phi-2-gpo-v7-i1 / trainer_state.json
lole25's picture
Model save
9c0f043 verified
raw
history blame
32.3 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9996190476190476,
"eval_steps": 500,
"global_step": 656,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 7.575757575757576e-08,
"logits/chosen": 0.04974596947431564,
"logits/rejected": 0.2963363230228424,
"logps/chosen": -446.36370849609375,
"logps/rejected": -275.23162841796875,
"loss": 0.3488,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.02,
"learning_rate": 7.575757575757576e-07,
"logits/chosen": 0.07917162775993347,
"logits/rejected": 0.2545989155769348,
"logps/chosen": -351.4217529296875,
"logps/rejected": -305.8712463378906,
"loss": 0.3403,
"rewards/accuracies": 0.3472222089767456,
"rewards/chosen": -1.3770493296760833e-06,
"rewards/margins": -5.5950724345166236e-05,
"rewards/rejected": 5.4573672969127074e-05,
"step": 10
},
{
"epoch": 0.03,
"learning_rate": 1.5151515151515152e-06,
"logits/chosen": 0.08743862062692642,
"logits/rejected": 0.2663661539554596,
"logps/chosen": -356.5379333496094,
"logps/rejected": -272.34771728515625,
"loss": 0.3613,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": 8.02798240329139e-05,
"rewards/margins": 0.0002525355666875839,
"rewards/rejected": -0.00017225573537871242,
"step": 20
},
{
"epoch": 0.05,
"learning_rate": 2.2727272727272728e-06,
"logits/chosen": 0.10032109916210175,
"logits/rejected": 0.2204272300004959,
"logps/chosen": -320.36749267578125,
"logps/rejected": -265.65338134765625,
"loss": 0.3542,
"rewards/accuracies": 0.44999998807907104,
"rewards/chosen": 0.00015481823356822133,
"rewards/margins": 7.609631575178355e-05,
"rewards/rejected": 7.872191054048017e-05,
"step": 30
},
{
"epoch": 0.06,
"learning_rate": 3.0303030303030305e-06,
"logits/chosen": 0.1070769876241684,
"logits/rejected": 0.2916509211063385,
"logps/chosen": -365.18646240234375,
"logps/rejected": -253.8933868408203,
"loss": 0.3552,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": 0.0006457852432504296,
"rewards/margins": 0.0006201790529303253,
"rewards/rejected": 2.560619941505138e-05,
"step": 40
},
{
"epoch": 0.08,
"learning_rate": 3.7878787878787882e-06,
"logits/chosen": 0.08771614730358124,
"logits/rejected": 0.28780585527420044,
"logps/chosen": -370.36212158203125,
"logps/rejected": -294.9586181640625,
"loss": 0.3338,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.0016288382466882467,
"rewards/margins": 0.0014908717712387443,
"rewards/rejected": 0.000137966446345672,
"step": 50
},
{
"epoch": 0.09,
"learning_rate": 4.5454545454545455e-06,
"logits/chosen": 0.06850675493478775,
"logits/rejected": 0.276347815990448,
"logps/chosen": -375.48883056640625,
"logps/rejected": -298.6650695800781,
"loss": 0.3422,
"rewards/accuracies": 0.7562500238418579,
"rewards/chosen": 0.004066063556820154,
"rewards/margins": 0.004035579971969128,
"rewards/rejected": 3.0483581213047728e-05,
"step": 60
},
{
"epoch": 0.11,
"learning_rate": 4.999432965739786e-06,
"logits/chosen": 0.1320158988237381,
"logits/rejected": 0.3291288912296295,
"logps/chosen": -330.6249084472656,
"logps/rejected": -303.25128173828125,
"loss": 0.3197,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": 0.007968532852828503,
"rewards/margins": 0.007708103861659765,
"rewards/rejected": 0.0002604298642836511,
"step": 70
},
{
"epoch": 0.12,
"learning_rate": 4.9930567839810125e-06,
"logits/chosen": 0.09887860715389252,
"logits/rejected": 0.31491416692733765,
"logps/chosen": -353.16217041015625,
"logps/rejected": -297.11651611328125,
"loss": 0.3248,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 0.01291077770292759,
"rewards/margins": 0.018067900091409683,
"rewards/rejected": -0.005157124251127243,
"step": 80
},
{
"epoch": 0.14,
"learning_rate": 4.979613761906212e-06,
