
"logits/chosen": -1.0824536085128784,
|
|
"logits/rejected": -1.0418331623077393,
|
|
"logps/chosen": -486.5535888671875,
|
|
"logps/rejected": -583.5165405273438,
|
|
"loss": 0.5399,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.476439952850342,
|
|
"rewards/margins": 0.8414661884307861,
|
|
"rewards/rejected": -3.317906379699707,
|
|
"step": 154
|
|
},
|
|
{
|
|
"epoch": 0.33128506545551695,
|
|
"grad_norm": 14.128652205737456,
|
|
"learning_rate": 6.764250595947458e-07,
|
|
"logits/chosen": -1.059680700302124,
|
|
"logits/rejected": -1.131382703781128,
|
|
"logps/chosen": -486.99127197265625,
|
|
"logps/rejected": -562.620361328125,
|
|
"loss": 0.4737,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.4245104789733887,
|
|
"rewards/margins": 0.9044227004051208,
|
|
"rewards/rejected": -3.328933000564575,
|
|
"step": 155
|
|
},
|
|
{
|
|
"epoch": 0.3334223884584558,
|
|
"grad_norm": 12.201086119166204,
|
|
"learning_rate": 6.742547496113737e-07,
|
|
"logits/chosen": -1.0942405462265015,
|
|
"logits/rejected": -1.1031134128570557,
|
|
"logps/chosen": -506.2625732421875,
|
|
"logps/rejected": -582.8713989257812,
|
|
"loss": 0.5059,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.9856767654418945,
|
|
"rewards/margins": 0.6826351284980774,
|
|
"rewards/rejected": -3.6683120727539062,
|
|
"step": 156
|
|
},
|
|
{
|
|
"epoch": 0.3355597114613946,
|
|
"grad_norm": 12.433943055598153,
|
|
"learning_rate": 6.720690951083677e-07,
|
|
"logits/chosen": -0.9618053436279297,
|
|
"logits/rejected": -0.948087215423584,
|
|
"logps/chosen": -620.4253540039062,
|
|
"logps/rejected": -752.8404541015625,
|
|
"loss": 0.4782,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.224658489227295,
|
|
"rewards/margins": 1.47016441822052,
|
|
"rewards/rejected": -4.694822311401367,
|
|
"step": 157
|
|
},
|
|
{
|
|
"epoch": 0.3376970344643334,
|
|
"grad_norm": 14.86954063543414,
|
|
"learning_rate": 6.698682183728298e-07,
|
|
"logits/chosen": -1.1300894021987915,
|
|
"logits/rejected": -1.039293646812439,
|
|
"logps/chosen": -464.7246398925781,
|
|
"logps/rejected": -609.8098754882812,
|
|
"loss": 0.4946,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.3228816986083984,
|
|
"rewards/margins": 1.281423807144165,
|
|
"rewards/rejected": -3.6043052673339844,
|
|
"step": 158
|
|
},
|
|
{
|
|
"epoch": 0.33983435746727225,
|
|
"grad_norm": 11.83277304769165,
|
|
"learning_rate": 6.676522425435432e-07,
|
|
"logits/chosen": -1.131430745124817,
|
|
"logits/rejected": -1.0974528789520264,
|
|
"logps/chosen": -469.0117492675781,
|
|
"logps/rejected": -556.792236328125,
|
|
"loss": 0.4858,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.6056880950927734,
|
|
"rewards/margins": 0.8633173704147339,
|
|
"rewards/rejected": -3.4690053462982178,
|
|
"step": 159
|
|
},
|
|
{
|
|
"epoch": 0.34197168047021104,
|
|
"grad_norm": 11.767019931678893,
|
|
"learning_rate": 6.654212916040845e-07,
|
|
"logits/chosen": -1.1052802801132202,
|
|
"logits/rejected": -1.058582067489624,
|
|
"logps/chosen": -557.347412109375,
|
|
"logps/rejected": -626.2900390625,
|
|
"loss": 0.4836,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.7133045196533203,
|
|
"rewards/margins": 0.6999992728233337,
|
|
"rewards/rejected": -3.413303852081299,
|
|
"step": 160
|
|
},
|
|
{
|
|
"epoch": 0.3441090034731499,
|
|
"grad_norm": 10.178256055891817,
|
|
"learning_rate": 6.63175490375885e-07,
|
|
"logits/chosen": -1.1756691932678223,
|
|
"logits/rejected": -1.103055715560913,
|
|
"logps/chosen": -481.7811584472656,
|
|
"logps/rejected": -647.7122192382812,
|
|
"loss": 0.4671,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -2.503532648086548,
|
|
"rewards/margins": 1.39969003200531,
|
|
"rewards/rejected": -3.9032225608825684,
|
|
"step": 161
|
|
},
|
|
{
|
|
"epoch": 0.3462463264760887,
|
|
"grad_norm": 11.090255106567847,
|
|
"learning_rate": 6.609149645112484e-07,
|
|
"logits/chosen": -1.0091067552566528,
|
|
"logits/rejected": -0.9871212840080261,
|
|
"logps/chosen": -494.93609619140625,
|
|
"logps/rejected": -592.403564453125,
|
|
"loss": 0.4872,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -2.7299017906188965,
|
|
"rewards/margins": 0.8612344861030579,
|
|
"rewards/rejected": -3.5911364555358887,
|
|
"step": 162
|
|
},
|
|
{
|
|
"epoch": 0.3483836494790275,
|
|
"grad_norm": 11.3289959517762,
|
|
"learning_rate": 6.586398404863198e-07,
|
|
"logits/chosen": -1.0473204851150513,
|
|
"logits/rejected": -1.0162488222122192,
|
|
"logps/chosen": -468.8031311035156,
|
|
"logps/rejected": -638.3246459960938,
|
|
"loss": 0.4592,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -2.419950485229492,
|
|
"rewards/margins": 1.7530534267425537,
|
|
"rewards/rejected": -4.173004150390625,
|
|
"step": 163
|
|
},
|
|
{
|
|
"epoch": 0.35052097248196634,
|
|
"grad_norm": 12.136113240636318,
|
|
"learning_rate": 6.563502455940096e-07,
|
|
"logits/chosen": -0.9164242148399353,
|
|
"logits/rejected": -0.9073900580406189,
|
|
"logps/chosen": -547.6764526367188,
|
|
"logps/rejected": -694.9613037109375,
|
|
"loss": 0.5092,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -2.788818597793579,
|
|
"rewards/margins": 1.4330434799194336,
|
|
"rewards/rejected": -4.221862316131592,
|
|
"step": 164
|
|
},
|
|
{
|
|
"epoch": 0.3526582954849052,
|
|
"grad_norm": 13.204127745290974,
|
|
"learning_rate": 6.540463079368714e-07,
|
|
"logits/chosen": -1.156178593635559,
|
|
"logits/rejected": -1.2097373008728027,
|
|
"logps/chosen": -669.5219116210938,
|
|
"logps/rejected": -734.2588500976562,
|
|
"loss": 0.4965,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.4145455360412598,
|
|
"rewards/margins": 0.7804650068283081,
|
|
"rewards/rejected": -4.195010662078857,
|
|
"step": 165
|
|
},
|
|
{
|
|
"epoch": 0.35479561848784397,
|
|
"grad_norm": 12.906217141464444,
|
|
"learning_rate": 6.51728156419935e-07,
|
|
"logits/chosen": -1.1758369207382202,
|
|
"logits/rejected": -1.1248809099197388,
|
|
"logps/chosen": -432.7223815917969,
|
|
"logps/rejected": -522.7244262695312,
|
|
"loss": 0.5121,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -2.4266903400421143,
|
|
"rewards/margins": 0.7254734635353088,
|
|
"rewards/rejected": -3.1521639823913574,
|
|
"step": 166
|
|
},
|
|
{
|
|
"epoch": 0.3569329414907828,
|
|
"grad_norm": 12.171039393472755,
|
|
"learning_rate": 6.493959207434934e-07,
|
|
"logits/chosen": -0.9444956183433533,
|
|
"logits/rejected": -1.0015169382095337,
|
|
"logps/chosen": -607.41650390625,
|
|
"logps/rejected": -728.7015380859375,
|
|
"loss": 0.4482,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.5008718967437744,
|
|
"rewards/margins": 1.2940125465393066,
|
|
"rewards/rejected": -4.794884204864502,
|
|
"step": 167
|
|
},
|
|
{
|
|
"epoch": 0.3590702644937216,
|
|
"grad_norm": 12.676709989313622,
|
|
"learning_rate": 6.470497313958472e-07,
|
|
"logits/chosen": -1.0418492555618286,
|
|
"logits/rejected": -0.9932193160057068,
|
|
"logps/chosen": -495.92938232421875,
|
|
"logps/rejected": -625.4636840820312,
|
|
"loss": 0.5167,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -2.6624858379364014,
|
|
"rewards/margins": 1.1294273138046265,
|
|
"rewards/rejected": -3.791912794113159,
|
|
"step": 168
|
|
},
|
|
{
|
|
"epoch": 0.36120758749666043,
|
|
"grad_norm": 12.231651860351334,
|
|
"learning_rate": 6.446897196460025e-07,
|
|
"logits/chosen": -1.0079303979873657,
|
|
"logits/rejected": -0.9732705950737,
|
|
"logps/chosen": -458.8643493652344,
|
|
"logps/rejected": -558.783447265625,
|
|
"loss": 0.4617,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.6645169258117676,
|
|
"rewards/margins": 0.922903299331665,
|
|
"rewards/rejected": -3.5874204635620117,
|
|
"step": 169
|
|
},
|
|
{
|
|
"epoch": 0.36334491049959927,
|
|
"grad_norm": 13.830086184878045,
|
|
"learning_rate": 6.42316017536328e-07,
|
|
"logits/chosen": -1.1729865074157715,
|
|
"logits/rejected": -1.1441071033477783,
|
|
"logps/chosen": -493.1350402832031,
|
|
"logps/rejected": -565.9179077148438,
|
|
"loss": 0.4823,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.748018741607666,
|
|
"rewards/margins": 0.6935210227966309,
|
|
"rewards/rejected": -3.4415395259857178,
|
|
"step": 170
|
|
},
|
|
{
|
|
"epoch": 0.36548223350253806,
|
|
"grad_norm": 13.82995654030433,
|
|
"learning_rate": 6.399287578751656e-07,
|
|
"logits/chosen": -1.0457535982131958,
|
|
"logits/rejected": -0.9317151308059692,
|
|
"logps/chosen": -545.84375,
|
|
"logps/rejected": -692.236328125,
|
|
"loss": 0.4293,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.1345977783203125,
|
|
"rewards/margins": 1.223220705986023,
|
|
"rewards/rejected": -4.357818603515625,
|
|
"step": 171
|
|
},
|
|
{
|
|
"epoch": 0.3676195565054769,
|
|
"grad_norm": 13.523650685418982,
|
|
"learning_rate": 6.375280742294006e-07,
|
|
"logits/chosen": -0.807671070098877,
|
|
"logits/rejected": -0.8415660858154297,
|
|
"logps/chosen": -581.8911743164062,
|
|
"logps/rejected": -669.5233764648438,
|
|
"loss": 0.5499,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.200932264328003,
|
|
"rewards/margins": 0.9946154356002808,
|
|
"rewards/rejected": -4.195547580718994,
|
|
"step": 172
|
|
},
|
|
{
|
|
"epoch": 0.36975687950841574,
|
|
"grad_norm": 12.874802080841212,
|
|
"learning_rate": 6.351141009169893e-07,
|
|
"logits/chosen": -1.1410176753997803,
|
|
"logits/rejected": -1.10181725025177,
|
|
"logps/chosen": -447.8720703125,
|
|
"logps/rejected": -564.0751953125,
|
|
"loss": 0.4725,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -2.5582220554351807,
|
|
"rewards/margins": 1.1303775310516357,
|
|
"rewards/rejected": -3.6885998249053955,
|
|
"step": 173
|
|
},
|
|
{
|
|
"epoch": 0.3718942025113545,
|
|
"grad_norm": 12.769700171240247,
|
|
"learning_rate": 6.326869729994423e-07,
|
|
"logits/chosen": -1.0553710460662842,
|
|
"logits/rejected": -1.0439517498016357,
|
|
"logps/chosen": -494.44775390625,
|
|
"logps/rejected": -636.4324951171875,
|
|
"loss": 0.4471,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -2.5291152000427246,
|
|
"rewards/margins": 1.314854383468628,
|
|
"rewards/rejected": -3.8439698219299316,
|
|
"step": 174
|
|
},
|
|
{
|
|
"epoch": 0.37403152551429336,
|
|
"grad_norm": 11.212264761027647,
|
|
"learning_rate": 6.302468262742694e-07,
|
|
"logits/chosen": -0.950715959072113,
|
|
"logits/rejected": -0.884972333908081,
|
|
"logps/chosen": -521.4841918945312,
|
|
"logps/rejected": -649.075927734375,
|
|
"loss": 0.4173,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.026653289794922,
|
|
"rewards/margins": 1.0717153549194336,
|
|
"rewards/rejected": -4.098368167877197,
|
|
"step": 175
|
|
},
|
|
{
|
|
"epoch": 0.37616884851723215,
|
|
"grad_norm": 11.18847853926289,
|
|
"learning_rate": 6.277937972673808e-07,
|
|
"logits/chosen": -1.1451900005340576,
|
|
"logits/rejected": -1.1354702711105347,
|
|
"logps/chosen": -544.9923095703125,
|
|
"logps/rejected": -680.9166259765625,
|
|
"loss": 0.4452,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.016019582748413,
|
|
"rewards/margins": 1.4868509769439697,
|
|
"rewards/rejected": -4.502870559692383,
|
|
"step": 176
|
|
},
|
|
{
|
|
"epoch": 0.378306171520171,
|
|
"grad_norm": 12.264246636952317,
|
|
"learning_rate": 6.253280232254487e-07,
|
|
"logits/chosen": -1.1703343391418457,
|
|
"logits/rejected": -1.1669647693634033,
|
|
"logps/chosen": -511.00799560546875,
|
|
"logps/rejected": -609.4747314453125,
|
|
"loss": 0.4253,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.0230469703674316,
|
|
"rewards/margins": 0.9169848561286926,
|
|
"rewards/rejected": -3.9400315284729004,
|
|
"step": 177
|
|
},
|
|
{
|
|
"epoch": 0.3804434945231098,
|
|
"grad_norm": 14.813100911794251,
|
|
"learning_rate": 6.228496421082289e-07,
|
|
"logits/chosen": -1.015668511390686,
|
|
"logits/rejected": -1.1096816062927246,
|
|
"logps/chosen": -636.0406494140625,
|
|
"logps/rejected": -721.57666015625,
|
|
"loss": 0.4504,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.538027286529541,
|
|
"rewards/margins": 1.1062912940979004,
|
|
"rewards/rejected": -4.6443190574646,
|
|
"step": 178
|
|
},
|
|
{
|
|
"epoch": 0.3825808175260486,
|
|
"grad_norm": 12.280974148700997,
|
|
"learning_rate": 6.20358792580841e-07,
|
|
"logits/chosen": -0.9031931757926941,
|
|
"logits/rejected": -0.9428293704986572,
|
|
"logps/chosen": -572.744384765625,
|
|
"logps/rejected": -719.7515869140625,
|
|
"loss": 0.4117,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.4223711490631104,
|
|
"rewards/margins": 1.4378111362457275,
|
|
"rewards/rejected": -4.860182285308838,
|
|
"step": 179
|
|
},
|
|
{
|
|
"epoch": 0.38471814052898745,
|
|
"grad_norm": 11.985112172297303,
|
|
"learning_rate": 6.178556140060108e-07,
|
|
"logits/chosen": -0.9790340662002563,
|
|
"logits/rejected": -1.0062155723571777,
|
|
"logps/chosen": -667.6234130859375,
|
|
"logps/rejected": -789.353759765625,
|
|
"loss": 0.4774,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.024725914001465,
|
|
"rewards/margins": 1.263332724571228,
|
|
"rewards/rejected": -5.288058757781982,
|
|
"step": 180
|
|
},
|
|
{
|
|
"epoch": 0.38685546353192624,
|
|
"grad_norm": 15.173768084164333,
|
|
"learning_rate": 6.153402464362729e-07,
|
|
"logits/chosen": -1.1106728315353394,
|
|
"logits/rejected": -1.1724138259887695,
|
|
"logps/chosen": -580.4508056640625,
|
|
"logps/rejected": -633.2486572265625,
|
|
"loss": 0.5298,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.401585340499878,
|
|
"rewards/margins": 0.4253509044647217,
|
|
"rewards/rejected": -3.8269362449645996,
|
|
"step": 181
|
|
},
|
|
{
|
|
"epoch": 0.3889927865348651,
|
|
"grad_norm": 14.282404953550225,
|
|
"learning_rate": 6.128128306061346e-07,
|
|
"logits/chosen": -1.279030203819275,
|
|
"logits/rejected": -1.2068696022033691,
|
|
"logps/chosen": -563.7296142578125,
|
|
"logps/rejected": -805.1969604492188,
|
|
"loss": 0.4137,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.0469908714294434,
|
|
"rewards/margins": 2.2822344303131104,
|
|
"rewards/rejected": -5.329225063323975,
|
|
"step": 182
|
|
},
|
|
{
|
|
"epoch": 0.3911301095378039,
|
|
"grad_norm": 12.48256287123551,
|
|
"learning_rate": 6.102735079242018e-07,
|
|
"logits/chosen": -1.091313123703003,
|
|
"logits/rejected": -1.0286011695861816,
|
|
"logps/chosen": -562.3391723632812,
|
|
"logps/rejected": -746.7239379882812,
|
|
"loss": 0.4182,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.107255697250366,
|
|
"rewards/margins": 1.946906328201294,
|
|
"rewards/rejected": -5.054161548614502,
|
|
"step": 183
|
|
},
|
|
{
|
|
"epoch": 0.3932674325407427,
|
|
"grad_norm": 12.225189138276614,
|
|
"learning_rate": 6.077224204652676e-07,
|
|
"logits/chosen": -1.2094743251800537,
|
|
"logits/rejected": -1.2332773208618164,
|
|
"logps/chosen": -541.5833740234375,
|
|
"logps/rejected": -624.4737548828125,
|
|
"loss": 0.439,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.0276694297790527,
|
|
"rewards/margins": 0.8412066698074341,
|
|
"rewards/rejected": -3.868875741958618,
|
|
"step": 184
|
|
},
|
|
{
|
|
"epoch": 0.39540475554368154,
|
|
"grad_norm": 14.957361436447231,
|
|
"learning_rate": 6.051597109623624e-07,
|
|
"logits/chosen": -1.028442144393921,
|
|
"logits/rejected": -1.0131793022155762,
|
|
"logps/chosen": -600.3471069335938,
|
|
"logps/rejected": -733.6820678710938,
|
|
"loss": 0.4948,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.375563859939575,
|
|
"rewards/margins": 1.3205593824386597,
|
|
"rewards/rejected": -4.6961236000061035,
|
|
"step": 185
|
|
},
|
|
{
|
|
"epoch": 0.3975420785466204,
|
|
"grad_norm": 13.214401719298836,
|
|
"learning_rate": 6.025855227987691e-07,
|
|
"logits/chosen": -1.1182265281677246,
|
|
"logits/rejected": -1.0665013790130615,
|
|
"logps/chosen": -583.1017456054688,
|
|
"logps/rejected": -770.8402709960938,
|
|
"loss": 0.4751,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.495480537414551,
|
|
"rewards/margins": 1.8332180976867676,
|
|
"rewards/rejected": -5.328698635101318,
|
|
"step": 186
|
|
},
|
|
{
|
|
"epoch": 0.39967940154955917,
|
|
"grad_norm": 13.595163023407325,
|
|
"learning_rate": 6e-07,
|
|
"logits/chosen": -1.070106863975525,
|
|
"logits/rejected": -1.0235223770141602,
|
|
"logps/chosen": -578.7149047851562,
|
|
"logps/rejected": -709.0172729492188,
|
|
"loss": 0.4339,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.577277183532715,
|
|
"rewards/margins": 1.1270668506622314,
|
|
"rewards/rejected": -4.704343795776367,
|
|
"step": 187
|
|
},
|
|
{
|
|
"epoch": 0.401816724552498,
|
|
"grad_norm": 16.48310592089836,
|
|
"learning_rate": 5.974032872257387e-07,
|
|
"logits/chosen": -1.1554845571517944,
|
|
"logits/rejected": -1.1422069072723389,
|
|
"logps/chosen": -742.5419921875,
|
|
"logps/rejected": -898.2552490234375,
|
|
"loss": 0.5009,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.372076988220215,
|
|
"rewards/margins": 1.6709426641464233,
|
|
"rewards/rejected": -6.043020248413086,
|
|
"step": 188
|
|
},
|
|
{
|
|
"epoch": 0.4039540475554368,
|
|
"grad_norm": 14.86949064709501,
|
|
"learning_rate": 5.947955297617469e-07,
|
|
"logits/chosen": -1.0644134283065796,
|
|
"logits/rejected": -1.0610729455947876,
|
|
"logps/chosen": -604.2228393554688,
|
|
"logps/rejected": -724.7699584960938,
|
|
"loss": 0.4967,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.4057650566101074,
|
|
"rewards/margins": 1.34103262424469,
|
|
"rewards/rejected": -4.746798038482666,
|
|
"step": 189
|
|
},
|
|
{
|
|
"epoch": 0.40609137055837563,
|
|
"grad_norm": 12.688678020056997,
|
|
"learning_rate": 5.92176873511735e-07,
|
|
"logits/chosen": -1.015690803527832,
|
|
"logits/rejected": -0.9889406561851501,
|
|
"logps/chosen": -544.1987915039062,
|
|
"logps/rejected": -630.72705078125,
|
|
"loss": 0.4642,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.410604953765869,
|
|
"rewards/margins": 0.8672491908073425,
|
|
"rewards/rejected": -4.277853965759277,
|
|
"step": 190
|
|
},
|
|
{
|
|
"epoch": 0.40822869356131447,
|
|
"grad_norm": 16.497268225970018,
|
|
"learning_rate": 5.895474649891994e-07,
|
|
"logits/chosen": -1.1754518747329712,
|
|
"logits/rejected": -1.1124107837677002,
|
|
"logps/chosen": -664.4409790039062,
|
|
"logps/rejected": -850.3493041992188,
|
|
"loss": 0.4583,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.937302350997925,
|
|
"rewards/margins": 1.6396393775939941,
|
|
"rewards/rejected": -5.576941967010498,
|
|
"step": 191
|
|
},
|
|
{
|
|
"epoch": 0.41036601656425326,
|
|
"grad_norm": 16.209730758284643,
|
|
"learning_rate": 5.869074513092248e-07,
|
|
"logits/chosen": -1.1026970148086548,
|
|
"logits/rejected": -1.0514788627624512,
|
|
"logps/chosen": -518.1434936523438,
|
|
"logps/rejected": -543.8851318359375,
|
|
"loss": 0.5024,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.1491613388061523,
|
|
"rewards/margins": 0.30625391006469727,
|
|
"rewards/rejected": -3.4554154872894287,
|
|
"step": 192
|
|
},
|
|
{
|
|
"epoch": 0.4125033395671921,
|
|
"grad_norm": 14.339086717238697,
|
|
"learning_rate": 5.842569801802529e-07,
|
|
"logits/chosen": -0.9835341572761536,
|
|
"logits/rejected": -0.9959213137626648,
|
|
"logps/chosen": -591.5449829101562,
|
|
"logps/rejected": -795.7198486328125,
|
|
"loss": 0.4859,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.573066473007202,
|
|
"rewards/margins": 1.810349464416504,
|
|
"rewards/rejected": -5.383416175842285,
|
|
"step": 193
|
|
},
|
|
{
|
|
"epoch": 0.41464066257013094,
