2200 lines
86 KiB
JSON
2200 lines
86 KiB
JSON
|
{
|
||
|
|
"best_global_step": null,
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 0.535833891493637,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 1000,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.0005358338914936369,
|
||
|
|
"grad_norm": 29.978843688964844,
|
||
|
|
"learning_rate": 0.0,
|
||
|
|
"logits/chosen": -2.52783203125,
|
||
|
|
"logits/rejected": -2.4970703125,
|
||
|
|
"logps/chosen": -277.75,
|
||
|
|
"logps/rejected": -223.0625,
|
||
|
|
"logps/weighted_chosen": -0.4293212890625,
|
||
|
|
"logps/weighted_rejected": -0.3968505859375,
|
||
|
|
"loss": 0.6914,
|
||
|
|
"rewards/accuracies": 0.0,
|
||
|
|
"rewards/chosen": 0.0,
|
||
|
|
"rewards/margins": 0.0,
|
||
|
|
"rewards/rejected": 0.0,
|
||
|
|
"rewards/weighted_accuracies": 0.0,
|
||
|
|
"rewards/weighted_chosen": 0.0,
|
||
|
|
"rewards/weighted_margins": 0.0,
|
||
|
|
"rewards/weighted_rejected": 0.0,
|
||
|
|
"step": 1
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0053583389149363695,
|
||
|
|
"grad_norm": 28.513858795166016,
|
||
|
|
"learning_rate": 4.81283422459893e-08,
|
||
|
|
"logits/chosen": -2.42626953125,
|
||
|
|
"logits/rejected": -2.397569417953491,
|
||
|
|
"logps/chosen": -195.53038024902344,
|
||
|
|
"logps/rejected": -195.91839599609375,
|
||
|
|
"logps/weighted_chosen": -0.4505276083946228,
|
||
|
|
"logps/weighted_rejected": -0.4737955629825592,
|
||
|
|
"loss": 0.6926,
|
||
|
|
"rewards/accuracies": 0.2743055522441864,
|
||
|
|
"rewards/chosen": 0.0321180559694767,
|
||
|
|
"rewards/margins": -0.0375434011220932,
|
||
|
|
"rewards/rejected": 0.0696614608168602,
|
||
|
|
"rewards/weighted_accuracies": 0.3402777910232544,
|
||
|
|
"rewards/weighted_chosen": 5.679660534951836e-05,
|
||
|
|
"rewards/weighted_margins": 3.984239447163418e-05,
|
||
|
|
"rewards/weighted_rejected": 1.695421087788418e-05,
|
||
|
|
"step": 10
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.010716677829872739,
|
||
|
|
"grad_norm": 20.803266525268555,
|
||
|
|
"learning_rate": 1.0160427807486631e-07,
|
||
|
|
"logits/chosen": -2.4007811546325684,
|
||
|
|
"logits/rejected": -2.3998780250549316,
|
||
|
|
"logps/chosen": -199.92343139648438,
|
||
|
|
"logps/rejected": -202.2101593017578,
|
||
|
|
"logps/weighted_chosen": -0.4370788633823395,
|
||
|
|
"logps/weighted_rejected": -0.464111328125,
|
||
|
|
"loss": 0.6927,
|
||
|
|
"rewards/accuracies": 0.34687501192092896,
|
||
|
|
"rewards/chosen": -0.02207031287252903,
|
||
|
|
"rewards/margins": 0.04130859300494194,
|
||
|
|
"rewards/rejected": -0.06337890774011612,
|
||
|
|
"rewards/weighted_accuracies": 0.36250001192092896,
|
||
|
|
"rewards/weighted_chosen": 0.0006473541143350303,
|
||
|
|
"rewards/weighted_margins": 7.362365431617945e-05,
|
||
|
|
"rewards/weighted_rejected": 0.0005735397571697831,
|
||
|
|
"step": 20
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.016075016744809108,
|
||
|
|
"grad_norm": 33.961883544921875,
|
||
|
|
"learning_rate": 1.5508021390374333e-07,
|
||
|
|
"logits/chosen": -2.32586669921875,
|
||
|
|
"logits/rejected": -2.300219774246216,
|
||
|
|
"logps/chosen": -198.45858764648438,
|
||
|
|
"logps/rejected": -212.0234375,
|
||
|
|
"logps/weighted_chosen": -0.45709228515625,
|
||
|
|
"logps/weighted_rejected": -0.45776671171188354,
|
||
|
|
"loss": 0.6925,
|
||
|
|
"rewards/accuracies": 0.38749998807907104,
|
||
|
|
"rewards/chosen": 0.1083984375,
|
||
|
|
"rewards/margins": 0.1162109375,
|
||
|
|
"rewards/rejected": -0.0078125,
|
||
|
|
"rewards/weighted_accuracies": 0.4312500059604645,
|
||
|
|
"rewards/weighted_chosen": 0.0002254486025776714,
|
||
|
|
"rewards/weighted_margins": 0.0002849578741006553,
|
||
|
|
"rewards/weighted_rejected": -5.970001075183973e-05,
|
||
|
|
"step": 30
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.021433355659745478,
|
||
|
|
"grad_norm": 44.94234848022461,
|
||
|
|
"learning_rate": 2.085561497326203e-07,
|
||
|
|
"logits/chosen": -2.3628907203674316,
|
||
|
|
"logits/rejected": -2.3779540061950684,
|
||
|
|
"logps/chosen": -192.63125610351562,
|
||
|
|
"logps/rejected": -207.24765014648438,
|
||
|
|
"logps/weighted_chosen": -0.45967406034469604,
|
||
|
|
"logps/weighted_rejected": -0.4842773377895355,
|
||
|
|
"loss": 0.6926,
|
||
|
|
"rewards/accuracies": 0.3687500059604645,
|
||
|
|
"rewards/chosen": -0.01308593712747097,
|
||
|
|
"rewards/margins": 0.04453124850988388,
|
||
|
|
"rewards/rejected": -0.0576171875,
|
||
|
|
"rewards/weighted_accuracies": 0.40937501192092896,
|
||
|
|
"rewards/weighted_chosen": 0.0007827758672647178,
|
||
|
|
"rewards/weighted_margins": 0.00011138916306663305,
|
||
|
|
"rewards/weighted_rejected": 0.00067138671875,
|
||
|
|
"step": 40
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.02679169457468185,
|
||
|
|
"grad_norm": 58.061580657958984,
|
||
|
|
"learning_rate": 2.620320855614973e-07,
|
||
|
|
"logits/chosen": -2.402636766433716,
|
||
|
|
"logits/rejected": -2.3846192359924316,
|
||
|
|
"logps/chosen": -213.23672485351562,
|
||
|
|
"logps/rejected": -228.9015655517578,
|
||
|
|
"logps/weighted_chosen": -0.4499572813510895,
|
||
|
|
"logps/weighted_rejected": -0.4646240174770355,
|
||
|
|
"loss": 0.6926,
|
||
|
|
"rewards/accuracies": 0.34687501192092896,
|
||
|
|
"rewards/chosen": -0.09589843451976776,
|
||
|
|
"rewards/margins": 0.0673828125,
|
||
|
|
"rewards/rejected": -0.16328124701976776,
|
||
|
|
"rewards/weighted_accuracies": 0.44999998807907104,
|
||
|
|
"rewards/weighted_chosen": 0.0027015686500817537,
|
||
|
|
"rewards/weighted_margins": 0.0005447387811727822,
|
||
|
|
"rewards/weighted_rejected": 0.0021568299271166325,
|
||
|
|
"step": 50
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.032150033489618215,
|
||
|
|
"grad_norm": 30.78350830078125,
|
||
|
|
"learning_rate": 3.155080213903743e-07,
|
||
|
|
"logits/chosen": -2.349609375,
|
||
|
|
"logits/rejected": -2.3248534202575684,
|
||
|
|
"logps/chosen": -211.83438110351562,
|
||
|
|
"logps/rejected": -218.11563110351562,
|
||
|
|
"logps/weighted_chosen": -0.44114989042282104,
|
||
|
|
"logps/weighted_rejected": -0.4686218202114105,
|
||
|
|
"loss": 0.6918,
|
||
|
|
"rewards/accuracies": 0.4124999940395355,
|
||
|
|
"rewards/chosen": -0.0025390624068677425,
|
||
|
|
"rewards/margins": 0.06503906100988388,
|
||
|
|
"rewards/rejected": -0.06757812201976776,
|
||
|
|
"rewards/weighted_accuracies": 0.48750001192092896,
|
||
|
|
"rewards/weighted_chosen": 0.005175781436264515,
|
||
|
|
"rewards/weighted_margins": 0.0019653320778161287,
|
||
|
|
"rewards/weighted_rejected": 0.0032104491256177425,
|
||
|
|
"step": 60
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03750837240455459,
|
||
|
|
"grad_norm": 23.173866271972656,
|
||
|
|
"learning_rate": 3.689839572192513e-07,
|
||
|
|
"logits/chosen": -2.458251953125,
|
||
|
|
"logits/rejected": -2.463671922683716,
|
||
|
|
"logps/chosen": -194.55624389648438,
|
||
|
|
"logps/rejected": -200.5437469482422,
|
||
|
|
"logps/weighted_chosen": -0.4099182188510895,
|
||
|
|
"logps/weighted_rejected": -0.4346679747104645,
|
||
|
|
"loss": 0.6923,
|
||
|
|
"rewards/accuracies": 0.4124999940395355,
|
||
|
|
"rewards/chosen": -0.16562500596046448,
|
||
|
|
"rewards/margins": 0.203125,
|
||
|
|
"rewards/rejected": -0.3687500059604645,
|
||
|
|
"rewards/weighted_accuracies": 0.503125011920929,
|
||
|
|
"rewards/weighted_chosen": 0.0095977783203125,
|
||
|
|
"rewards/weighted_margins": 0.0012573242420330644,
|
||
|
|
"rewards/weighted_rejected": 0.00834045372903347,
|
||
|
|
"step": 70
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.042866711319490956,
|
||
|
|
"grad_norm": 27.169227600097656,
|
||
|
|
"learning_rate": 4.2245989304812833e-07,
|
||
|
|
"logits/chosen": -2.44384765625,
|
||
|
|
"logits/rejected": -2.4704346656799316,
|
||
|
|
"logps/chosen": -219.31405639648438,
|
||
|
|
"logps/rejected": -240.16641235351562,
|
||
|
|
"logps/weighted_chosen": -0.4397827088832855,
|
||
|
|
"logps/weighted_rejected": -0.46443480253219604,
|
||
|
|
"loss": 0.6915,
|
||
|
|
"rewards/accuracies": 0.4375,
|
||
|
|
"rewards/chosen": 0.02402343787252903,
|
||
|
|
"rewards/margins": 0.2919921875,
|
||
|
|
"rewards/rejected": -0.2679687440395355,
|
||
|
|
"rewards/weighted_accuracies": 0.47187501192092896,
|
||
|
|
"rewards/weighted_chosen": 0.012739372439682484,
|
||
|
|
"rewards/weighted_margins": 0.003216362092643976,
|
||
|
|
"rewards/weighted_rejected": 0.00952300988137722,
|
||
|
|
"step": 80
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04822505023442733,
|
||
|
|
"grad_norm": 24.797061920166016,
|
||
|
|
"learning_rate": 4.7593582887700533e-07,
|
||
|
|
"logits/chosen": -2.46044921875,
|
||
|
|
"logits/rejected": -2.4736571311950684,
|
||
|
|
"logps/chosen": -195.90390014648438,
|
||
|
|
"logps/rejected": -208.1320343017578,
|
||
|
|
"logps/weighted_chosen": -0.42793577909469604,
|
||
|
|
"logps/weighted_rejected": -0.447784423828125,
|
||
|
|
"loss": 0.6922,
|
||
|
|
"rewards/accuracies": 0.48750001192092896,
|
||
|
|
"rewards/chosen": -0.599609375,
|
||
|
|
"rewards/margins": 0.3626953065395355,
|
||
|
|
"rewards/rejected": -0.9623047113418579,
|
||
|
|
"rewards/weighted_accuracies": 0.5062500238418579,
|
||
|
|
"rewards/weighted_chosen": 0.00842132605612278,
|
||
|
|
"rewards/weighted_margins": 0.0021730423904955387,
|
||
|
|
"rewards/weighted_rejected": 0.00624923687428236,
|
||
|
|
"step": 90
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0535833891493637,
|
||
|
|
"grad_norm": 13.923910140991211,
|
||
|
|
"learning_rate": 5.294117647058823e-07,
|
||
|
|
"logits/chosen": -2.4405760765075684,
|
||
|
|
"logits/rejected": -2.425537109375,
|
||
|
|
"logps/chosen": -224.88516235351562,
|
||
|
|
"logps/rejected": -229.6984405517578,
|
||
|
|
"logps/weighted_chosen": -0.4417480528354645,
|
||
|
|
"logps/weighted_rejected": -0.454376220703125,
|
||
|
|
"loss": 0.6949,
|
||
|
|
"rewards/accuracies": 0.4625000059604645,
|
||
|
|
"rewards/chosen": -1.443750023841858,
|
||
|
|
"rewards/margins": 0.4691406190395355,
|
||
|
|
"rewards/rejected": -1.9128906726837158,
|
||
|
|
"rewards/weighted_accuracies": 0.518750011920929,
|
||
|
|
"rewards/weighted_chosen": 0.001766204833984375,
|
||
|
|
"rewards/weighted_margins": -0.0024765015114098787,
|
||
|
|
"rewards/weighted_rejected": 0.0042396546341478825,
|
||
|
|
"step": 100
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.058941728064300064,
|
||
|
|
"grad_norm": 43.69069290161133,
|
||
|
|
"learning_rate": 5.828877005347593e-07,
|
||
|
|
"logits/chosen": -2.4117431640625,
|
||
|
|
"logits/rejected": -2.413012742996216,
|
||
|
|
"logps/chosen": -217.89453125,
|
||
|
|
"logps/rejected": -225.7703094482422,
|
||
|
|
"logps/weighted_chosen": -0.45955199003219604,
|
||
|
|
"logps/weighted_rejected": -0.4866271913051605,
|
||
|
|
"loss": 0.6916,
|
||
|
|
"rewards/accuracies": 0.4781250059604645,
|
||
|
|
"rewards/chosen": -2.8353514671325684,
|
||
|
|
"rewards/margins": 0.3775390684604645,
|
||
|
|
"rewards/rejected": -3.212890625,
|
||
|
|
"rewards/weighted_accuracies": 0.5093749761581421,
|
||
|
|
"rewards/weighted_chosen": 0.0031097412575036287,
|
||
|
|
"rewards/weighted_margins": 0.004855346865952015,
|
||
|
|
"rewards/weighted_rejected": -0.0017486572032794356,
|
||
|
|
"step": 110
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06430006697923643,
|
||
|
|