"logits/chosen": 0.17029765248298645,
"logits/rejected": 0.260185182094574,
"logps/chosen": -352.32379150390625,
"logps/rejected": -288.6719665527344,
"loss": 0.2939,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": 0.007807808928191662,
"rewards/margins": 0.023743372410535812,
"rewards/rejected": -0.015935566276311874,
"step": 90
},
{
"epoch": 0.15,
"learning_rate": 4.959142005221991e-06,
"logits/chosen": 0.1353963315486908,
"logits/rejected": 0.2721942365169525,
"logps/chosen": -363.1653137207031,
"logps/rejected": -354.89312744140625,
"loss": 0.2981,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.01541908085346222,
"rewards/margins": 0.04708936810493469,
"rewards/rejected": -0.06250844895839691,
"step": 100
},
{
"epoch": 0.17,
"learning_rate": 4.931699543346854e-06,
"logits/chosen": 0.20096346735954285,
"logits/rejected": 0.3208036422729492,
"logps/chosen": -412.08306884765625,
"logps/rejected": -410.16162109375,
"loss": 0.2759,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.06705033034086227,
"rewards/margins": 0.07482309639453888,
"rewards/rejected": -0.14187341928482056,
"step": 110
},
{
"epoch": 0.18,
"learning_rate": 4.897364164920515e-06,
"logits/chosen": 0.15477022528648376,
"logits/rejected": 0.24951288104057312,
"logps/chosen": -500.6244201660156,
"logps/rejected": -552.2005004882812,
"loss": 0.284,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.15765079855918884,
"rewards/margins": 0.10079488903284073,
"rewards/rejected": -0.258445680141449,
"step": 120
},
{
"epoch": 0.2,
"learning_rate": 4.8562331973035396e-06,
"logits/chosen": 0.15251179039478302,
"logits/rejected": 0.2451750487089157,
"logps/chosen": -500.58416748046875,
"logps/rejected": -580.9321899414062,
"loss": 0.2784,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.15634217858314514,
"rewards/margins": 0.12097392231225967,
"rewards/rejected": -0.27731606364250183,
"step": 130
},
{
"epoch": 0.21,
"learning_rate": 4.808423230692374e-06,
"logits/chosen": 0.14251364767551422,
"logits/rejected": 0.3138013184070587,
"logps/chosen": -508.4844665527344,
"logps/rejected": -568.8987426757812,
"loss": 0.2638,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.15639568865299225,
"rewards/margins": 0.1208159551024437,
"rewards/rejected": -0.27721160650253296,
"step": 140
},
{
"epoch": 0.23,
"learning_rate": 4.754069787631761e-06,
"logits/chosen": 0.2015986144542694,
"logits/rejected": 0.28302785754203796,
"logps/chosen": -529.3695068359375,
"logps/rejected": -604.30859375,
"loss": 0.251,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.18331554532051086,
"rewards/margins": 0.1465958058834076,
"rewards/rejected": -0.32991132140159607,
"step": 150
},
{
"epoch": 0.24,
"learning_rate": 4.693326938861367e-06,
"logits/chosen": 0.19963108003139496,
"logits/rejected": 0.2948302626609802,
"logps/chosen": -566.9232788085938,
"logps/rejected": -621.8565673828125,
"loss": 0.2514,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.2004656344652176,
"rewards/margins": 0.12107620388269424,
"rewards/rejected": -0.32154184579849243,
"step": 160
},
{
"epoch": 0.26,
"learning_rate": 4.626366866585528e-06,
"logits/chosen": 0.20774176716804504,
"logits/rejected": 0.31210097670555115,
"logps/chosen": -525.9526977539062,
"logps/rejected": -598.7525634765625,
"loss": 0.2523,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.2060777246952057,
"rewards/margins": 0.1383783519268036,
"rewards/rejected": -0.3444560766220093,
"step": 170
},
{
"epoch": 0.27,
"learning_rate": 4.553379376404085e-06,
"logits/chosen": 0.1882271021604538,
"logits/rejected": 0.2884698808193207,
"logps/chosen": -622.5642700195312,
"logps/rejected": -703.97998046875,
"loss": 0.2686,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.2503646910190582,