|
|
"grad_norm": 14.717147219609627,
|
|
"learning_rate": 5.815961998958187e-07,
|
|
"logits/chosen": -1.2837733030319214,
|
|
"logits/rejected": -1.2531107664108276,
|
|
"logps/chosen": -598.62841796875,
|
|
"logps/rejected": -655.9907836914062,
|
|
"loss": 0.5017,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -3.0715248584747314,
|
|
"rewards/margins": 0.7950742244720459,
|
|
"rewards/rejected": -3.8665990829467773,
|
|
"step": 194
|
|
},
|
|
{
|
|
"epoch": 0.4167779855730697,
|
|
"grad_norm": 13.661497341383276,
|
|
"learning_rate": 5.78925259326253e-07,
|
|
"logits/chosen": -1.2736696004867554,
|
|
"logits/rejected": -1.2564581632614136,
|
|
"logps/chosen": -678.5160522460938,
|
|
"logps/rejected": -836.7633666992188,
|
|
"loss": 0.4046,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.7295868396759033,
|
|
"rewards/margins": 1.5906658172607422,
|
|
"rewards/rejected": -5.320252418518066,
|
|
"step": 195
|
|
},
|
|
{
|
|
"epoch": 0.41891530857600856,
|
|
"grad_norm": 17.358547092004624,
|
|
"learning_rate": 5.762443079103534e-07,
|
|
"logits/chosen": -1.1058768033981323,
|
|
"logits/rejected": -1.1880860328674316,
|
|
"logps/chosen": -632.157470703125,
|
|
"logps/rejected": -689.973876953125,
|
|
"loss": 0.6262,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.6461615562438965,
|
|
"rewards/margins": 0.6898024082183838,
|
|
"rewards/rejected": -4.335964202880859,
|
|
"step": 196
|
|
},
|
|
{
|
|
"epoch": 0.42105263157894735,
|
|
"grad_norm": 11.719355560175847,
|
|
"learning_rate": 5.735534956470232e-07,
|
|
"logits/chosen": -1.1999759674072266,
|
|
"logits/rejected": -1.228690505027771,
|
|
"logps/chosen": -673.6649169921875,
|
|
"logps/rejected": -844.5328369140625,
|
|
"loss": 0.4206,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.677556037902832,
|
|
"rewards/margins": 1.6347838640213013,
|
|
"rewards/rejected": -5.312339782714844,
|
|
"step": 197
|
|
},
|
|
{
|
|
"epoch": 0.4231899545818862,
|
|
"grad_norm": 10.672786717708536,
|
|
"learning_rate": 5.708529730868787e-07,
|
|
"logits/chosen": -1.203027606010437,
|
|
"logits/rejected": -1.1142572164535522,
|
|
"logps/chosen": -541.4075317382812,
|
|
"logps/rejected": -631.228759765625,
|
|
"loss": 0.4455,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.15077805519104,
|
|
"rewards/margins": 0.7978792190551758,
|
|
"rewards/rejected": -3.9486567974090576,
|
|
"step": 198
|
|
},
|
|
{
|
|
"epoch": 0.425327277584825,
|
|
"grad_norm": 11.74204286722898,
|
|
"learning_rate": 5.681428913238262e-07,
|
|
"logits/chosen": -1.0974667072296143,
|
|
"logits/rejected": -1.1039204597473145,
|
|
"logps/chosen": -562.8824462890625,
|
|
"logps/rejected": -656.3973388671875,
|
|
"loss": 0.4138,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.360422134399414,
|
|
"rewards/margins": 1.0035967826843262,
|
|
"rewards/rejected": -4.36401891708374,
|
|
"step": 199
|
|
},
|
|
{
|
|
"epoch": 0.4274646005877638,
|
|
"grad_norm": 12.842353670192367,
|
|
"learning_rate": 5.654234019866083e-07,
|
|
"logits/chosen": -1.1496554613113403,
|
|
"logits/rejected": -1.1662920713424683,
|
|
"logps/chosen": -515.6434326171875,
|
|
"logps/rejected": -629.1980590820312,
|
|
"loss": 0.5011,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -3.022861957550049,
|
|
"rewards/margins": 0.9866237640380859,
|
|
"rewards/rejected": -4.009485244750977,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.4274646005877638,
|
|
"eval_logits/chosen": -1.2976744174957275,
|
|
"eval_logits/rejected": -1.3147474527359009,
|
|
"eval_logps/chosen": -629.65478515625,
|
|
"eval_logps/rejected": -736.8406372070312,
|
|
"eval_loss": 0.4427746832370758,
|
|
"eval_rewards/accuracies": 0.7903226017951965,
|
|
"eval_rewards/chosen": -3.5014870166778564,
|
|
"eval_rewards/margins": 1.131365180015564,
|
|
"eval_rewards/rejected": -4.632852554321289,
|
|
"eval_runtime": 127.7594,
|
|
"eval_samples_per_second": 15.349,
|
|
"eval_steps_per_second": 0.485,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.42960192359070265,
|
|
"grad_norm": 14.704888625045939,
|
|
"learning_rate": 5.626946572303201e-07,
|
|
"logits/chosen": -1.217873454093933,
|
|
"logits/rejected": -1.1787673234939575,
|
|
"logps/chosen": -640.8902587890625,
|
|
"logps/rejected": -832.7512817382812,
|
|
"loss": 0.438,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.6388046741485596,
|
|
"rewards/margins": 2.023084878921509,
|
|
"rewards/rejected": -5.661889553070068,
|
|
"step": 201
|
|
},
|
|
{
|
|
"epoch": 0.4317392465936415,
|
|
"grad_norm": 12.551626779621763,
|
|
"learning_rate": 5.599568097278963e-07,
|
|
"logits/chosen": -1.1497160196304321,
|
|
"logits/rejected": -1.150457501411438,
|
|
"logps/chosen": -672.3145141601562,
|
|
"logps/rejected": -819.0764770507812,
|
|
"loss": 0.4017,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.58868145942688,
|
|
"rewards/margins": 1.576560616493225,
|
|
"rewards/rejected": -5.165241718292236,
|
|
"step": 202
|
|
},
|
|
{
|
|
"epoch": 0.4338765695965803,
|
|
"grad_norm": 13.096747763619007,
|
|
"learning_rate": 5.572100126615694e-07,
|
|
"logits/chosen": -1.1709344387054443,
|
|
"logits/rejected": -1.1770344972610474,
|
|
"logps/chosen": -610.950439453125,
|
|
"logps/rejected": -775.8607788085938,
|
|
"loss": 0.4092,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.6900205612182617,
|
|
"rewards/margins": 1.358374834060669,
|
|
"rewards/rejected": -5.048395156860352,
|
|
"step": 203
|
|
},
|
|
{
|
|
"epoch": 0.4360138925995191,
|
|
"grad_norm": 16.19055766301351,
|
|
"learning_rate": 5.544544197142984e-07,
|
|
"logits/chosen": -1.1622633934020996,
|
|
"logits/rejected": -1.0976923704147339,
|
|
"logps/chosen": -718.2122802734375,
|
|
"logps/rejected": -918.7157592773438,
|
|
"loss": 0.4207,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.543436050415039,
|
|
"rewards/margins": 1.6670641899108887,
|
|
"rewards/rejected": -6.210500717163086,
|
|
"step": 204
|
|
},
|
|
{
|
|
"epoch": 0.4381512156024579,
|
|
"grad_norm": 13.688669492121473,
|
|
"learning_rate": 5.516901850611714e-07,
|
|
"logits/chosen": -1.1796329021453857,
|
|
"logits/rejected": -1.1708921194076538,
|
|
"logps/chosen": -581.9102172851562,
|
|
"logps/rejected": -728.2744750976562,
|
|
"loss": 0.4615,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.4006991386413574,
|
|
"rewards/margins": 1.4271738529205322,
|
|
"rewards/rejected": -4.827873229980469,
|
|
"step": 205
|
|
},
|
|
{
|
|
"epoch": 0.44028853860539674,
|
|
"grad_norm": 20.309923000737594,
|
|
"learning_rate": 5.489174633607781e-07,
|
|
"logits/chosen": -0.9230062961578369,
|
|
"logits/rejected": -1.0179659128189087,
|
|
"logps/chosen": -700.2884521484375,
|
|
"logps/rejected": -722.5196533203125,
|
|
"loss": 0.5692,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.400253772735596,
|
|
"rewards/margins": 0.43511492013931274,
|
|
"rewards/rejected": -4.835369110107422,
|
|
"step": 206
|
|
},
|
|
{
|
|
"epoch": 0.4424258616083356,
|
|
"grad_norm": 20.545691501495256,
|
|
"learning_rate": 5.461364097465581e-07,
|
|
"logits/chosen": -1.1641998291015625,
|
|
"logits/rejected": -1.1823948621749878,
|
|
"logps/chosen": -649.5260009765625,
|
|
"logps/rejected": -736.3587036132812,
|
|
"loss": 0.5013,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.8688461780548096,
|
|
"rewards/margins": 0.9938517808914185,
|
|
"rewards/rejected": -4.862697601318359,
|
|
"step": 207
|
|
},
|
|
{
|
|
"epoch": 0.44456318461127436,
|
|
"grad_norm": 13.254346937801454,
|
|
"learning_rate": 5.4334717981812e-07,
|
|
"logits/chosen": -1.212602972984314,
|
|
"logits/rejected": -1.293573260307312,
|
|
"logps/chosen": -660.5288696289062,
|
|
"logps/rejected": -745.7517700195312,
|
|
"loss": 0.4214,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -4.027283668518066,
|
|
"rewards/margins": 0.8199220895767212,
|
|
"rewards/rejected": -4.8472065925598145,
|
|
"step": 208
|
|
},
|
|
{
|
|
"epoch": 0.4467005076142132,
|
|
"grad_norm": 13.832436176790107,
|
|
"learning_rate": 5.405499296325371e-07,
|
|
"logits/chosen": -1.112457513809204,
|
|
"logits/rejected": -1.1108170747756958,
|
|
"logps/chosen": -542.69140625,
|
|
"logps/rejected": -657.5791015625,
|
|
"loss": 0.4021,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.3648681640625,
|
|
"rewards/margins": 1.0020387172698975,
|
|
"rewards/rejected": -4.366907119750977,
|
|
"step": 209
|
|
},
|
|
{
|
|
"epoch": 0.448837830617152,
|
|
"grad_norm": 16.15754914343496,
|
|
"learning_rate": 5.377448156956139e-07,
|
|
"logits/chosen": -1.2227798700332642,
|
|
"logits/rejected": -1.1601148843765259,
|
|
"logps/chosen": -826.6924438476562,
|
|
"logps/rejected": -967.8865966796875,
|
|
"loss": 0.4399,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -5.01950740814209,
|
|
"rewards/margins": 1.297555923461914,
|
|
"rewards/rejected": -6.317063331604004,
|
|
"step": 210
|
|
},
|
|
{
|
|
"epoch": 0.45097515362009083,
|
|
"grad_norm": 14.676898780340569,
|
|
"learning_rate": 5.34931994953132e-07,
|
|
"logits/chosen": -1.2065531015396118,
|
|
"logits/rejected": -1.2652815580368042,
|
|
"logps/chosen": -682.3113403320312,
|
|
"logps/rejected": -843.3710327148438,
|
|
"loss": 0.3938,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.9389686584472656,
|
|
"rewards/margins": 1.772455096244812,
|
|
"rewards/rejected": -5.711423397064209,
|
|
"step": 211
|
|
},
|
|
{
|
|
"epoch": 0.45311247662302967,
|
|
"grad_norm": 14.111414144546993,
|
|
"learning_rate": 5.321116247820669e-07,
|
|
"logits/chosen": -1.1172921657562256,
|
|
"logits/rejected": -1.1278425455093384,
|
|
"logps/chosen": -585.2348022460938,
|
|
"logps/rejected": -633.3766479492188,
|
|
"loss": 0.4125,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -3.4476540088653564,
|
|
"rewards/margins": 0.7052488327026367,
|
|
"rewards/rejected": -4.152902603149414,
|
|
"step": 212
|
|
},
|
|
{
|
|
"epoch": 0.45524979962596845,
|
|
"grad_norm": 16.54663848569769,
|
|
"learning_rate": 5.29283862981784e-07,
|
|
"logits/chosen": -1.1069424152374268,
|
|
"logits/rejected": -1.0816022157669067,
|
|
"logps/chosen": -623.0894775390625,
|
|
"logps/rejected": -707.4627685546875,
|
|
"loss": 0.5382,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.8370444774627686,
|
|
"rewards/margins": 0.6275334358215332,
|
|
"rewards/rejected": -4.464578151702881,
|
|
"step": 213
|
|
},
|
|
{
|
|
"epoch": 0.4573871226289073,
|
|
"grad_norm": 15.2093168956623,
|
|
"learning_rate": 5.264488677652097e-07,
|
|
"logits/chosen": -1.2023216485977173,
|
|
"logits/rejected": -1.1567494869232178,
|
|
"logps/chosen": -548.3652954101562,
|
|
"logps/rejected": -656.767578125,
|
|
"loss": 0.3832,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.367328643798828,
|
|
"rewards/margins": 0.951412558555603,
|
|
"rewards/rejected": -4.318741321563721,
|
|
"step": 214
|
|
},
|
|
{
|
|
"epoch": 0.45952444563184613,
|
|
"grad_norm": 19.223112891790837,
|
|
"learning_rate": 5.23606797749979e-07,
|
|
"logits/chosen": -1.100671410560608,
|
|
"logits/rejected": -1.0285398960113525,
|
|
"logps/chosen": -586.305419921875,
|
|
"logps/rejected": -665.7771606445312,
|
|
"loss": 0.5119,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -3.6576530933380127,
|
|
"rewards/margins": 0.7183351516723633,
|
|
"rewards/rejected": -4.375988483428955,
|
|
"step": 215
|
|
},
|
|
{
|
|
"epoch": 0.4616617686347849,
|
|
"grad_norm": 12.33986287311908,
|
|
"learning_rate": 5.207578119495608e-07,
|
|
"logits/chosen": -1.2366998195648193,
|
|
"logits/rejected": -1.2117341756820679,
|
|
"logps/chosen": -629.59716796875,
|
|
"logps/rejected": -807.658935546875,
|
|
"loss": 0.4152,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.4987330436706543,
|
|
"rewards/margins": 1.8127856254577637,
|
|
"rewards/rejected": -5.311518669128418,
|
|
"step": 216
|
|
},
|
|
{
|
|
"epoch": 0.46379909163772376,
|
|
"grad_norm": 12.965958762194157,
|
|
"learning_rate": 5.179020697643617e-07,
|
|
"logits/chosen": -1.2369261980056763,
|
|
"logits/rejected": -1.2110192775726318,
|
|
"logps/chosen": -524.5565795898438,
|
|
"logps/rejected": -678.670654296875,
|
|
"loss": 0.4466,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -2.8585205078125,
|
|
"rewards/margins": 1.3874884843826294,
|
|
"rewards/rejected": -4.246009349822998,
|
|
"step": 217
|
|
},
|
|
{
|
|
"epoch": 0.46593641464066254,
|
|
"grad_norm": 16.584181179977357,
|
|
"learning_rate": 5.150397309728068e-07,
|
|
"logits/chosen": -1.3417975902557373,
|
|
"logits/rejected": -1.3138681650161743,
|
|
"logps/chosen": -658.9564208984375,
|
|
"logps/rejected": -744.6614990234375,
|
|
"loss": 0.5277,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.620779275894165,
|
|
"rewards/margins": 0.9568281769752502,
|
|
"rewards/rejected": -4.577607154846191,
|
|
"step": 218
|
|
},
|
|
{
|
|
"epoch": 0.4680737376436014,
|
|
"grad_norm": 14.322701228154292,
|
|
"learning_rate": 5.121709557224011e-07,
|
|
"logits/chosen": -1.0924996137619019,
|
|
"logits/rejected": -1.0888468027114868,
|
|
"logps/chosen": -669.5311279296875,
|
|
"logps/rejected": -813.560546875,
|
|
"loss": 0.4291,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.6334965229034424,
|
|
"rewards/margins": 1.4997690916061401,
|
|
"rewards/rejected": -5.133265495300293,
|
|
"step": 219
|
|
},
|
|
{
|
|
"epoch": 0.4702110606465402,
|
|
"grad_norm": 17.034367007461572,
|
|
"learning_rate": 5.09295904520768e-07,
|
|
"logits/chosen": -1.1301909685134888,
|
|
"logits/rejected": -1.140136480331421,
|
|
"logps/chosen": -595.55126953125,
|
|
"logps/rejected": -707.5346069335938,
|
|
"loss": 0.4488,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.697270154953003,
|
|
"rewards/margins": 1.1222831010818481,
|
|
"rewards/rejected": -4.819552898406982,
|
|
"step": 220
|
|
},
|
|
{
|
|
"epoch": 0.472348383649479,
|
|
"grad_norm": 18.144002765573468,
|
|
"learning_rate": 5.0641473822667e-07,
|
|
"logits/chosen": -1.1868654489517212,
|
|
"logits/rejected": -1.1960182189941406,
|
|
"logps/chosen": -459.4822692871094,
|
|
"logps/rejected": -537.1638793945312,
|
|
"loss": 0.395,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -2.789658546447754,
|
|
"rewards/margins": 0.8773064017295837,
|
|
"rewards/rejected": -3.6669647693634033,
|
|
"step": 221
|
|
},
|
|
{
|
|
"epoch": 0.47448570665241785,
|
|
"grad_norm": 14.14471098052339,
|
|
"learning_rate": 5.035276180410083e-07,
|
|
"logits/chosen": -1.1389260292053223,
|
|
"logits/rejected": -1.1852952241897583,
|
|
"logps/chosen": -611.3282470703125,
|
|
"logps/rejected": -692.434326171875,
|
|
"loss": 0.426,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.461778402328491,
|
|
"rewards/margins": 0.9153075814247131,
|
|
"rewards/rejected": -4.3770856857299805,
|
|
"step": 222
|
|
},
|
|
{
|
|
"epoch": 0.4766230296553567,
|
|
"grad_norm": 13.32254209502914,
|
|
"learning_rate": 5.006347054978034e-07,
|
|
"logits/chosen": -1.1993439197540283,
|
|
"logits/rejected": -1.1267905235290527,
|
|
"logps/chosen": -672.0899658203125,
|
|
"logps/rejected": -908.1744384765625,
|
|
"loss": 0.3713,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.8041088581085205,
|
|
"rewards/margins": 2.0054619312286377,
|
|
"rewards/rejected": -5.809570789337158,
|
|
"step": 223
|
|
},
|
|
{
|
|
"epoch": 0.4787603526582955,
|
|
"grad_norm": 18.83052351653402,
|
|
"learning_rate": 4.977361624551576e-07,
|
|
"logits/chosen": -1.265091896057129,
|
|
"logits/rejected": -1.2223862409591675,
|
|
"logps/chosen": -536.5931396484375,
|
|
"logps/rejected": -670.3464965820312,
|
|
"loss": 0.4369,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.259114980697632,
|
|
"rewards/margins": 1.1988860368728638,
|
|
"rewards/rejected": -4.458001136779785,
|
|
"step": 224
|
|
},
|
|
{
|
|
"epoch": 0.4808976756612343,
|
|
"grad_norm": 17.558770227833907,
|
|
"learning_rate": 4.94832151086199e-07,
|
|
"logits/chosen": -1.2254117727279663,
|
|
"logits/rejected": -1.198996663093567,
|
|
"logps/chosen": -693.514404296875,
|
|
"logps/rejected": -909.830322265625,
|
|
"loss": 0.4987,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.9128432273864746,
|
|
"rewards/margins": 1.8729009628295898,
|
|
"rewards/rejected": -5.7857441902160645,
|
|
"step": 225
|
|
},
|
|
{
|
|
"epoch": 0.4830349986641731,
|
|
"grad_norm": 16.275797205270916,
|
|
"learning_rate": 4.919228338700078e-07,
|
|
"logits/chosen": -1.158968210220337,
|
|
"logits/rejected": -1.1183545589447021,
|
|
"logps/chosen": -637.254150390625,
|
|
"logps/rejected": -791.1282958984375,
|
|
"loss": 0.4417,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.6901886463165283,
|
|
"rewards/margins": 1.393813967704773,
|
|
"rewards/rejected": -5.084002494812012,
|
|
"step": 226
|
|
},
|
|
{
|
|
"epoch": 0.48517232166711194,
|
|
"grad_norm": 15.28807242884776,
|
|
"learning_rate": 4.890083735825257e-07,
|
|
"logits/chosen": -1.1542084217071533,
|
|
"logits/rejected": -1.1580474376678467,
|
|
"logps/chosen": -645.986572265625,
|
|
"logps/rejected": -748.2597045898438,
|
|
"loss": 0.4419,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.941164016723633,
|
|
"rewards/margins": 1.0907334089279175,
|
|
"rewards/rejected": -5.03189754486084,
|
|
"step": 227
|
|
},
|
|
{
|
|
"epoch": 0.4873096446700508,
|
|
"grad_norm": 21.17554543529619,
|
|
"learning_rate": 4.860889332874488e-07,
|
|
"logits/chosen": -1.1863421201705933,
|
|
"logits/rejected": -1.2254602909088135,
|
|
"logps/chosen": -577.6887817382812,
|
|
"logps/rejected": -713.466796875,
|
|
"loss": 0.4874,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.2465620040893555,
|
|
"rewards/margins": 1.4701895713806152,
|
|
"rewards/rejected": -4.7167510986328125,
|
|
"step": 228
|
|
},
|
|
{
|
|
"epoch": 0.48944696767298956,
|
|
"grad_norm": 16.327068285344875,
|
|
"learning_rate": 4.831646763271037e-07,
|
|
"logits/chosen": -1.1210421323776245,
|
|
"logits/rejected": -1.0809756517410278,
|
|
"logps/chosen": -746.8450927734375,
|
|
"logps/rejected": -985.5284423828125,
|
|
"loss": 0.368,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.552522659301758,
|
|
"rewards/margins": 2.365595579147339,
|
|
"rewards/rejected": -6.918118000030518,
|
|
"step": 229
|
|
},
|
|
{
|
|
"epoch": 0.4915842906759284,
|
|
"grad_norm": 17.677696465125493,
|
|
"learning_rate": 4.802357663133091e-07,
|
|
"logits/chosen": -1.232605218887329,
|
|
"logits/rejected": -1.1589381694793701,
|
|
"logps/chosen": -744.6641235351562,
|
|
"logps/rejected": -832.0184326171875,
|
|
"loss": 0.4541,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -4.563028335571289,
|
|
"rewards/margins": 0.9094908833503723,
|
|
"rewards/rejected": -5.472519874572754,
|
|
"step": 230
|
|
},
|
|
{
|
|
"epoch": 0.49372161367886724,
|
|
"grad_norm": 18.66106887116432,
|
|
"learning_rate": 4.773023671182213e-07,
|
|
"logits/chosen": -1.2479596138000488,
|
|
"logits/rejected": -1.1852896213531494,
|
|
"logps/chosen": -643.7618408203125,
|
|
"logps/rejected": -795.2108764648438,
|
|
"loss": 0.428,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.981501579284668,
|
|
"rewards/margins": 1.6247977018356323,
|
|
"rewards/rejected": -5.60629940032959,
|
|
"step": 231
|
|
},
|
|
{
|
|
"epoch": 0.49585893668180603,
|
|
"grad_norm": 18.591705682457018,
|
|
"learning_rate": 4.743646428651659e-07,
|
|
"logits/chosen": -1.1132831573486328,
|
|
"logits/rejected": -1.1119723320007324,
|
|
"logps/chosen": -640.3826904296875,
|
|
"logps/rejected": -761.2876586914062,
|
|