"grad_norm": 24.36446762084961,
|
||
|
|
"learning_rate": 6.363636363636363e-07,
|
||
|
|
"logits/chosen": -2.4598631858825684,
|
||
|
|
"logits/rejected": -2.4720215797424316,
|
||
|
|
"logps/chosen": -188.3484344482422,
|
||
|
|
"logps/rejected": -226.6437530517578,
|
||
|
|
"logps/weighted_chosen": -0.4706787168979645,
|
||
|
|
"logps/weighted_rejected": -0.551287829875946,
|
||
|
|
"loss": 0.6893,
|
||
|
|
"rewards/accuracies": 0.47187501192092896,
|
||
|
|
"rewards/chosen": -4.0625,
|
||
|
|
"rewards/margins": 0.947460949420929,
|
||
|
|
"rewards/rejected": -5.010156154632568,
|
||
|
|
"rewards/weighted_accuracies": 0.528124988079071,
|
||
|
|
"rewards/weighted_chosen": -0.018326569348573685,
|
||
|
|
"rewards/weighted_margins": 0.010852813720703125,
|
||
|
|
"rewards/weighted_rejected": -0.02917938306927681,
|
||
|
|
"step": 120
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06965840589417281,
|
||
|
|
"grad_norm": 35.46619415283203,
|
||
|
|
"learning_rate": 6.898395721925134e-07,
|
||
|
|
"logits/chosen": -2.566601514816284,
|
||
|
|
"logits/rejected": -2.5640625953674316,
|
||
|
|
"logps/chosen": -233.8390655517578,
|
||
|
|
"logps/rejected": -243.31875610351562,
|
||
|
|
"logps/weighted_chosen": -0.46611326932907104,
|
||
|
|
"logps/weighted_rejected": -0.48984986543655396,
|
||
|
|
"loss": 0.7012,
|
||
|
|
"rewards/accuracies": 0.574999988079071,
|
||
|
|
"rewards/chosen": -5.547656059265137,
|
||
|
|
"rewards/margins": 1.3078124523162842,
|
||
|
|
"rewards/rejected": -6.855859279632568,
|
||
|
|
"rewards/weighted_accuracies": 0.49687498807907104,
|
||
|
|
"rewards/weighted_chosen": -0.01891174353659153,
|
||
|
|
"rewards/weighted_margins": -0.0069145201705396175,
|
||
|
|
"rewards/weighted_rejected": -0.012065887451171875,
|
||
|
|
"step": 130
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07501674480910918,
|
||
|
|
"grad_norm": 18.125104904174805,
|
||
|
|
"learning_rate": 7.433155080213903e-07,
|
||
|
|
"logits/chosen": -2.602294921875,
|
||
|
|
"logits/rejected": -2.599804639816284,
|
||
|
|
"logps/chosen": -202.98281860351562,
|
||
|
|
"logps/rejected": -215.4796905517578,
|
||
|
|
"logps/weighted_chosen": -0.4594665467739105,
|
||
|
|
"logps/weighted_rejected": -0.5065857172012329,
|
||
|
|
"loss": 0.6858,
|
||
|
|
"rewards/accuracies": 0.559374988079071,
|
||
|
|
"rewards/chosen": -8.157422065734863,
|
||
|
|
"rewards/margins": 2.392578125,
|
||
|
|
"rewards/rejected": -10.550000190734863,
|
||
|
|
"rewards/weighted_accuracies": 0.5531250238418579,
|
||
|
|
"rewards/weighted_chosen": -0.02419433556497097,
|
||
|
|
"rewards/weighted_margins": 0.01836395263671875,
|
||
|
|
"rewards/weighted_rejected": -0.04254303127527237,
|
||
|
|
"step": 140
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08037508372404555,
|
||
|
|
"grad_norm": 44.02063751220703,
|
||
|
|
"learning_rate": 7.967914438502673e-07,
|
||
|
|
"logits/chosen": -2.6329102516174316,
|
||
|
|
"logits/rejected": -2.630322217941284,
|
||
|
|
"logps/chosen": -221.5109405517578,
|
||
|
|
"logps/rejected": -245.00155639648438,
|
||
|
|
"logps/weighted_chosen": -0.522930920124054,
|
||
|
|
"logps/weighted_rejected": -0.535870373249054,
|
||
|
|
"loss": 0.6933,
|
||
|
|
"rewards/accuracies": 0.5625,
|
||
|
|
"rewards/chosen": -11.2197265625,
|
||
|
|
"rewards/margins": 2.375,
|
||
|
|
"rewards/rejected": -13.596288681030273,
|
||
|
|
"rewards/weighted_accuracies": 0.5406249761581421,
|
||
|
|
"rewards/weighted_chosen": -0.07070770114660263,
|
||
|
|
"rewards/weighted_margins": 0.00907135009765625,
|
||
|
|
"rewards/weighted_rejected": -0.07977227866649628,
|
||
|
|
"step": 150
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08573342263898191,
|
||
|
|
"grad_norm": 30.984752655029297,
|
||
|
|
"learning_rate": 8.502673796791443e-07,
|
||
|
|
"logits/chosen": -2.552539110183716,
|
||
|
|
"logits/rejected": -2.5354247093200684,
|
||
|
|
"logps/chosen": -207.53280639648438,
|
||
|
|
"logps/rejected": -236.8117218017578,
|
||
|
|
"logps/weighted_chosen": -0.5689147710800171,
|
||
|
|
"logps/weighted_rejected": -0.5829712152481079,
|
||
|
|
"loss": 0.6936,
|
||
|
|
"rewards/accuracies": 0.534375011920929,
|
||
|
|
"rewards/chosen": -13.640625,
|
||
|
|
"rewards/margins": 1.2478516101837158,
|
||
|
|
"rewards/rejected": -14.890039443969727,
|
||
|
|
"rewards/weighted_accuracies": 0.5625,
|
||
|
|
"rewards/weighted_chosen": -0.11262588202953339,
|
||
|
|
"rewards/weighted_margins": 0.01301422156393528,
|
||
|
|
"rewards/weighted_rejected": -0.12571564316749573,
|
||
|
|
"step": 160
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09109176155391828,
|
||
|
|
"grad_norm": 29.657917022705078,
|
||
|
|
"learning_rate": 9.037433155080213e-07,
|
||
|
|
"logits/chosen": -2.5386719703674316,
|
||
|
|
"logits/rejected": -2.550463914871216,
|
||
|
|
"logps/chosen": -213.2921905517578,
|
||
|
|
"logps/rejected": -224.1906280517578,
|
||
|
|
"logps/weighted_chosen": -0.561798095703125,
|
||
|
|
"logps/weighted_rejected": -0.5876098871231079,
|
||
|
|
"loss": 0.6873,
|
||
|
|
"rewards/accuracies": 0.565625011920929,
|
||
|
|
"rewards/chosen": -15.824999809265137,
|
||
|
|
"rewards/margins": 3.636914014816284,
|
||
|
|
"rewards/rejected": -19.457422256469727,
|
||
|
|
"rewards/weighted_accuracies": 0.534375011920929,
|
||
|
|
"rewards/weighted_chosen": -0.11036185920238495,
|
||
|
|
"rewards/weighted_margins": 0.027300262823700905,
|
||
|
|
"rewards/weighted_rejected": -0.13774414360523224,
|
||
|
|
"step": 170
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09645010046885466,
|
||
|
|
"grad_norm": 16.24775505065918,
|
||
|
|
"learning_rate": 9.572192513368984e-07,
|
||
|
|
"logits/chosen": -2.5972657203674316,
|
||
|
|
"logits/rejected": -2.5906739234924316,
|
||
|
|
"logps/chosen": -229.1453094482422,
|
||
|
|
"logps/rejected": -249.83749389648438,
|
||
|
|
"logps/weighted_chosen": -0.607281506061554,
|
||
|
|
"logps/weighted_rejected": -0.6552734375,
|
||
|
|
"loss": 0.688,
|
||
|
|
"rewards/accuracies": 0.5375000238418579,
|
||
|
|
"rewards/chosen": -19.492773056030273,
|
||
|
|
"rewards/margins": 5.0849609375,
|
||
|
|
"rewards/rejected": -24.581249237060547,
|
||
|
|
"rewards/weighted_accuracies": 0.528124988079071,
|
||
|
|
"rewards/weighted_chosen": -0.16547851264476776,
|
||
|
|
"rewards/weighted_margins": 0.02171630784869194,
|
||
|
|
"rewards/weighted_rejected": -0.1871211975812912,
|
||
|
|
"step": 180
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10180843938379103,
|
||
|
|
"grad_norm": 22.05425262451172,
|
||
|
|
"learning_rate": 9.999965031204306e-07,
|
||
|
|
"logits/chosen": -2.589916944503784,
|
||
|
|
"logits/rejected": -2.578906297683716,
|
||
|
|
"logps/chosen": -224.77499389648438,
|
||
|
|
"logps/rejected": -224.9296875,
|
||
|
|
"logps/weighted_chosen": -0.6605224609375,
|
||
|
|
"logps/weighted_rejected": -0.7260376214981079,
|
||
|
|
"loss": 0.6892,
|
||
|
|
"rewards/accuracies": 0.6000000238418579,
|
||
|
|
"rewards/chosen": -23.936328887939453,
|
||
|
|
"rewards/margins": 2.9306640625,
|
||
|
|
"rewards/rejected": -26.86328125,
|
||
|
|
"rewards/weighted_accuracies": 0.5406249761581421,
|
||
|
|
"rewards/weighted_chosen": -0.21764373779296875,
|
||
|
|
"rewards/weighted_margins": 0.04151611402630806,
|
||
|
|
"rewards/weighted_rejected": -0.25913238525390625,
|
||
|
|
"step": 190
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1071667782987274,
|
||
|
|
"grad_norm": 23.82738494873047,
|
||
|
|
"learning_rate": 9.998741174712533e-07,
|
||
|
|
"logits/chosen": -2.669238328933716,
|
||
|
|
"logits/rejected": -2.6612305641174316,
|
||
|
|
"logps/chosen": -255.4250030517578,
|
||
|
|
"logps/rejected": -270.40313720703125,
|
||
|
|
"logps/weighted_chosen": -0.611083984375,
|
||
|
|
"logps/weighted_rejected": -0.6286376714706421,
|
||
|
|
"loss": 0.7003,
|
||
|
|
"rewards/accuracies": 0.5249999761581421,
|
||
|
|
"rewards/chosen": -27.870702743530273,
|
||
|
|
"rewards/margins": 2.9964842796325684,
|
||
|
|
"rewards/rejected": -30.857812881469727,
|
||
|
|
"rewards/weighted_accuracies": 0.5,
|
||
|
|
"rewards/weighted_chosen": -0.15414123237133026,
|
||
|
|
"rewards/weighted_margins": 0.0028816224075853825,
|
||
|
|
"rewards/weighted_rejected": -0.15706482529640198,
|
||
|
|
"step": 200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11252511721366376,
|
||
|
|
"grad_norm": 28.775007247924805,
|
||
|
|
"learning_rate": 9.995769367531952e-07,
|
||
|
|
"logits/chosen": -2.7216796875,
|
||
|
|
"logits/rejected": -2.700000047683716,
|
||
|
|
"logps/chosen": -248.61093139648438,
|
||
|
|
"logps/rejected": -247.4812469482422,
|
||
|
|
"logps/weighted_chosen": -0.5626465082168579,
|
||
|
|
"logps/weighted_rejected": -0.5739685297012329,
|
||
|
|
"loss": 0.7034,
|
||
|
|
"rewards/accuracies": 0.559374988079071,
|
||
|
|
"rewards/chosen": -30.315235137939453,
|
||
|
|
"rewards/margins": 1.091210961341858,
|
||
|
|
"rewards/rejected": -31.409765243530273,
|
||
|
|
"rewards/weighted_accuracies": 0.559374988079071,
|
||
|
|
"rewards/weighted_chosen": -0.14006805419921875,
|
||
|
|
"rewards/weighted_margins": -0.0037292479537427425,
|
||
|
|
"rewards/weighted_rejected": -0.13644561171531677,
|
||
|
|
"step": 210
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11788345612860013,
|
||
|
|
"grad_norm": 32.326786041259766,
|
||
|
|
"learning_rate": 9.991050648838675e-07,
|
||
|
|
"logits/chosen": -2.8436036109924316,
|
||
|
|
"logits/rejected": -2.8392577171325684,
|
||
|
|
"logps/chosen": -242.8984375,
|
||
|
|
"logps/rejected": -257.7124938964844,
|
||
|
|
"logps/weighted_chosen": -0.616424560546875,
|
||
|
|
"logps/weighted_rejected": -0.65435791015625,
|
||
|
|
"loss": 0.6882,
|
||
|
|
"rewards/accuracies": 0.5718749761581421,
|
||
|
|
"rewards/chosen": -38.271095275878906,
|
||
|
|
"rewards/margins": 1.724218726158142,
|
||
|
|
"rewards/rejected": -39.98163986206055,
|
||
|
|
"rewards/weighted_accuracies": 0.546875,
|
||
|
|
"rewards/weighted_chosen": -0.15750885009765625,
|
||
|
|
"rewards/weighted_margins": 0.02539825439453125,
|
||
|
|
"rewards/weighted_rejected": -0.18288879096508026,
|
||
|
|
"step": 220
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12324179504353651,
|
||
|
|
"grad_norm": 19.367631912231445,
|
||
|
|
"learning_rate": 9.98458666866564e-07,
|
||
|
|
"logits/chosen": -2.869873046875,
|
||
|
|
"logits/rejected": -2.870898485183716,
|
||
|
|
"logps/chosen": -243.66561889648438,
|
||
|
|
"logps/rejected": -263.89373779296875,
|
||
|
|
"logps/weighted_chosen": -0.7010498046875,
|
||
|
|
"logps/weighted_rejected": -0.750701904296875,
|
||
|
|
"loss": 0.6841,
|
||
|
|
"rewards/accuracies": 0.5562499761581421,
|
||
|
|
"rewards/chosen": -47.01171875,
|
||
|
|
"rewards/margins": 7.957421779632568,
|
||
|
|
"rewards/rejected": -54.978515625,
|
||
|
|
"rewards/weighted_accuracies": 0.565625011920929,
|
||
|
|
"rewards/weighted_chosen": -0.24614867568016052,
|
||
|
|
"rewards/weighted_margins": 0.04576721042394638,
|
||
|
|
"rewards/weighted_rejected": -0.2919418215751648,
|
||
|
|
"step": 230
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12860013395847286,
|
||
|
|
"grad_norm": 13.939057350158691,
|
||
|
|
"learning_rate": 9.97637968732563e-07,
|
||
|
|
"logits/chosen": -2.9676756858825684,
|
||
|
|
"logits/rejected": -2.950244188308716,
|
||
|
|
"logps/chosen": -248.04843139648438,
|
||
|
|
"logps/rejected": -263.4750061035156,