"rewards/margins": 0.14028559625148773,
"rewards/rejected": -0.39065021276474,
"step": 180
},
{
"epoch": 0.29,
"learning_rate": 4.474571359287791e-06,
"logits/chosen": 0.1948501169681549,
"logits/rejected": 0.3517269492149353,
"logps/chosen": -605.8263549804688,
"logps/rejected": -685.9619750976562,
"loss": 0.2324,
"rewards/accuracies": 0.7562500238418579,
"rewards/chosen": -0.21189594268798828,
"rewards/margins": 0.19469039142131805,
"rewards/rejected": -0.40658634901046753,
"step": 190
},
{
"epoch": 0.3,
"learning_rate": 4.3901662051233755e-06,
"logits/chosen": 0.1501813530921936,
"logits/rejected": 0.29967957735061646,
"logps/chosen": -588.287841796875,
"logps/rejected": -692.4317626953125,
"loss": 0.232,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.21489660441875458,
"rewards/margins": 0.20667514204978943,
"rewards/rejected": -0.4215717315673828,
"step": 200
},
{
"epoch": 0.32,
"learning_rate": 4.30040316949064e-06,
"logits/chosen": 0.1845458298921585,
"logits/rejected": 0.25728127360343933,
"logps/chosen": -593.4507446289062,
"logps/rejected": -647.9249267578125,
"loss": 0.268,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.2259581983089447,
"rewards/margins": 0.1573750227689743,
"rewards/rejected": -0.3833332657814026,
"step": 210
},
{
"epoch": 0.34,
"learning_rate": 4.205536695466524e-06,
"logits/chosen": 0.19619445502758026,
"logits/rejected": 0.289485901594162,
"logps/chosen": -596.2908325195312,
"logps/rejected": -673.9340209960938,
"loss": 0.2279,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.20733025670051575,
"rewards/margins": 0.1769087016582489,
"rewards/rejected": -0.38423892855644226,
"step": 220
},
{
"epoch": 0.35,
"learning_rate": 4.105835692378557e-06,
"logits/chosen": 0.18007412552833557,
"logits/rejected": 0.23454514145851135,
"logps/chosen": -605.221923828125,
"logps/rejected": -682.0065307617188,
"loss": 0.2477,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.2464592009782791,
"rewards/margins": 0.1562257707118988,
"rewards/rejected": -0.4026849865913391,
"step": 230
},
{
"epoch": 0.37,
"learning_rate": 4.001582773552153e-06,
"logits/chosen": 0.17503593862056732,
"logits/rejected": 0.26662883162498474,
"logps/chosen": -545.9107666015625,
"logps/rejected": -640.1486206054688,
"loss": 0.2225,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.23691585659980774,
"rewards/margins": 0.16106419265270233,
"rewards/rejected": -0.39798006415367126,
"step": 240
},
{
"epoch": 0.38,
"learning_rate": 3.893073455212438e-06,
"logits/chosen": 0.200395867228508,
"logits/rejected": 0.3584834933280945,
"logps/chosen": -608.1827392578125,
"logps/rejected": -740.0325317382812,
"loss": 0.2478,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.26550978422164917,
"rewards/margins": 0.1945229470729828,
"rewards/rejected": -0.46003276109695435,
"step": 250
},
{
"epoch": 0.4,
"learning_rate": 3.7806153188114027e-06,
"logits/chosen": 0.1786949336528778,
"logits/rejected": 0.2547721266746521,
"logps/chosen": -599.2467041015625,
"logps/rejected": -727.934326171875,
"loss": 0.2601,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.22916042804718018,
"rewards/margins": 0.17522968351840973,
"rewards/rejected": -0.40439003705978394,
"step": 260
},
{
"epoch": 0.41,
"learning_rate": 3.6645271391548542e-06,
"logits/chosen": 0.18216630816459656,
"logits/rejected": 0.24754111468791962,
"logps/chosen": -618.3128051757812,
"logps/rejected": -640.1773681640625,
"loss": 0.2761,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.22649279236793518,
"rewards/margins": 0.14542898535728455,
"rewards/rejected": -0.37192174792289734,
"step": 270
},
{
"epoch": 0.43,
"learning_rate": 3.5451379808006014e-06,
"logits/chosen": 0.19754137098789215,
"logits/rejected": 0.27186593413352966,
"logps/chosen": -573.2451171875,