"loss": 0.4324,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.8130741119384766,
|
|
"rewards/margins": 1.1951515674591064,
|
|
"rewards/rejected": -5.008225440979004,
|
|
"step": 232
|
|
},
|
|
{
|
|
"epoch": 0.49799625968474487,
|
|
"grad_norm": 16.36807630859828,
|
|
"learning_rate": 4.7142275791945465e-07,
|
|
"logits/chosen": -1.1380739212036133,
|
|
"logits/rejected": -1.056833028793335,
|
|
"logps/chosen": -574.717529296875,
|
|
"logps/rejected": -770.259033203125,
|
|
"loss": 0.4061,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.5353829860687256,
|
|
"rewards/margins": 1.7000670433044434,
|
|
"rewards/rejected": -5.23544979095459,
|
|
"step": 233
|
|
},
|
|
{
|
|
"epoch": 0.5001335826876837,
|
|
"grad_norm": 20.982879510548624,
|
|
"learning_rate": 4.684768768791901e-07,
|
|
"logits/chosen": -1.2214264869689941,
|
|
"logits/rejected": -1.2726728916168213,
|
|
"logps/chosen": -561.918701171875,
|
|
"logps/rejected": -653.3917236328125,
|
|
"loss": 0.5084,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.465195655822754,
|
|
"rewards/margins": 0.9199050664901733,
|
|
"rewards/rejected": -4.385100841522217,
|
|
"step": 234
|
|
},
|
|
{
|
|
"epoch": 0.5022709056906225,
|
|
"grad_norm": 15.699299624363588,
|
|
"learning_rate": 4.6552716456605507e-07,
|
|
"logits/chosen": -1.0644322633743286,
|
|
"logits/rejected": -1.093167781829834,
|
|
"logps/chosen": -685.47021484375,
|
|
"logps/rejected": -693.9439697265625,
|
|
"loss": 0.3964,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -4.20876932144165,
|
|
"rewards/margins": 0.20507556200027466,
|
|
"rewards/rejected": -4.413845062255859,
|
|
"step": 235
|
|
},
|
|
{
|
|
"epoch": 0.5044082286935613,
|
|
"grad_norm": 15.274480085299626,
|
|
"learning_rate": 4.6257378601609237e-07,
|
|
"logits/chosen": -1.2189481258392334,
|
|
"logits/rejected": -1.183915376663208,
|
|
"logps/chosen": -632.9229736328125,
|
|
"logps/rejected": -755.5510864257812,
|
|
"loss": 0.4118,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.502497434616089,
|
|
"rewards/margins": 1.3661162853240967,
|
|
"rewards/rejected": -4.8686137199401855,
|
|
"step": 236
|
|
},
|
|
{
|
|
"epoch": 0.5065455516965002,
|
|
"grad_norm": 18.902872440282195,
|
|
"learning_rate": 4.596169064704697e-07,
|
|
"logits/chosen": -1.134831428527832,
|
|
"logits/rejected": -1.0569604635238647,
|
|
"logps/chosen": -603.5774536132812,
|
|
"logps/rejected": -1207.822998046875,
|
|
"loss": 0.4843,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -3.4753661155700684,
|
|
"rewards/margins": 0.5529882907867432,
|
|
"rewards/rejected": -4.028354644775391,
|
|
"step": 237
|
|
},
|
|
{
|
|
"epoch": 0.508682874699439,
|
|
"grad_norm": 14.821420949979046,
|
|
"learning_rate": 4.5665669136623563e-07,
|
|
"logits/chosen": -1.0976612567901611,
|
|
"logits/rejected": -1.0576744079589844,
|
|
"logps/chosen": -658.7794189453125,
|
|
"logps/rejected": -707.6447143554688,
|
|
"loss": 0.468,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.8431081771850586,
|
|
"rewards/margins": 0.5369940996170044,
|
|
"rewards/rejected": -4.380102157592773,
|
|
"step": 238
|
|
},
|
|
{
|
|
"epoch": 0.5108201977023777,
|
|
"grad_norm": 13.617631823910862,
|
|
"learning_rate": 4.536933063270622e-07,
|
|
"logits/chosen": -1.2480061054229736,
|
|
"logits/rejected": -1.1350898742675781,
|
|
"logps/chosen": -675.2293090820312,
|
|
"logps/rejected": -976.1577758789062,
|
|
"loss": 0.3874,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.117259979248047,
|
|
"rewards/margins": 2.769744634628296,
|
|
"rewards/rejected": -6.887004852294922,
|
|
"step": 239
|
|
},
|
|
{
|
|
"epoch": 0.5129575207053166,
|
|
"grad_norm": 14.571710656068836,
|
|
"learning_rate": 4.5072691715397937e-07,
|
|
"logits/chosen": -1.1107938289642334,
|
|
"logits/rejected": -1.0369246006011963,
|
|
"logps/chosen": -516.5548095703125,
|
|
"logps/rejected": -683.7294921875,
|
|
"loss": 0.4148,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.2809367179870605,
|
|
"rewards/margins": 1.5373725891113281,
|
|
"rewards/rejected": -4.818309307098389,
|
|
"step": 240
|
|
},
|
|
{
|
|
"epoch": 0.5150948437082554,
|
|
"grad_norm": 18.45094154637716,
|
|
"learning_rate": 4.477576898160976e-07,
|
|
"logits/chosen": -1.1420975923538208,
|
|
"logits/rejected": -1.1292245388031006,
|
|
"logps/chosen": -711.1873779296875,
|
|
"logps/rejected": -837.0042114257812,
|
|
"loss": 0.3439,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.162815570831299,
|
|
"rewards/margins": 1.0814441442489624,
|
|
"rewards/rejected": -5.244259834289551,
|
|
"step": 241
|
|
},
|
|
{
|
|
"epoch": 0.5172321667111942,
|
|
"grad_norm": 16.29215056433998,
|
|
"learning_rate": 4.447857904413231e-07,
|
|
"logits/chosen": -1.145432710647583,
|
|
"logits/rejected": -1.0742875337600708,
|
|
"logps/chosen": -512.1181030273438,
|
|
"logps/rejected": -708.3970947265625,
|
|
"loss": 0.3978,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -2.8924291133880615,
|
|
"rewards/margins": 1.711695909500122,
|
|
"rewards/rejected": -4.604125022888184,
|
|
"step": 242
|
|
},
|
|
{
|
|
"epoch": 0.5193694897141331,
|
|
"grad_norm": 16.775923990492764,
|
|
"learning_rate": 4.418113853070614e-07,
|
|
"logits/chosen": -1.1763932704925537,
|
|
"logits/rejected": -1.1778051853179932,
|
|
"logps/chosen": -601.1099853515625,
|
|
"logps/rejected": -744.1261596679688,
|
|
"loss": 0.4368,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.8077492713928223,
|
|
"rewards/margins": 1.3345444202423096,
|
|
"rewards/rejected": -5.1422929763793945,
|
|
"step": 243
|
|
},
|
|
{
|
|
"epoch": 0.5215068127170719,
|
|
"grad_norm": 15.43010582256506,
|
|
"learning_rate": 4.388346408309153e-07,
|
|
"logits/chosen": -1.029261589050293,
|
|
"logits/rejected": -0.904821515083313,
|
|
"logps/chosen": -537.7769775390625,
|
|
"logps/rejected": -740.9544677734375,
|
|
"loss": 0.4082,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.179456949234009,
|
|
"rewards/margins": 1.8058176040649414,
|
|
"rewards/rejected": -4.985274791717529,
|
|
"step": 244
|
|
},
|
|
{
|
|
"epoch": 0.5236441357200107,
|
|
"grad_norm": 16.998724541332553,
|
|
"learning_rate": 4.358557235613734e-07,
|
|
"logits/chosen": -1.211896300315857,
|
|
"logits/rejected": -1.17788565158844,
|
|
"logps/chosen": -609.4110107421875,
|
|
"logps/rejected": -725.498779296875,
|
|
"loss": 0.4229,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.361295223236084,
|
|
"rewards/margins": 1.2065562009811401,
|
|
"rewards/rejected": -4.5678510665893555,
|
|
"step": 245
|
|
},
|
|
{
|
|
"epoch": 0.5257814587229495,
|
|
"grad_norm": 15.563821028699236,
|
|
"learning_rate": 4.3287480016849214e-07,
|
|
"logits/chosen": -1.0585182905197144,
|
|
"logits/rejected": -1.056843876838684,
|
|
"logps/chosen": -631.421630859375,
|
|
"logps/rejected": -787.7451782226562,
|
|
"loss": 0.4121,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.867764472961426,
|
|
"rewards/margins": 1.3012657165527344,
|
|
"rewards/rejected": -5.169030666351318,
|
|
"step": 246
|
|
},
|
|
{
|
|
"epoch": 0.5279187817258884,
|
|
"grad_norm": 15.0812037854319,
|
|
"learning_rate": 4.298920374345697e-07,
|
|
"logits/chosen": -1.1332510709762573,
|
|
"logits/rejected": -1.1722335815429688,
|
|
"logps/chosen": -611.515625,
|
|
"logps/rejected": -733.9335327148438,
|
|
"loss": 0.4044,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.460312843322754,
|
|
"rewards/margins": 1.2320371866226196,
|
|
"rewards/rejected": -4.692350387573242,
|
|
"step": 247
|
|
},
|
|
{
|
|
"epoch": 0.5300561047288271,
|
|
"grad_norm": 19.373390619498192,
|
|
"learning_rate": 4.2690760224481585e-07,
|
|
"logits/chosen": -1.25339674949646,
|
|
"logits/rejected": -1.2282085418701172,
|
|
"logps/chosen": -507.8250732421875,
|
|
"logps/rejected": -594.3472290039062,
|
|
"loss": 0.5068,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -2.9802823066711426,
|
|
"rewards/margins": 0.9198769330978394,
|
|
"rewards/rejected": -3.9001595973968506,
|
|
"step": 248
|
|
},
|
|
{
|
|
"epoch": 0.5321934277317659,
|
|
"grad_norm": 18.365430353743545,
|
|
"learning_rate": 4.239216615780137e-07,
|
|
"logits/chosen": -1.0999282598495483,
|
|
"logits/rejected": -1.1215919256210327,
|
|
"logps/chosen": -714.6709594726562,
|
|
"logps/rejected": -876.0394897460938,
|
|
"loss": 0.4225,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.1298370361328125,
|
|
"rewards/margins": 1.5860515832901,
|
|
"rewards/rejected": -5.715888977050781,
|
|
"step": 249
|
|
},
|
|
{
|
|
"epoch": 0.5343307507347048,
|
|
"grad_norm": 20.578219385486992,
|
|
"learning_rate": 4.2093438249717754e-07,
|
|
"logits/chosen": -1.3405359983444214,
|
|
"logits/rejected": -1.3141045570373535,
|
|
"logps/chosen": -621.981201171875,
|
|
"logps/rejected": -748.8320922851562,
|
|
"loss": 0.5387,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.775174140930176,
|
|
"rewards/margins": 1.1837245225906372,
|
|
"rewards/rejected": -4.958899021148682,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 0.5364680737376436,
|
|
"grad_norm": 13.688542006328998,
|
|
"learning_rate": 4.17945932140206e-07,
|
|
"logits/chosen": -1.0330238342285156,
|
|
"logits/rejected": -0.9786776900291443,
|
|
"logps/chosen": -804.1392822265625,
|
|
"logps/rejected": -984.1054077148438,
|
|
"loss": 0.3583,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.976142406463623,
|
|
"rewards/margins": 1.77129065990448,
|
|
"rewards/rejected": -6.747433662414551,
|
|
"step": 251
|
|
},
|
|
{
|
|
"epoch": 0.5386053967405824,
|
|
"grad_norm": 23.49388537702244,
|
|
"learning_rate": 4.1495647771053034e-07,
|
|
"logits/chosen": -0.9389646649360657,
|
|
"logits/rejected": -1.0423595905303955,
|
|
"logps/chosen": -592.633544921875,
|
|
"logps/rejected": -688.3473510742188,
|
|
"loss": 0.4508,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -3.8458104133605957,
|
|
"rewards/margins": 0.9122557640075684,
|
|
"rewards/rejected": -4.758066654205322,
|
|
"step": 252
|
|
},
|
|
{
|
|
"epoch": 0.5407427197435213,
|
|
"grad_norm": 22.3306378383826,
|
|
"learning_rate": 4.119661864677594e-07,
|
|
"logits/chosen": -1.3090245723724365,
|
|
"logits/rejected": -1.293576955795288,
|
|
"logps/chosen": -752.04296875,
|
|
"logps/rejected": -861.8148193359375,
|
|
"loss": 0.4564,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.430123805999756,
|
|
"rewards/margins": 1.2242711782455444,
|
|
"rewards/rejected": -5.65439510345459,
|
|
"step": 253
|
|
},
|
|
{
|
|
"epoch": 0.5428800427464601,
|
|
"grad_norm": 14.971016637567855,
|
|
"learning_rate": 4.08975225718322e-07,
|
|
"logits/chosen": -1.172256350517273,
|
|
"logits/rejected": -1.1566345691680908,
|
|
"logps/chosen": -750.8662719726562,
|
|
"logps/rejected": -982.0726318359375,
|
|
"loss": 0.4086,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -4.331751823425293,
|
|
"rewards/margins": 2.4119136333465576,
|
|
"rewards/rejected": -6.74366569519043,
|
|
"step": 254
|
|
},
|
|
{
|
|
"epoch": 0.5450173657493989,
|
|
"grad_norm": 16.74852123402871,
|
|
"learning_rate": 4.0598376280610545e-07,
|
|
"logits/chosen": -1.1496988534927368,
|
|
"logits/rejected": -1.1922340393066406,
|
|
"logps/chosen": -729.1366577148438,
|
|
"logps/rejected": -863.9114990234375,
|
|
"loss": 0.4343,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.9464735984802246,
|
|
"rewards/margins": 1.6140131950378418,
|
|
"rewards/rejected": -5.560486793518066,
|
|
"step": 255
|
|
},
|
|
{
|
|
"epoch": 0.5471546887523377,
|
|
"grad_norm": 21.036518560558875,
|
|
"learning_rate": 4.029919651030932e-07,
|
|
"logits/chosen": -1.2442394495010376,
|
|
"logits/rejected": -1.224663496017456,
|
|
"logps/chosen": -604.781005859375,
|
|
"logps/rejected": -661.07763671875,
|
|
"loss": 0.4658,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.6011922359466553,
|
|
"rewards/margins": 0.4141632914543152,
|
|
"rewards/rejected": -4.015355587005615,
|
|
"step": 256
|
|
},
|
|
{
|
|
"epoch": 0.5492920117552765,
|
|
"grad_norm": 21.112582633149223,
|
|
"learning_rate": 4e-07,
|
|
"logits/chosen": -1.2898122072219849,
|
|
"logits/rejected": -1.235668659210205,
|
|
"logps/chosen": -665.9544677734375,
|
|
"logps/rejected": -848.035400390625,
|
|
"loss": 0.4932,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.9917383193969727,
|
|
"rewards/margins": 1.5706775188446045,
|
|
"rewards/rejected": -5.562415599822998,
|
|
"step": 257
|
|
},
|
|
{
|
|
"epoch": 0.5514293347582153,
|
|
"grad_norm": 17.267156100092052,
|
|
"learning_rate": 3.9700803489690677e-07,
|
|
"logits/chosen": -1.2663730382919312,
|
|
"logits/rejected": -1.2143464088439941,
|
|
"logps/chosen": -638.0128173828125,
|
|
"logps/rejected": -827.65966796875,
|
|
"loss": 0.3859,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.7429115772247314,
|
|
"rewards/margins": 1.5405060052871704,
|
|
"rewards/rejected": -5.283417701721191,
|
|
"step": 258
|
|
},
|
|
{
|
|
"epoch": 0.5535666577611541,
|
|
"grad_norm": 14.88755803006458,
|
|
"learning_rate": 3.940162371938946e-07,
|
|
"logits/chosen": -1.1717605590820312,
|
|
"logits/rejected": -1.1811717748641968,
|
|
"logps/chosen": -575.023681640625,
|
|
"logps/rejected": -674.9644165039062,
|
|
"loss": 0.385,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.434030771255493,
|
|
"rewards/margins": 0.9599059224128723,
|
|
"rewards/rejected": -4.393936634063721,
|
|
"step": 259
|
|
},
|
|
{
|
|
"epoch": 0.555703980764093,
|
|
"grad_norm": 15.467774533794707,
|
|
"learning_rate": 3.91024774281678e-07,
|
|
"logits/chosen": -1.231236457824707,
|
|
"logits/rejected": -1.3122860193252563,
|
|
"logps/chosen": -754.0248413085938,
|
|
"logps/rejected": -851.9044799804688,
|
|
"loss": 0.4214,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.264395713806152,
|
|
"rewards/margins": 1.528761625289917,
|
|
"rewards/rejected": -5.79315710067749,
|
|
"step": 260
|
|
},
|
|
{
|
|
"epoch": 0.5578413037670318,
|
|
"grad_norm": 14.748190228015728,
|
|
"learning_rate": 3.880338135322407e-07,
|
|
"logits/chosen": -0.9601567983627319,
|
|
"logits/rejected": -0.9594780206680298,
|
|
"logps/chosen": -532.7817993164062,
|
|
"logps/rejected": -641.3465576171875,
|
|
"loss": 0.4361,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -3.3210856914520264,
|
|
"rewards/margins": 1.0201408863067627,
|
|
"rewards/rejected": -4.341227054595947,
|
|
"step": 261
|
|
},
|
|
{
|
|
"epoch": 0.5599786267699706,
|
|
"grad_norm": 18.112744314127806,
|
|
"learning_rate": 3.850435222894698e-07,
|
|
"logits/chosen": -1.0205025672912598,
|
|
"logits/rejected": -1.062712550163269,
|
|
"logps/chosen": -724.7273559570312,
|
|
"logps/rejected": -883.1221923828125,
|
|
"loss": 0.4313,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.406254291534424,
|
|
"rewards/margins": 1.7283343076705933,
|
|
"rewards/rejected": -6.134588718414307,
|
|
"step": 262
|
|
},
|
|
{
|
|
"epoch": 0.5621159497729095,
|
|
"grad_norm": 18.973955414256828,
|
|
"learning_rate": 3.8205406785979413e-07,
|
|
"logits/chosen": -1.1698824167251587,
|
|
"logits/rejected": -1.245084524154663,
|
|
"logps/chosen": -756.6258544921875,
|
|
"logps/rejected": -888.836669921875,
|
|
"loss": 0.4413,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.1967387199401855,
|
|
"rewards/margins": 1.4100241661071777,
|
|
"rewards/rejected": -5.606762886047363,
|
|
"step": 263
|
|
},
|
|
{
|
|
"epoch": 0.5642532727758482,
|
|
"grad_norm": 18.24827242372365,
|
|
"learning_rate": 3.790656175028224e-07,
|
|
"logits/chosen": -1.0585461854934692,
|
|
"logits/rejected": -1.0999590158462524,
|
|
"logps/chosen": -639.7376708984375,
|
|
"logps/rejected": -804.4436645507812,
|
|
"loss": 0.4411,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.9402523040771484,
|
|
"rewards/margins": 1.6609243154525757,
|
|
"rewards/rejected": -5.601176738739014,
|
|
"step": 264
|
|
},
|
|
{
|
|
"epoch": 0.566390595778787,
|
|
"grad_norm": 24.668076315985935,
|
|
"learning_rate": 3.760783384219863e-07,
|
|
"logits/chosen": -1.0429140329360962,
|
|
"logits/rejected": -1.0325287580490112,
|
|
"logps/chosen": -712.508544921875,
|
|
"logps/rejected": -817.262451171875,
|
|
"loss": 0.5696,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -4.626794815063477,
|
|
"rewards/margins": 1.1756452322006226,
|
|
"rewards/rejected": -5.802440166473389,
|
|
"step": 265
|
|
},
|
|
{
|
|
"epoch": 0.5685279187817259,
|
|
"grad_norm": 20.420005320382554,
|
|
"learning_rate": 3.730923977551841e-07,
|
|
"logits/chosen": -1.0772805213928223,
|
|
"logits/rejected": -1.1414085626602173,
|
|
"logps/chosen": -486.30316162109375,
|
|
"logps/rejected": -534.7408447265625,
|
|
"loss": 0.4484,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.166795492172241,
|
|
"rewards/margins": 0.49554774165153503,
|
|
"rewards/rejected": -3.6623430252075195,
|
|
"step": 266
|
|
},
|
|
{
|
|
"epoch": 0.5706652417846647,
|
|
"grad_norm": 16.033046055991687,
|
|
"learning_rate": 3.7010796256543027e-07,
|
|
"logits/chosen": -1.0211303234100342,
|
|
"logits/rejected": -1.028752088546753,
|
|
"logps/chosen": -680.7797241210938,
|
|
"logps/rejected": -796.4379272460938,
|
|
"loss": 0.4204,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -4.561736583709717,
|
|
"rewards/margins": 1.135624647140503,
|
|
"rewards/rejected": -5.697361469268799,
|
|
"step": 267
|
|
},
|
|
{
|
|
"epoch": 0.5728025647876035,
|
|
"grad_norm": 13.506751879581227,
|
|
"learning_rate": 3.671251998315079e-07,
|
|
"logits/chosen": -1.143643856048584,
|
|
"logits/rejected": -1.1033873558044434,
|
|
"logps/chosen": -630.0643920898438,
|
|
"logps/rejected": -847.8607177734375,
|
|
"loss": 0.3377,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -3.909426689147949,
|
|
"rewards/margins": 2.0678324699401855,
|
|
"rewards/rejected": -5.977258682250977,
|
|
"step": 268
|
|
},
|
|
{
|
|
"epoch": 0.5749398877905424,
|
|
"grad_norm": 23.078993663211875,
|
|
"learning_rate": 3.641442764386266e-07,
|
|
"logits/chosen": -1.2684578895568848,
|
|
"logits/rejected": -1.2815937995910645,
|
|
"logps/chosen": -547.9085693359375,
|
|
"logps/rejected": -619.15673828125,
|
|
"loss": 0.5744,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -3.3864290714263916,
|
|
"rewards/margins": 0.6473042964935303,
|
|
"rewards/rejected": -4.033733367919922,
|
|
"step": 269
|
|
},
|
|
{
|
|
"epoch": 0.5770772107934812,
|
|
"grad_norm": 16.148785025450067,
|
|
"learning_rate": 3.611653591690847e-07,
|
|
"logits/chosen": -1.0950298309326172,
|
|
"logits/rejected": -1.1259742975234985,
|
|
"logps/chosen": -606.2462158203125,
|
|
"logps/rejected": -697.3421020507812,
|
|
"loss": 0.4529,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.8272175788879395,
|
|
"rewards/margins": 0.8501311540603638,
|
|
"rewards/rejected": -4.677348613739014,
|
|
"step": 270
|
|
},
|
|
{
|
|
"epoch": 0.57921453379642,
|
|
"grad_norm": 14.989894099730115,
|
|
"learning_rate": 3.5818861469293866e-07,
|
|
"logits/chosen": -1.1437263488769531,
|
|
"logits/rejected": -1.1254221200942993,
|
|
"logps/chosen": -622.5841674804688,
|
|
"logps/rejected": -733.7385864257812,
|
|
"loss": 0.3795,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.9022326469421387,
|
|
"rewards/margins": 1.0598499774932861,
|
|
"rewards/rejected": -4.962083339691162,
|
|
"step": 271
|
|
},
|
|
{
|
|
"epoch": 0.5813518567993589,
|
|
"grad_norm": 16.156205336224875,
|
|
"learning_rate": 3.5521420955867687e-07,
|
|
"logits/chosen": -1.0730040073394775,
|
|
"logits/rejected": -1.0788663625717163,
|
|
"logps/chosen": -613.4228515625,