|
||
|
|
"logps/weighted_chosen": -0.6573241949081421,
|
||
|
|
"logps/weighted_rejected": -0.648510754108429,
|
||
|
|
"loss": 0.7232,
|
||
|
|
"rewards/accuracies": 0.5562499761581421,
|
||
|
|
"rewards/chosen": -43.94921875,
|
||
|
|
"rewards/margins": 5.845312595367432,
|
||
|
|
"rewards/rejected": -49.795310974121094,
|
||
|
|
"rewards/weighted_accuracies": 0.512499988079071,
|
||
|
|
"rewards/weighted_chosen": -0.22567901015281677,
|
||
|
|
"rewards/weighted_margins": -0.026827240362763405,
|
||
|
|
"rewards/weighted_rejected": -0.19892653822898865,
|
||
|
|
"step": 240
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13395847287340926,
|
||
|
|
"grad_norm": 12.663073539733887,
|
||
|
|
"learning_rate": 9.966432574620906e-07,
|
||
|
|
"logits/chosen": -3.0162110328674316,
|
||
|
|
"logits/rejected": -3.0179200172424316,
|
||
|
|
"logps/chosen": -242.93905639648438,
|
||
|
|
"logps/rejected": -270.6875,
|
||
|
|
"logps/weighted_chosen": -0.628466784954071,
|
||
|
|
"logps/weighted_rejected": -0.67413330078125,
|
||
|
|
"loss": 0.696,
|
||
|
|
"rewards/accuracies": 0.5249999761581421,
|
||
|
|
"rewards/chosen": -48.034568786621094,
|
||
|
|
"rewards/margins": 3.3388671875,
|
||
|
|
"rewards/rejected": -51.372657775878906,
|
||
|
|
"rewards/weighted_accuracies": 0.534375011920929,
|
||
|
|
"rewards/weighted_chosen": -0.21130982041358948,
|
||
|
|
"rewards/weighted_margins": 0.009045409969985485,
|
||
|
|
"rewards/weighted_rejected": -0.22044983506202698,
|
||
|
|
"step": 250
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13931681178834562,
|
||
|
|
"grad_norm": 14.56610107421875,
|
||
|
|
"learning_rate": 9.954748808839674e-07,
|
||
|
|
"logits/chosen": -2.9679198265075684,
|
||
|
|
"logits/rejected": -2.976855516433716,
|
||
|
|
"logps/chosen": -277.53436279296875,
|
||
|
|
"logps/rejected": -275.7718811035156,
|
||
|
|
"logps/weighted_chosen": -0.63873291015625,
|
||
|
|
"logps/weighted_rejected": -0.6825927495956421,
|
||
|
|
"loss": 0.6844,
|
||
|
|
"rewards/accuracies": 0.528124988079071,
|
||
|
|
"rewards/chosen": -50.900779724121094,
|
||
|
|
"rewards/margins": 4.017773628234863,
|
||
|
|
"rewards/rejected": -54.93476486206055,
|
||
|
|
"rewards/weighted_accuracies": 0.559374988079071,
|
||
|
|
"rewards/weighted_chosen": -0.18876799941062927,
|
||
|
|
"rewards/weighted_margins": 0.03838195651769638,
|
||
|
|
"rewards/weighted_rejected": -0.2271774262189865,
|
||
|
|
"step": 260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.144675150703282,
|
||
|
|
"grad_norm": 12.570405006408691,
|
||
|
|
"learning_rate": 9.941332475539824e-07,
|
||
|
|
"logits/chosen": -3.037890672683716,
|
||
|
|
"logits/rejected": -3.046337842941284,
|
||
|
|
"logps/chosen": -263.0531311035156,
|
||
|
|
"logps/rejected": -299.34375,
|
||
|
|
"logps/weighted_chosen": -0.689990222454071,
|
||
|
|
"logps/weighted_rejected": -0.7867187261581421,
|
||
|
|
"loss": 0.6742,
|
||
|
|
"rewards/accuracies": 0.6187499761581421,
|
||
|
|
"rewards/chosen": -52.885154724121094,
|
||
|
|
"rewards/margins": 9.96875,
|
||
|
|
"rewards/rejected": -62.86054611206055,
|
||
|
|
"rewards/weighted_accuracies": 0.6343749761581421,
|
||
|
|
"rewards/weighted_chosen": -0.21307983994483948,
|
||
|
|
"rewards/weighted_margins": 0.06503143161535263,
|
||
|
|
"rewards/weighted_rejected": -0.2780609130859375,
|
||
|
|
"step": 270
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15003348961821836,
|
||
|
|
"grad_norm": 17.126554489135742,
|
||
|
|
"learning_rate": 9.926188266120295e-07,
|
||
|
|
"logits/chosen": -3.169921875,
|
||
|
|
"logits/rejected": -3.1683592796325684,
|
||
|
|
"logps/chosen": -246.58438110351562,
|
||
|
|
"logps/rejected": -262.4078063964844,
|
||
|
|
"logps/weighted_chosen": -0.681854248046875,
|
||
|
|
"logps/weighted_rejected": -0.746777355670929,
|
||
|
|
"loss": 0.6785,
|
||
|
|
"rewards/accuracies": 0.59375,
|
||
|
|
"rewards/chosen": -50.88359451293945,
|
||
|
|
"rewards/margins": 8.079882621765137,
|
||
|
|
"rewards/rejected": -58.97187423706055,
|
||
|
|
"rewards/weighted_accuracies": 0.53125,
|
||
|
|
"rewards/weighted_chosen": -0.2380828857421875,
|
||
|
|
"rewards/weighted_margins": 0.05399169772863388,
|
||
|
|
"rewards/weighted_rejected": -0.291983038187027,
|
||
|
|
"step": 280
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15539182853315472,
|
||
|
|
"grad_norm": 17.43288230895996,
|
||
|
|
"learning_rate": 9.909321476180591e-07,
|
||
|
|
"logits/chosen": -3.285205125808716,
|
||
|
|
"logits/rejected": -3.2626953125,
|
||
|
|
"logps/chosen": -269.65155029296875,
|
||
|
|
"logps/rejected": -298.9703063964844,
|
||
|
|
"logps/weighted_chosen": -0.7649291753768921,
|
||
|
|
"logps/weighted_rejected": -0.828442394733429,
|
||
|
|
"loss": 0.6972,
|
||
|
|
"rewards/accuracies": 0.518750011920929,
|
||
|
|
"rewards/chosen": -66.34883117675781,
|
||
|
|
"rewards/margins": 6.403515815734863,
|
||
|
|
"rewards/rejected": -72.75663757324219,
|
||
|
|
"rewards/weighted_accuracies": 0.5093749761581421,
|
||
|
|
"rewards/weighted_chosen": -0.307464599609375,
|
||
|
|
"rewards/weighted_margins": 0.0262603759765625,
|
||
|
|
"rewards/weighted_rejected": -0.333578497171402,
|
||
|
|
"step": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1607501674480911,
|
||
|
|
"grad_norm": 16.17060089111328,
|
||
|
|
"learning_rate": 9.890738003669027e-07,
|
||
|
|
"logits/chosen": -3.3641114234924316,
|
||
|
|
"logits/rejected": -3.360107421875,
|
||
|
|
"logps/chosen": -260.75,
|
||
|
|
"logps/rejected": -282.5093688964844,
|
||
|
|
"logps/weighted_chosen": -0.743237316608429,
|
||
|
|
"logps/weighted_rejected": -0.8062988519668579,
|
||
|
|
"loss": 0.6858,
|
||
|
|
"rewards/accuracies": 0.609375,
|
||
|
|
"rewards/chosen": -65.58906555175781,
|
||
|
|
"rewards/margins": 8.375,
|
||
|
|
"rewards/rejected": -73.9820327758789,
|
||
|
|
"rewards/weighted_accuracies": 0.534375011920929,
|
||
|
|
"rewards/weighted_chosen": -0.294107049703598,
|
||
|
|
"rewards/weighted_margins": 0.04635162279009819,
|
||
|
|
"rewards/weighted_rejected": -0.34059447050094604,
|
||
|
|
"step": 300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16610850636302746,
|
||
|
|
"grad_norm": 19.679088592529297,
|
||
|
|
"learning_rate": 9.870444346820348e-07,
|
||
|
|
"logits/chosen": -3.3017578125,
|
||
|
|
"logits/rejected": -3.2752928733825684,
|
||
|
|
"logps/chosen": -277.9906311035156,
|
||
|
|
"logps/rejected": -299.7265625,
|
||
|
|
"logps/weighted_chosen": -0.76312255859375,
|
||
|
|
"logps/weighted_rejected": -0.793139636516571,
|
||
|
|
"loss": 0.6994,
|
||
|
|
"rewards/accuracies": 0.6156250238418579,
|
||
|
|
"rewards/chosen": -64.0882797241211,
|
||
|
|
"rewards/margins": 10.8203125,
|
||
|
|
"rewards/rejected": -74.90898132324219,
|
||
|
|
"rewards/weighted_accuracies": 0.5218750238418579,
|
||
|
|
"rewards/weighted_chosen": -0.310202032327652,
|
||
|
|
"rewards/weighted_margins": 0.007830810733139515,
|
||
|
|
"rewards/weighted_rejected": -0.3179565370082855,
|
||
|
|
"step": 310
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17146684527796383,
|
||
|
|
"grad_norm": 18.09251594543457,
|
||
|
|
"learning_rate": 9.848447601883433e-07,
|
||
|
|
"logits/chosen": -3.251171827316284,
|
||
|
|
"logits/rejected": -3.2337403297424316,
|
||
|
|
"logps/chosen": -255.09530639648438,
|
||
|
|
"logps/rejected": -293.9375,
|
||
|
|
"logps/weighted_chosen": -0.7666381597518921,
|
||
|
|
"logps/weighted_rejected": -0.833050549030304,
|
||
|
|
"loss": 0.6918,
|
||
|
|
"rewards/accuracies": 0.574999988079071,
|
||
|
|
"rewards/chosen": -58.306251525878906,
|
||
|
|
"rewards/margins": 15.045312881469727,
|
||
|
|
"rewards/rejected": -73.3515625,
|
||
|
|
"rewards/weighted_accuracies": 0.565625011920929,
|
||
|
|
"rewards/weighted_chosen": -0.332742303609848,
|
||
|
|
"rewards/weighted_margins": 0.03985900804400444,
|
||
|
|
"rewards/weighted_rejected": -0.37254637479782104,
|
||
|
|
"step": 320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1768251841929002,
|
||
|
|
"grad_norm": 25.393939971923828,
|
||
|
|
"learning_rate": 9.824755460639899e-07,
|
||
|
|
"logits/chosen": -3.203418016433716,
|
||
|
|
"logits/rejected": -3.19873046875,
|
||
|
|
"logps/chosen": -261.8515625,
|
||
|
|
"logps/rejected": -305.0328063964844,
|
||
|
|
"logps/weighted_chosen": -0.838146984577179,
|
||
|
|
"logps/weighted_rejected": -0.897717297077179,
|
||
|
|
"loss": 0.6862,
|
||
|
|
"rewards/accuracies": 0.5843750238418579,
|
||
|
|
"rewards/chosen": -70.07734680175781,
|
||
|
|
"rewards/margins": 16.4833984375,
|
||
|
|
"rewards/rejected": -86.54609680175781,
|
||
|
|
"rewards/weighted_accuracies": 0.5531250238418579,
|
||
|
|
"rewards/weighted_chosen": -0.4094604551792145,
|
||
|
|
"rewards/weighted_margins": 0.05315857008099556,
|
||
|
|
"rewards/weighted_rejected": -0.4626617431640625,
|
||
|
|
"step": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18218352310783656,
|
||
|
|
"grad_norm": 15.981916427612305,
|
||
|
|
"learning_rate": 9.799376207714444e-07,
|
||
|
|
"logits/chosen": -3.241992235183716,
|
||
|
|
"logits/rejected": -3.2372069358825684,
|
||
|
|
"logps/chosen": -283.8984375,
|
||
|
|
"logps/rejected": -301.45782470703125,
|
||
|
|
"logps/weighted_chosen": -0.940869152545929,
|
||
|
|
"logps/weighted_rejected": -1.0183837413787842,
|
||
|
|
"loss": 0.6807,
|
||
|
|
"rewards/accuracies": 0.5531250238418579,
|
||
|
|
"rewards/chosen": -82.82109069824219,
|
||
|
|
"rewards/margins": 8.819140434265137,
|
||
|
|
"rewards/rejected": -91.63594055175781,
|
||
|
|
"rewards/weighted_accuracies": 0.5625,
|
||
|
|
"rewards/weighted_chosen": -0.47734373807907104,
|
||
|
|
"rewards/weighted_margins": 0.06443405151367188,
|
||
|
|
"rewards/weighted_rejected": -0.5418640375137329,
|
||
|
|
"step": 340
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18754186202277295,
|
||
|
|
"grad_norm": 11.714133262634277,
|
||
|
|
"learning_rate": 9.772318717677903e-07,
|
||
|
|
"logits/chosen": -3.29248046875,
|
||
|
|
"logits/rejected": -3.2901368141174316,
|
||
|
|
"logps/chosen": -281.28125,
|
||
|
|
"logps/rejected": -298.8671875,
|
||
|
|
"logps/weighted_chosen": -0.8809448480606079,
|
||
|
|
"logps/weighted_rejected": -0.9219726324081421,
|
||
|
|
"loss": 0.6935,
|
||
|
|
"rewards/accuracies": 0.6000000238418579,
|
||
|
|
"rewards/chosen": -81.07421875,
|
||
|
|
"rewards/margins": 11.681249618530273,
|
||
|
|
"rewards/rejected": -92.7164077758789,
|
||
|
|
"rewards/weighted_accuracies": 0.574999988079071,
|
||
|
|
"rewards/weighted_chosen": -0.4380859434604645,
|
||
|
|
"rewards/weighted_margins": 0.02567749097943306,
|
||
|
|
"rewards/weighted_rejected": -0.46390992403030396,
|
||
|
|
"step": 350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19290020093770932,
|
||
|
|
"grad_norm": 17.111141204833984,
|
||
|
|
"learning_rate": 9.743592451943998e-07,
|
||
|
|
"logits/chosen": -3.371386766433716,
|
||
|
|
"logits/rejected": -3.3677735328674316,
|
||
|
|
"logps/chosen": -301.2203063964844,
|
||
|
|
"logps/rejected": -326.7718811035156,
|
||
|
|
"logps/weighted_chosen": -0.884228527545929,
|
||
|
|
"logps/weighted_rejected": -0.8854614496231079,
|
||
|
|
"loss": 0.6983,
|
||
|
|
"rewards/accuracies": 0.581250011920929,
|
||
|
|
"rewards/chosen": -89.6175765991211,
|
||
|
|
"rewards/margins": 10.710546493530273,
|
||
|
|
"rewards/rejected": -100.3335952758789,
|
||
|
|