"logps/rejected": -707.2796020507812,
"loss": 0.2209,
"rewards/accuracies": 0.7562500238418579,
"rewards/chosen": -0.2302563190460205,
"rewards/margins": 0.19838322699069977,
"rewards/rejected": -0.42863956093788147,
"step": 280
},
{
"epoch": 0.44,
"learning_rate": 3.4227862652892106e-06,
"logits/chosen": 0.16008315980434418,
"logits/rejected": 0.27827057242393494,
"logps/chosen": -662.4371948242188,
"logps/rejected": -759.597900390625,
"loss": 0.2408,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.2687516212463379,
"rewards/margins": 0.17680145800113678,
"rewards/rejected": -0.4455530643463135,
"step": 290
},
{
"epoch": 0.46,
"learning_rate": 3.2978188118513814e-06,
"logits/chosen": 0.17593641579151154,
"logits/rejected": 0.2789141535758972,
"logps/chosen": -642.6029052734375,
"logps/rejected": -753.4479370117188,
"loss": 0.2304,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.27796030044555664,
"rewards/margins": 0.1859740912914276,
"rewards/rejected": -0.46393436193466187,
"step": 300
},
{
"epoch": 0.47,
"learning_rate": 3.1705898543111576e-06,
"logits/chosen": 0.13969933986663818,
"logits/rejected": 0.3079048693180084,
"logps/chosen": -630.0859375,
"logps/rejected": -765.4630737304688,
"loss": 0.2384,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.2862771153450012,
"rewards/margins": 0.19564750790596008,
"rewards/rejected": -0.4819245934486389,
"step": 310
},
{
"epoch": 0.49,
"learning_rate": 3.041460036971664e-06,
"logits/chosen": 0.14904941618442535,
"logits/rejected": 0.27918586134910583,
"logps/chosen": -619.8759765625,
"logps/rejected": -759.2853393554688,
"loss": 0.2347,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.27361705899238586,
"rewards/margins": 0.19538012146949768,
"rewards/rejected": -0.4689972400665283,
"step": 320
},
{
"epoch": 0.5,
"learning_rate": 2.910795392329649e-06,
"logits/chosen": 0.1377524435520172,
"logits/rejected": 0.2504701614379883,
"logps/chosen": -647.4981689453125,
"logps/rejected": -789.5301513671875,
"loss": 0.2352,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": -0.2706158757209778,
"rewards/margins": 0.18975581228733063,
"rewards/rejected": -0.460371732711792,
"step": 330
},
{
"epoch": 0.52,
"learning_rate": 2.7789663035166035e-06,
"logits/chosen": 0.18703623116016388,
"logits/rejected": 0.3294333815574646,
"logps/chosen": -656.440673828125,
"logps/rejected": -793.1935424804688,
"loss": 0.2072,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.2833554744720459,
"rewards/margins": 0.21432232856750488,
"rewards/rejected": -0.4976778030395508,
"step": 340
},
{
"epoch": 0.53,
"learning_rate": 2.6463464544075344e-06,
"logits/chosen": 0.15668293833732605,
"logits/rejected": 0.2470695674419403,
"logps/chosen": -603.0676879882812,
"logps/rejected": -736.9429931640625,
"loss": 0.227,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.27748292684555054,
"rewards/margins": 0.1968688815832138,
"rewards/rejected": -0.4743518829345703,
"step": 350
},
{
"epoch": 0.55,
"learning_rate": 2.513311770373421e-06,
"logits/chosen": 0.09050649404525757,
"logits/rejected": 0.27548736333847046,
"logps/chosen": -645.054443359375,
"logps/rejected": -780.7376708984375,
"loss": 0.2277,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.27279263734817505,
"rewards/margins": 0.2085564136505127,
"rewards/rejected": -0.48134905099868774,
"step": 360
},
{
"epoch": 0.56,
"learning_rate": 2.380239352679908e-06,
"logits/chosen": 0.11263048648834229,
"logits/rejected": 0.23795035481452942,
"logps/chosen": -635.0528564453125,
"logps/rejected": -752.4202270507812,
"loss": 0.2295,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": -0.28419169783592224,
"rewards/margins": 0.18686431646347046,
"rewards/rejected": -0.4710559844970703,
"step": 370
},
{
"epoch": 0.58,