|
|
"logps/rejected": -743.5269165039062,
|
|
"loss": 0.392,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.553554058074951,
|
|
"rewards/margins": 1.3978602886199951,
|
|
"rewards/rejected": -4.951415061950684,
|
|
"step": 272
|
|
},
|
|
{
|
|
"epoch": 0.5834891798022976,
|
|
"grad_norm": 16.15777168809771,
|
|
"learning_rate": 3.5224231018390234e-07,
|
|
"logits/chosen": -1.2778537273406982,
|
|
"logits/rejected": -1.1532944440841675,
|
|
"logps/chosen": -565.4224853515625,
|
|
"logps/rejected": -756.1430053710938,
|
|
"loss": 0.4587,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.2680768966674805,
|
|
"rewards/margins": 1.7718642950057983,
|
|
"rewards/rejected": -5.039941787719727,
|
|
"step": 273
|
|
},
|
|
{
|
|
"epoch": 0.5856265028052364,
|
|
"grad_norm": 14.465499992590448,
|
|
"learning_rate": 3.492730828460207e-07,
|
|
"logits/chosen": -1.1964144706726074,
|
|
"logits/rejected": -1.0841807126998901,
|
|
"logps/chosen": -606.995849609375,
|
|
"logps/rejected": -809.55224609375,
|
|
"loss": 0.4038,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.509310245513916,
|
|
"rewards/margins": 1.7641215324401855,
|
|
"rewards/rejected": -5.273431777954102,
|
|
"step": 274
|
|
},
|
|
{
|
|
"epoch": 0.5877638258081752,
|
|
"grad_norm": 18.62595847705199,
|
|
"learning_rate": 3.4630669367293793e-07,
|
|
"logits/chosen": -1.322818398475647,
|
|
"logits/rejected": -1.2504830360412598,
|
|
"logps/chosen": -557.3622436523438,
|
|
"logps/rejected": -669.48974609375,
|
|
"loss": 0.4694,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.3967947959899902,
|
|
"rewards/margins": 0.9858072996139526,
|
|
"rewards/rejected": -4.382602214813232,
|
|
"step": 275
|
|
},
|
|
{
|
|
"epoch": 0.5899011488111141,
|
|
"grad_norm": 14.846250971855795,
|
|
"learning_rate": 3.4334330863376444e-07,
|
|
"logits/chosen": -1.0945661067962646,
|
|
"logits/rejected": -1.0822480916976929,
|
|
"logps/chosen": -538.973388671875,
|
|
"logps/rejected": -668.257568359375,
|
|
"loss": 0.3706,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.1467766761779785,
|
|
"rewards/margins": 1.3210586309432983,
|
|
"rewards/rejected": -4.467835903167725,
|
|
"step": 276
|
|
},
|
|
{
|
|
"epoch": 0.5920384718140529,
|
|
"grad_norm": 15.039517764071677,
|
|
"learning_rate": 3.4038309352953014e-07,
|
|
"logits/chosen": -1.1040856838226318,
|
|
"logits/rejected": -1.1212596893310547,
|
|
"logps/chosen": -417.47808837890625,
|
|
"logps/rejected": -508.0076599121094,
|
|
"loss": 0.3713,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -2.398800849914551,
|
|
"rewards/margins": 0.8262351751327515,
|
|
"rewards/rejected": -3.225036382675171,
|
|
"step": 277
|
|
},
|
|
{
|
|
"epoch": 0.5941757948169917,
|
|
"grad_norm": 13.350623045669444,
|
|
"learning_rate": 3.3742621398390754e-07,
|
|
"logits/chosen": -1.0822662115097046,
|
|
"logits/rejected": -1.0757577419281006,
|
|
"logps/chosen": -532.8475341796875,
|
|
"logps/rejected": -611.3041381835938,
|
|
"loss": 0.3868,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -3.1749584674835205,
|
|
"rewards/margins": 0.6507718563079834,
|
|
"rewards/rejected": -3.8257298469543457,
|
|
"step": 278
|
|
},
|
|
{
|
|
"epoch": 0.5963131178199306,
|
|
"grad_norm": 15.78335853431774,
|
|
"learning_rate": 3.344728354339449e-07,
|
|
"logits/chosen": -1.2665464878082275,
|
|
"logits/rejected": -1.2659828662872314,
|
|
"logps/chosen": -554.4942016601562,
|
|
"logps/rejected": -625.689208984375,
|
|
"loss": 0.3992,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -3.3500773906707764,
|
|
"rewards/margins": 0.8534866571426392,
|
|
"rewards/rejected": -4.203563690185547,
|
|
"step": 279
|
|
},
|
|
{
|
|
"epoch": 0.5984504408228694,
|
|
"grad_norm": 18.33665050808218,
|
|
"learning_rate": 3.315231231208099e-07,
|
|
"logits/chosen": -1.3685458898544312,
|
|
"logits/rejected": -1.233042597770691,
|
|
"logps/chosen": -728.152099609375,
|
|
"logps/rejected": -849.6878662109375,
|
|
"loss": 0.4596,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.0944743156433105,
|
|
"rewards/margins": 1.1558749675750732,
|
|
"rewards/rejected": -5.250349998474121,
|
|
"step": 280
|
|
},
|
|
{
|
|
"epoch": 0.6005877638258081,
|
|
"grad_norm": 18.245751326222873,
|
|
"learning_rate": 3.2857724208054536e-07,
|
|
"logits/chosen": -1.3227458000183105,
|
|
"logits/rejected": -1.2658473253250122,
|
|
"logps/chosen": -644.74755859375,
|
|
"logps/rejected": -807.0831909179688,
|
|
"loss": 0.4902,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.4471018314361572,
|
|
"rewards/margins": 1.4783265590667725,
|
|
"rewards/rejected": -4.92542839050293,
|
|
"step": 281
|
|
},
|
|
{
|
|
"epoch": 0.602725086828747,
|
|
"grad_norm": 15.332405651097067,
|
|
"learning_rate": 3.256353571348342e-07,
|
|
"logits/chosen": -1.2925728559494019,
|
|
"logits/rejected": -1.2592719793319702,
|
|
"logps/chosen": -593.9208984375,
|
|
"logps/rejected": -715.414306640625,
|
|
"loss": 0.4138,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.4808356761932373,
|
|
"rewards/margins": 1.198320984840393,
|
|
"rewards/rejected": -4.679156303405762,
|
|
"step": 282
|
|
},
|
|
{
|
|
"epoch": 0.6048624098316858,
|
|
"grad_norm": 20.021176403122144,
|
|
"learning_rate": 3.2269763288177875e-07,
|
|
"logits/chosen": -1.150704026222229,
|
|
"logits/rejected": -1.1762549877166748,
|
|
"logps/chosen": -602.2947387695312,
|
|
"logps/rejected": -703.0159912109375,
|
|
"loss": 0.3747,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.3265883922576904,
|
|
"rewards/margins": 1.0499697923660278,
|
|
"rewards/rejected": -4.376558303833008,
|
|
"step": 283
|
|
},
|
|
{
|
|
"epoch": 0.6069997328346246,
|
|
"grad_norm": 19.739728018611924,
|
|
"learning_rate": 3.197642336866909e-07,
|
|
"logits/chosen": -1.1246461868286133,
|
|
"logits/rejected": -1.1267664432525635,
|
|
"logps/chosen": -591.5923461914062,
|
|
"logps/rejected": -673.4954833984375,
|
|
"loss": 0.4876,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -3.6883199214935303,
|
|
"rewards/margins": 0.7502920627593994,
|
|
"rewards/rejected": -4.43861198425293,
|
|
"step": 284
|
|
},
|
|
{
|
|
"epoch": 0.6091370558375635,
|
|
"grad_norm": 23.876230746394956,
|
|
"learning_rate": 3.168353236728963e-07,
|
|
"logits/chosen": -1.2377270460128784,
|
|
"logits/rejected": -1.242301106452942,
|
|
"logps/chosen": -601.439697265625,
|
|
"logps/rejected": -642.472412109375,
|
|
"loss": 0.5947,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -3.569728374481201,
|
|
"rewards/margins": 0.6284288167953491,
|
|
"rewards/rejected": -4.19815731048584,
|
|
"step": 285
|
|
},
|
|
{
|
|
"epoch": 0.6112743788405023,
|
|
"grad_norm": 22.209485552755492,
|
|
"learning_rate": 3.139110667125512e-07,
|
|
"logits/chosen": -0.9830377697944641,
|
|
"logits/rejected": -0.9108029007911682,
|
|
"logps/chosen": -562.501708984375,
|
|
"logps/rejected": -705.535400390625,
|
|
"loss": 0.4883,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.516263961791992,
|
|
"rewards/margins": 1.286893367767334,
|
|
"rewards/rejected": -4.803157806396484,
|
|
"step": 286
|
|
},
|
|
{
|
|
"epoch": 0.6134117018434411,
|
|
"grad_norm": 19.053259171815892,
|
|
"learning_rate": 3.1099162641747425e-07,
|
|
"logits/chosen": -0.9869133830070496,
|
|
"logits/rejected": -1.0198793411254883,
|
|
"logps/chosen": -587.336669921875,
|
|
"logps/rejected": -673.4232788085938,
|
|
"loss": 0.4925,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.481856107711792,
|
|
"rewards/margins": 0.7810866832733154,
|
|
"rewards/rejected": -4.262942314147949,
|
|
"step": 287
|
|
},
|
|
{
|
|
"epoch": 0.6155490248463799,
|
|
"grad_norm": 20.90972346201698,
|
|
"learning_rate": 3.0807716612999227e-07,
|
|
"logits/chosen": -1.2177681922912598,
|
|
"logits/rejected": -1.2541393041610718,
|
|
"logps/chosen": -608.8695068359375,
|
|
"logps/rejected": -738.279052734375,
|
|
"loss": 0.4164,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.7167322635650635,
|
|
"rewards/margins": 1.185416340827942,
|
|
"rewards/rejected": -4.902148723602295,
|
|
"step": 288
|
|
},
|
|
{
|
|
"epoch": 0.6176863478493188,
|
|
"grad_norm": 17.392772948265563,
|
|
"learning_rate": 3.05167848913801e-07,
|
|
"logits/chosen": -1.1927785873413086,
|
|
"logits/rejected": -1.101136326789856,
|
|
"logps/chosen": -615.4705810546875,
|
|
"logps/rejected": -807.6544799804688,
|
|
"loss": 0.4141,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.5115911960601807,
|
|
"rewards/margins": 1.9201233386993408,
|
|
"rewards/rejected": -5.4317145347595215,
|
|
"step": 289
|
|
},
|
|
{
|
|
"epoch": 0.6198236708522575,
|
|
"grad_norm": 39.464356706878,
|
|
"learning_rate": 3.022638375448423e-07,
|
|
"logits/chosen": -1.198436975479126,
|
|
"logits/rejected": -1.1444064378738403,
|
|
"logps/chosen": -669.1489868164062,
|
|
"logps/rejected": -752.60009765625,
|
|
"loss": 0.4269,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -4.052773952484131,
|
|
"rewards/margins": 0.8161685466766357,
|
|
"rewards/rejected": -4.868942737579346,
|
|
"step": 290
|
|
},
|
|
{
|
|
"epoch": 0.6219609938551963,
|
|
"grad_norm": 14.841377854168094,
|
|
"learning_rate": 2.993652945021966e-07,
|
|
"logits/chosen": -1.2626550197601318,
|
|
"logits/rejected": -1.2754631042480469,
|
|
"logps/chosen": -603.2761840820312,
|
|
"logps/rejected": -811.0333251953125,
|
|
"loss": 0.3819,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.527158737182617,
|
|
"rewards/margins": 1.902860403060913,
|
|
"rewards/rejected": -5.430018424987793,
|
|
"step": 291
|
|
},
|
|
{
|
|
"epoch": 0.6240983168581352,
|
|
"grad_norm": 19.276604227233815,
|
|
"learning_rate": 2.9647238195899166e-07,
|
|
"logits/chosen": -1.097237467765808,
|
|
"logits/rejected": -1.037533164024353,
|
|
"logps/chosen": -494.391845703125,
|
|
"logps/rejected": -708.1276245117188,
|
|
"loss": 0.3576,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -2.6569738388061523,
|
|
"rewards/margins": 2.002774238586426,
|
|
"rewards/rejected": -4.659748077392578,
|
|
"step": 292
|
|
},
|
|
{
|
|
"epoch": 0.626235639861074,
|
|
"grad_norm": 18.95704234795035,
|
|
"learning_rate": 2.9358526177332995e-07,
|
|
"logits/chosen": -1.1904467344284058,
|
|
"logits/rejected": -1.1917550563812256,
|
|
"logps/chosen": -709.7570190429688,
|
|
"logps/rejected": -801.443359375,
|
|
"loss": 0.4692,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.291625022888184,
|
|
"rewards/margins": 0.9095453023910522,
|
|
"rewards/rejected": -5.201169967651367,
|
|
"step": 293
|
|
},
|
|
{
|
|
"epoch": 0.6283729628640128,
|
|
"grad_norm": 18.243219081118724,
|
|
"learning_rate": 2.90704095479232e-07,
|
|
"logits/chosen": -1.002289056777954,
|
|
"logits/rejected": -1.0251970291137695,
|
|
"logps/chosen": -596.549560546875,
|
|
"logps/rejected": -699.5868530273438,
|
|
"loss": 0.5095,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.7809488773345947,
|
|
"rewards/margins": 0.9690990447998047,
|
|
"rewards/rejected": -4.7500481605529785,
|
|
"step": 294
|
|
},
|
|
{
|
|
"epoch": 0.6305102858669517,
|
|
"grad_norm": 17.459161625382947,
|
|
"learning_rate": 2.8782904427759893e-07,
|
|
"logits/chosen": -1.3322062492370605,
|
|
"logits/rejected": -1.3123010396957397,
|
|
"logps/chosen": -506.9045715332031,
|
|
"logps/rejected": -793.8304443359375,
|
|
"loss": 0.4079,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -2.7689664363861084,
|
|
"rewards/margins": 2.5288219451904297,
|
|
"rewards/rejected": -5.297788619995117,
|
|
"step": 295
|
|
},
|
|
{
|
|
"epoch": 0.6326476088698905,
|
|
"grad_norm": 16.674107473408924,
|
|
"learning_rate": 2.849602690271931e-07,
|
|
"logits/chosen": -1.151752233505249,
|
|
"logits/rejected": -1.1323444843292236,
|
|
"logps/chosen": -626.8087768554688,
|
|
"logps/rejected": -798.4351806640625,
|
|
"loss": 0.3672,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -3.600851058959961,
|
|
"rewards/margins": 1.5587782859802246,
|
|
"rewards/rejected": -5.1596293449401855,
|
|
"step": 296
|
|
},
|
|
{
|
|
"epoch": 0.6347849318728293,
|
|
"grad_norm": 17.154415301310323,
|
|
"learning_rate": 2.820979302356383e-07,
|
|
"logits/chosen": -1.213531494140625,
|
|
"logits/rejected": -1.1952483654022217,
|
|
"logps/chosen": -578.425537109375,
|
|
"logps/rejected": -722.8228759765625,
|
|
"loss": 0.3553,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.1612937450408936,
|
|
"rewards/margins": 1.449401617050171,
|
|
"rewards/rejected": -4.6106953620910645,
|
|
"step": 297
|
|
},
|
|
{
|
|
"epoch": 0.6369222548757681,
|
|
"grad_norm": 18.96014182785911,
|
|
"learning_rate": 2.792421880504392e-07,
|
|
"logits/chosen": -1.2606010437011719,
|
|
"logits/rejected": -1.236906886100769,
|
|
"logps/chosen": -544.4500122070312,
|
|
"logps/rejected": -641.5679931640625,
|
|
"loss": 0.3772,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.2772369384765625,
|
|
"rewards/margins": 0.9185816049575806,
|
|
"rewards/rejected": -4.195818901062012,
|
|
"step": 298
|
|
},
|
|
{
|
|
"epoch": 0.6390595778787069,
|
|
"grad_norm": 23.422220984702424,
|
|
"learning_rate": 2.7639320225002107e-07,
|
|
"logits/chosen": -1.2836658954620361,
|
|
"logits/rejected": -1.2603038549423218,
|
|
"logps/chosen": -521.1630859375,
|
|
"logps/rejected": -686.9148559570312,
|
|
"loss": 0.4465,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.0996432304382324,
|
|
"rewards/margins": 1.5820202827453613,
|
|
"rewards/rejected": -4.681663990020752,
|
|
"step": 299
|
|
},
|
|
{
|
|
"epoch": 0.6411969008816457,
|
|
"grad_norm": 17.358355183824074,
|
|
"learning_rate": 2.735511322347904e-07,
|
|
"logits/chosen": -1.08761727809906,
|
|
"logits/rejected": -1.0740941762924194,
|
|
"logps/chosen": -684.2798461914062,
|
|
"logps/rejected": -888.6907958984375,
|
|
"loss": 0.3663,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.057728290557861,
|
|
"rewards/margins": 2.032811164855957,
|
|
"rewards/rejected": -6.090539455413818,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.6411969008816457,
|
|
"eval_logits/chosen": -1.3138450384140015,
|
|
"eval_logits/rejected": -1.3326525688171387,
|
|
"eval_logps/chosen": -668.366943359375,
|
|
"eval_logps/rejected": -808.5509033203125,
|
|
"eval_loss": 0.40117064118385315,
|
|
"eval_rewards/accuracies": 0.8387096524238586,
|
|
"eval_rewards/chosen": -3.888610601425171,
|
|
"eval_rewards/margins": 1.4613444805145264,
|
|
"eval_rewards/rejected": -5.349954605102539,
|
|
"eval_runtime": 127.6244,
|
|
"eval_samples_per_second": 15.365,
|
|
"eval_steps_per_second": 0.486,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.6433342238845846,
|
|
"grad_norm": 20.781279404805886,
|
|
"learning_rate": 2.7071613701821607e-07,
|
|
"logits/chosen": -1.1530544757843018,
|
|
"logits/rejected": -1.2095261812210083,
|
|
"logps/chosen": -620.2159423828125,
|
|
"logps/rejected": -744.0250854492188,
|
|
"loss": 0.4255,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.827709436416626,
|
|
"rewards/margins": 1.2618365287780762,
|
|
"rewards/rejected": -5.089546203613281,
|
|
"step": 301
|
|
},
|
|
{
|
|
"epoch": 0.6454715468875234,
|
|
"grad_norm": 20.821176231506904,
|
|
"learning_rate": 2.678883752179333e-07,
|
|
"logits/chosen": -1.2675704956054688,
|
|
"logits/rejected": -1.3029217720031738,
|
|
"logps/chosen": -689.5173950195312,
|
|
"logps/rejected": -833.730712890625,
|
|
"loss": 0.4655,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.1408796310424805,
|
|
"rewards/margins": 1.3784571886062622,
|
|
"rewards/rejected": -5.519336223602295,
|
|
"step": 302
|
|
},
|
|
{
|
|
"epoch": 0.6476088698904622,
|
|
"grad_norm": 20.312995975775525,
|
|
"learning_rate": 2.65068005046868e-07,
|
|
"logits/chosen": -1.3904988765716553,
|
|
"logits/rejected": -1.4191157817840576,
|
|
"logps/chosen": -761.3741455078125,
|
|
"logps/rejected": -1012.2886962890625,
|
|
"loss": 0.4047,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -4.49802303314209,
|
|
"rewards/margins": 2.586608409881592,
|
|
"rewards/rejected": -7.08463191986084,
|
|
"step": 303
|
|
},
|
|
{
|
|
"epoch": 0.649746192893401,
|
|
"grad_norm": 18.605444040069756,
|
|
"learning_rate": 2.622551843043859e-07,
|
|
"logits/chosen": -1.2163584232330322,
|
|
"logits/rejected": -1.2040042877197266,
|
|
"logps/chosen": -582.2776489257812,
|
|
"logps/rejected": -685.1724853515625,
|
|
"loss": 0.4572,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.6959495544433594,
|
|
"rewards/margins": 0.9104414582252502,
|
|
"rewards/rejected": -4.606390953063965,
|
|
"step": 304
|
|
},
|
|
{
|
|
"epoch": 0.6518835158963399,
|
|
"grad_norm": 16.54860058502452,
|
|
"learning_rate": 2.5945007036746285e-07,
|
|
"logits/chosen": -1.2848327159881592,
|
|
"logits/rejected": -1.2738027572631836,
|
|
"logps/chosen": -659.4707641601562,
|
|
"logps/rejected": -836.122802734375,
|
|
"loss": 0.4086,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.9217731952667236,
|
|
"rewards/margins": 1.8700674772262573,
|
|
"rewards/rejected": -5.791840553283691,
|
|
"step": 305
|
|
},
|
|
{
|
|
"epoch": 0.6540208388992786,
|
|
"grad_norm": 18.88108514365039,
|
|
"learning_rate": 2.5665282018187986e-07,
|
|
"logits/chosen": -1.2504708766937256,
|
|
"logits/rejected": -1.2493597269058228,
|
|
"logps/chosen": -616.8862915039062,
|
|
"logps/rejected": -682.9422607421875,
|
|
"loss": 0.4277,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.915069818496704,
|
|
"rewards/margins": 0.7035645246505737,
|
|
"rewards/rejected": -4.618634223937988,
|
|
"step": 306
|
|
},
|
|
{
|
|
"epoch": 0.6561581619022174,
|
|
"grad_norm": 26.06295428477828,
|
|
"learning_rate": 2.5386359025344197e-07,
|
|
"logits/chosen": -1.2800426483154297,
|
|
"logits/rejected": -1.2985005378723145,
|
|
"logps/chosen": -688.5096435546875,
|
|
"logps/rejected": -803.2349243164062,
|
|
"loss": 0.4146,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.173797130584717,
|
|
"rewards/margins": 1.3077186346054077,
|
|
"rewards/rejected": -5.481515884399414,
|
|
"step": 307
|
|
},
|
|
{
|
|
"epoch": 0.6582954849051563,
|
|
"grad_norm": 14.94353300613803,
|
|
"learning_rate": 2.510825366392219e-07,
|
|
"logits/chosen": -1.27797269821167,
|
|
"logits/rejected": -1.263561725616455,
|
|
"logps/chosen": -560.7982788085938,
|
|
"logps/rejected": -704.850341796875,
|
|
"loss": 0.3413,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.493176221847534,
|
|
"rewards/margins": 1.3737064599990845,
|
|
"rewards/rejected": -4.86688232421875,
|
|
"step": 308
|
|
},
|
|
{
|
|
"epoch": 0.6604328079080951,
|
|
"grad_norm": 20.334800907510225,
|
|
"learning_rate": 2.4830981493882865e-07,
|
|
"logits/chosen": -1.2430211305618286,
|
|
"logits/rejected": -1.275850534439087,
|
|
"logps/chosen": -584.9478759765625,
|
|
"logps/rejected": -739.5841674804688,
|
|
"loss": 0.464,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.6128716468811035,
|
|
"rewards/margins": 1.7369210720062256,
|
|
"rewards/rejected": -5.34979248046875,
|
|
"step": 309
|
|
},
|
|
{
|
|
"epoch": 0.6625701309110339,
|
|
"grad_norm": 17.83873419658186,
|
|
"learning_rate": 2.455455802857015e-07,
|
|
"logits/chosen": -1.1412709951400757,
|
|
"logits/rejected": -1.2062087059020996,
|
|
"logps/chosen": -765.4309692382812,
|
|
"logps/rejected": -982.59765625,
|
|
"loss": 0.3761,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.830318927764893,
|
|
"rewards/margins": 1.9539070129394531,
|
|
"rewards/rejected": -6.784224987030029,
|
|
"step": 310
|
|
},
|
|
{
|
|