"rewards/weighted_accuracies": 0.5562499761581421,
|
||
|
|
"rewards/weighted_chosen": -0.4132934510707855,
|
||
|
|
"rewards/weighted_margins": 0.013682556338608265,
|
||
|
|
"rewards/weighted_rejected": -0.427154541015625,
|
||
|
|
"step": 360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1982585398526457,
|
||
|
|
"grad_norm": 13.4055814743042,
|
||
|
|
"learning_rate": 9.713207455460892e-07,
|
||
|
|
"logits/chosen": -3.321484327316284,
|
||
|
|
"logits/rejected": -3.3426756858825684,
|
||
|
|
"logps/chosen": -296.109375,
|
||
|
|
"logps/rejected": -313.0843811035156,
|
||
|
|
"logps/weighted_chosen": -0.8559204339981079,
|
||
|
|
"logps/weighted_rejected": -0.9296630620956421,
|
||
|
|
"loss": 0.6728,
|
||
|
|
"rewards/accuracies": 0.6031249761581421,
|
||
|
|
"rewards/chosen": -90.3265609741211,
|
||
|
|
"rewards/margins": 7.080859184265137,
|
||
|
|
"rewards/rejected": -97.3773422241211,
|
||
|
|
"rewards/weighted_accuracies": 0.5843750238418579,
|
||
|
|
"rewards/weighted_chosen": -0.365234375,
|
||
|
|
"rewards/weighted_margins": 0.06043701246380806,
|
||
|
|
"rewards/weighted_rejected": -0.4256530702114105,
|
||
|
|
"step": 370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20361687876758205,
|
||
|
|
"grad_norm": 12.659865379333496,
|
||
|
|
"learning_rate": 9.681174353198686e-07,
|
||
|
|
"logits/chosen": -3.418164014816284,
|
||
|
|
"logits/rejected": -3.4073243141174316,
|
||
|
|
"logps/chosen": -303.66717529296875,
|
||
|
|
"logps/rejected": -334.4281311035156,
|
||
|
|
"logps/weighted_chosen": -0.7992187738418579,
|
||
|
|
"logps/weighted_rejected": -0.8946533203125,
|
||
|
|
"loss": 0.6715,
|
||
|
|
"rewards/accuracies": 0.59375,
|
||
|
|
"rewards/chosen": -92.99609375,
|
||
|
|
"rewards/margins": 13.2685546875,
|
||
|
|
"rewards/rejected": -106.234375,
|
||
|
|
"rewards/weighted_accuracies": 0.581250011920929,
|
||
|
|
"rewards/weighted_chosen": -0.362588495016098,
|
||
|
|
"rewards/weighted_margins": 0.07253112643957138,
|
||
|
|
"rewards/weighted_rejected": -0.4350738525390625,
|
||
|
|
"step": 380
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20897521768251842,
|
||
|
|
"grad_norm": 33.69499969482422,
|
||
|
|
"learning_rate": 9.647504346434103e-07,
|
||
|
|
"logits/chosen": -3.559375047683716,
|
||
|
|
"logits/rejected": -3.565234422683716,
|
||
|
|
"logps/chosen": -328.44219970703125,
|
||
|
|
"logps/rejected": -352.2484436035156,
|
||
|
|
"logps/weighted_chosen": -0.8904174566268921,
|
||
|
|
"logps/weighted_rejected": -0.944140613079071,
|
||
|
|
"loss": 0.688,
|
||
|
|
"rewards/accuracies": 0.5687500238418579,
|
||
|
|
"rewards/chosen": -107.42109680175781,
|
||
|
|
"rewards/margins": 25.632421493530273,
|
||
|
|
"rewards/rejected": -133.03515625,
|
||
|
|
"rewards/weighted_accuracies": 0.581250011920929,
|
||
|
|
"rewards/weighted_chosen": -0.430419921875,
|
||
|
|
"rewards/weighted_margins": 0.04887695237994194,
|
||
|
|
"rewards/weighted_rejected": -0.47917479276657104,
|
||
|
|
"step": 390
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2143335565974548,
|
||
|
|
"grad_norm": 21.384435653686523,
|
||
|
|
"learning_rate": 9.612209208833646e-07,
|
||
|
|
"logits/chosen": -3.5692381858825684,
|
||
|
|
"logits/rejected": -3.5658202171325684,
|
||
|
|
"logps/chosen": -310.74530029296875,
|
||
|
|
"logps/rejected": -319.8671875,
|
||
|
|
"logps/weighted_chosen": -0.885510265827179,
|
||
|
|
"logps/weighted_rejected": -0.952441394329071,
|
||
|
|
"loss": 0.6817,
|
||
|
|
"rewards/accuracies": 0.5718749761581421,
|
||
|
|
"rewards/chosen": -110.94140625,
|
||
|
|
"rewards/margins": 7.463281154632568,
|
||
|
|
"rewards/rejected": -118.3929672241211,
|
||
|
|
"rewards/weighted_accuracies": 0.5718749761581421,
|
||
|
|
"rewards/weighted_chosen": -0.43572998046875,
|
||
|
|
"rewards/weighted_margins": 0.05572357028722763,
|
||
|
|
"rewards/weighted_rejected": -0.4914703369140625,
|
||
|
|
"step": 400
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21969189551239116,
|
||
|
|
"grad_norm": 41.982784271240234,
|
||
|
|
"learning_rate": 9.5753012823366e-07,
|
||
|
|
"logits/chosen": -3.6553711891174316,
|
||
|
|
"logits/rejected": -3.646191358566284,
|
||
|
|
"logps/chosen": -335.3968811035156,
|
||
|
|
"logps/rejected": -370.7281188964844,
|
||
|
|
"logps/weighted_chosen": -0.857287585735321,
|
||
|
|
"logps/weighted_rejected": -0.8953613042831421,
|
||
|
|
"loss": 0.6959,
|
||
|
|
"rewards/accuracies": 0.609375,
|
||
|
|
"rewards/chosen": -123.1953125,
|
||
|
|
"rewards/margins": 24.586328506469727,
|
||
|
|
"rewards/rejected": -147.77499389648438,
|
||
|
|
"rewards/weighted_accuracies": 0.5874999761581421,
|
||
|
|
"rewards/weighted_chosen": -0.421051025390625,
|
||
|
|
"rewards/weighted_margins": 0.0330963134765625,
|
||
|
|
"rewards/weighted_rejected": -0.4541381895542145,
|
||
|
|
"step": 410
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22505023442732752,
|
||
|
|
"grad_norm": 30.137876510620117,
|
||
|
|
"learning_rate": 9.536793472839324e-07,
|
||
|
|
"logits/chosen": -3.714648485183716,
|
||
|
|
"logits/rejected": -3.7328124046325684,
|
||
|
|
"logps/chosen": -386.20001220703125,
|
||
|
|
"logps/rejected": -426.3656311035156,
|
||
|
|
"logps/weighted_chosen": -0.8975464105606079,
|
||
|
|
"logps/weighted_rejected": -0.970263659954071,
|
||
|
|
"loss": 0.6699,
|
||
|
|
"rewards/accuracies": 0.5687500238418579,
|
||
|
|
"rewards/chosen": -155.86874389648438,
|
||
|
|
"rewards/margins": 28.117578506469727,
|
||
|
|
"rewards/rejected": -183.94296264648438,
|
||
|
|
"rewards/weighted_accuracies": 0.5687500238418579,
|
||
|
|
"rewards/weighted_chosen": -0.43191832304000854,
|
||
|
|
"rewards/weighted_margins": 0.07274474948644638,
|
||
|
|
"rewards/weighted_rejected": -0.5046783685684204,
|
||
|
|
"step": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2304085733422639,
|
||
|
|
"grad_norm": 27.828502655029297,
|
||
|
|
"learning_rate": 9.496699245682351e-07,
|
||
|
|
"logits/chosen": -3.798535108566284,
|
||
|
|
"logits/rejected": -3.790332078933716,
|
||
|
|
"logps/chosen": -396.24688720703125,
|
||
|
|
"logps/rejected": -460.65625,
|
||
|
|
"logps/weighted_chosen": -0.987683117389679,
|
||
|
|
"logps/weighted_rejected": -1.0972778797149658,
|
||
|
|
"loss": 0.6717,
|
||
|
|
"rewards/accuracies": 0.6031249761581421,
|
||
|
|
"rewards/chosen": -175.32421875,
|
||
|
|
"rewards/margins": 44.544532775878906,
|
||
|
|
"rewards/rejected": -219.8874969482422,
|
||
|
|
"rewards/weighted_accuracies": 0.5718749761581421,
|
||
|
|
"rewards/weighted_chosen": -0.5299133062362671,
|
||
|
|
"rewards/weighted_margins": 0.08442840725183487,
|
||
|
|
"rewards/weighted_rejected": -0.6142212152481079,
|
||
|
|
"step": 430
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23576691225720026,
|
||
|
|
"grad_norm": 25.154258728027344,
|
||
|
|
"learning_rate": 9.455032620941839e-07,
|
||
|
|
"logits/chosen": -3.865527391433716,
|
||
|
|
"logits/rejected": -3.831738233566284,
|
||
|
|
"logps/chosen": -364.0687561035156,
|
||
|
|
"logps/rejected": -457.44061279296875,
|
||
|
|
"logps/weighted_chosen": -1.0288207530975342,
|
||
|
|
"logps/weighted_rejected": -1.1377685070037842,
|
||
|
|
"loss": 0.6717,
|
||
|
|
"rewards/accuracies": 0.628125011920929,
|
||
|
|
"rewards/chosen": -169.9953155517578,
|
||
|
|
"rewards/margins": 61.437110900878906,
|
||
|
|
"rewards/rejected": -231.46328735351562,
|
||
|
|
"rewards/weighted_accuracies": 0.5843750238418579,
|
||
|
|
"rewards/weighted_chosen": -0.5935516357421875,
|
||
|
|
"rewards/weighted_margins": 0.09948424994945526,
|
||
|
|
"rewards/weighted_rejected": -0.692797839641571,
|
||
|
|
"step": 440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24112525117213665,
|
||
|
|
"grad_norm": 56.727012634277344,
|
||
|
|
"learning_rate": 9.411808168527066e-07,
|
||
|
|
"logits/chosen": -3.862499952316284,
|
||
|
|
"logits/rejected": -3.863964796066284,
|
||
|
|
"logps/chosen": -430.85626220703125,
|
||
|
|
"logps/rejected": -484.0703125,
|
||
|
|
"logps/weighted_chosen": -1.107214331626892,
|
||
|
|
"logps/weighted_rejected": -1.182470679283142,
|
||
|
|
"loss": 0.6926,
|
||
|
|
"rewards/accuracies": 0.609375,
|
||
|
|
"rewards/chosen": -225.39218139648438,
|
||
|
|
"rewards/margins": 39.25664138793945,
|
||
|
|
"rewards/rejected": -264.46405029296875,
|
||
|
|
"rewards/weighted_accuracies": 0.550000011920929,
|
||
|
|
"rewards/weighted_chosen": -0.6642822027206421,
|
||
|
|
"rewards/weighted_margins": 0.06092224270105362,
|
||
|
|
"rewards/weighted_rejected": -0.725268542766571,
|
||
|
|
"step": 450
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24648359008707302,
|
||
|
|
"grad_norm": 22.511943817138672,
|
||
|
|
"learning_rate": 9.367041003085648e-07,
|
||
|
|
"logits/chosen": -3.80322265625,
|
||
|
|
"logits/rejected": -3.7904295921325684,
|
||
|
|
"logps/chosen": -523.4312744140625,
|
||
|
|
"logps/rejected": -489.375,
|
||
|
|
"logps/weighted_chosen": -1.161865234375,
|
||
|
|
"logps/weighted_rejected": -1.247900366783142,
|
||
|
|
"loss": 0.6807,
|
||
|
|
"rewards/accuracies": 0.546875,
|
||
|
|
"rewards/chosen": -304.58203125,
|
||
|
|
"rewards/margins": -29.098827362060547,
|
||
|
|
"rewards/rejected": -275.4828186035156,
|
||
|
|
"rewards/weighted_accuracies": 0.559374988079071,
|
||
|
|
"rewards/weighted_chosen": -0.703125,
|
||
|
|
"rewards/weighted_margins": 0.069793701171875,
|
||
|
|
"rewards/weighted_rejected": -0.7724975347518921,
|
||
|
|
"step": 460
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25184192900200936,
|
||
|
|
"grad_norm": 26.90446662902832,
|
||
|
|
"learning_rate": 9.320746778718274e-07,
|
||
|
|
"logits/chosen": -3.8857421875,
|
||
|
|
"logits/rejected": -3.8814454078674316,
|
||
|
|
"logps/chosen": -421.92498779296875,
|
||
|
|
"logps/rejected": -482.390625,
|
||
|
|
"logps/weighted_chosen": -1.198461890220642,
|
||
|
|
"logps/weighted_rejected": -1.3473632335662842,
|
||
|
|
"loss": 0.6521,
|
||
|
|
"rewards/accuracies": 0.574999988079071,
|
||
|
|
"rewards/chosen": -234.4382781982422,
|
||
|
|
"rewards/margins": 44.27734375,
|
||
|
|
"rewards/rejected": -278.60626220703125,
|
||
|
|
"rewards/weighted_accuracies": 0.578125,
|
||
|
|
"rewards/weighted_chosen": -0.731597900390625,
|
||
|
|
"rewards/weighted_margins": 0.13102111220359802,
|
||
|
|
"rewards/weighted_rejected": -0.862780749797821,
|
||
|
|
"step": 470
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2572002679169457,
|
||
|
|
"grad_norm": 39.92988586425781,
|
||
|
|
"learning_rate": 9.272941683504808e-07,
|
||
|
|
"logits/chosen": -3.947265625,
|
||
|
|
"logits/rejected": -3.962695360183716,
|
||
|
|
"logps/chosen": -564.7000122070312,
|
||
|
|
"logps/rejected": -586.8125,
|
||
|
|
"logps/weighted_chosen": -1.2533447742462158,
|
||
|
|
"logps/weighted_rejected": -1.4404296875,
|
||
|
|
"loss": 0.6616,
|
||
|
|
"rewards/accuracies": 0.5718749761581421,
|
||
|
|
"rewards/chosen": -346.68280029296875,
|
||
|
|
"rewards/margins": 19.563282012939453,
|
||
|
|
"rewards/rejected": -366.34686279296875,
|
||
|
|
"rewards/weighted_accuracies": 0.612500011920929,
|
||
|
|
"rewards/weighted_chosen": -0.810443103313446,
|
||
|
|
"rewards/weighted_margins": 0.157379150390625,
|
||
|
|
"rewards/weighted_rejected": -0.967883288860321,
|
||
|
|
"step": 480
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2625586068318821,
|
||
|
|
"grad_norm": 41.0818977355957,
|
||