"learning_rate": 2.247506409552795e-06,
"logits/chosen": 0.18179914355278015,
"logits/rejected": 0.2870942950248718,
"logps/chosen": -632.6707153320312,
"logps/rejected": -745.6337890625,
"loss": 0.21,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.2822064459323883,
"rewards/margins": 0.18581680953502655,
"rewards/rejected": -0.4680232107639313,
"step": 380
},
{
"epoch": 0.59,
"learning_rate": 2.1154891869403436e-06,
"logits/chosen": 0.1858624517917633,
"logits/rejected": 0.24371235072612762,
"logps/chosen": -612.2597045898438,
"logps/rejected": -773.7772216796875,
"loss": 0.24,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.2878335416316986,
"rewards/margins": 0.1961011290550232,
"rewards/rejected": -0.4839346408843994,
"step": 390
},
{
"epoch": 0.61,
"learning_rate": 1.9845619020032552e-06,
"logits/chosen": 0.11868009716272354,
"logits/rejected": 0.2442179173231125,
"logps/chosen": -628.7079467773438,
"logps/rejected": -782.9298095703125,
"loss": 0.2128,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -0.2861320376396179,
"rewards/margins": 0.2013685703277588,
"rewards/rejected": -0.4875006079673767,
"step": 400
},
{
"epoch": 0.62,
"learning_rate": 1.8550956823554708e-06,
"logits/chosen": 0.15941022336483002,
"logits/rejected": 0.26665717363357544,
"logps/chosen": -611.2957763671875,
"logps/rejected": -745.6875610351562,
"loss": 0.2419,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.26627546548843384,
"rewards/margins": 0.20409245789051056,
"rewards/rejected": -0.4703678488731384,
"step": 410
},
{
"epoch": 0.64,
"learning_rate": 1.7274575140626318e-06,
"logits/chosen": 0.12057618796825409,
"logits/rejected": 0.2439500391483307,
"logps/chosen": -563.8010864257812,
"logps/rejected": -753.3555908203125,
"loss": 0.2241,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.22670452296733856,
"rewards/margins": 0.22902043163776398,
"rewards/rejected": -0.45572495460510254,
"step": 420
},
{
"epoch": 0.66,
"learning_rate": 1.6020092013802002e-06,
"logits/chosen": 0.1292915642261505,
"logits/rejected": 0.27418118715286255,
"logps/chosen": -623.4757080078125,
"logps/rejected": -712.2957153320312,
"loss": 0.2196,
"rewards/accuracies": 0.7562500238418579,
"rewards/chosen": -0.24320180714130402,
"rewards/margins": 0.19605949521064758,
"rewards/rejected": -0.4392613470554352,
"step": 430
},
{
"epoch": 0.67,
"learning_rate": 1.4791063411799938e-06,
"logits/chosen": 0.10819476842880249,
"logits/rejected": 0.2511535882949829,
"logps/chosen": -570.8446044921875,
"logps/rejected": -685.5501708984375,
"loss": 0.2124,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.22663471102714539,
"rewards/margins": 0.20201578736305237,
"rewards/rejected": -0.42865046858787537,
"step": 440
},
{
"epoch": 0.69,
"learning_rate": 1.3590973149722103e-06,
"logits/chosen": 0.15888772904872894,
"logits/rejected": 0.2368732988834381,
"logps/chosen": -589.4254150390625,
"logps/rejected": -735.7716674804688,
"loss": 0.2038,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": -0.26341313123703003,
"rewards/margins": 0.1875002086162567,
"rewards/rejected": -0.45091336965560913,
"step": 450
},
{
"epoch": 0.7,
"learning_rate": 1.2423223013801946e-06,
"logits/chosen": 0.14813843369483948,
"logits/rejected": 0.21993982791900635,
"logps/chosen": -589.8951416015625,
"logps/rejected": -729.8585205078125,
"loss": 0.2504,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.2774823307991028,
"rewards/margins": 0.1887308657169342,
"rewards/rejected": -0.4662131667137146,
"step": 460
},
{
"epoch": 0.72,
"learning_rate": 1.1291123118671665e-06,
"logits/chosen": 0.09890522062778473,
"logits/rejected": 0.2421140968799591,
"logps/chosen": -639.2291870117188,
"logps/rejected": -729.2130126953125,
"loss": 0.1871,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.2574080526828766,