"epoch": 0.6647074539139728,
|
|
"grad_norm": 16.263277573360906,
|
|
"learning_rate": 2.4278998733843056e-07,
|
|
"logits/chosen": -1.149855136871338,
|
|
"logits/rejected": -1.1428004503250122,
|
|
"logps/chosen": -699.1431274414062,
|
|
"logps/rejected": -828.4837036132812,
|
|
"loss": 0.428,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.370693206787109,
|
|
"rewards/margins": 1.2527273893356323,
|
|
"rewards/rejected": -5.623420715332031,
|
|
"step": 311
|
|
},
|
|
{
|
|
"epoch": 0.6668447769169116,
|
|
"grad_norm": 21.249199073124323,
|
|
"learning_rate": 2.4004319027210367e-07,
|
|
"logits/chosen": -1.0749962329864502,
|
|
"logits/rejected": -1.0791423320770264,
|
|
"logps/chosen": -673.4476318359375,
|
|
"logps/rejected": -822.49169921875,
|
|
"loss": 0.3787,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.9068007469177246,
|
|
"rewards/margins": 1.6602363586425781,
|
|
"rewards/rejected": -5.567037105560303,
|
|
"step": 312
|
|
},
|
|
{
|
|
"epoch": 0.6689820999198504,
|
|
"grad_norm": 19.210337957305978,
|
|
"learning_rate": 2.3730534276967998e-07,
|
|
"logits/chosen": -1.2630716562271118,
|
|
"logits/rejected": -1.2258106470108032,
|
|
"logps/chosen": -718.524658203125,
|
|
"logps/rejected": -831.8648681640625,
|
|
"loss": 0.4599,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.357652187347412,
|
|
"rewards/margins": 1.134694218635559,
|
|
"rewards/rejected": -5.492346286773682,
|
|
"step": 313
|
|
},
|
|
{
|
|
"epoch": 0.6711194229227893,
|
|
"grad_norm": 19.848471724389256,
|
|
"learning_rate": 2.345765980133918e-07,
|
|
"logits/chosen": -1.2012803554534912,
|
|
"logits/rejected": -1.1929750442504883,
|
|
"logps/chosen": -779.7426147460938,
|
|
"logps/rejected": -1021.65625,
|
|
"loss": 0.4261,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.710476875305176,
|
|
"rewards/margins": 2.3127245903015137,
|
|
"rewards/rejected": -7.023200988769531,
|
|
"step": 314
|
|
},
|
|
{
|
|
"epoch": 0.673256745925728,
|
|
"grad_norm": 16.96894863528391,
|
|
"learning_rate": 2.3185710867617386e-07,
|
|
"logits/chosen": -1.13607656955719,
|
|
"logits/rejected": -1.1218336820602417,
|
|
"logps/chosen": -734.45751953125,
|
|
"logps/rejected": -819.6937866210938,
|
|
"loss": 0.4107,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.487407684326172,
|
|
"rewards/margins": 0.7835354804992676,
|
|
"rewards/rejected": -5.2709431648254395,
|
|
"step": 315
|
|
},
|
|
{
|
|
"epoch": 0.6753940689286668,
|
|
"grad_norm": 23.53945549697492,
|
|
"learning_rate": 2.2914702691312144e-07,
|
|
"logits/chosen": -1.2212194204330444,
|
|
"logits/rejected": -1.2105563879013062,
|
|
"logps/chosen": -656.0445556640625,
|
|
"logps/rejected": -824.4144287109375,
|
|
"loss": 0.4747,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.8364951610565186,
|
|
"rewards/margins": 1.6471456289291382,
|
|
"rewards/rejected": -5.483640670776367,
|
|
"step": 316
|
|
},
|
|
{
|
|
"epoch": 0.6775313919316056,
|
|
"grad_norm": 22.00641078072797,
|
|
"learning_rate": 2.2644650435297675e-07,
|
|
"logits/chosen": -0.9918270707130432,
|
|
"logits/rejected": -1.0709521770477295,
|
|
"logps/chosen": -635.0587768554688,
|
|
"logps/rejected": -808.44140625,
|
|
"loss": 0.4351,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.9043281078338623,
|
|
"rewards/margins": 1.8593040704727173,
|
|
"rewards/rejected": -5.763632297515869,
|
|
"step": 317
|
|
},
|
|
{
|
|
"epoch": 0.6796687149345445,
|
|
"grad_norm": 18.51407185653388,
|
|
"learning_rate": 2.2375569208964645e-07,
|
|
"logits/chosen": -1.2991502285003662,
|
|
"logits/rejected": -1.2963578701019287,
|
|
"logps/chosen": -552.6873779296875,
|
|
"logps/rejected": -702.3699951171875,
|
|
"loss": 0.4147,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.1100082397460938,
|
|
"rewards/margins": 1.3412725925445557,
|
|
"rewards/rejected": -4.45128059387207,
|
|
"step": 318
|
|
},
|
|
{
|
|
"epoch": 0.6818060379374833,
|
|
"grad_norm": 21.46578371453758,
|
|
"learning_rate": 2.210747406737469e-07,
|
|
"logits/chosen": -1.1598680019378662,
|
|
"logits/rejected": -1.0896382331848145,
|
|
"logps/chosen": -679.51171875,
|
|
"logps/rejected": -972.2852783203125,
|
|
"loss": 0.3608,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -4.203690052032471,
|
|
"rewards/margins": 2.7421700954437256,
|
|
"rewards/rejected": -6.945860862731934,
|
|
"step": 319
|
|
},
|
|
{
|
|
"epoch": 0.6839433609404221,
|
|
"grad_norm": 15.531933254870697,
|
|
"learning_rate": 2.1840380010418128e-07,
|
|
"logits/chosen": -1.2312465906143188,
|
|
"logits/rejected": -1.2830405235290527,
|
|
"logps/chosen": -637.8733520507812,
|
|
"logps/rejected": -837.0244140625,
|
|
"loss": 0.4024,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.5754833221435547,
|
|
"rewards/margins": 1.8828397989273071,
|
|
"rewards/rejected": -5.458322525024414,
|
|
"step": 320
|
|
},
|
|
{
|
|
"epoch": 0.686080683943361,
|
|
"grad_norm": 21.09966403146541,
|
|
"learning_rate": 2.1574301981974715e-07,
|
|
"logits/chosen": -1.0287922620773315,
|
|
"logits/rejected": -1.1071122884750366,
|
|
"logps/chosen": -648.6826171875,
|
|
"logps/rejected": -769.0919799804688,
|
|
"loss": 0.4074,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.858161211013794,
|
|
"rewards/margins": 1.2575337886810303,
|
|
"rewards/rejected": -5.115694999694824,
|
|
"step": 321
|
|
},
|
|
{
|
|
"epoch": 0.6882180069462998,
|
|
"grad_norm": 18.679551175728633,
|
|
"learning_rate": 2.1309254869077518e-07,
|
|
"logits/chosen": -1.123620629310608,
|
|
"logits/rejected": -1.142130732536316,
|
|
"logps/chosen": -643.7474365234375,
|
|
"logps/rejected": -805.8037719726562,
|
|
"loss": 0.3653,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.7344915866851807,
|
|
"rewards/margins": 1.7917336225509644,
|
|
"rewards/rejected": -5.526225566864014,
|
|
"step": 322
|
|
},
|
|
{
|
|
"epoch": 0.6903553299492385,
|
|
"grad_norm": 20.66758893901287,
|
|
"learning_rate": 2.1045253501080054e-07,
|
|
"logits/chosen": -1.1394745111465454,
|
|
"logits/rejected": -1.1683499813079834,
|
|
"logps/chosen": -561.175537109375,
|
|
"logps/rejected": -684.5661010742188,
|
|
"loss": 0.401,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.2739458084106445,
|
|
"rewards/margins": 1.2987473011016846,
|
|
"rewards/rejected": -4.57269287109375,
|
|
"step": 323
|
|
},
|
|
{
|
|
"epoch": 0.6924926529521774,
|
|
"grad_norm": 18.77702538784566,
|
|
"learning_rate": 2.0782312648826505e-07,
|
|
"logits/chosen": -1.2171223163604736,
|
|
"logits/rejected": -1.2107038497924805,
|
|
"logps/chosen": -609.377197265625,
|
|
"logps/rejected": -789.9899291992188,
|
|
"loss": 0.3615,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.611778736114502,
|
|
"rewards/margins": 1.7298355102539062,
|
|
"rewards/rejected": -5.34161376953125,
|
|
"step": 324
|
|
},
|
|
{
|
|
"epoch": 0.6946299759551162,
|
|
"grad_norm": 17.599881691021295,
|
|
"learning_rate": 2.0520447023825306e-07,
|
|
"logits/chosen": -1.1523199081420898,
|
|
"logits/rejected": -1.1357042789459229,
|
|
"logps/chosen": -502.8484802246094,
|
|
"logps/rejected": -595.9307250976562,
|
|
"loss": 0.3407,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.2537741661071777,
|
|
"rewards/margins": 0.8426163792610168,
|
|
"rewards/rejected": -4.096390247344971,
|
|
"step": 325
|
|
},
|
|
{
|
|
"epoch": 0.696767298958055,
|
|
"grad_norm": 15.783181757197172,
|
|
"learning_rate": 2.0259671277426123e-07,
|
|
"logits/chosen": -1.0857826471328735,
|
|
"logits/rejected": -1.0804615020751953,
|
|
"logps/chosen": -541.2989501953125,
|
|
"logps/rejected": -728.1956176757812,
|
|
"loss": 0.3597,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.1643340587615967,
|
|
"rewards/margins": 1.8594626188278198,
|
|
"rewards/rejected": -5.023797035217285,
|
|
"step": 326
|
|
},
|
|
{
|
|
"epoch": 0.6989046219609939,
|
|
"grad_norm": 18.584132337966185,
|
|
"learning_rate": 2.0000000000000007e-07,
|
|
"logits/chosen": -1.1965287923812866,
|
|
"logits/rejected": -1.1642347574234009,
|
|
"logps/chosen": -604.3222045898438,
|
|
"logps/rejected": -733.7069091796875,
|
|
"loss": 0.443,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.5162549018859863,
|
|
"rewards/margins": 1.2821638584136963,
|
|
"rewards/rejected": -4.7984185218811035,
|
|
"step": 327
|
|
},
|
|
{
|
|
"epoch": 0.7010419449639327,
|
|
"grad_norm": 20.799891467119032,
|
|
"learning_rate": 1.9741447720123095e-07,
|
|
"logits/chosen": -1.184598684310913,
|
|
"logits/rejected": -1.1836549043655396,
|
|
"logps/chosen": -607.91748046875,
|
|
"logps/rejected": -766.5301513671875,
|
|
"loss": 0.4253,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.7808310985565186,
|
|
"rewards/margins": 1.377992033958435,
|
|
"rewards/rejected": -5.158823013305664,
|
|
"step": 328
|
|
},
|
|
{
|
|
"epoch": 0.7031792679668715,
|
|
"grad_norm": 17.876290923345074,
|
|
"learning_rate": 1.9484028903763754e-07,
|
|
"logits/chosen": -1.3433024883270264,
|
|
"logits/rejected": -1.3774957656860352,
|
|
"logps/chosen": -657.9730224609375,
|
|
"logps/rejected": -829.743896484375,
|
|
"loss": 0.4266,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.6865885257720947,
|
|
"rewards/margins": 1.7604457139968872,
|
|
"rewards/rejected": -5.4470343589782715,
|
|
"step": 329
|
|
},
|
|
{
|
|
"epoch": 0.7053165909698104,
|
|
"grad_norm": 19.62276873807506,
|
|
"learning_rate": 1.9227757953473233e-07,
|
|
"logits/chosen": -1.1646440029144287,
|
|
"logits/rejected": -1.1711232662200928,
|
|
"logps/chosen": -685.2410888671875,
|
|
"logps/rejected": -812.1082763671875,
|
|
"loss": 0.3925,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.885378360748291,
|
|
"rewards/margins": 1.3792006969451904,
|
|
"rewards/rejected": -5.2645792961120605,
|
|
"step": 330
|
|
},
|
|
{
|
|
"epoch": 0.7074539139727491,
|
|
"grad_norm": 20.713783128170522,
|
|
"learning_rate": 1.8972649207579807e-07,
|
|
"logits/chosen": -1.1773666143417358,
|
|
"logits/rejected": -1.183628797531128,
|
|
"logps/chosen": -571.4490356445312,
|
|
"logps/rejected": -727.8609619140625,
|
|
"loss": 0.3542,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.21671462059021,
|
|
"rewards/margins": 1.5842726230621338,
|
|
"rewards/rejected": -4.800987720489502,
|
|
"step": 331
|
|
},
|
|
{
|
|
"epoch": 0.7095912369756879,
|
|
"grad_norm": 14.323520360594955,
|
|
"learning_rate": 1.871871693938654e-07,
|
|
"logits/chosen": -1.2549258470535278,
|
|
"logits/rejected": -1.218357801437378,
|
|
"logps/chosen": -706.3805541992188,
|
|
"logps/rejected": -884.6824951171875,
|
|
"loss": 0.3146,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.322238922119141,
|
|
"rewards/margins": 1.728010892868042,
|
|
"rewards/rejected": -6.050250053405762,
|
|
"step": 332
|
|
},
|
|
{
|
|
"epoch": 0.7117285599786267,
|
|
"grad_norm": 15.481665401011659,
|
|
"learning_rate": 1.8465975356372713e-07,
|
|
"logits/chosen": -1.2304145097732544,
|
|
"logits/rejected": -1.216366171836853,
|
|
"logps/chosen": -635.7239379882812,
|
|
"logps/rejected": -711.3768920898438,
|
|
"loss": 0.4016,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.948883056640625,
|
|
"rewards/margins": 0.8189201951026917,
|
|
"rewards/rejected": -4.767803192138672,
|
|
"step": 333
|
|
},
|
|
{
|
|
"epoch": 0.7138658829815656,
|
|
"grad_norm": 17.737870919716684,
|
|
"learning_rate": 1.8214438599398915e-07,
|
|
"logits/chosen": -1.159693956375122,
|
|
"logits/rejected": -1.2034831047058105,
|
|
"logps/chosen": -673.4019775390625,
|
|
"logps/rejected": -883.9022216796875,
|
|
"loss": 0.4045,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.365835666656494,
|
|
"rewards/margins": 1.9925627708435059,
|
|
"rewards/rejected": -6.358397960662842,
|
|
"step": 334
|
|
},
|
|
{
|
|
"epoch": 0.7160032059845044,
|
|
"grad_norm": 25.705664070692066,
|
|
"learning_rate": 1.7964120741915902e-07,
|
|
"logits/chosen": -1.2710121870040894,
|
|
"logits/rejected": -1.1944791078567505,
|
|
"logps/chosen": -608.1783447265625,
|
|
"logps/rejected": -811.2279052734375,
|
|
"loss": 0.4087,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.5328927040100098,
|
|
"rewards/margins": 1.9209730625152588,
|
|
"rewards/rejected": -5.453866004943848,
|
|
"step": 335
|
|
},
|
|
{
|
|
"epoch": 0.7181405289874432,
|
|
"grad_norm": 18.1742169248362,
|
|
"learning_rate": 1.771503578917711e-07,
|
|
"logits/chosen": -1.1713533401489258,
|
|
"logits/rejected": -1.240103006362915,
|
|
"logps/chosen": -669.6802368164062,
|
|
"logps/rejected": -841.9739990234375,
|
|
"loss": 0.427,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.10801362991333,
|
|
"rewards/margins": 1.8291915655136108,
|
|
"rewards/rejected": -5.9372053146362305,
|
|
"step": 336
|
|
},
|
|
{
|
|
"epoch": 0.7202778519903821,
|
|
"grad_norm": 18.465855653993856,
|
|
"learning_rate": 1.7467197677455116e-07,
|
|
"logits/chosen": -1.1570916175842285,
|
|
"logits/rejected": -1.1773279905319214,
|
|
"logps/chosen": -621.7543334960938,
|
|
"logps/rejected": -863.5875854492188,
|
|
"loss": 0.318,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.662381172180176,
|
|
"rewards/margins": 2.4044947624206543,
|
|
"rewards/rejected": -6.066876411437988,
|
|
"step": 337
|
|
},
|
|
{
|
|
"epoch": 0.7224151749933209,
|
|
"grad_norm": 18.086374212134224,
|
|
"learning_rate": 1.7220620273261922e-07,
|
|
"logits/chosen": -1.2305190563201904,
|
|
"logits/rejected": -1.1802289485931396,
|
|
"logps/chosen": -574.2893676757812,
|
|
"logps/rejected": -654.89501953125,
|
|
"loss": 0.3703,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.308969020843506,
|
|
"rewards/margins": 0.8964172601699829,
|
|
"rewards/rejected": -4.205386161804199,
|
|
"step": 338
|
|
},
|
|
{
|
|
"epoch": 0.7245524979962596,
|
|
"grad_norm": 21.686278920021145,
|
|
"learning_rate": 1.6975317372573065e-07,
|
|
"logits/chosen": -0.9897362589836121,
|
|
"logits/rejected": -1.0513758659362793,
|
|
"logps/chosen": -718.6741943359375,
|
|
"logps/rejected": -755.384765625,
|
|
"loss": 0.392,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -4.776792526245117,
|
|
"rewards/margins": 0.4688657820224762,
|
|
"rewards/rejected": -5.2456583976745605,
|
|
"step": 339
|
|
},
|
|
{
|
|
"epoch": 0.7266898209991985,
|
|
"grad_norm": 17.6310482182181,
|
|
"learning_rate": 1.6731302700055777e-07,
|
|
"logits/chosen": -1.2428513765335083,
|
|
"logits/rejected": -1.2316035032272339,
|
|
"logps/chosen": -669.8923950195312,
|
|
"logps/rejected": -796.8135986328125,
|
|
"loss": 0.4199,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.199666976928711,
|
|
"rewards/margins": 1.1926888227462769,
|
|
"rewards/rejected": -5.392355442047119,
|
|
"step": 340
|
|
},
|
|
{
|
|
"epoch": 0.7288271440021373,
|
|
"grad_norm": 15.542714019081663,
|
|
"learning_rate": 1.6488589908301078e-07,
|
|
"logits/chosen": -1.090733289718628,
|
|
"logits/rejected": -1.0847265720367432,
|
|
"logps/chosen": -755.2330932617188,
|
|
"logps/rejected": -932.0684204101562,
|
|
"loss": 0.3997,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -5.0632171630859375,
|
|
"rewards/margins": 1.7293668985366821,
|
|
"rewards/rejected": -6.792583465576172,
|
|
"step": 341
|
|
},
|
|
{
|
|
"epoch": 0.7309644670050761,
|
|
"grad_norm": 21.28545495220758,
|
|
"learning_rate": 1.6247192577059941e-07,
|
|
"logits/chosen": -1.232560634613037,
|
|
"logits/rejected": -1.2369352579116821,
|
|
"logps/chosen": -646.4221801757812,
|
|
"logps/rejected": -811.7963256835938,
|
|
"loss": 0.3937,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -4.2266082763671875,
|
|
"rewards/margins": 1.3389954566955566,
|
|
"rewards/rejected": -5.565604209899902,
|
|
"step": 342
|
|
},
|
|
{
|
|
"epoch": 0.733101790008015,
|
|
"grad_norm": 19.319434903860703,
|
|
"learning_rate": 1.600712421248345e-07,
|
|
"logits/chosen": -1.1512730121612549,
|
|
"logits/rejected": -1.1143797636032104,
|
|
"logps/chosen": -672.3272705078125,
|
|
"logps/rejected": -789.5181274414062,
|
|
"loss": 0.4251,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.444333076477051,
|
|
"rewards/margins": 1.1563409566879272,
|
|
"rewards/rejected": -5.600674629211426,
|
|
"step": 343
|
|
},
|
|
{
|
|
"epoch": 0.7352391130109538,
|
|
"grad_norm": 22.15460617482266,
|
|
"learning_rate": 1.5768398246367207e-07,
|
|
"logits/chosen": -1.113842248916626,
|
|
"logits/rejected": -1.1331862211227417,
|
|
"logps/chosen": -731.4851684570312,
|
|
"logps/rejected": -965.7476196289062,
|
|
"loss": 0.3713,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -4.364319801330566,
|
|
"rewards/margins": 2.2552897930145264,
|
|
"rewards/rejected": -6.6196088790893555,
|
|
"step": 344
|
|
},
|
|
{
|
|
"epoch": 0.7373764360138926,
|
|
"grad_norm": 16.990002139173328,
|
|
"learning_rate": 1.553102803539974e-07,
|
|
"logits/chosen": -1.2558772563934326,
|
|
"logits/rejected": -1.2412264347076416,
|
|
"logps/chosen": -637.6004028320312,
|
|
"logps/rejected": -920.36865234375,
|
|
"loss": 0.3857,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.0147528648376465,
|
|
"rewards/margins": 2.530723810195923,
|
|
"rewards/rejected": -6.54547643661499,
|
|
"step": 345
|
|
},
|
|
{
|
|
"epoch": 0.7395137590168315,
|
|
"grad_norm": 23.667341921272293,
|
|
"learning_rate": 1.529502686041529e-07,
|
|
"logits/chosen": -1.15803861618042,
|
|
"logits/rejected": -1.2394546270370483,
|
|
"logps/chosen": -798.1978759765625,
|
|
"logps/rejected": -771.4423217773438,
|
|
"loss": 0.4965,
|
|
"rewards/accuracies": 0.4375,
|
|
"rewards/chosen": -5.245590686798096,
|
|
"rewards/margins": -0.053717248141765594,
|
|
"rewards/rejected": -5.191873073577881,
|
|
"step": 346
|
|
},
|
|
{
|
|
"epoch": 0.7416510820197703,
|
|
"grad_norm": 17.49003980577173,
|
|
"learning_rate": 1.506040792565066e-07,
|
|
"logits/chosen": -1.2386159896850586,
|
|
"logits/rejected": -1.2004777193069458,
|
|
"logps/chosen": -696.9265747070312,
|
|
"logps/rejected": -820.576171875,
|
|
"loss": 0.3834,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.919370651245117,
|
|
"rewards/margins": 1.2635618448257446,
|
|
"rewards/rejected": -5.1829328536987305,
|
|
"step": 347
|
|
},
|
|
{
|
|
"epoch": 0.743788405022709,
|
|
"grad_norm": 26.239386727495823,
|
|
"learning_rate": 1.4827184358006508e-07,
|
|
"logits/chosen": -1.0444031953811646,
|
|
"logits/rejected": -1.0797080993652344,
|
|
"logps/chosen": -661.2929077148438,
|
|
"logps/rejected": -858.2059936523438,
|
|
"loss": 0.416,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.284615993499756,
|
|
"rewards/margins": 1.9141457080841064,
|
|
"rewards/rejected": -6.198761940002441,
|
|
"step": 348
|
|
},
|
|
{
|
|
"epoch": 0.7459257280256478,
|
|
"grad_norm": 22.633565158197506,
|
|
"learning_rate": 1.459536920631285e-07,
|
|
"logits/chosen": -1.2120310068130493,
|
|
"logits/rejected": -1.2144323587417603,
|
|
"logps/chosen": -793.6173095703125,
|
|
"logps/rejected": -964.0809326171875,
|
|
"loss": 0.3976,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -5.077376842498779,
|
|
"rewards/margins": 1.8791519403457642,
|
|
"rewards/rejected": -6.956528663635254,
|
|
"step": 349
|
|
},
|
|
{
|
|
"epoch": 0.7480630510285867,
|
|
"grad_norm": 21.615429147306983,
|
|
"learning_rate": 1.4364975440599033e-07,
|
|
"logits/chosen": -1.2455774545669556,
|
|
"logits/rejected": -1.2984986305236816,
|
|
"logps/chosen": -720.0133056640625,
|
|
"logps/rejected": -948.4733276367188,
|
|
"loss": 0.412,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.872766494750977,
|
|
"rewards/margins": 2.1328089237213135,