|
|
"learning_rate": 9.223642433843679e-07,
|
||
|
|
"logits/chosen": -4.0751953125,
|
||
|
|
"logits/rejected": -4.055956840515137,
|
||
|
|
"logps/chosen": -527.2843627929688,
|
||
|
|
"logps/rejected": -571.0406494140625,
|
||
|
|
"logps/weighted_chosen": -1.251977562904358,
|
||
|
|
"logps/weighted_rejected": -1.4073975086212158,
|
||
|
|
"loss": 0.6527,
|
||
|
|
"rewards/accuracies": 0.6156250238418579,
|
||
|
|
"rewards/chosen": -325.359375,
|
||
|
|
"rewards/margins": 38.974220275878906,
|
||
|
|
"rewards/rejected": -364.4312438964844,
|
||
|
|
"rewards/weighted_accuracies": 0.6343749761581421,
|
||
|
|
"rewards/weighted_chosen": -0.7925049066543579,
|
||
|
|
"rewards/weighted_margins": 0.14805908501148224,
|
||
|
|
"rewards/weighted_rejected": -0.94061279296875,
|
||
|
|
"step": 490
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2679169457468185,
|
||
|
|
"grad_norm": 27.582181930541992,
|
||
|
|
"learning_rate": 9.172866268606513e-07,
|
||
|
|
"logits/chosen": -4.211133003234863,
|
||
|
|
"logits/rejected": -4.199023246765137,
|
||
|
|
"logps/chosen": -616.1749877929688,
|
||
|
|
"logps/rejected": -701.46875,
|
||
|
|
"logps/weighted_chosen": -1.4478759765625,
|
||
|
|
"logps/weighted_rejected": -1.623193383216858,
|
||
|
|
"loss": 0.6735,
|
||
|
|
"rewards/accuracies": 0.609375,
|
||
|
|
"rewards/chosen": -401.2093811035156,
|
||
|
|
"rewards/margins": 72.16484069824219,
|
||
|
|
"rewards/rejected": -473.28436279296875,
|
||
|
|
"rewards/weighted_accuracies": 0.581250011920929,
|
||
|
|
"rewards/weighted_chosen": -0.987841784954071,
|
||
|
|
"rewards/weighted_margins": 0.13963623344898224,
|
||
|
|
"rewards/weighted_rejected": -1.127478003501892,
|
||
|
|
"step": 500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2679169457468185,
|
||
|
|
"eval_logits/chosen": -4.276368618011475,
|
||
|
|
"eval_logits/rejected": -4.278684139251709,
|
||
|
|
"eval_logps/chosen": -607.260986328125,
|
||
|
|
"eval_logps/rejected": -683.6949462890625,
|
||
|
|
"eval_logps/weighted_chosen": -1.5014480352401733,
|
||
|
|
"eval_logps/weighted_rejected": -1.661415457725525,
|
||
|
|
"eval_loss": 0.6799212694168091,
|
||
|
|
"eval_rewards/accuracies": 0.5794392228126526,
|
||
|
|
"eval_rewards/chosen": -404.2022705078125,
|
||
|
|
"eval_rewards/margins": 63.045310974121094,
|
||
|
|
"eval_rewards/rejected": -467.23162841796875,
|
||
|
|
"eval_rewards/weighted_accuracies": 0.5817757248878479,
|
||
|
|
"eval_rewards/weighted_chosen": -1.0436044931411743,
|
||
|
|
"eval_rewards/weighted_margins": 0.14398153126239777,
|
||
|
|
"eval_rewards/weighted_rejected": -1.1875860691070557,
|
||
|
|
"eval_runtime": 2401.7026,
|
||
|
|
"eval_samples_per_second": 1.247,
|
||
|
|
"eval_steps_per_second": 0.312,
|
||
|
|
"step": 500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2732752846617549,
|
||
|
|
"grad_norm": 54.40583801269531,
|
||
|
|
"learning_rate": 9.120630943110077e-07,
|
||
|
|
"logits/chosen": -4.304296970367432,
|
||
|
|
"logits/rejected": -4.307519435882568,
|
||
|
|
"logps/chosen": -611.9562377929688,
|
||
|
|
"logps/rejected": -747.1343994140625,
|
||
|
|
"logps/weighted_chosen": -1.560327172279358,
|
||
|
|
"logps/weighted_rejected": -1.7308349609375,
|
||
|
|
"loss": 0.6867,
|
||
|
|
"rewards/accuracies": 0.6156250238418579,
|
||
|
|
"rewards/chosen": -421.84844970703125,
|
||
|
|
"rewards/margins": 114.833984375,
|
||
|
|
"rewards/rejected": -536.4578247070312,
|
||
|
|
"rewards/weighted_accuracies": 0.574999988079071,
|
||
|
|
"rewards/weighted_chosen": -1.1248657703399658,
|
||
|
|
"rewards/weighted_margins": 0.15375670790672302,
|
||
|
|
"rewards/weighted_rejected": -1.278204321861267,
|
||
|
|
"step": 510
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27863362357669125,
|
||
|
|
"grad_norm": 34.89970397949219,
|
||
|
|
"learning_rate": 9.066954722907638e-07,
|
||
|
|
"logits/chosen": -4.209668159484863,
|
||
|
|
"logits/rejected": -4.206738471984863,
|
||
|
|
"logps/chosen": -670.203125,
|
||
|
|
"logps/rejected": -714.546875,
|
||
|
|
"logps/weighted_chosen": -1.442846655845642,
|
||
|
|
"logps/weighted_rejected": -1.5417969226837158,
|
||
|
|
"loss": 0.6897,
|
||
|
|
"rewards/accuracies": 0.59375,
|
||
|
|
"rewards/chosen": -450.8140563964844,
|
||
|
|
"rewards/margins": 37.505859375,
|
||
|
|
"rewards/rejected": -488.6499938964844,
|
||
|
|
"rewards/weighted_accuracies": 0.5562499761581421,
|
||
|
|
"rewards/weighted_chosen": -0.9898437261581421,
|
||
|
|
"rewards/weighted_margins": 0.09036865085363388,
|
||
|
|
"rewards/weighted_rejected": -1.080297827720642,
|
||
|
|
"step": 520
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2839919624916276,
|
||
|
|
"grad_norm": 23.328060150146484,
|
||
|
|
"learning_rate": 9.01185637740189e-07,
|
||
|
|
"logits/chosen": -4.20751953125,
|
||
|
|
"logits/rejected": -4.211133003234863,
|
||
|
|
"logps/chosen": -556.3812255859375,
|
||
|
|
"logps/rejected": -589.7468872070312,
|
||
|
|
"logps/weighted_chosen": -1.385229468345642,
|
||
|
|
"logps/weighted_rejected": -1.45050048828125,
|
||
|
|
"loss": 0.7116,
|
||
|
|
"rewards/accuracies": 0.543749988079071,
|
||
|
|
"rewards/chosen": -350.5453186035156,
|
||
|
|
"rewards/margins": 26.184375762939453,
|
||
|
|
"rewards/rejected": -376.70001220703125,
|
||
|
|
"rewards/weighted_accuracies": 0.528124988079071,
|
||
|
|
"rewards/weighted_chosen": -0.9394775629043579,
|
||
|
|
"rewards/weighted_margins": 0.040618896484375,
|
||
|
|
"rewards/weighted_rejected": -0.9805663824081421,
|
||
|
|
"step": 530
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.289350301406564,
|
||
|
|
"grad_norm": 16.991273880004883,
|
||
|
|
"learning_rate": 8.955355173281707e-07,
|
||
|
|
"logits/chosen": -4.077343940734863,
|
||
|
|
"logits/rejected": -4.068749904632568,
|
||
|
|
"logps/chosen": -523.546875,
|
||
|
|
"logps/rejected": -587.75,
|
||
|
|
"logps/weighted_chosen": -1.2385742664337158,
|
||
|
|
"logps/weighted_rejected": -1.4024169445037842,
|
||
|
|
"loss": 0.664,
|
||
|
|
"rewards/accuracies": 0.5874999761581421,
|
||
|
|
"rewards/chosen": -325.93438720703125,
|
||
|
|
"rewards/margins": 49.041404724121094,
|
||
|
|
"rewards/rejected": -374.98748779296875,
|
||
|
|
"rewards/weighted_accuracies": 0.5718749761581421,
|
||
|
|
"rewards/weighted_chosen": -0.800976574420929,
|
||
|
|
"rewards/weighted_margins": 0.12630614638328552,
|
||
|
|
"rewards/weighted_rejected": -0.927136242389679,
|
||
|
|
"step": 540
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29470864032150035,
|
||
|
|
"grad_norm": 27.356420516967773,
|
||
|
|
"learning_rate": 8.897470867785002e-07,
|
||
|
|
"logits/chosen": -3.99658203125,
|
||
|
|
"logits/rejected": -3.965039014816284,
|
||
|
|
"logps/chosen": -568.8343505859375,
|
||
|
|
"logps/rejected": -617.6062622070312,
|
||
|
|
"logps/weighted_chosen": -1.497778296470642,
|
||
|
|
"logps/weighted_rejected": -1.623876929283142,
|
||
|
|
"loss": 0.6754,
|
||
|
|
"rewards/accuracies": 0.625,
|
||
|
|
"rewards/chosen": -357.1234436035156,
|
||
|
|
"rewards/margins": 26.792577743530273,
|
||
|
|
"rewards/rejected": -383.796875,
|
||
|
|
"rewards/weighted_accuracies": 0.59375,
|
||
|
|
"rewards/weighted_chosen": -1.0353882312774658,
|
||
|
|
"rewards/weighted_margins": 0.11194457858800888,
|
||
|
|
"rewards/weighted_rejected": -1.147424340248108,
|
||
|
|
"step": 550
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3000669792364367,
|
||
|
|
"grad_norm": 25.5198974609375,
|
||
|
|
"learning_rate": 8.838223701790055e-07,
|
||
|
|
"logits/chosen": -4.065625190734863,
|
||
|
|
"logits/rejected": -4.058691501617432,
|
||
|
|
"logps/chosen": -624.9249877929688,
|
||
|
|
"logps/rejected": -750.0968627929688,
|
||
|
|
"logps/weighted_chosen": -1.3964111804962158,
|
||
|
|
"logps/weighted_rejected": -1.536376953125,
|
||
|
|
"loss": 0.6737,
|
||
|
|
"rewards/accuracies": 0.5843750238418579,
|
||
|
|
"rewards/chosen": -375.62188720703125,
|
||
|
|
"rewards/margins": 91.9671859741211,
|
||
|
|
"rewards/rejected": -467.7281188964844,
|
||
|
|
"rewards/weighted_accuracies": 0.581250011920929,
|
||
|
|
"rewards/weighted_chosen": -0.920971691608429,
|
||
|
|
"rewards/weighted_margins": 0.10718383640050888,
|
||
|
|
"rewards/weighted_rejected": -1.027990698814392,
|
||
|
|
"step": 560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3054253181513731,
|
||
|
|
"grad_norm": 25.556983947753906,
|
||
|
|
"learning_rate": 8.777634392737718e-07,
|
||
|
|
"logits/chosen": -4.264452934265137,
|
||
|
|
"logits/rejected": -4.25390625,
|
||
|
|
"logps/chosen": -525.6281127929688,
|
||
|
|
"logps/rejected": -586.3875122070312,
|
||
|
|
"logps/weighted_chosen": -1.3518555164337158,
|
||
|
|
"logps/weighted_rejected": -1.4489257335662842,
|
||
|
|
"loss": 0.6742,
|
||
|
|
"rewards/accuracies": 0.625,
|
||
|
|
"rewards/chosen": -333.21563720703125,
|
||
|
|
"rewards/margins": 48.41523361206055,
|
||
|
|
"rewards/rejected": -381.6578063964844,
|
||
|
|
"rewards/weighted_accuracies": 0.625,
|
||
|
|
"rewards/weighted_chosen": -0.872875988483429,
|
||
|
|
"rewards/weighted_margins": 0.09407959133386612,
|
||
|
|
"rewards/weighted_rejected": -0.966778576374054,
|
||
|
|
"step": 570
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31078365706630945,
|
||
|
|
"grad_norm": 24.922456741333008,
|
||
|
|
"learning_rate": 8.71572412738697e-07,
|
||
|
|
"logits/chosen": -4.380566596984863,
|
||
|
|
"logits/rejected": -4.354784965515137,
|
||
|
|
"logps/chosen": -615.390625,
|
||
|
|
"logps/rejected": -714.2687377929688,
|
||
|
|
"logps/weighted_chosen": -1.4163818359375,
|
||
|
|
"logps/weighted_rejected": -1.651708960533142,
|
||
|
|
"loss": 0.6381,
|
||
|
|
"rewards/accuracies": 0.596875011920929,
|
||
|
|
"rewards/chosen": -407.7984313964844,
|
||
|
|
"rewards/margins": 89.12968444824219,
|
||
|
|
"rewards/rejected": -497.0406188964844,
|
||
|
|
"rewards/weighted_accuracies": 0.59375,
|
||
|
|
"rewards/weighted_chosen": -0.986492931842804,
|
||
|
|
"rewards/weighted_margins": 0.215484619140625,
|
||
|
|
"rewards/weighted_rejected": -1.201416015625,
|
||
|
|
"step": 580
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3161419959812458,
|
||
|
|
"grad_norm": 33.28370666503906,
|
||
|
|
"learning_rate": 8.652514554406387e-07,
|
||
|
|
"logits/chosen": -4.452832221984863,
|
||
|
|
"logits/rejected": -4.452929496765137,
|
||
|
|
"logps/chosen": -663.5750122070312,
|
||
|
|
"logps/rejected": -756.0187377929688,
|
||
|
|
"logps/weighted_chosen": -1.5756347179412842,
|
||
|
|
"logps/weighted_rejected": -1.753076195716858,
|
||
|
|
"loss": 0.6673,
|
||
|
|
"rewards/accuracies": 0.581250011920929,
|
||
|
|
"rewards/chosen": -458.00469970703125,
|
||
|
|
"rewards/margins": 74.2398452758789,
|
||
|
|
"rewards/rejected": -532.1781005859375,
|
||
|
|
"rewards/weighted_accuracies": 0.590624988079071,
|
||
|
|
"rewards/weighted_chosen": -1.1391479969024658,
|
||
|
|
"rewards/weighted_margins": 0.15434570610523224,
|
||
|
|
"rewards/weighted_rejected": -1.294091820716858,
|
||
|
|
"step": 590
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3215003348961822,
|
||
|
|
"grad_norm": 35.64254379272461,
|
||
|
|
"learning_rate": 8.588027776804058e-07,
|
||
|
|
"logits/chosen": -4.769824028015137,
|
||
|
|
"logits/rejected": -4.750781059265137,
|
||
|
|
"logps/chosen": -647.7312622070312,
|
||
|
|
"logps/rejected": -909.8531494140625,
|
||
|
|
"logps/weighted_chosen": -1.617163062095642,