"rewards/margins": 0.20436222851276398,
"rewards/rejected": -0.46177029609680176,
"step": 470
},
{
"epoch": 0.73,
"learning_rate": 1.019788252448267e-06,
"logits/chosen": 0.1626007854938507,
"logits/rejected": 0.2534940242767334,
"logps/chosen": -614.7600708007812,
"logps/rejected": -789.6011352539062,
"loss": 0.202,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.27494579553604126,
"rewards/margins": 0.21227261424064636,
"rewards/rejected": -0.48721843957901,
"step": 480
},
{
"epoch": 0.75,
"learning_rate": 9.146600140475945e-07,
"logits/chosen": 0.18194663524627686,
"logits/rejected": 0.2571627199649811,
"logps/chosen": -695.4674682617188,
"logps/rejected": -824.3109130859375,
"loss": 0.2242,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": -0.31008249521255493,
"rewards/margins": 0.20844268798828125,
"rewards/rejected": -0.518525242805481,
"step": 490
},
{
"epoch": 0.76,
"learning_rate": 8.140255940787059e-07,
"logits/chosen": 0.1600276529788971,
"logits/rejected": 0.24592892825603485,
"logps/chosen": -621.3486328125,
"logps/rejected": -787.8779296875,
"loss": 0.2379,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.2887078523635864,
"rewards/margins": 0.21262690424919128,
"rewards/rejected": -0.5013347864151001,
"step": 500
},
{
"epoch": 0.78,
"learning_rate": 7.181702517385789e-07,
"logits/chosen": 0.11802725493907928,
"logits/rejected": 0.24163658916950226,
"logps/chosen": -626.5403442382812,
"logps/rejected": -774.00830078125,
"loss": 0.219,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.27267104387283325,
"rewards/margins": 0.21761374175548553,
"rewards/rejected": -0.4902847409248352,
"step": 510
},
{
"epoch": 0.79,
"learning_rate": 6.273656994094232e-07,
"logits/chosen": 0.1450013965368271,
"logits/rejected": 0.23478934168815613,
"logps/chosen": -608.7362060546875,
"logps/rejected": -739.3218994140625,
"loss": 0.2294,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.2396283596754074,
"rewards/margins": 0.21917715668678284,
"rewards/rejected": -0.45880550146102905,
"step": 520
},
{
"epoch": 0.81,
"learning_rate": 5.418693324604082e-07,
"logits/chosen": 0.16014720499515533,
"logits/rejected": 0.25679388642311096,
"logps/chosen": -598.508056640625,
"logps/rejected": -752.1389770507812,
"loss": 0.2315,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.26550406217575073,
"rewards/margins": 0.20078308880329132,
"rewards/rejected": -0.46628713607788086,
"step": 530
},
{
"epoch": 0.82,
"learning_rate": 4.619234996325314e-07,
"logits/chosen": 0.10893194377422333,
"logits/rejected": 0.18960845470428467,
"logps/chosen": -589.2452392578125,
"logps/rejected": -712.2008666992188,
"loss": 0.2139,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": -0.2257637232542038,
"rewards/margins": 0.20952418446540833,
"rewards/rejected": -0.4352878928184509,
"step": 540
},
{
"epoch": 0.84,
"learning_rate": 3.877548160747768e-07,
"logits/chosen": 0.1536407321691513,
"logits/rejected": 0.25428491830825806,
"logps/chosen": -667.2605590820312,
"logps/rejected": -712.8155517578125,
"loss": 0.2543,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.26264914870262146,
"rewards/margins": 0.1710439920425415,
"rewards/rejected": -0.4336931109428406,
"step": 550
},
{
"epoch": 0.85,
"learning_rate": 3.195735209788528e-07,
"logits/chosen": 0.1331530511379242,
"logits/rejected": 0.25019171833992004,
"logps/chosen": -614.9050903320312,
"logps/rejected": -703.8809814453125,
"loss": 0.2492,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.27054914832115173,
"rewards/margins": 0.14634455740451813,
"rewards/rejected": -0.4168936610221863,
"step": 560
},
{
"epoch": 0.87,
"learning_rate": 2.5757288163336806e-07,
"logits/chosen": 0.15042927861213684,
"logits/rejected": 0.2752520442008972,