|
|
"rewards/rejected": -7.005575180053711,
|
|
"step": 350
|
|
},
|
|
{
|
|
"epoch": 0.7502003740315255,
|
|
"grad_norm": 17.381380445589752,
|
|
"learning_rate": 1.4136015951368018e-07,
|
|
"logits/chosen": -1.2067402601242065,
|
|
"logits/rejected": -1.2139760255813599,
|
|
"logps/chosen": -676.5394897460938,
|
|
"logps/rejected": -823.6031494140625,
|
|
"loss": 0.3261,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -4.276626110076904,
|
|
"rewards/margins": 1.5800349712371826,
|
|
"rewards/rejected": -5.856661796569824,
|
|
"step": 351
|
|
},
|
|
{
|
|
"epoch": 0.7523376970344643,
|
|
"grad_norm": 21.729828693203185,
|
|
"learning_rate": 1.3908503548875167e-07,
|
|
"logits/chosen": -1.2517892122268677,
|
|
"logits/rejected": -1.1772764921188354,
|
|
"logps/chosen": -711.4560546875,
|
|
"logps/rejected": -869.8018798828125,
|
|
"loss": 0.3411,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.649675369262695,
|
|
"rewards/margins": 1.567895531654358,
|
|
"rewards/rejected": -6.21757173538208,
|
|
"step": 352
|
|
},
|
|
{
|
|
"epoch": 0.7544750200374032,
|
|
"grad_norm": 18.37581288755564,
|
|
"learning_rate": 1.3682450962411495e-07,
|
|
"logits/chosen": -1.202965259552002,
|
|
"logits/rejected": -1.1767388582229614,
|
|
"logps/chosen": -735.4443359375,
|
|
"logps/rejected": -899.7939453125,
|
|
"loss": 0.4046,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.456427097320557,
|
|
"rewards/margins": 1.780933141708374,
|
|
"rewards/rejected": -6.23736047744751,
|
|
"step": 353
|
|
},
|
|
{
|
|
"epoch": 0.756612343040342,
|
|
"grad_norm": 25.687813258400997,
|
|
"learning_rate": 1.3457870839591556e-07,
|
|
"logits/chosen": -1.1969988346099854,
|
|
"logits/rejected": -1.190890908241272,
|
|
"logps/chosen": -668.3444213867188,
|
|
"logps/rejected": -812.265625,
|
|
"loss": 0.3904,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -4.523505687713623,
|
|
"rewards/margins": 1.2760380506515503,
|
|
"rewards/rejected": -5.799543857574463,
|
|
"step": 354
|
|
},
|
|
{
|
|
"epoch": 0.7587496660432808,
|
|
"grad_norm": 19.70525548630737,
|
|
"learning_rate": 1.3234775745645682e-07,
|
|
"logits/chosen": -1.298694133758545,
|
|
"logits/rejected": -1.2099275588989258,
|
|
"logps/chosen": -629.9318237304688,
|
|
"logps/rejected": -793.7557373046875,
|
|
"loss": 0.3677,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.134130001068115,
|
|
"rewards/margins": 1.51906418800354,
|
|
"rewards/rejected": -5.653193950653076,
|
|
"step": 355
|
|
},
|
|
{
|
|
"epoch": 0.7608869890462197,
|
|
"grad_norm": 20.97352305735656,
|
|
"learning_rate": 1.3013178162717026e-07,
|
|
"logits/chosen": -1.1704061031341553,
|
|
"logits/rejected": -1.1709686517715454,
|
|
"logps/chosen": -619.7793579101562,
|
|
"logps/rejected": -770.8734130859375,
|
|
"loss": 0.3622,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.893303632736206,
|
|
"rewards/margins": 1.5143392086029053,
|
|
"rewards/rejected": -5.407642841339111,
|
|
"step": 356
|
|
},
|
|
{
|
|
"epoch": 0.7630243120491584,
|
|
"grad_norm": 19.640911151476274,
|
|
"learning_rate": 1.2793090489163218e-07,
|
|
"logits/chosen": -1.1160629987716675,
|
|
"logits/rejected": -1.2110121250152588,
|
|
"logps/chosen": -786.0536499023438,
|
|
"logps/rejected": -1036.768310546875,
|
|
"loss": 0.4379,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.929718017578125,
|
|
"rewards/margins": 2.5798377990722656,
|
|
"rewards/rejected": -7.509555816650391,
|
|
"step": 357
|
|
},
|
|
{
|
|
"epoch": 0.7651616350520972,
|
|
"grad_norm": 18.712384438599756,
|
|
"learning_rate": 1.257452503886263e-07,
|
|
"logits/chosen": -1.2526037693023682,
|
|
"logits/rejected": -1.232513666152954,
|
|
"logps/chosen": -785.4493408203125,
|
|
"logps/rejected": -969.44140625,
|
|
"loss": 0.3292,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -5.247096538543701,
|
|
"rewards/margins": 1.8201085329055786,
|
|
"rewards/rejected": -7.067205429077148,
|
|
"step": 358
|
|
},
|
|
{
|
|
"epoch": 0.7672989580550361,
|
|
"grad_norm": 22.206627398708065,
|
|
"learning_rate": 1.2357494040525414e-07,
|
|
"logits/chosen": -1.0485000610351562,
|
|
"logits/rejected": -1.0508546829223633,
|
|
"logps/chosen": -662.2064819335938,
|
|
"logps/rejected": -915.0498046875,
|
|
"loss": 0.4551,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.379815101623535,
|
|
"rewards/margins": 2.327033758163452,
|
|
"rewards/rejected": -6.706849575042725,
|
|
"step": 359
|
|
},
|
|
{
|
|
"epoch": 0.7694362810579749,
|
|
"grad_norm": 23.59419747338851,
|
|
"learning_rate": 1.2142009637009333e-07,
|
|
"logits/chosen": -1.1591897010803223,
|
|
"logits/rejected": -1.090525507926941,
|
|
"logps/chosen": -831.9130859375,
|
|
"logps/rejected": -980.3768920898438,
|
|
"loss": 0.4204,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -5.667198181152344,
|
|
"rewards/margins": 1.3969204425811768,
|
|
"rewards/rejected": -7.0641188621521,
|
|
"step": 360
|
|
},
|
|
{
|
|
"epoch": 0.7715736040609137,
|
|
"grad_norm": 20.325203409090374,
|
|
"learning_rate": 1.192808388464034e-07,
|
|
"logits/chosen": -1.1007847785949707,
|
|
"logits/rejected": -0.9627750515937805,
|
|
"logps/chosen": -626.9066162109375,
|
|
"logps/rejected": -841.2676391601562,
|
|
"loss": 0.3843,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.2317962646484375,
|
|
"rewards/margins": 1.9466259479522705,
|
|
"rewards/rejected": -6.178422451019287,
|
|
"step": 361
|
|
},
|
|
{
|
|
"epoch": 0.7737109270638525,
|
|
"grad_norm": 18.105397272000737,
|
|
"learning_rate": 1.17157287525381e-07,
|
|
"logits/chosen": -1.155585765838623,
|
|
"logits/rejected": -1.0961203575134277,
|
|
"logps/chosen": -684.6275024414062,
|
|
"logps/rejected": -876.870849609375,
|
|
"loss": 0.3921,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.383594512939453,
|
|
"rewards/margins": 1.803823471069336,
|
|
"rewards/rejected": -6.187417507171631,
|
|
"step": 362
|
|
},
|
|
{
|
|
"epoch": 0.7758482500667914,
|
|
"grad_norm": 26.052189317294125,
|
|
"learning_rate": 1.1504956121946214e-07,
|
|
"logits/chosen": -1.0707626342773438,
|
|
"logits/rejected": -1.1878434419631958,
|
|
"logps/chosen": -714.69384765625,
|
|
"logps/rejected": -863.8623046875,
|
|
"loss": 0.5443,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.7078094482421875,
|
|
"rewards/margins": 1.6244311332702637,
|
|
"rewards/rejected": -6.332241058349609,
|
|
"step": 363
|
|
},
|
|
{
|
|
"epoch": 0.7779855730697302,
|
|
"grad_norm": 19.3332723154738,
|
|
"learning_rate": 1.1295777785567522e-07,
|
|
"logits/chosen": -1.2466063499450684,
|
|
"logits/rejected": -1.3214343786239624,
|
|
"logps/chosen": -628.520751953125,
|
|
"logps/rejected": -765.861083984375,
|
|
"loss": 0.3409,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.307971477508545,
|
|
"rewards/margins": 1.5903899669647217,
|
|
"rewards/rejected": -5.8983612060546875,
|
|
"step": 364
|
|
},
|
|
{
|
|
"epoch": 0.7801228960726689,
|
|
"grad_norm": 16.65631148968948,
|
|
"learning_rate": 1.1088205446904342e-07,
|
|
"logits/chosen": -1.2889045476913452,
|
|
"logits/rejected": -1.2453293800354004,
|
|
"logps/chosen": -632.6788940429688,
|
|
"logps/rejected": -860.2509155273438,
|
|
"loss": 0.3204,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.8525776863098145,
|
|
"rewards/margins": 2.0945329666137695,
|
|
"rewards/rejected": -5.947110652923584,
|
|
"step": 365
|
|
},
|
|
{
|
|
"epoch": 0.7822602190756078,
|
|
"grad_norm": 25.617856733947676,
|
|
"learning_rate": 1.0882250719603567e-07,
|
|
"logits/chosen": -1.1964340209960938,
|
|
"logits/rejected": -1.23922598361969,
|
|
"logps/chosen": -930.8063354492188,
|
|
"logps/rejected": -1066.240478515625,
|
|
"loss": 0.4511,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -5.721828937530518,
|
|
"rewards/margins": 1.5414330959320068,
|
|
"rewards/rejected": -7.263261795043945,
|
|
"step": 366
|
|
},
|
|
{
|
|
"epoch": 0.7843975420785466,
|
|
"grad_norm": 24.227338760419844,
|
|
"learning_rate": 1.0677925126806955e-07,
|
|
"logits/chosen": -1.169182300567627,
|
|
"logits/rejected": -1.2425594329833984,
|
|
"logps/chosen": -638.6636962890625,
|
|
"logps/rejected": -791.4046020507812,
|
|
"loss": 0.3979,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.070646286010742,
|
|
"rewards/margins": 1.7463229894638062,
|
|
"rewards/rejected": -5.816969394683838,
|
|
"step": 367
|
|
},
|
|
{
|
|
"epoch": 0.7865348650814854,
|
|
"grad_norm": 16.35871784445141,
|
|
"learning_rate": 1.0475240100506377e-07,
|
|
"logits/chosen": -1.0509607791900635,
|
|
"logits/rejected": -1.0016944408416748,
|
|
"logps/chosen": -731.5370483398438,
|
|
"logps/rejected": -868.427978515625,
|
|
"loss": 0.2992,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.792571067810059,
|
|
"rewards/margins": 1.3596065044403076,
|
|
"rewards/rejected": -6.152177333831787,
|
|
"step": 368
|
|
},
|
|
{
|
|
"epoch": 0.7886721880844243,
|
|
"grad_norm": 20.50758693822622,
|
|
"learning_rate": 1.0274206980904226e-07,
|
|
"logits/chosen": -1.0662883520126343,
|
|
"logits/rejected": -1.0809623003005981,
|
|
"logps/chosen": -701.0050048828125,
|
|
"logps/rejected": -845.4842529296875,
|
|
"loss": 0.3898,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.612939834594727,
|
|
"rewards/margins": 1.4648727178573608,
|
|
"rewards/rejected": -6.077812194824219,
|
|
"step": 369
|
|
},
|
|
{
|
|
"epoch": 0.7908095110873631,
|
|
"grad_norm": 21.295312469199402,
|
|
"learning_rate": 1.0074837015778919e-07,
|
|
"logits/chosen": -1.1983091831207275,
|
|
"logits/rejected": -1.2084715366363525,
|
|
"logps/chosen": -812.4208374023438,
|
|
"logps/rejected": -999.982177734375,
|
|
"loss": 0.4268,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -5.570307731628418,
|
|
"rewards/margins": 1.8499504327774048,
|
|
"rewards/rejected": -7.420258522033691,
|
|
"step": 370
|
|
},
|
|
{
|
|
"epoch": 0.7929468340903019,
|
|
"grad_norm": 23.822573821786133,
|
|
"learning_rate": 9.877141359855565e-08,
|
|
"logits/chosen": -1.206276774406433,
|
|
"logits/rejected": -1.2356822490692139,
|
|
"logps/chosen": -775.7623901367188,
|
|
"logps/rejected": -1005.5828247070312,
|
|
"loss": 0.4341,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -4.983241558074951,
|
|
"rewards/margins": 2.1637260913848877,
|
|
"rewards/rejected": -7.14696741104126,
|
|
"step": 371
|
|
},
|
|
{
|
|
"epoch": 0.7950841570932408,
|
|
"grad_norm": 17.234697083614194,
|
|
"learning_rate": 9.681131074181874e-08,
|
|
"logits/chosen": -1.3584322929382324,
|
|
"logits/rejected": -1.299919605255127,
|
|
"logps/chosen": -613.3756103515625,
|
|
"logps/rejected": -737.3634033203125,
|
|
"loss": 0.3485,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.7099380493164062,
|
|
"rewards/margins": 1.276181697845459,
|
|
"rewards/rejected": -4.986119747161865,
|
|
"step": 372
|
|
},
|
|
{
|
|
"epoch": 0.7972214800961795,
|
|
"grad_norm": 17.48652928766399,
|
|
"learning_rate": 9.486817125509339e-08,
|
|
"logits/chosen": -1.1015208959579468,
|
|
"logits/rejected": -1.0562043190002441,
|
|
"logps/chosen": -806.1483764648438,
|
|
"logps/rejected": -1005.850341796875,
|
|
"loss": 0.3802,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -5.184823036193848,
|
|
"rewards/margins": 1.9502785205841064,
|
|
"rewards/rejected": -7.135101795196533,
|
|
"step": 373
|
|
},
|
|
{
|
|
"epoch": 0.7993588030991183,
|
|
"grad_norm": 23.159224310703724,
|
|
"learning_rate": 9.294210385679555e-08,
|
|
"logits/chosen": -1.20077645778656,
|
|
"logits/rejected": -1.168664574623108,
|
|
"logps/chosen": -732.3262329101562,
|
|
"logps/rejected": -890.5650634765625,
|
|
"loss": 0.4008,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.771186828613281,
|
|
"rewards/margins": 1.5593047142028809,
|
|
"rewards/rejected": -6.33049201965332,
|
|
"step": 374
|
|
},
|
|
{
|
|
"epoch": 0.8014961261020572,
|
|
"grad_norm": 18.102274669415063,
|
|
"learning_rate": 9.103321631016023e-08,
|
|
"logits/chosen": -1.157024621963501,
|
|
"logits/rejected": -1.1546276807785034,
|
|
"logps/chosen": -793.4083251953125,
|
|
"logps/rejected": -937.8966674804688,
|
|
"loss": 0.4136,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.976907253265381,
|
|
"rewards/margins": 1.4234803915023804,
|
|
"rewards/rejected": -6.400387763977051,
|
|
"step": 375
|
|
},
|
|
{
|
|
"epoch": 0.803633449104996,
|
|
"grad_norm": 22.18700340122903,
|
|
"learning_rate": 8.914161541721163e-08,
|
|
"logits/chosen": -1.2537552118301392,
|
|
"logits/rejected": -1.237257480621338,
|
|
"logps/chosen": -648.7177124023438,
|
|
"logps/rejected": -805.5570678710938,
|
|
"loss": 0.4648,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.536306381225586,
|
|
"rewards/margins": 1.4047529697418213,
|
|
"rewards/rejected": -5.941059112548828,
|
|
"step": 376
|
|
},
|
|
{
|
|
"epoch": 0.8057707721079348,
|
|
"grad_norm": 17.300986839334925,
|
|
"learning_rate": 8.726740701278807e-08,
|
|
"logits/chosen": -1.3532252311706543,
|
|
"logits/rejected": -1.290065884590149,
|
|
"logps/chosen": -593.5990600585938,
|
|
"logps/rejected": -786.1945190429688,
|
|
"loss": 0.377,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.6258466243743896,
|
|
"rewards/margins": 1.8245813846588135,
|
|
"rewards/rejected": -5.450427532196045,
|
|
"step": 377
|
|
},
|
|
{
|
|
"epoch": 0.8079080951108736,
|
|
"grad_norm": 20.165978429403715,
|
|
"learning_rate": 8.541069595862009e-08,
|
|
"logits/chosen": -1.2440727949142456,
|
|
"logits/rejected": -1.14741849899292,
|
|
"logps/chosen": -717.24951171875,
|
|
"logps/rejected": -860.7356567382812,
|
|
"loss": 0.4126,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.245814800262451,
|
|
"rewards/margins": 1.338247537612915,
|
|
"rewards/rejected": -5.584062576293945,
|
|
"step": 378
|
|
},
|
|
{
|
|
"epoch": 0.8100454181138125,
|
|
"grad_norm": 16.719380649306434,
|
|
"learning_rate": 8.35715861374636e-08,
|
|
"logits/chosen": -1.1465423107147217,
|
|
"logits/rejected": -1.1313426494598389,
|
|
"logps/chosen": -563.7252197265625,
|
|
"logps/rejected": -776.3715209960938,
|
|
"loss": 0.363,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -3.422579288482666,
|
|
"rewards/margins": 1.9151452779769897,
|
|
"rewards/rejected": -5.337724208831787,
|
|
"step": 379
|
|
},
|
|
{
|
|
"epoch": 0.8121827411167513,
|
|
"grad_norm": 21.079445701366463,
|
|
"learning_rate": 8.175018044728762e-08,
|
|
"logits/chosen": -1.2566273212432861,
|
|
"logits/rejected": -1.2379119396209717,
|
|
"logps/chosen": -588.2570190429688,
|
|
"logps/rejected": -726.7328491210938,
|
|
"loss": 0.4063,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.6017940044403076,
|
|
"rewards/margins": 1.2587138414382935,
|
|
"rewards/rejected": -4.860507965087891,
|
|
"step": 380
|
|
},
|
|
{
|
|
"epoch": 0.81432006411969,
|
|
"grad_norm": 15.615292244997176,
|
|
"learning_rate": 7.994658079551752e-08,
|
|
"logits/chosen": -0.9509527683258057,
|
|
"logits/rejected": -0.928947925567627,
|
|
"logps/chosen": -749.95556640625,
|
|
"logps/rejected": -985.079345703125,
|
|
"loss": 0.3506,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.5622477531433105,
|
|
"rewards/margins": 2.3395752906799316,
|
|
"rewards/rejected": -6.901823043823242,
|
|
"step": 381
|
|
},
|
|
{
|
|
"epoch": 0.8164573871226289,
|
|
"grad_norm": 21.72824821202356,
|
|
"learning_rate": 7.816088809333265e-08,
|
|
"logits/chosen": -1.112214207649231,
|
|
"logits/rejected": -1.0684149265289307,
|
|
"logps/chosen": -764.0916137695312,
|
|
"logps/rejected": -841.7941284179688,
|
|
"loss": 0.4301,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.703119277954102,
|
|
"rewards/margins": 0.840286374092102,
|
|
"rewards/rejected": -5.543406009674072,
|
|
"step": 382
|
|
},
|
|
{
|
|
"epoch": 0.8185947101255677,
|
|
"grad_norm": 17.6012054275616,
|
|
"learning_rate": 7.639320225002106e-08,
|
|
"logits/chosen": -1.123322606086731,
|
|
"logits/rejected": -1.0792958736419678,
|
|
"logps/chosen": -656.8797607421875,
|
|
"logps/rejected": -841.3156127929688,
|
|
"loss": 0.4329,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -4.335733890533447,
|
|
"rewards/margins": 1.6224541664123535,
|
|
"rewards/rejected": -5.958188056945801,
|
|
"step": 383
|
|
},
|
|
{
|
|
"epoch": 0.8207320331285065,
|
|
"grad_norm": 20.681275510903472,
|
|
"learning_rate": 7.464362216738882e-08,
|
|
"logits/chosen": -1.1937505006790161,
|
|
"logits/rejected": -1.172591209411621,
|
|
"logps/chosen": -640.5665283203125,
|
|
"logps/rejected": -749.8336181640625,
|
|
"loss": 0.4055,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.004388809204102,
|
|
"rewards/margins": 1.1090167760849,
|
|
"rewards/rejected": -5.113406181335449,
|
|
"step": 384
|
|
},
|
|
{
|
|
"epoch": 0.8228693561314454,
|
|
"grad_norm": 20.783011075052453,
|
|
"learning_rate": 7.291224573422746e-08,
|
|
"logits/chosen": -1.0488053560256958,
|
|
"logits/rejected": -1.0833758115768433,
|
|
"logps/chosen": -651.447998046875,
|
|
"logps/rejected": -798.273681640625,
|
|
"loss": 0.3371,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.223069190979004,
|
|
"rewards/margins": 1.4724674224853516,
|
|
"rewards/rejected": -5.695535659790039,
|
|
"step": 385
|
|
},
|
|
{
|
|
"epoch": 0.8250066791343842,
|
|
"grad_norm": 19.706489967433907,
|
|
"learning_rate": 7.119916982083606e-08,
|
|
"logits/chosen": -1.1488205194473267,
|
|
"logits/rejected": -1.1003917455673218,
|
|
"logps/chosen": -594.006591796875,
|
|
"logps/rejected": -760.5867309570312,
|
|
"loss": 0.3778,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.641071319580078,
|
|
"rewards/margins": 1.4457651376724243,
|
|
"rewards/rejected": -5.086835861206055,
|
|
"step": 386
|
|
},
|
|
{
|
|
"epoch": 0.827144002137323,
|
|
"grad_norm": 21.521620560904775,
|
|
"learning_rate": 6.950449027360213e-08,
|
|
"logits/chosen": -1.1488525867462158,
|
|
"logits/rejected": -1.1080577373504639,
|
|
"logps/chosen": -554.5031127929688,
|
|
"logps/rejected": -725.948974609375,
|
|
"loss": 0.3668,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -3.534379005432129,
|
|
"rewards/margins": 1.5944782495498657,
|
|
"rewards/rejected": -5.128857135772705,
|
|
"step": 387
|
|
},
|
|
{
|
|
"epoch": 0.8292813251402619,
|
|
"grad_norm": 18.836097701043112,
|
|
"learning_rate": 6.782830190963849e-08,
|
|
"logits/chosen": -1.0808215141296387,
|
|
"logits/rejected": -1.1298860311508179,
|
|
"logps/chosen": -559.855224609375,
|
|
"logps/rejected": -634.6371459960938,
|
|
"loss": 0.4358,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.5141537189483643,
|
|
"rewards/margins": 0.8853537440299988,
|
|
"rewards/rejected": -4.399507522583008,
|
|
"step": 388
|
|
},
|
|
{
|
|
"epoch": 0.8314186481432007,
|
|
"grad_norm": 17.005948701459808,
|
|
"learning_rate": 6.617069851147893e-08,
|
|
"logits/chosen": -1.1958553791046143,
|
|
"logits/rejected": -1.1517484188079834,
|
|
"logps/chosen": -775.6495971679688,
|
|
"logps/rejected": -891.900390625,
|
|
"loss": 0.3866,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -4.379716873168945,
|
|
"rewards/margins": 1.2400314807891846,
|
|
"rewards/rejected": -5.619749069213867,
|
|
"step": 389
|
|
},
|
|
{
|
|
"epoch": 0.8335559711461394,
|
|
"grad_norm": 18.782389894036523,
|
|
"learning_rate": 6.453177282183043e-08,
|
|
"logits/chosen": -1.0847554206848145,
|
|
"logits/rejected": -1.1218299865722656,
|
|
"logps/chosen": -553.1158447265625,
|
|
"logps/rejected": -640.9627075195312,
|
|
"loss": 0.395,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.2679800987243652,