|
||
|
|
"logps/weighted_rejected": -1.894140601158142,
|
||
|
|
"loss": 0.6271,
|
||
|
|
"rewards/accuracies": 0.625,
|
||
|
|
"rewards/chosen": -460.3999938964844,
|
||
|
|
"rewards/margins": 230.3953094482422,
|
||
|
|
"rewards/rejected": -690.9593505859375,
|
||
|
|
"rewards/weighted_accuracies": 0.612500011920929,
|
||
|
|
"rewards/weighted_chosen": -1.1735718250274658,
|
||
|
|
"rewards/weighted_margins": 0.250540167093277,
|
||
|
|
"rewards/weighted_rejected": -1.423791527748108,
|
||
|
|
"step": 600
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.32685867381111855,
|
||
|
|
"grad_norm": 40.923988342285156,
|
||
|
|
"learning_rate": 8.522286344198657e-07,
|
||
|
|
"logits/chosen": -4.929883003234863,
|
||
|
|
"logits/rejected": -4.950390815734863,
|
||
|
|
"logps/chosen": -829.8375244140625,
|
||
|
|
"logps/rejected": -974.1187744140625,
|
||
|
|
"logps/weighted_chosen": -1.932958960533142,
|
||
|
|
"logps/weighted_rejected": -2.1434082984924316,
|
||
|
|
"loss": 0.6941,
|
||
|
|
"rewards/accuracies": 0.6031249761581421,
|
||
|
|
"rewards/chosen": -633.2562255859375,
|
||
|
|
"rewards/margins": 127.22265625,
|
||
|
|
"rewards/rejected": -760.5562744140625,
|
||
|
|
"rewards/weighted_accuracies": 0.581250011920929,
|
||
|
|
"rewards/weighted_chosen": -1.479040503501892,
|
||
|
|
"rewards/weighted_margins": 0.19188232719898224,
|
||
|
|
"rewards/weighted_rejected": -1.670751929283142,
|
||
|
|
"step": 610
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3322170127260549,
|
||
|
|
"grad_norm": 36.736900329589844,
|
||
|
|
"learning_rate": 8.455313244934324e-07,
|
||
|
|
"logits/chosen": -4.833398342132568,
|
||
|
|
"logits/rejected": -4.824902534484863,
|
||
|
|
"logps/chosen": -797.4406127929688,
|
||
|
|
"logps/rejected": -925.0250244140625,
|
||
|
|
"logps/weighted_chosen": -1.9183349609375,
|
||
|
|
"logps/weighted_rejected": -2.187915086746216,
|
||
|
|
"loss": 0.6334,
|
||
|
|
"rewards/accuracies": 0.612500011920929,
|
||
|
|
"rewards/chosen": -607.0437622070312,
|
||
|
|
"rewards/margins": 120.53047180175781,
|
||
|
|
"rewards/rejected": -727.6281127929688,
|
||
|
|
"rewards/weighted_accuracies": 0.640625,
|
||
|
|
"rewards/weighted_chosen": -1.450830101966858,
|
||
|
|
"rewards/weighted_margins": 0.269805908203125,
|
||
|
|
"rewards/weighted_rejected": -1.720971703529358,
|
||
|
|
"step": 620
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3375753516409913,
|
||
|
|
"grad_norm": 31.050582885742188,
|
||
|
|
"learning_rate": 8.38713189804215e-07,
|
||
|
|
"logits/chosen": -4.622754096984863,
|
||
|
|
"logits/rejected": -4.619921684265137,
|
||
|
|
"logps/chosen": -849.8062744140625,
|
||
|
|
"logps/rejected": -945.5875244140625,
|
||
|
|
"logps/weighted_chosen": -1.9373047351837158,
|
||
|
|
"logps/weighted_rejected": -2.010424852371216,
|
||
|
|
"loss": 0.7012,
|
||
|
|
"rewards/accuracies": 0.534375011920929,
|
||
|
|
"rewards/chosen": -641.3187255859375,
|
||
|
|
"rewards/margins": 92.91719055175781,
|
||
|
|
"rewards/rejected": -734.2515869140625,
|
||
|
|
"rewards/weighted_accuracies": 0.5843750238418579,
|
||
|
|
"rewards/weighted_chosen": -1.4818847179412842,
|
||
|
|
"rewards/weighted_margins": 0.08740844577550888,
|
||
|
|
"rewards/weighted_rejected": -1.569421410560608,
|
||
|
|
"step": 630
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34293369055592765,
|
||
|
|
"grad_norm": 40.31166076660156,
|
||
|
|
"learning_rate": 8.317766145051057e-07,
|
||
|
|
"logits/chosen": -4.409081935882568,
|
||
|
|
"logits/rejected": -4.413866996765137,
|
||
|
|
"logps/chosen": -700.421875,
|
||
|
|
"logps/rejected": -798.5531005859375,
|
||
|
|
"logps/weighted_chosen": -1.5460937023162842,
|
||
|
|
"logps/weighted_rejected": -1.7400634288787842,
|
||
|
|
"loss": 0.6505,
|
||
|
|
"rewards/accuracies": 0.5874999761581421,
|
||
|
|
"rewards/chosen": -500.640625,
|
||
|
|
"rewards/margins": 82.5933609008789,
|
||
|
|
"rewards/rejected": -583.515625,
|
||
|
|
"rewards/weighted_accuracies": 0.6031249761581421,
|
||
|
|
"rewards/weighted_chosen": -1.10516357421875,
|
||
|
|
"rewards/weighted_margins": 0.16577759385108948,
|
||
|
|
"rewards/weighted_rejected": -1.271386742591858,
|
||
|
|
"step": 640
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.348292029470864,
|
||
|
|
"grad_norm": 48.715423583984375,
|
||
|
|
"learning_rate": 8.247240241650917e-07,
|
||
|
|
"logits/chosen": -4.5068359375,
|
||
|
|
"logits/rejected": -4.5087890625,
|
||
|
|
"logps/chosen": -747.8687744140625,
|
||
|
|
"logps/rejected": -850.5374755859375,
|
||
|
|
"logps/weighted_chosen": -1.6713135242462158,
|
||
|
|
"logps/weighted_rejected": -1.8655273914337158,
|
||
|
|
"loss": 0.6506,
|
||
|
|
"rewards/accuracies": 0.612500011920929,
|
||
|
|
"rewards/chosen": -539.6968994140625,
|
||
|
|
"rewards/margins": 100.546875,
|
||
|
|
"rewards/rejected": -640.5562744140625,
|
||
|
|
"rewards/weighted_accuracies": 0.621874988079071,
|
||
|
|
"rewards/weighted_chosen": -1.21282958984375,
|
||
|
|
"rewards/weighted_margins": 0.18698731064796448,
|
||
|
|
"rewards/weighted_rejected": -1.400488257408142,
|
||
|
|
"step": 650
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3536503683858004,
|
||
|
|
"grad_norm": 46.12395477294922,
|
||
|
|
"learning_rate": 8.175578849210894e-07,
|
||
|
|
"logits/chosen": -4.743945121765137,
|
||
|
|
"logits/rejected": -4.784277439117432,
|
||
|
|
"logps/chosen": -885.8312377929688,
|
||
|
|
"logps/rejected": -1061.4625244140625,
|
||
|
|
"logps/weighted_chosen": -1.837182641029358,
|
||
|
|
"logps/weighted_rejected": -2.1468505859375,
|
||
|
|
"loss": 0.6184,
|
||
|
|
"rewards/accuracies": 0.6031249761581421,
|
||
|
|
"rewards/chosen": -676.6203002929688,
|
||
|
|
"rewards/margins": 169.16796875,
|
||
|
|
"rewards/rejected": -845.6687622070312,
|
||
|
|
"rewards/weighted_accuracies": 0.668749988079071,
|
||
|
|
"rewards/weighted_chosen": -1.385009765625,
|
||
|
|
"rewards/weighted_margins": 0.30499267578125,
|
||
|
|
"rewards/weighted_rejected": -1.688879370689392,
|
||
|
|
"step": 660
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35900870730073675,
|
||
|
|
"grad_norm": 135.67051696777344,
|
||
|
|
"learning_rate": 8.102807026155873e-07,
|
||
|
|
"logits/chosen": -4.988379001617432,
|
||
|
|
"logits/rejected": -5.020312309265137,
|
||
|
|
"logps/chosen": -1107.8218994140625,
|
||
|
|
"logps/rejected": -1233.659423828125,
|
||
|
|
"logps/weighted_chosen": -2.1700682640075684,
|
||
|
|
"logps/weighted_rejected": -2.523388624191284,
|
||
|
|
"loss": 0.6695,
|
||
|
|
"rewards/accuracies": 0.5687500238418579,
|
||
|
|
"rewards/chosen": -899.3546752929688,
|
||
|
|
"rewards/margins": 124.86640930175781,
|
||
|
|
"rewards/rejected": -1024.395263671875,
|
||
|
|
"rewards/weighted_accuracies": 0.637499988079071,
|
||
|
|
"rewards/weighted_chosen": -1.724951148033142,
|
||
|
|
"rewards/weighted_margins": 0.35167235136032104,
|
||
|
|
"rewards/weighted_rejected": -2.076953172683716,
|
||
|
|
"step": 670
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3643670462156731,
|
||
|
|
"grad_norm": 35.317413330078125,
|
||
|
|
"learning_rate": 8.028950219204099e-07,
|
||
|
|
"logits/chosen": -4.710644721984863,
|
||
|
|
"logits/rejected": -4.72119140625,
|
||
|
|
"logps/chosen": -753.5906372070312,
|
||
|
|
"logps/rejected": -938.3656005859375,
|
||
|
|
"logps/weighted_chosen": -1.5244872570037842,
|
||
|
|
"logps/weighted_rejected": -1.7093017101287842,
|
||
|
|
"loss": 0.6609,
|
||
|
|
"rewards/accuracies": 0.5874999761581421,
|
||
|
|
"rewards/chosen": -544.7249755859375,
|
||
|
|
"rewards/margins": 165.43359375,
|
||
|
|
"rewards/rejected": -710.1968994140625,
|
||
|
|
"rewards/weighted_accuracies": 0.581250011920929,
|
||
|
|
"rewards/weighted_chosen": -1.093054175376892,
|
||
|
|
"rewards/weighted_margins": 0.16083984076976776,
|
||
|
|
"rewards/weighted_rejected": -1.254003882408142,
|
||
|
|
"step": 680
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3697253851306095,
|
||
|
|
"grad_norm": 32.520843505859375,
|
||
|
|
"learning_rate": 7.954034254469e-07,
|
||
|
|
"logits/chosen": -4.69140625,
|
||
|
|
"logits/rejected": -4.702832221984863,
|
||
|
|
"logps/chosen": -823.9593505859375,
|
||
|
|
"logps/rejected": -1088.659423828125,
|
||
|
|
"logps/weighted_chosen": -1.5863525867462158,
|
||
|
|
"logps/weighted_rejected": -1.8434326648712158,
|
||
|
|
"loss": 0.6454,
|
||
|
|
"rewards/accuracies": 0.581250011920929,
|
||
|
|
"rewards/chosen": -615.2468872070312,
|
||
|
|
"rewards/margins": 252.4890594482422,
|
||
|
|
"rewards/rejected": -867.453125,
|
||
|
|
"rewards/weighted_accuracies": 0.637499988079071,
|
||
|
|
"rewards/weighted_chosen": -1.138574242591858,
|
||
|
|
"rewards/weighted_margins": 0.23261718451976776,
|
||
|
|
"rewards/weighted_rejected": -1.371545433998108,
|
||
|
|
"step": 690
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3750837240455459,
|
||
|
|
"grad_norm": 46.228511810302734,
|
||
|
|
"learning_rate": 7.878085328428368e-07,
|
||
|
|
"logits/chosen": -4.722070217132568,
|
||
|
|
"logits/rejected": -4.739160060882568,
|
||
|
|
"logps/chosen": -1027.971923828125,
|
||
|
|
"logps/rejected": -1167.421875,
|
||
|
|
"logps/weighted_chosen": -1.942407250404358,
|
||
|
|
"logps/weighted_rejected": -2.2004151344299316,
|
||
|
|
"loss": 0.648,
|
||
|
|
"rewards/accuracies": 0.6000000238418579,
|
||
|
|
"rewards/chosen": -807.4937744140625,
|
||
|
|
"rewards/margins": 135.06015014648438,
|
||
|
|
"rewards/rejected": -942.3468627929688,
|
||
|
|
"rewards/weighted_accuracies": 0.659375011920929,
|
||
|
|
"rewards/weighted_chosen": -1.4796142578125,
|
||
|
|
"rewards/weighted_margins": 0.23435059189796448,
|
||
|
|
"rewards/weighted_rejected": -1.7139403820037842,
|
||
|
|
"step": 700
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3804420629604823,
|
||
|
|
"grad_norm": 78.81863403320312,
|
||
|
|
"learning_rate": 7.801129998764014e-07,
|
||
|
|
"logits/chosen": -5.2890625,
|
||
|
|
"logits/rejected": -5.337304592132568,
|
||
|
|
"logps/chosen": -1257.909423828125,
|
||
|
|
"logps/rejected": -1334.559326171875,
|
||
|
|
"logps/weighted_chosen": -2.3622069358825684,
|
||
|
|
"logps/weighted_rejected": -2.719970703125,
|
||
|
|
"loss": 0.6136,
|
||
|
|
"rewards/accuracies": 0.5625,
|
||
|
|
"rewards/chosen": -1042.5062255859375,
|
||
|
|
"rewards/margins": 75.5640640258789,
|
||
|
|
"rewards/rejected": -1118.1546630859375,
|
||
|
|
"rewards/weighted_accuracies": 0.6312500238418579,
|
||
|
|
"rewards/weighted_chosen": -1.926733374595642,
|
||
|
|
"rewards/weighted_margins": 0.32972413301467896,
|
||
|
|
"rewards/weighted_rejected": -2.256054639816284,
|
||
|
|
"step": 710
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38580040187541864,
|
||
|
|
"grad_norm": 61.859046936035156,
|
||
|
|
"learning_rate": 7.723195175075135e-07,
|
||
|
|
"logits/chosen": -5.721093654632568,
|
||
|
|
"logits/rejected": -5.796093940734863,
|
||
|
|
"logps/chosen": -1532.6312255859375,
|
||
|
|
"logps/rejected": -1961.4312744140625,
|
||
|
|
"logps/weighted_chosen": -3.167187452316284,
|
||
|
|
"logps/weighted_rejected": -3.689257860183716,
|
||
|
|
"loss": 0.6426,
|
||
|
|
"rewards/accuracies": 0.640625,
|
||
|
|
"rewards/chosen": -1346.1624755859375,
|
||
|
|
"rewards/margins": 397.6070251464844,
|
||
|
|
"rewards/rejected": -1743.856201171875,
|
||
|
|
"rewards/weighted_accuracies": 0.659375011920929,