"logps/chosen": -605.6216430664062,
"logps/rejected": -749.6766357421875,
"loss": 0.2349,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": -0.2648725211620331,
"rewards/margins": 0.18051207065582275,
"rewards/rejected": -0.44538459181785583,
"step": 570
},
{
"epoch": 0.88,
"learning_rate": 2.019286455866981e-07,
"logits/chosen": 0.09744317829608917,
"logits/rejected": 0.30391809344291687,
"logps/chosen": -656.5165405273438,
"logps/rejected": -782.3225708007812,
"loss": 0.2347,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.26627975702285767,
"rewards/margins": 0.2162231206893921,
"rewards/rejected": -0.48250293731689453,
"step": 580
},
{
"epoch": 0.9,
"learning_rate": 1.5279854247146703e-07,
"logits/chosen": 0.10827809572219849,
"logits/rejected": 0.23128509521484375,
"logps/chosen": -597.4005126953125,
"logps/rejected": -727.5601196289062,
"loss": 0.2356,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.2508590817451477,
"rewards/margins": 0.19355928897857666,
"rewards/rejected": -0.44441837072372437,
"step": 590
},
{
"epoch": 0.91,
"learning_rate": 1.1032183690276754e-07,
"logits/chosen": 0.1284581571817398,
"logits/rejected": 0.22490856051445007,
"logps/chosen": -608.4317016601562,
"logps/rejected": -756.5374755859375,
"loss": 0.2282,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.27711886167526245,
"rewards/margins": 0.19119976460933685,
"rewards/rejected": -0.4683186411857605,
"step": 600
},
{
"epoch": 0.93,
"learning_rate": 7.46189337174788e-08,
"logits/chosen": 0.1757223904132843,
"logits/rejected": 0.26904186606407166,
"logps/chosen": -603.6378173828125,
"logps/rejected": -660.9766235351562,
"loss": 0.2443,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.24105675518512726,
"rewards/margins": 0.1722828447818756,
"rewards/rejected": -0.4133395552635193,
"step": 610
},
{
"epoch": 0.94,
"learning_rate": 4.579103667367385e-08,
"logits/chosen": 0.10849970579147339,
"logits/rejected": 0.19708193838596344,
"logps/chosen": -565.9713134765625,
"logps/rejected": -679.5427856445312,
"loss": 0.2211,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.2512280344963074,
"rewards/margins": 0.17560149729251862,
"rewards/rejected": -0.4268294870853424,
"step": 620
},
{
"epoch": 0.96,
"learning_rate": 2.3919861577572924e-08,
"logits/chosen": 0.13952800631523132,
"logits/rejected": 0.20136961340904236,
"logps/chosen": -621.4708862304688,
"logps/rejected": -748.71240234375,
"loss": 0.2234,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": -0.2641890048980713,
"rewards/margins": 0.19575798511505127,
"rewards/rejected": -0.45994701981544495,
"step": 630
},
{
"epoch": 0.98,
"learning_rate": 9.067404651211808e-09,
"logits/chosen": 0.18144121766090393,
"logits/rejected": 0.23935365676879883,
"logps/chosen": -601.0186767578125,
"logps/rejected": -719.6766357421875,
"loss": 0.2458,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": -0.26298612356185913,
"rewards/margins": 0.16590112447738647,
"rewards/rejected": -0.428887277841568,
"step": 640
},
{
"epoch": 0.99,
"learning_rate": 1.2757667974155896e-09,
"logits/chosen": 0.13537321984767914,
"logits/rejected": 0.2640915811061859,
"logps/chosen": -576.1346435546875,
"logps/rejected": -705.7242431640625,
"loss": 0.2243,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.22819241881370544,
"rewards/margins": 0.1957322061061859,
"rewards/rejected": -0.42392459511756897,
"step": 650
},
{
"epoch": 1.0,
"step": 656,
"total_flos": 0.0,
"train_loss": 0.19985946376876132,
"train_runtime": 6634.0696,
"train_samples_per_second": 3.165,
"train_steps_per_second": 0.099
}
],
"logging_steps": 10,
"max_steps": 656,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 100,
"total_flos": 0.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}