|
|
"rewards/margins": 1.0876916646957397,
|
|
"rewards/rejected": -4.3556718826293945,
|
|
"step": 390
|
|
},
|
|
{
|
|
"epoch": 0.8356932941490782,
|
|
"grad_norm": 20.020475926403833,
|
|
"learning_rate": 6.291161653838433e-08,
|
|
"logits/chosen": -1.1652432680130005,
|
|
"logits/rejected": -1.1427745819091797,
|
|
"logps/chosen": -619.0631713867188,
|
|
"logps/rejected": -852.297607421875,
|
|
"loss": 0.347,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.763627052307129,
|
|
"rewards/margins": 2.3782565593719482,
|
|
"rewards/rejected": -6.141883850097656,
|
|
"step": 391
|
|
},
|
|
{
|
|
"epoch": 0.8378306171520171,
|
|
"grad_norm": 17.938841226698447,
|
|
"learning_rate": 6.131032030868634e-08,
|
|
"logits/chosen": -1.3010265827178955,
|
|
"logits/rejected": -1.2798787355422974,
|
|
"logps/chosen": -648.106201171875,
|
|
"logps/rejected": -922.129638671875,
|
|
"loss": 0.3221,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.8143765926361084,
|
|
"rewards/margins": 2.5941810607910156,
|
|
"rewards/rejected": -6.408557415008545,
|
|
"step": 392
|
|
},
|
|
{
|
|
"epoch": 0.8399679401549559,
|
|
"grad_norm": 18.079175693306084,
|
|
"learning_rate": 5.972797372506421e-08,
|
|
"logits/chosen": -1.3528358936309814,
|
|
"logits/rejected": -1.3354519605636597,
|
|
"logps/chosen": -690.3780517578125,
|
|
"logps/rejected": -912.6945190429688,
|
|
"loss": 0.3283,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.9751172065734863,
|
|
"rewards/margins": 1.9700416326522827,
|
|
"rewards/rejected": -5.945158958435059,
|
|
"step": 393
|
|
},
|
|
{
|
|
"epoch": 0.8421052631578947,
|
|
"grad_norm": 16.39904375972069,
|
|
"learning_rate": 5.8164665319615416e-08,
|
|
"logits/chosen": -1.3244935274124146,
|
|
"logits/rejected": -1.250441074371338,
|
|
"logps/chosen": -579.123779296875,
|
|
"logps/rejected": -744.0313720703125,
|
|
"loss": 0.3762,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.270462989807129,
|
|
"rewards/margins": 1.4447557926177979,
|
|
"rewards/rejected": -4.715218544006348,
|
|
"step": 394
|
|
},
|
|
{
|
|
"epoch": 0.8442425861608336,
|
|
"grad_norm": 21.900567022185207,
|
|
"learning_rate": 5.662048255925356e-08,
|
|
"logits/chosen": -1.1036981344223022,
|
|
"logits/rejected": -1.073646068572998,
|
|
"logps/chosen": -705.447509765625,
|
|
"logps/rejected": -859.1611938476562,
|
|
"loss": 0.4035,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -4.186145782470703,
|
|
"rewards/margins": 1.4756170511245728,
|
|
"rewards/rejected": -5.661762714385986,
|
|
"step": 395
|
|
},
|
|
{
|
|
"epoch": 0.8463799091637724,
|
|
"grad_norm": 24.440062318949717,
|
|
"learning_rate": 5.5095511840814957e-08,
|
|
"logits/chosen": -1.2547982931137085,
|
|
"logits/rejected": -1.2235480546951294,
|
|
"logps/chosen": -632.1664428710938,
|
|
"logps/rejected": -789.1854858398438,
|
|
"loss": 0.4916,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.523899555206299,
|
|
"rewards/margins": 1.5776091814041138,
|
|
"rewards/rejected": -5.101509094238281,
|
|
"step": 396
|
|
},
|
|
{
|
|
"epoch": 0.8485172321667112,
|
|
"grad_norm": 17.709481773760306,
|
|
"learning_rate": 5.358983848622451e-08,
|
|
"logits/chosen": -1.1620187759399414,
|
|
"logits/rejected": -1.1197460889816284,
|
|
"logps/chosen": -677.14697265625,
|
|
"logps/rejected": -884.9602661132812,
|
|
"loss": 0.3381,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.283136367797852,
|
|
"rewards/margins": 2.155815601348877,
|
|
"rewards/rejected": -6.43895149230957,
|
|
"step": 397
|
|
},
|
|
{
|
|
"epoch": 0.85065455516965,
|
|
"grad_norm": 21.95309780747416,
|
|
"learning_rate": 5.210354673772186e-08,
|
|
"logits/chosen": -1.1960110664367676,
|
|
"logits/rejected": -1.1809446811676025,
|
|
"logps/chosen": -591.931640625,
|
|
"logps/rejected": -729.0454711914062,
|
|
"loss": 0.3959,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.7348904609680176,
|
|
"rewards/margins": 1.1620283126831055,
|
|
"rewards/rejected": -4.896918296813965,
|
|
"step": 398
|
|
},
|
|
{
|
|
"epoch": 0.8527918781725888,
|
|
"grad_norm": 18.231544991301615,
|
|
"learning_rate": 5.063671975314814e-08,
|
|
"logits/chosen": -1.1651999950408936,
|
|
"logits/rejected": -1.1458827257156372,
|
|
"logps/chosen": -544.5778198242188,
|
|
"logps/rejected": -645.9469604492188,
|
|
"loss": 0.4343,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.4847066402435303,
|
|
"rewards/margins": 0.870603084564209,
|
|
"rewards/rejected": -4.355309963226318,
|
|
"step": 399
|
|
},
|
|
{
|
|
"epoch": 0.8549292011755276,
|
|
"grad_norm": 16.85803441290228,
|
|
"learning_rate": 4.918943960129338e-08,
|
|
"logits/chosen": -1.356669306755066,
|
|
"logits/rejected": -1.3354345560073853,
|
|
"logps/chosen": -568.3810424804688,
|
|
"logps/rejected": -700.6995239257812,
|
|
"loss": 0.3856,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.211237907409668,
|
|
"rewards/margins": 1.343750238418579,
|
|
"rewards/rejected": -4.554987907409668,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 0.8549292011755276,
|
|
"eval_logits/chosen": -1.3413100242614746,
|
|
"eval_logits/rejected": -1.3605502843856812,
|
|
"eval_logps/chosen": -735.5136108398438,
|
|
"eval_logps/rejected": -898.4371337890625,
|
|
"eval_loss": 0.38407519459724426,
|
|
"eval_rewards/accuracies": 0.8588709831237793,
|
|
"eval_rewards/chosen": -4.560075759887695,
|
|
"eval_rewards/margins": 1.6887413263320923,
|
|
"eval_rewards/rejected": -6.2488179206848145,
|
|
"eval_runtime": 127.6843,
|
|
"eval_samples_per_second": 15.358,
|
|
"eval_steps_per_second": 0.486,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 0.8570665241784665,
|
|
"grad_norm": 19.09808858732997,
|
|
"learning_rate": 4.7761787257304796e-08,
|
|
"logits/chosen": -1.1470953226089478,
|
|
"logits/rejected": -1.167830467224121,
|
|
"logps/chosen": -727.25341796875,
|
|
"logps/rejected": -1019.1134643554688,
|
|
"loss": 0.3153,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -4.293584823608398,
|
|
"rewards/margins": 2.9339499473571777,
|
|
"rewards/rejected": -7.227534294128418,
|
|
"step": 401
|
|
},
|
|
{
|
|
"epoch": 0.8592038471814053,
|
|
"grad_norm": 20.031183507985155,
|
|
"learning_rate": 4.6353842598156134e-08,
|
|
"logits/chosen": -1.2042540311813354,
|
|
"logits/rejected": -1.2306770086288452,
|
|
"logps/chosen": -832.0037841796875,
|
|
"logps/rejected": -949.708251953125,
|
|
"loss": 0.4025,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -5.143182277679443,
|
|
"rewards/margins": 1.2332004308700562,
|
|
"rewards/rejected": -6.376382827758789,
|
|
"step": 402
|
|
},
|
|
{
|
|
"epoch": 0.8613411701843441,
|
|
"grad_norm": 20.27761411702487,
|
|
"learning_rate": 4.496568439817836e-08,
|
|
"logits/chosen": -1.0823955535888672,
|
|
"logits/rejected": -1.0055339336395264,
|
|
"logps/chosen": -647.8933715820312,
|
|
"logps/rejected": -846.7052001953125,
|
|
"loss": 0.4049,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -4.089991092681885,
|
|
"rewards/margins": 1.9131741523742676,
|
|
"rewards/rejected": -6.003165245056152,
|
|
"step": 403
|
|
},
|
|
{
|
|
"epoch": 0.863478493187283,
|
|
"grad_norm": 20.06260268066064,
|
|
"learning_rate": 4.359739032465288e-08,
|
|
"logits/chosen": -1.238147258758545,
|
|
"logits/rejected": -1.2503734827041626,
|
|
"logps/chosen": -628.3074340820312,
|
|
"logps/rejected": -690.2806396484375,
|
|
"loss": 0.4391,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.6560134887695312,
|
|
"rewards/margins": 0.6559101343154907,
|
|
"rewards/rejected": -4.311923503875732,
|
|
"step": 404
|
|
},
|
|
{
|
|
"epoch": 0.8656158161902218,
|
|
"grad_norm": 24.04539917692329,
|
|
"learning_rate": 4.224903693346533e-08,
|
|
"logits/chosen": -1.1764211654663086,
|
|
"logits/rejected": -1.1727336645126343,
|
|
"logps/chosen": -837.36767578125,
|
|
"logps/rejected": -985.0379638671875,
|
|
"loss": 0.4175,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -5.274961471557617,
|
|
"rewards/margins": 1.683783769607544,
|
|
"rewards/rejected": -6.95874547958374,
|
|
"step": 405
|
|
},
|
|
{
|
|
"epoch": 0.8677531391931605,
|
|
"grad_norm": 16.471866635220206,
|
|
"learning_rate": 4.092069966482272e-08,
|
|
"logits/chosen": -1.0595924854278564,
|
|
"logits/rejected": -1.1007946729660034,
|
|
"logps/chosen": -696.4332275390625,
|
|
"logps/rejected": -854.4127197265625,
|
|
"loss": 0.3502,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.544949531555176,
|
|
"rewards/margins": 1.6854453086853027,
|
|
"rewards/rejected": -6.23039436340332,
|
|
"step": 406
|
|
},
|
|
{
|
|
"epoch": 0.8698904621960993,
|
|
"grad_norm": 18.880730560681034,
|
|
"learning_rate": 3.9612452839032386e-08,
|
|
"logits/chosen": -1.1478521823883057,
|
|
"logits/rejected": -1.1282038688659668,
|
|
"logps/chosen": -612.4228515625,
|
|
"logps/rejected": -871.350341796875,
|
|
"loss": 0.3887,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -4.034952163696289,
|
|
"rewards/margins": 2.548003911972046,
|
|
"rewards/rejected": -6.582956314086914,
|
|
"step": 407
|
|
},
|
|
{
|
|
"epoch": 0.8720277851990382,
|
|
"grad_norm": 24.18209624562828,
|
|
"learning_rate": 3.8324369652343954e-08,
|
|
"logits/chosen": -1.2262625694274902,
|
|
"logits/rejected": -1.1821433305740356,
|
|
"logps/chosen": -582.0433959960938,
|
|
"logps/rejected": -733.720458984375,
|
|
"loss": 0.4053,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.865800142288208,
|
|
"rewards/margins": 1.6720855236053467,
|
|
"rewards/rejected": -5.537886142730713,
|
|
"step": 408
|
|
},
|
|
{
|
|
"epoch": 0.874165108201977,
|
|
"grad_norm": 18.382867885499415,
|
|
"learning_rate": 3.7056522172854045e-08,
|
|
"logits/chosen": -1.2061108350753784,
|
|
"logits/rejected": -1.2035906314849854,
|
|
"logps/chosen": -787.7473754882812,
|
|
"logps/rejected": -904.2359008789062,
|
|
"loss": 0.3439,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -5.003305435180664,
|
|
"rewards/margins": 1.1130136251449585,
|
|
"rewards/rejected": -6.116319179534912,
|
|
"step": 409
|
|
},
|
|
{
|
|
"epoch": 0.8763024312049158,
|
|
"grad_norm": 23.170208730820974,
|
|
"learning_rate": 3.580898133647365e-08,
|
|
"logits/chosen": -1.1570967435836792,
|
|
"logits/rejected": -1.1725834608078003,
|
|
"logps/chosen": -491.8000793457031,
|
|
"logps/rejected": -648.202880859375,
|
|
"loss": 0.438,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.1505000591278076,
|
|
"rewards/margins": 1.3966213464736938,
|
|
"rewards/rejected": -4.547121524810791,
|
|
"step": 410
|
|
},
|
|
{
|
|
"epoch": 0.8784397542078547,
|
|
"grad_norm": 13.414426080076618,
|
|
"learning_rate": 3.4581816942959606e-08,
|
|
"logits/chosen": -1.2170116901397705,
|
|
"logits/rejected": -1.1329779624938965,
|
|
"logps/chosen": -799.8605346679688,
|
|
"logps/rejected": -1069.72412109375,
|
|
"loss": 0.3279,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -4.736441612243652,
|
|
"rewards/margins": 2.4817426204681396,
|
|
"rewards/rejected": -7.218183994293213,
|
|
"step": 411
|
|
},
|
|
{
|
|
"epoch": 0.8805770772107935,
|
|
"grad_norm": 25.164272834377723,
|
|
"learning_rate": 3.337509765200952e-08,
|
|
"logits/chosen": -1.263645887374878,
|
|
"logits/rejected": -1.2011065483093262,
|
|
"logps/chosen": -733.5821533203125,
|
|
"logps/rejected": -922.4378662109375,
|
|
"loss": 0.4522,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.364438533782959,
|
|
"rewards/margins": 1.6717685461044312,
|
|
"rewards/rejected": -6.0362067222595215,
|
|
"step": 412
|
|
},
|
|
{
|
|
"epoch": 0.8827144002137323,
|
|
"grad_norm": 21.972539663439804,
|
|
"learning_rate": 3.218889097941977e-08,
|
|
"logits/chosen": -1.1184589862823486,
|
|
"logits/rejected": -1.0591604709625244,
|
|
"logps/chosen": -735.7826538085938,
|
|
"logps/rejected": -919.8989868164062,
|
|
"loss": 0.4028,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.92265510559082,
|
|
"rewards/margins": 1.8032639026641846,
|
|
"rewards/rejected": -6.725918769836426,
|
|
"step": 413
|
|
},
|
|
{
|
|
"epoch": 0.8848517232166712,
|
|
"grad_norm": 19.5992524976479,
|
|
"learning_rate": 3.102326329330833e-08,
|
|
"logits/chosen": -1.180734395980835,
|
|
"logits/rejected": -1.1732540130615234,
|
|
"logps/chosen": -568.0210571289062,
|
|
"logps/rejected": -707.9583740234375,
|
|
"loss": 0.3788,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -3.657954454421997,
|
|
"rewards/margins": 1.3774219751358032,
|
|
"rewards/rejected": -5.03537654876709,
|
|
"step": 414
|
|
},
|
|
{
|
|
"epoch": 0.88698904621961,
|
|
"grad_norm": 19.912191558357517,
|
|
"learning_rate": 2.987827981040132e-08,
|
|
"logits/chosen": -1.0525703430175781,
|
|
"logits/rejected": -1.0452289581298828,
|
|
"logps/chosen": -732.814453125,
|
|
"logps/rejected": -772.8712158203125,
|
|
"loss": 0.3975,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -4.660562515258789,
|
|
"rewards/margins": 0.4499680995941162,
|
|
"rewards/rejected": -5.110530853271484,
|
|
"step": 415
|
|
},
|
|
{
|
|
"epoch": 0.8891263692225487,
|
|
"grad_norm": 26.727536724022972,
|
|
"learning_rate": 2.875400459238446e-08,
|
|
"logits/chosen": -1.2559760808944702,
|
|
"logits/rejected": -1.2618238925933838,
|
|
"logps/chosen": -765.4171142578125,
|
|
"logps/rejected": -1030.8968505859375,
|
|
"loss": 0.453,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.885884761810303,
|
|
"rewards/margins": 2.478846549987793,
|
|
"rewards/rejected": -7.364731788635254,
|
|
"step": 416
|
|
},
|
|
{
|
|
"epoch": 0.8912636922254876,
|
|
"grad_norm": 22.005485483200033,
|
|
"learning_rate": 2.7650500542318344e-08,
|
|
"logits/chosen": -1.2461367845535278,
|
|
"logits/rejected": -1.2258386611938477,
|
|
"logps/chosen": -687.5975952148438,
|
|
"logps/rejected": -837.7894287109375,
|
|
"loss": 0.3724,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.603135108947754,
|
|
"rewards/margins": 1.5509103536605835,
|
|
"rewards/rejected": -6.154045104980469,
|
|
"step": 417
|
|
},
|
|
{
|
|
"epoch": 0.8934010152284264,
|
|
"grad_norm": 20.686911824626822,
|
|
"learning_rate": 2.6567829401119346e-08,
|
|
"logits/chosen": -1.2208616733551025,
|
|
"logits/rejected": -1.208647608757019,
|
|
"logps/chosen": -694.425537109375,
|
|
"logps/rejected": -811.75927734375,
|
|
"loss": 0.3835,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.29609489440918,
|
|
"rewards/margins": 1.2039227485656738,
|
|
"rewards/rejected": -5.500017166137695,
|
|
"step": 418
|
|
},
|
|
{
|
|
"epoch": 0.8955383382313652,
|
|
"grad_norm": 21.443800944172835,
|
|
"learning_rate": 2.5506051744105117e-08,
|
|
"logits/chosen": -1.1272634267807007,
|
|
"logits/rejected": -1.1442179679870605,
|
|
"logps/chosen": -639.4913330078125,
|
|
"logps/rejected": -763.7130737304688,
|
|
"loss": 0.4385,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.8574483394622803,
|
|
"rewards/margins": 1.4068204164505005,
|
|
"rewards/rejected": -5.26426887512207,
|
|
"step": 419
|
|
},
|
|
{
|
|
"epoch": 0.897675661234304,
|
|
"grad_norm": 25.09735847833568,
|
|
"learning_rate": 2.4465226977605558e-08,
|
|
"logits/chosen": -1.192622423171997,
|
|
"logits/rejected": -1.1234114170074463,
|
|
"logps/chosen": -763.729248046875,
|
|
"logps/rejected": -975.58935546875,
|
|
"loss": 0.4638,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -4.789132118225098,
|
|
"rewards/margins": 1.976550579071045,
|
|
"rewards/rejected": -6.765683174133301,
|
|
"step": 420
|
|
},
|
|
{
|
|
"epoch": 0.8998129842372429,
|
|
"grad_norm": 19.301280601090053,
|
|
"learning_rate": 2.3445413335638763e-08,
|
|
"logits/chosen": -1.3491016626358032,
|
|
"logits/rejected": -1.3572145700454712,
|
|
"logps/chosen": -705.7742919921875,
|
|
"logps/rejected": -887.14013671875,
|
|
"loss": 0.3918,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.3721818923950195,
|
|
"rewards/margins": 1.478896975517273,
|
|
"rewards/rejected": -5.851078510284424,
|
|
"step": 421
|
|
},
|
|
{
|
|
"epoch": 0.9019503072401817,
|
|
"grad_norm": 23.607839241733036,
|
|
"learning_rate": 2.244666787665297e-08,
|
|
"logits/chosen": -1.2374908924102783,
|
|
"logits/rejected": -1.2936606407165527,
|
|
"logps/chosen": -756.2825927734375,
|
|
"logps/rejected": -903.6665649414062,
|
|
"loss": 0.4241,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.921727180480957,
|
|
"rewards/margins": 1.6834334135055542,
|
|
"rewards/rejected": -6.605160713195801,
|
|
"step": 422
|
|
},
|
|
{
|
|
"epoch": 0.9040876302431204,
|
|
"grad_norm": 16.860532940493467,
|
|
"learning_rate": 2.1469046480334296e-08,
|
|
"logits/chosen": -1.228078842163086,
|
|
"logits/rejected": -1.1509662866592407,
|
|
"logps/chosen": -617.8320922851562,
|
|
"logps/rejected": -804.2755737304688,
|
|
"loss": 0.3241,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -4.25843620300293,
|
|
"rewards/margins": 1.6883177757263184,
|
|
"rewards/rejected": -5.946753978729248,
|
|
"step": 423
|
|
},
|
|
{
|
|
"epoch": 0.9062249532460593,
|
|
"grad_norm": 19.717442168166148,
|
|
"learning_rate": 2.05126038444801e-08,
|
|
"logits/chosen": -1.047011137008667,
|
|
"logits/rejected": -1.0633037090301514,
|
|
"logps/chosen": -587.2730712890625,
|
|
"logps/rejected": -760.9138793945312,
|
|
"loss": 0.411,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.5641701221466064,
|
|
"rewards/margins": 1.6567331552505493,
|
|
"rewards/rejected": -5.220903396606445,
|
|
"step": 424
|
|
},
|
|
{
|
|
"epoch": 0.9083622762489981,
|
|
"grad_norm": 19.401646167891172,
|
|
"learning_rate": 1.9577393481938586e-08,
|
|
"logits/chosen": -1.1300030946731567,
|
|
"logits/rejected": -1.1926459074020386,
|
|
"logps/chosen": -668.7816772460938,
|
|
"logps/rejected": -773.44677734375,
|
|
"loss": 0.371,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.893263578414917,
|
|
"rewards/margins": 1.1991028785705566,
|
|
"rewards/rejected": -5.0923662185668945,
|
|
"step": 425
|
|
},
|
|
{
|
|
"epoch": 0.9104995992519369,
|
|
"grad_norm": 19.48240593398746,
|
|
"learning_rate": 1.8663467717614955e-08,
|
|
"logits/chosen": -1.2391318082809448,
|
|
"logits/rejected": -1.218917727470398,
|
|
"logps/chosen": -715.0926513671875,
|
|
"logps/rejected": -927.725830078125,
|
|
"loss": 0.337,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.81939697265625,
|
|
"rewards/margins": 2.132520914077759,
|
|
"rewards/rejected": -6.95191764831543,
|
|
"step": 426
|
|
},
|
|
{
|
|
"epoch": 0.9126369222548758,
|
|
"grad_norm": 18.414154188749755,
|
|
"learning_rate": 1.7770877685543685e-08,
|
|
"logits/chosen": -0.9984086155891418,
|
|
"logits/rejected": -0.9543211460113525,
|
|
"logps/chosen": -630.6248168945312,
|
|
"logps/rejected": -776.2060546875,
|
|
"loss": 0.3846,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.3436994552612305,
|
|
"rewards/margins": 1.370991587638855,
|
|
"rewards/rejected": -5.714690685272217,
|
|
"step": 427
|
|
},
|
|
{
|
|
"epoch": 0.9147742452578146,
|
|
"grad_norm": 28.47516460996172,
|
|
"learning_rate": 1.689967332602782e-08,
|
|
"logits/chosen": -1.2434695959091187,
|
|
"logits/rejected": -1.2322406768798828,
|
|
"logps/chosen": -648.59375,
|
|
"logps/rejected": -814.1904907226562,
|
|
"loss": 0.4201,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.107604026794434,
|
|
"rewards/margins": 1.6489994525909424,
|
|
"rewards/rejected": -5.756603717803955,
|
|
"step": 428
|
|
},
|
|
{
|
|
"epoch": 0.9169115682607534,
|
|
"grad_norm": 20.787778312181622,
|
|
"learning_rate": 1.604990338284442e-08,
|
|
"logits/chosen": -1.2705280780792236,