|
||
|
|
"rewards/weighted_chosen": -2.710205078125,
|
||
|
|
"rewards/weighted_margins": 0.50091552734375,
|
||
|
|
"rewards/weighted_rejected": -3.210693359375,
|
||
|
|
"step": 720
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.391158740790355,
|
||
|
|
"grad_norm": 41.49199295043945,
|
||
|
|
"learning_rate": 7.644308109468609e-07,
|
||
|
|
"logits/chosen": -4.699120998382568,
|
||
|
|
"logits/rejected": -4.692578315734863,
|
||
|
|
"logps/chosen": -1019.1875,
|
||
|
|
"logps/rejected": -1183.715576171875,
|
||
|
|
"logps/weighted_chosen": -2.080761671066284,
|
||
|
|
"logps/weighted_rejected": -2.3409667015075684,
|
||
|
|
"loss": 0.6433,
|
||
|
|
"rewards/accuracies": 0.5874999761581421,
|
||
|
|
"rewards/chosen": -815.9515380859375,
|
||
|
|
"rewards/margins": 163.99960327148438,
|
||
|
|
"rewards/rejected": -979.8297119140625,
|
||
|
|
"rewards/weighted_accuracies": 0.65625,
|
||
|
|
"rewards/weighted_chosen": -1.63037109375,
|
||
|
|
"rewards/weighted_margins": 0.25758057832717896,
|
||
|
|
"rewards/weighted_rejected": -1.888208031654358,
|
||
|
|
"step": 730
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3965170797052914,
|
||
|
|
"grad_norm": 42.90148162841797,
|
||
|
|
"learning_rate": 7.564496387029531e-07,
|
||
|
|
"logits/chosen": -4.489648342132568,
|
||
|
|
"logits/rejected": -4.511328220367432,
|
||
|
|
"logps/chosen": -835.9718627929688,
|
||
|
|
"logps/rejected": -982.0,
|
||
|
|
"logps/weighted_chosen": -1.939965844154358,
|
||
|
|
"logps/weighted_rejected": -2.176513671875,
|
||
|
|
"loss": 0.6367,
|
||
|
|
"rewards/accuracies": 0.637499988079071,
|
||
|
|
"rewards/chosen": -630.2437744140625,
|
||
|
|
"rewards/margins": 136.02578735351562,
|
||
|
|
"rewards/rejected": -766.3656005859375,
|
||
|
|
"rewards/weighted_accuracies": 0.6187499761581421,
|
||
|
|
"rewards/weighted_chosen": -1.4935791492462158,
|
||
|
|
"rewards/weighted_margins": 0.223297119140625,
|
||
|
|
"rewards/weighted_rejected": -1.7169189453125,
|
||
|
|
"step": 740
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.40187541862022774,
|
||
|
|
"grad_norm": 59.750553131103516,
|
||
|
|
"learning_rate": 7.483787916175306e-07,
|
||
|
|
"logits/chosen": -4.936230659484863,
|
||
|
|
"logits/rejected": -4.978906154632568,
|
||
|
|
"logps/chosen": -1336.675048828125,
|
||
|
|
"logps/rejected": -1601.0625,
|
||
|
|
"logps/weighted_chosen": -2.7215819358825684,
|
||
|
|
"logps/weighted_rejected": -3.0896973609924316,
|
||
|
|
"loss": 0.6144,
|
||
|
|
"rewards/accuracies": 0.6031249761581421,
|
||
|
|
"rewards/chosen": -1120.268798828125,
|
||
|
|
"rewards/margins": 255.3640594482422,
|
||
|
|
"rewards/rejected": -1375.2125244140625,
|
||
|
|
"rewards/weighted_accuracies": 0.6781250238418579,
|
||
|
|
"rewards/weighted_chosen": -2.2766356468200684,
|
||
|
|
"rewards/weighted_margins": 0.36883544921875,
|
||
|
|
"rewards/weighted_rejected": -2.64599609375,
|
||
|
|
"step": 750
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4072337575351641,
|
||
|
|
"grad_norm": 46.1683235168457,
|
||
|
|
"learning_rate": 7.402210918896689e-07,
|
||
|
|
"logits/chosen": -5.019238471984863,
|
||
|
|
"logits/rejected": -5.020898342132568,
|
||
|
|
"logps/chosen": -1522.1875,
|
||
|
|
"logps/rejected": -1768.2249755859375,
|
||
|
|
"logps/weighted_chosen": -3.219921827316284,
|
||
|
|
"logps/weighted_rejected": -3.6636719703674316,
|
||
|
|
"loss": 0.6513,
|
||
|
|
"rewards/accuracies": 0.6031249761581421,
|
||
|
|
"rewards/chosen": -1321.1937255859375,
|
||
|
|
"rewards/margins": 242.9265594482422,
|
||
|
|
"rewards/rejected": -1564.503173828125,
|
||
|
|
"rewards/weighted_accuracies": 0.625,
|
||
|
|
"rewards/weighted_chosen": -2.7684082984924316,
|
||
|
|
"rewards/weighted_margins": 0.41920167207717896,
|
||
|
|
"rewards/weighted_rejected": -3.188183546066284,
|
||
|
|
"step": 760
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4125920964501005,
|
||
|
|
"grad_norm": 63.735076904296875,
|
||
|
|
"learning_rate": 7.31979392088917e-07,
|
||
|
|
"logits/chosen": -4.508008003234863,
|
||
|
|
"logits/rejected": -4.523144721984863,
|
||
|
|
"logps/chosen": -1159.2125244140625,
|
||
|
|
"logps/rejected": -1374.90625,
|
||
|
|
"logps/weighted_chosen": -2.1778321266174316,
|
||
|
|
"logps/weighted_rejected": -2.497607469558716,
|
||
|
|
"loss": 0.6347,
|
||
|
|
"rewards/accuracies": 0.581250011920929,
|
||
|
|
"rewards/chosen": -934.84375,
|
||
|
|
"rewards/margins": 200.5187530517578,
|
||
|
|
"rewards/rejected": -1135.8812255859375,
|
||
|
|
"rewards/weighted_accuracies": 0.643750011920929,
|
||
|
|
"rewards/weighted_chosen": -1.7113769054412842,
|
||
|
|
"rewards/weighted_margins": 0.29365235567092896,
|
||
|
|
"rewards/weighted_rejected": -2.00531005859375,
|
||
|
|
"step": 770
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41795043536503684,
|
||
|
|
"grad_norm": 59.270362854003906,
|
||
|
|
"learning_rate": 7.236565741578162e-07,
|
||
|
|
"logits/chosen": -4.526562690734863,
|
||
|
|
"logits/rejected": -4.559374809265137,
|
||
|
|
"logps/chosen": -1011.7062377929688,
|
||
|
|
"logps/rejected": -1285.487548828125,
|
||
|
|
"logps/weighted_chosen": -2.357372999191284,
|
||
|
|
"logps/weighted_rejected": -2.668701171875,
|
||
|
|
"loss": 0.6262,
|
||
|
|
"rewards/accuracies": 0.637499988079071,
|
||
|
|
"rewards/chosen": -831.0437622070312,
|
||
|
|
"rewards/margins": 246.8093719482422,
|
||
|
|
"rewards/rejected": -1078.125,
|
||
|
|
"rewards/weighted_accuracies": 0.6781250238418579,
|
||
|
|
"rewards/weighted_chosen": -1.8983886241912842,
|
||
|
|
"rewards/weighted_margins": 0.29566651582717896,
|
||
|
|
"rewards/weighted_rejected": -2.19390869140625,
|
||
|
|
"step": 780
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4233087742799732,
|
||
|
|
"grad_norm": 52.86665725708008,
|
||
|
|
"learning_rate": 7.152555484041475e-07,
|
||
|
|
"logits/chosen": -4.317285060882568,
|
||
|
|
"logits/rejected": -4.292870998382568,
|
||
|
|
"logps/chosen": -1603.3062744140625,
|
||
|
|
"logps/rejected": -1803.6937255859375,
|
||
|
|
"logps/weighted_chosen": -3.095752000808716,
|
||
|
|
"logps/weighted_rejected": -3.362353563308716,
|
||
|
|
"loss": 0.6784,
|
||
|
|
"rewards/accuracies": 0.565625011920929,
|
||
|
|
"rewards/chosen": -1380.3187255859375,
|
||
|
|
"rewards/margins": 201.5476531982422,
|
||
|
|
"rewards/rejected": -1581.324951171875,
|
||
|
|
"rewards/weighted_accuracies": 0.625,
|
||
|
|
"rewards/weighted_chosen": -2.6065917015075684,
|
||
|
|
"rewards/weighted_margins": 0.2591919004917145,
|
||
|
|
"rewards/weighted_rejected": -2.8663573265075684,
|
||
|
|
"step": 790
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4286671131949096,
|
||
|
|
"grad_norm": 62.58599853515625,
|
||
|
|
"learning_rate": 7.067792524832603e-07,
|
||
|
|
"logits/chosen": -3.805224657058716,
|
||
|
|
"logits/rejected": -3.7125487327575684,
|
||
|
|
"logps/chosen": -1634.5562744140625,
|
||
|
|
"logps/rejected": -2077.981201171875,
|
||
|
|
"logps/weighted_chosen": -3.627148389816284,
|
||
|
|
"logps/weighted_rejected": -4.034033298492432,
|
||
|
|
"loss": 0.615,
|
||
|
|
"rewards/accuracies": 0.6468750238418579,
|
||
|
|
"rewards/chosen": -1447.7249755859375,
|
||
|
|
"rewards/margins": 420.54376220703125,
|
||
|
|
"rewards/rejected": -1869.34375,
|
||
|
|
"rewards/weighted_accuracies": 0.653124988079071,
|
||
|
|
"rewards/weighted_chosen": -3.1629881858825684,
|
||
|
|
"rewards/weighted_margins": 0.39582520723342896,
|
||
|
|
"rewards/weighted_rejected": -3.559765577316284,
|
||
|
|
"step": 800
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.43402545210984594,
|
||
|
|
"grad_norm": 50.020774841308594,
|
||
|
|
"learning_rate": 6.982306503708387e-07,
|
||
|
|
"logits/chosen": -3.9469237327575684,
|
||
|
|
"logits/rejected": -3.803515672683716,
|
||
|
|
"logps/chosen": -2404.94384765625,
|
||
|
|
"logps/rejected": -2957.58740234375,
|
||
|
|
"logps/weighted_chosen": -4.365234375,
|
||
|
|
"logps/weighted_rejected": -4.791406154632568,
|
||
|
|
"loss": 0.6656,
|
||
|
|
"rewards/accuracies": 0.5874999761581421,
|
||
|
|
"rewards/chosen": -2192.418701171875,
|
||
|
|
"rewards/margins": 530.7859497070312,
|
||
|
|
"rewards/rejected": -2723.253173828125,
|
||
|
|
"rewards/weighted_accuracies": 0.6625000238418579,
|
||
|
|
"rewards/weighted_chosen": -3.917187452316284,
|
||
|
|
"rewards/weighted_margins": 0.39775389432907104,
|
||
|
|
"rewards/weighted_rejected": -4.31494140625,
|
||
|
|
"step": 810
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4393837910247823,
|
||
|
|
"grad_norm": 58.94343948364258,
|
||
|
|
"learning_rate": 6.896127313264642e-07,
|
||
|
|
"logits/chosen": -3.968212842941284,
|
||
|
|
"logits/rejected": -3.8505859375,
|
||
|
|
"logps/chosen": -2033.018798828125,
|
||
|
|
"logps/rejected": -2514.77490234375,
|
||
|
|
"logps/weighted_chosen": -3.811718702316284,
|
||
|
|
"logps/weighted_rejected": -4.425683498382568,
|
||
|
|
"loss": 0.593,
|
||
|
|
"rewards/accuracies": 0.59375,
|
||
|
|
"rewards/chosen": -1811.596923828125,
|
||
|
|
"rewards/margins": 460.2734375,
|
||
|
|
"rewards/rejected": -2272.324951171875,
|
||
|
|
"rewards/weighted_accuracies": 0.699999988079071,
|
||
|
|
"rewards/weighted_chosen": -3.3623046875,
|
||
|
|
"rewards/weighted_margins": 0.5855957269668579,
|
||
|
|
"rewards/weighted_rejected": -3.9478516578674316,
|
||
|
|
"step": 820
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4447421299397187,
|
||
|
|
"grad_norm": 125.08843994140625,
|
||
|
|
"learning_rate": 6.809285088483361e-07,
|
||
|
|
"logits/chosen": -3.9242186546325684,
|
||
|
|
"logits/rejected": -3.902148485183716,
|
||
|
|
"logps/chosen": -1946.675048828125,
|
||
|
|
"logps/rejected": -2531.72509765625,
|
||
|
|
"logps/weighted_chosen": -4.324951171875,
|
||
|
|
"logps/weighted_rejected": -4.77001953125,
|
||
|
|
"loss": 0.6535,
|
||
|
|
"rewards/accuracies": 0.606249988079071,
|
||
|
|
"rewards/chosen": -1747.137451171875,
|
||
|
|
"rewards/margins": 569.9296875,
|
||
|
|
"rewards/rejected": -2317.175048828125,
|
||
|
|
"rewards/weighted_accuracies": 0.625,
|
||
|
|
"rewards/weighted_chosen": -3.8677978515625,
|
||
|
|
"rewards/weighted_margins": 0.4211181700229645,
|
||
|
|
"rewards/weighted_rejected": -4.289404392242432,
|
||
|
|
"step": 830
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45010046885465504,
|
||
|
|
"grad_norm": 139.7412109375,
|
||
|
|
"learning_rate": 6.721810196195174e-07,
|
||
|
|
"logits/chosen": -4.127343654632568,
|
||
|
|
"logits/rejected": -4.019824028015137,
|
||
|
|
"logps/chosen": -1883.1812744140625,
|
||
|
|
"logps/rejected": -2426.175048828125,
|
||
|
|
"logps/weighted_chosen": -4.117968559265137,
|
||
|
|
"logps/weighted_rejected": -4.751269340515137,
|
||
|
|
"loss": 0.5946,
|
||
|
|
"rewards/accuracies": 0.609375,
|
||
|
|
"rewards/chosen": -1689.3062744140625,
|
||
|
|
"rewards/margins": 523.2828369140625,
|
||
|
|
"rewards/rejected": -2212.34375,
|
||
|
|
"rewards/weighted_accuracies": 0.659375011920929,
|
||
|
|
"rewards/weighted_chosen": -3.679248094558716,
|
||
|
|
"rewards/weighted_margins": 0.62701416015625,
|
||
|
|
"rewards/weighted_rejected": -4.307226657867432,
|
||
|
|
"step": 840
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4554588077695914,
|
||
|
|
"grad_norm": 52.51661682128906,
|
||
|
|
"learning_rate": 6.633733224460737e-07,
|
||
|
|
"logits/chosen": -4.39306640625,