|
|
"logits/rejected": -1.1843757629394531,
|
|
"logps/chosen": -623.3156127929688,
|
|
"logps/rejected": -877.883544921875,
|
|
"loss": 0.3917,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.032170295715332,
|
|
"rewards/margins": 2.4661622047424316,
|
|
"rewards/rejected": -6.498332500457764,
|
|
"step": 429
|
|
},
|
|
{
|
|
"epoch": 0.9190488912636923,
|
|
"grad_norm": 23.557666816240015,
|
|
"learning_rate": 1.522161540051763e-08,
|
|
"logits/chosen": -1.318623423576355,
|
|
"logits/rejected": -1.2060602903366089,
|
|
"logps/chosen": -841.9174194335938,
|
|
"logps/rejected": -1067.8670654296875,
|
|
"loss": 0.3666,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -5.210690975189209,
|
|
"rewards/margins": 2.1968445777893066,
|
|
"rewards/rejected": -7.407536029815674,
|
|
"step": 430
|
|
},
|
|
{
|
|
"epoch": 0.921186214266631,
|
|
"grad_norm": 19.13785898561301,
|
|
"learning_rate": 1.4414855721658703e-08,
|
|
"logits/chosen": -1.153996467590332,
|
|
"logits/rejected": -1.1101250648498535,
|
|
"logps/chosen": -573.8594970703125,
|
|
"logps/rejected": -784.9390869140625,
|
|
"loss": 0.3734,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.4458227157592773,
|
|
"rewards/margins": 2.165468692779541,
|
|
"rewards/rejected": -5.611291408538818,
|
|
"step": 431
|
|
},
|
|
{
|
|
"epoch": 0.9233235372695698,
|
|
"grad_norm": 24.984469469230163,
|
|
"learning_rate": 1.362966948437272e-08,
|
|
"logits/chosen": -1.161950707435608,
|
|
"logits/rejected": -1.1117969751358032,
|
|
"logps/chosen": -742.8123779296875,
|
|
"logps/rejected": -1053.392578125,
|
|
"loss": 0.4537,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -4.41086483001709,
|
|
"rewards/margins": 2.8615894317626953,
|
|
"rewards/rejected": -7.272454261779785,
|
|
"step": 432
|
|
},
|
|
{
|
|
"epoch": 0.9254608602725087,
|
|
"grad_norm": 17.14897188522608,
|
|
"learning_rate": 1.2866100619733434e-08,
|
|
"logits/chosen": -1.1124742031097412,
|
|
"logits/rejected": -1.0683679580688477,
|
|
"logps/chosen": -692.58056640625,
|
|
"logps/rejected": -865.1836547851562,
|
|
"loss": 0.4003,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.52462911605835,
|
|
"rewards/margins": 1.5883965492248535,
|
|
"rewards/rejected": -6.113025665283203,
|
|
"step": 433
|
|
},
|
|
{
|
|
"epoch": 0.9275981832754475,
|
|
"grad_norm": 18.976385617816387,
|
|
"learning_rate": 1.212419184932525e-08,
|
|
"logits/chosen": -1.283325433731079,
|
|
"logits/rejected": -1.2627527713775635,
|
|
"logps/chosen": -848.5104370117188,
|
|
"logps/rejected": -1140.6717529296875,
|
|
"loss": 0.323,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -5.1217360496521,
|
|
"rewards/margins": 2.514495611190796,
|
|
"rewards/rejected": -7.636231899261475,
|
|
"step": 434
|
|
},
|
|
{
|
|
"epoch": 0.9297355062783863,
|
|
"grad_norm": 18.525625944510605,
|
|
"learning_rate": 1.1403984682852997e-08,
|
|
"logits/chosen": -1.1664470434188843,
|
|
"logits/rejected": -1.1477528810501099,
|
|
"logps/chosen": -637.484619140625,
|
|
"logps/rejected": -859.3595581054688,
|
|
"loss": 0.3552,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.328897953033447,
|
|
"rewards/margins": 1.8881185054779053,
|
|
"rewards/rejected": -6.217016220092773,
|
|
"step": 435
|
|
},
|
|
{
|
|
"epoch": 0.9318728292813251,
|
|
"grad_norm": 23.166677968022782,
|
|
"learning_rate": 1.0705519415819431e-08,
|
|
"logits/chosen": -1.2317030429840088,
|
|
"logits/rejected": -1.1913528442382812,
|
|
"logps/chosen": -722.13330078125,
|
|
"logps/rejected": -883.4630126953125,
|
|
"loss": 0.385,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.459592819213867,
|
|
"rewards/margins": 1.4444059133529663,
|
|
"rewards/rejected": -5.903999328613281,
|
|
"step": 436
|
|
},
|
|
{
|
|
"epoch": 0.934010152284264,
|
|
"grad_norm": 29.360763259322653,
|
|
"learning_rate": 1.0028835127270552e-08,
|
|
"logits/chosen": -1.041462779045105,
|
|
"logits/rejected": -1.1673197746276855,
|
|
"logps/chosen": -612.159912109375,
|
|
"logps/rejected": -679.0078125,
|
|
"loss": 0.5247,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -3.783984422683716,
|
|
"rewards/margins": 0.7272745370864868,
|
|
"rewards/rejected": -4.511259078979492,
|
|
"step": 437
|
|
},
|
|
{
|
|
"epoch": 0.9361474752872028,
|
|
"grad_norm": 19.605084291231115,
|
|
"learning_rate": 9.373969677609483e-09,
|
|
"logits/chosen": -1.3655058145523071,
|
|
"logits/rejected": -1.294546127319336,
|
|
"logps/chosen": -698.7198486328125,
|
|
"logps/rejected": -940.9436645507812,
|
|
"loss": 0.427,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.112486839294434,
|
|
"rewards/margins": 2.187129497528076,
|
|
"rewards/rejected": -6.299617290496826,
|
|
"step": 438
|
|
},
|
|
{
|
|
"epoch": 0.9382847982901416,
|
|
"grad_norm": 22.063411073782135,
|
|
"learning_rate": 8.740959706477725e-09,
|
|
"logits/chosen": -1.077052354812622,
|
|
"logits/rejected": -1.151231050491333,
|
|
"logps/chosen": -696.8163452148438,
|
|
"logps/rejected": -933.444091796875,
|
|
"loss": 0.4013,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -4.566803455352783,
|
|
"rewards/margins": 2.404057025909424,
|
|
"rewards/rejected": -6.970860481262207,
|
|
"step": 439
|
|
},
|
|
{
|
|
"epoch": 0.9404221212930804,
|
|
"grad_norm": 25.49031368931362,
|
|
"learning_rate": 8.129840630705809e-09,
|
|
"logits/chosen": -1.328292727470398,
|
|
"logits/rejected": -1.3197101354599,
|
|
"logps/chosen": -745.0568237304688,
|
|
"logps/rejected": -901.1380615234375,
|
|
"loss": 0.4558,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.428404808044434,
|
|
"rewards/margins": 1.5418949127197266,
|
|
"rewards/rejected": -5.970300674438477,
|
|
"step": 440
|
|
},
|
|
{
|
|
"epoch": 0.9425594442960192,
|
|
"grad_norm": 19.90218785765348,
|
|
"learning_rate": 7.540646642330984e-09,
|
|
"logits/chosen": -1.204488754272461,
|
|
"logits/rejected": -1.0937587022781372,
|
|
"logps/chosen": -753.1057739257812,
|
|
"logps/rejected": -1030.1650390625,
|
|
"loss": 0.4377,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.8104987144470215,
|
|
"rewards/margins": 2.6143453121185303,
|
|
"rewards/rejected": -7.424844741821289,
|
|
"step": 441
|
|
},
|
|
{
|
|
"epoch": 0.944696767298958,
|
|
"grad_norm": 17.97300546078058,
|
|
"learning_rate": 6.9734107066846906e-09,
|
|
"logits/chosen": -1.2257647514343262,
|
|
"logits/rejected": -1.260422945022583,
|
|
"logps/chosen": -596.8286743164062,
|
|
"logps/rejected": -672.286376953125,
|
|
"loss": 0.3596,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -3.532057523727417,
|
|
"rewards/margins": 0.8501084446907043,
|
|
"rewards/rejected": -4.382165908813477,
|
|
"step": 442
|
|
},
|
|
{
|
|
"epoch": 0.9468340903018969,
|
|
"grad_norm": 17.202047962190797,
|
|
"learning_rate": 6.428164560548133e-09,
|
|
"logits/chosen": -1.1206276416778564,
|
|
"logits/rejected": -1.157448172569275,
|
|
"logps/chosen": -758.1364135742188,
|
|
"logps/rejected": -882.7911376953125,
|
|
"loss": 0.3883,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.609854698181152,
|
|
"rewards/margins": 1.4294791221618652,
|
|
"rewards/rejected": -6.039333820343018,
|
|
"step": 443
|
|
},
|
|
{
|
|
"epoch": 0.9489714133048357,
|
|
"grad_norm": 23.28751615124104,
|
|
"learning_rate": 5.9049387103763175e-09,
|
|
"logits/chosen": -1.0286593437194824,
|
|
"logits/rejected": -1.0789514780044556,
|
|
"logps/chosen": -584.1970825195312,
|
|
"logps/rejected": -675.3173828125,
|
|
"loss": 0.4016,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.8274412155151367,
|
|
"rewards/margins": 1.028598427772522,
|
|
"rewards/rejected": -4.856039047241211,
|
|
"step": 444
|
|
},
|
|
{
|
|
"epoch": 0.9511087363077745,
|
|
"grad_norm": 21.061524434011588,
|
|
"learning_rate": 5.403762430591463e-09,
|
|
"logits/chosen": -1.3204259872436523,
|
|
"logits/rejected": -1.2254197597503662,
|
|
"logps/chosen": -608.9866943359375,
|
|
"logps/rejected": -819.9737548828125,
|
|
"loss": 0.3407,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.7939043045043945,
|
|
"rewards/margins": 1.8012745380401611,
|
|
"rewards/rejected": -5.595179080963135,
|
|
"step": 445
|
|
},
|
|
{
|
|
"epoch": 0.9532460593107134,
|
|
"grad_norm": 24.59026604382506,
|
|
"learning_rate": 4.924663761944936e-09,
|
|
"logits/chosen": -1.1305930614471436,
|
|
"logits/rejected": -1.1460232734680176,
|
|
"logps/chosen": -762.6463623046875,
|
|
"logps/rejected": -925.34326171875,
|
|
"loss": 0.3586,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -4.521890163421631,
|
|
"rewards/margins": 1.8425782918930054,
|
|
"rewards/rejected": -6.364468574523926,
|
|
"step": 446
|
|
},
|
|
{
|
|
"epoch": 0.9553833823136522,
|
|
"grad_norm": 20.723204696998593,
|
|
"learning_rate": 4.467669509948591e-09,
|
|
"logits/chosen": -1.22001314163208,
|
|
"logits/rejected": -1.2475132942199707,
|
|
"logps/chosen": -567.6207275390625,
|
|
"logps/rejected": -686.4295043945312,
|
|
"loss": 0.363,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -3.614631175994873,
|
|
"rewards/margins": 1.1865804195404053,
|
|
"rewards/rejected": -4.801211833953857,
|
|
"step": 447
|
|
},
|
|
{
|
|
"epoch": 0.957520705316591,
|
|
"grad_norm": 17.735725565956283,
|
|
"learning_rate": 4.032805243374815e-09,
|
|
"logits/chosen": -1.0866748094558716,
|
|
"logits/rejected": -1.0737552642822266,
|
|
"logps/chosen": -707.5775146484375,
|
|
"logps/rejected": -885.0850830078125,
|
|
"loss": 0.3426,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.333130836486816,
|
|
"rewards/margins": 1.6789565086364746,
|
|
"rewards/rejected": -6.012087345123291,
|
|
"step": 448
|
|
},
|
|
{
|
|
"epoch": 0.9596580283195298,
|
|
"grad_norm": 17.3793434143283,
|
|
"learning_rate": 3.62009529282612e-09,
|
|
"logits/chosen": -1.2599151134490967,
|
|
"logits/rejected": -1.234681487083435,
|
|
"logps/chosen": -582.177978515625,
|
|
"logps/rejected": -735.9927978515625,
|
|
"loss": 0.4105,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.6418380737304688,
|
|
"rewards/margins": 1.4252443313598633,
|
|
"rewards/rejected": -5.06708288192749,
|
|
"step": 449
|
|
},
|
|
{
|
|
"epoch": 0.9617953513224686,
|
|
"grad_norm": 24.52461330654904,
|
|
"learning_rate": 3.2295627493735155e-09,
|
|
"logits/chosen": -1.2312730550765991,
|
|
"logits/rejected": -1.1974250078201294,
|
|
"logps/chosen": -790.2763671875,
|
|
"logps/rejected": -950.1840209960938,
|
|
"loss": 0.4455,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.5439605712890625,
|
|
"rewards/margins": 1.6306798458099365,
|
|
"rewards/rejected": -6.174641132354736,
|
|
"step": 450
|
|
},
|
|
{
|
|
"epoch": 0.9639326743254074,
|
|
"grad_norm": 22.5202862560615,
|
|
"learning_rate": 2.8612294632650578e-09,
|
|
"logits/chosen": -1.2211391925811768,
|
|
"logits/rejected": -1.226765751838684,
|
|
"logps/chosen": -697.6963500976562,
|
|
"logps/rejected": -789.454833984375,
|
|
"loss": 0.4469,
|
|
"rewards/accuracies": 0.5625,
|
|
"rewards/chosen": -4.249999046325684,
|
|
"rewards/margins": 0.9787368774414062,
|
|
"rewards/rejected": -5.22873592376709,
|
|
"step": 451
|
|
},
|
|
{
|
|
"epoch": 0.9660699973283462,
|
|
"grad_norm": 21.041642949979845,
|
|
"learning_rate": 2.5151160427029583e-09,
|
|
"logits/chosen": -1.341503381729126,
|
|
"logits/rejected": -1.2913732528686523,
|
|
"logps/chosen": -760.9803466796875,
|
|
"logps/rejected": -895.4586181640625,
|
|
"loss": 0.4508,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -4.8633880615234375,
|
|
"rewards/margins": 1.360411524772644,
|
|
"rewards/rejected": -6.223799228668213,
|
|
"step": 452
|
|
},
|
|
{
|
|
"epoch": 0.9682073203312851,
|
|
"grad_norm": 24.476456599792517,
|
|
"learning_rate": 2.191241852690684e-09,
|
|
"logits/chosen": -1.2209526300430298,
|
|
"logits/rejected": -1.3040820360183716,
|
|
"logps/chosen": -729.9022216796875,
|
|
"logps/rejected": -835.0201416015625,
|
|
"loss": 0.4371,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -4.68671989440918,
|
|
"rewards/margins": 1.158172369003296,
|
|
"rewards/rejected": -5.8448920249938965,
|
|
"step": 453
|
|
},
|
|
{
|
|
"epoch": 0.9703446433342239,
|
|
"grad_norm": 17.467256417005025,
|
|
"learning_rate": 1.8896250139494252e-09,
|
|
"logits/chosen": -1.133383870124817,
|
|
"logits/rejected": -1.1557976007461548,
|
|
"logps/chosen": -641.5121459960938,
|
|
"logps/rejected": -720.6392211914062,
|
|
"loss": 0.3563,
|
|
"rewards/accuracies": 0.6875,
|
|
"rewards/chosen": -4.076579570770264,
|
|
"rewards/margins": 0.7415927648544312,
|
|
"rewards/rejected": -4.818172454833984,
|
|
"step": 454
|
|
},
|
|
{
|
|
"epoch": 0.9724819663371627,
|
|
"grad_norm": 29.06168949637391,
|
|
"learning_rate": 1.6102824019043725e-09,
|
|
"logits/chosen": -1.3402838706970215,
|
|
"logits/rejected": -1.2556098699569702,
|
|
"logps/chosen": -714.4627685546875,
|
|
"logps/rejected": -948.2529907226562,
|
|
"loss": 0.4449,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.323197364807129,
|
|
"rewards/margins": 2.1209869384765625,
|
|
"rewards/rejected": -6.444183826446533,
|
|
"step": 455
|
|
},
|
|
{
|
|
"epoch": 0.9746192893401016,
|
|
"grad_norm": 24.172794750012795,
|
|
"learning_rate": 1.3532296457404946e-09,
|
|
"logits/chosen": -1.0889981985092163,
|
|
"logits/rejected": -1.0795581340789795,
|
|
"logps/chosen": -892.805419921875,
|
|
"logps/rejected": -1116.1790771484375,
|
|
"loss": 0.4206,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -5.471851825714111,
|
|
"rewards/margins": 2.0194764137268066,
|
|
"rewards/rejected": -7.491328239440918,
|
|
"step": 456
|
|
},
|
|
{
|
|
"epoch": 0.9767566123430403,
|
|
"grad_norm": 23.800223087589952,
|
|
"learning_rate": 1.1184811275279481e-09,
|
|
"logits/chosen": -1.1701947450637817,
|
|
"logits/rejected": -1.1236315965652466,
|
|
"logps/chosen": -692.0287475585938,
|
|
"logps/rejected": -1029.572509765625,
|
|
"loss": 0.3649,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -4.509798526763916,
|
|
"rewards/margins": 3.4749834537506104,
|
|
"rewards/rejected": -7.9847822189331055,
|
|
"step": 457
|
|
},
|
|
{
|
|
"epoch": 0.9788939353459791,
|
|
"grad_norm": 16.492942853907678,
|
|
"learning_rate": 9.06049981417567e-10,
|
|
"logits/chosen": -1.2626256942749023,
|
|
"logits/rejected": -1.289964199066162,
|
|
"logps/chosen": -730.7406616210938,
|
|
"logps/rejected": -926.16015625,
|
|
"loss": 0.4024,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.51789665222168,
|
|
"rewards/margins": 2.0349254608154297,
|
|
"rewards/rejected": -6.552822113037109,
|
|
"step": 458
|
|
},
|
|
{
|
|
"epoch": 0.981031258348918,
|
|
"grad_norm": 23.671232760661084,
|
|
"learning_rate": 7.159480929059381e-10,
|
|
"logits/chosen": -1.2181737422943115,
|
|
"logits/rejected": -1.1888567209243774,
|
|
"logps/chosen": -670.6812744140625,
|
|
"logps/rejected": -897.4075317382812,
|
|
"loss": 0.4605,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -3.946488380432129,
|
|
"rewards/margins": 2.147352933883667,
|
|
"rewards/rejected": -6.093842029571533,
|
|
"step": 459
|
|
},
|
|
{
|
|
"epoch": 0.9831685813518568,
|
|
"grad_norm": 18.779798535200392,
|
|
"learning_rate": 5.481860981704667e-10,
|
|
"logits/chosen": -1.1538286209106445,
|
|
"logits/rejected": -1.1552972793579102,
|
|
"logps/chosen": -690.300537109375,
|
|
"logps/rejected": -903.190673828125,
|
|
"loss": 0.3621,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -4.287991523742676,
|
|
"rewards/margins": 1.8346328735351562,
|
|
"rewards/rejected": -6.122623920440674,
|
|
"step": 460
|
|
},
|
|
{
|
|
"epoch": 0.9853059043547956,
|
|
"grad_norm": 17.340195720499324,
|
|
"learning_rate": 4.027733834741642e-10,
|
|
"logits/chosen": -1.245958924293518,
|
|
"logits/rejected": -1.314367413520813,
|
|
"logps/chosen": -711.3773193359375,
|
|
"logps/rejected": -859.24462890625,
|
|
"loss": 0.3816,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -4.259920120239258,
|
|
"rewards/margins": 1.5348095893859863,
|
|
"rewards/rejected": -5.794729709625244,
|
|
"step": 461
|
|
},
|
|
{
|
|
"epoch": 0.9874432273577345,
|
|
"grad_norm": 20.215944038876593,
|
|
"learning_rate": 2.7971808464055667e-10,
|
|
"logits/chosen": -1.160940408706665,
|
|
"logits/rejected": -1.1195350885391235,
|
|
"logps/chosen": -534.329833984375,
|
|
"logps/rejected": -804.3992919921875,
|
|
"loss": 0.3931,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -3.372945785522461,
|
|
"rewards/margins": 2.39996337890625,
|
|
"rewards/rejected": -5.772909641265869,
|
|
"step": 462
|
|
},
|
|
{
|
|
"epoch": 0.9895805503606733,
|
|
"grad_norm": 44.280226028243526,
|
|
"learning_rate": 1.7902708659867095e-10,
|
|
"logits/chosen": -1.2246938943862915,
|
|
"logits/rejected": -1.2264604568481445,
|
|
"logps/chosen": -670.49169921875,
|
|
"logps/rejected": -792.8765869140625,
|
|
"loss": 0.3134,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": -4.106105327606201,
|
|
"rewards/margins": 1.2818169593811035,
|
|
"rewards/rejected": -5.3879218101501465,
|
|
"step": 463
|
|
},
|
|
{
|
|
"epoch": 0.9917178733636121,
|
|
"grad_norm": 21.939913072475736,
|
|
"learning_rate": 1.0070602299738773e-10,
|
|
"logits/chosen": -1.3647425174713135,
|
|
"logits/rejected": -1.348206639289856,
|
|
"logps/chosen": -645.3593139648438,
|
|
"logps/rejected": -828.7014770507812,
|
|
"loss": 0.3997,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -4.110527038574219,
|
|
"rewards/margins": 1.7152307033538818,
|
|
"rewards/rejected": -5.8257575035095215,
|
|
"step": 464
|
|
},
|
|
{
|
|
"epoch": 0.9938551963665508,
|
|
"grad_norm": 20.19763187664986,
|
|
"learning_rate": 4.475927589062678e-11,
|
|
"logits/chosen": -1.1594457626342773,
|
|
"logits/rejected": -1.1328028440475464,
|
|
"logps/chosen": -565.1810913085938,
|
|
"logps/rejected": -730.758056640625,
|
|
"loss": 0.3701,
|
|
"rewards/accuracies": 0.875,
|
|
"rewards/chosen": -3.548121213912964,
|
|
"rewards/margins": 1.580583095550537,
|
|
"rewards/rejected": -5.128703594207764,
|
|
"step": 465
|
|
},
|
|
{
|
|
"epoch": 0.9959925193694897,
|
|
"grad_norm": 19.096874791193336,
|
|
"learning_rate": 1.1189975492076343e-11,
|
|
"logits/chosen": -1.142935037612915,
|
|
"logits/rejected": -1.178402304649353,
|
|
"logps/chosen": -515.046875,
|
|
"logps/rejected": -609.339111328125,
|
|
"loss": 0.381,
|
|
"rewards/accuracies": 0.75,
|
|
"rewards/chosen": -3.373872756958008,
|
|
"rewards/margins": 0.8627163767814636,
|
|
"rewards/rejected": -4.236588954925537,
|
|
"step": 466
|
|
},
|
|
{
|
|
"epoch": 0.9981298423724285,
|
|
"grad_norm": 15.024893493481045,
|
|
"learning_rate": 0.0,
|
|
"logits/chosen": -1.2568109035491943,
|
|
"logits/rejected": -1.233332872390747,
|
|
"logps/chosen": -832.8156127929688,
|
|
"logps/rejected": -1123.501708984375,
|
|
"loss": 0.3511,
|
|
"rewards/accuracies": 0.8125,
|
|
"rewards/chosen": -5.070691108703613,
|
|
"rewards/margins": 3.0254790782928467,
|
|
"rewards/rejected": -8.096170425415039,
|
|
"step": 467
|
|
},
|
|
{
|
|
"epoch": 0.9981298423724285,
|
|
"step": 467,
|
|
"total_flos": 0.0,
|
|
"train_loss": 0.47420934423389477,
|
|
"train_runtime": 9989.1556,
|
|
"train_samples_per_second": 5.994,
|
|
"train_steps_per_second": 0.047
|
|
}
|
|
],
|
|
"logging_steps": 1,
|
|
"max_steps": 467,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 1,
|
|
"save_steps": 100,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": true
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 0.0,
|
|
"train_batch_size": 2,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|