|
||
|
|
"logits/rejected": -4.395898342132568,
|
||
|
|
"logps/chosen": -1563.7125244140625,
|
||
|
|
"logps/rejected": -2087.175048828125,
|
||
|
|
"logps/weighted_chosen": -3.411572217941284,
|
||
|
|
"logps/weighted_rejected": -3.908007860183716,
|
||
|
|
"loss": 0.5817,
|
||
|
|
"rewards/accuracies": 0.640625,
|
||
|
|
"rewards/chosen": -1360.9937744140625,
|
||
|
|
"rewards/margins": 503.0453186035156,
|
||
|
|
"rewards/rejected": -1863.5687255859375,
|
||
|
|
"rewards/weighted_accuracies": 0.671875,
|
||
|
|
"rewards/weighted_chosen": -2.9693846702575684,
|
||
|
|
"rewards/weighted_margins": 0.48114013671875,
|
||
|
|
"rewards/weighted_rejected": -3.4501953125,
|
||
|
|
"step": 850
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4608171466845278,
|
||
|
|
"grad_norm": 75.17611694335938,
|
||
|
|
"learning_rate": 6.545084971874736e-07,
|
||
|
|
"logits/chosen": -4.860058784484863,
|
||
|
|
"logits/rejected": -4.897070407867432,
|
||
|
|
"logps/chosen": -2070.018798828125,
|
||
|
|
"logps/rejected": -2581.47509765625,
|
||
|
|
"logps/weighted_chosen": -4.260986328125,
|
||
|
|
"logps/weighted_rejected": -4.812890529632568,
|
||
|
|
"loss": 0.6117,
|
||
|
|
"rewards/accuracies": 0.6468750238418579,
|
||
|
|
"rewards/chosen": -1878.949951171875,
|
||
|
|
"rewards/margins": 492.6968688964844,
|
||
|
|
"rewards/rejected": -2372.15625,
|
||
|
|
"rewards/weighted_accuracies": 0.6499999761581421,
|
||
|
|
"rewards/weighted_chosen": -3.812451124191284,
|
||
|
|
"rewards/weighted_margins": 0.538012683391571,
|
||
|
|
"rewards/weighted_rejected": -4.350634574890137,
|
||
|
|
"step": 860
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.46617548559946415,
|
||
|
|
"grad_norm": 61.795143127441406,
|
||
|
|
"learning_rate": 6.455896436796313e-07,
|
||
|
|
"logits/chosen": -4.861230373382568,
|
||
|
|
"logits/rejected": -4.895312309265137,
|
||
|
|
"logps/chosen": -2298.987548828125,
|
||
|
|
"logps/rejected": -2865.375,
|
||
|
|
"logps/weighted_chosen": -4.515429496765137,
|
||
|
|
"logps/weighted_rejected": -5.017285346984863,
|
||
|
|
"loss": 0.6405,
|
||
|
|
"rewards/accuracies": 0.612500011920929,
|
||
|
|
"rewards/chosen": -2101.11865234375,
|
||
|
|
"rewards/margins": 552.28125,
|
||
|
|
"rewards/rejected": -2653.13134765625,
|
||
|
|
"rewards/weighted_accuracies": 0.625,
|
||
|
|
"rewards/weighted_chosen": -4.060351371765137,
|
||
|
|
"rewards/weighted_margins": 0.4820556640625,
|
||
|
|
"rewards/weighted_rejected": -4.539941310882568,
|
||
|
|
"step": 870
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4715338245144005,
|
||
|
|
"grad_norm": 58.134342193603516,
|
||
|
|
"learning_rate": 6.3661988065096e-07,
|
||
|
|
"logits/chosen": -4.554785251617432,
|
||
|
|
"logits/rejected": -4.533789157867432,
|
||
|
|
"logps/chosen": -1841.206298828125,
|
||
|
|
"logps/rejected": -2302.1875,
|
||
|
|
"logps/weighted_chosen": -3.680468797683716,
|
||
|
|
"logps/weighted_rejected": -4.158984184265137,
|
||
|
|
"loss": 0.6052,
|
||
|
|
"rewards/accuracies": 0.6625000238418579,
|
||
|
|
"rewards/chosen": -1644.3125,
|
||
|
|
"rewards/margins": 452.23748779296875,
|
||
|
|
"rewards/rejected": -2097.24365234375,
|
||
|
|
"rewards/weighted_accuracies": 0.659375011920929,
|
||
|
|
"rewards/weighted_chosen": -3.228564500808716,
|
||
|
|
"rewards/weighted_margins": 0.45640867948532104,
|
||
|
|
"rewards/weighted_rejected": -3.6844725608825684,
|
||
|
|
"step": 880
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4768921634293369,
|
||
|
|
"grad_norm": 47.12223815917969,
|
||
|
|
"learning_rate": 6.276023446318213e-07,
|
||
|
|
"logits/chosen": -4.478906154632568,
|
||
|
|
"logits/rejected": -4.466796875,
|
||
|
|
"logps/chosen": -2029.737548828125,
|
||
|
|
"logps/rejected": -2337.10009765625,
|
||
|
|
"logps/weighted_chosen": -3.8642578125,
|
||
|
|
"logps/weighted_rejected": -4.384863376617432,
|
||
|
|
"loss": 0.5771,
|
||
|
|
"rewards/accuracies": 0.653124988079071,
|
||
|
|
"rewards/chosen": -1830.9156494140625,
|
||
|
|
"rewards/margins": 296.99530029296875,
|
||
|
|
"rewards/rejected": -2128.512451171875,
|
||
|
|
"rewards/weighted_accuracies": 0.684374988079071,
|
||
|
|
"rewards/weighted_chosen": -3.408935546875,
|
||
|
|
"rewards/weighted_margins": 0.4983886778354645,
|
||
|
|
"rewards/weighted_rejected": -3.9090819358825684,
|
||
|
|
"step": 890
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4822505023442733,
|
||
|
|
"grad_norm": 67.17890930175781,
|
||
|
|
"learning_rate": 6.185401888577487e-07,
|
||
|
|
"logits/chosen": -4.597363471984863,
|
||
|
|
"logits/rejected": -4.523828029632568,
|
||
|
|
"logps/chosen": -2203.487548828125,
|
||
|
|
"logps/rejected": -3171.262451171875,
|
||
|
|
"logps/weighted_chosen": -4.45849609375,
|
||
|
|
"logps/weighted_rejected": -5.061327934265137,
|
||
|
|
"loss": 0.5889,
|
||
|
|
"rewards/accuracies": 0.621874988079071,
|
||
|
|
"rewards/chosen": -2014.4625244140625,
|
||
|
|
"rewards/margins": 938.859375,
|
||
|
|
"rewards/rejected": -2952.96875,
|
||
|
|
"rewards/weighted_accuracies": 0.653124988079071,
|
||
|
|
"rewards/weighted_chosen": -4.013134956359863,
|
||
|
|
"rewards/weighted_margins": 0.587145984172821,
|
||
|
|
"rewards/weighted_rejected": -4.601904392242432,
|
||
|
|
"step": 900
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48760884125920967,
|
||
|
|
"grad_norm": 56.467411041259766,
|
||
|
|
"learning_rate": 6.094365821668307e-07,
|
||
|
|
"logits/chosen": -4.489843845367432,
|
||
|
|
"logits/rejected": -4.480322360992432,
|
||
|
|
"logps/chosen": -1832.862548828125,
|
||
|
|
"logps/rejected": -2382.074951171875,
|
||
|
|
"logps/weighted_chosen": -3.7759766578674316,
|
||
|
|
"logps/weighted_rejected": -4.449414253234863,
|
||
|
|
"loss": 0.5499,
|
||
|
|
"rewards/accuracies": 0.65625,
|
||
|
|
"rewards/chosen": -1650.199951171875,
|
||
|
|
"rewards/margins": 529.1953125,
|
||
|
|
"rewards/rejected": -2178.621826171875,
|
||
|
|
"rewards/weighted_accuracies": 0.7437499761581421,
|
||
|
|
"rewards/weighted_chosen": -3.3441405296325684,
|
||
|
|
"rewards/weighted_margins": 0.637280285358429,
|
||
|
|
"rewards/weighted_rejected": -3.982177734375,
|
||
|
|
"step": 910
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49296718017414604,
|
||
|
|
"grad_norm": 53.93306350708008,
|
||
|
|
"learning_rate": 6.002947078916364e-07,
|
||
|
|
"logits/chosen": -4.535009860992432,
|
||
|
|
"logits/rejected": -4.543847560882568,
|
||
|
|
"logps/chosen": -1841.824951171875,
|
||
|
|
"logps/rejected": -2362.90625,
|
||
|
|
"logps/weighted_chosen": -3.4969725608825684,
|
||
|
|
"logps/weighted_rejected": -3.9888672828674316,
|
||
|
|
"loss": 0.6163,
|
||
|
|
"rewards/accuracies": 0.6343749761581421,
|
||
|
|
"rewards/chosen": -1628.5625,
|
||
|
|
"rewards/margins": 496.4624938964844,
|
||
|
|
"rewards/rejected": -2125.612548828125,
|
||
|
|
"rewards/weighted_accuracies": 0.6781250238418579,
|
||
|
|
"rewards/weighted_chosen": -3.064746141433716,
|
||
|
|
"rewards/weighted_margins": 0.46923828125,
|
||
|
|
"rewards/weighted_rejected": -3.53564453125,
|
||
|
|
"step": 920
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4983255190890824,
|
||
|
|
"grad_norm": 38.40058898925781,
|
||
|
|
"learning_rate": 5.911177627460738e-07,
|
||
|
|
"logits/chosen": -4.379052639007568,
|
||
|
|
"logits/rejected": -4.362548828125,
|
||
|
|
"logps/chosen": -1661.074951171875,
|
||
|
|
"logps/rejected": -2101.175048828125,
|
||
|
|
"logps/weighted_chosen": -3.29833984375,
|
||
|
|
"logps/weighted_rejected": -3.7914061546325684,
|
||
|
|
"loss": 0.6162,
|
||
|
|
"rewards/accuracies": 0.606249988079071,
|
||
|
|
"rewards/chosen": -1442.7750244140625,
|
||
|
|
"rewards/margins": 429.57342529296875,
|
||
|
|
"rewards/rejected": -1872.356201171875,
|
||
|
|
"rewards/weighted_accuracies": 0.659375011920929,
|
||
|
|
"rewards/weighted_chosen": -2.8355956077575684,
|
||
|
|
"rewards/weighted_margins": 0.46900635957717896,
|
||
|
|
"rewards/weighted_rejected": -3.3045411109924316,
|
||
|
|
"step": 930
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5036838580040187,
|
||
|
|
"grad_norm": 62.751251220703125,
|
||
|
|
"learning_rate": 5.819089557075688e-07,
|
||
|
|
"logits/chosen": -4.216992378234863,
|
||
|
|
"logits/rejected": -4.262597560882568,
|
||
|
|
"logps/chosen": -1618.6812744140625,
|
||
|
|
"logps/rejected": -2067.5,
|
||
|
|
"logps/weighted_chosen": -3.3895020484924316,
|
||
|
|
"logps/weighted_rejected": -4.025586128234863,
|
||
|
|
"loss": 0.5534,
|
||
|
|
"rewards/accuracies": 0.6468750238418579,
|
||
|
|
"rewards/chosen": -1416.8843994140625,
|
||
|
|
"rewards/margins": 434.48748779296875,
|
||
|
|
"rewards/rejected": -1851.956298828125,
|
||
|
|
"rewards/weighted_accuracies": 0.7250000238418579,
|
||
|
|
"rewards/weighted_chosen": -2.9775390625,
|
||
|
|
"rewards/weighted_margins": 0.608203113079071,
|
||
|
|
"rewards/weighted_rejected": -3.585742235183716,
|
||
|
|
"step": 940
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5090421969189551,
|
||
|
|
"grad_norm": 129.74171447753906,
|
||
|
|
"learning_rate": 5.726715068949564e-07,
|
||
|
|
"logits/chosen": -4.279882907867432,
|
||
|
|
"logits/rejected": -4.284081935882568,
|
||
|
|
"logps/chosen": -2506.074951171875,
|
||
|
|
"logps/rejected": -2878.94384765625,
|
||
|
|
"logps/weighted_chosen": -5.016894340515137,
|
||
|
|
"logps/weighted_rejected": -5.711621284484863,
|
||
|
|
"loss": 0.5989,
|
||
|
|
"rewards/accuracies": 0.6156250238418579,
|
||
|
|
"rewards/chosen": -2290.46875,
|
||
|
|
"rewards/margins": 367.1187438964844,
|
||
|
|
"rewards/rejected": -2656.796875,
|
||
|
|
"rewards/weighted_accuracies": 0.6812499761581421,
|
||
|
|
"rewards/weighted_chosen": -4.538281440734863,
|
||
|
|
"rewards/weighted_margins": 0.670092761516571,
|
||
|
|
"rewards/weighted_rejected": -5.206250190734863,
|
||
|
|
"step": 950
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5144005358338914,
|
||
|
|
"grad_norm": 55.60570526123047,
|
||
|
|
"learning_rate": 5.634086464424742e-07,
|
||
|
|
"logits/chosen": -4.012402534484863,
|
||
|
|
"logits/rejected": -3.98974609375,
|
||
|
|
"logps/chosen": -2225.10009765625,
|
||
|
|
"logps/rejected": -2796.268798828125,
|
||
|
|
"logps/weighted_chosen": -4.904882907867432,
|
||
|
|
"logps/weighted_rejected": -5.526464939117432,
|
||
|
|
"loss": 0.5626,
|
||
|
|
"rewards/accuracies": 0.6187499761581421,
|
||
|
|
"rewards/chosen": -2037.143798828125,
|
||
|
|
"rewards/margins": 551.8984375,
|
||
|
|
"rewards/rejected": -2589.043701171875,
|
||
|
|
"rewards/weighted_accuracies": 0.71875,
|
||
|
|
"rewards/weighted_chosen": -4.45361328125,
|
||
|
|
"rewards/weighted_margins": 0.5914306640625,
|
||
|
|
"rewards/weighted_rejected": -5.044140815734863,
|
||
|
|
"step": 960
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5197588747488279,
|
||
|
|
"grad_norm": 47.078128814697266,
|
||
|
|
"learning_rate": 5.54123613370256e-07,
|
||
|
|
"logits/chosen": -3.900390625,
|
||
|
|
"logits/rejected": -3.8998045921325684,
|
||
|
|
"logps/chosen": -1623.7125244140625,
|
||
|
|
"logps/rejected": -1898.550048828125,
|
||
|
|
"logps/weighted_chosen": -3.878124952316284,
|
||
|
|
"logps/weighted_rejected": -4.330664157867432,
|
||
|
|
"loss": 0.582,
|
||
|
|
"rewards/accuracies": 0.609375,
|
||
|
|
"rewards/chosen": -1428.956298828125,
|
||
|
|
"rewards/margins": 268.56561279296875,
|
||
|
|
"rewards/rejected": -1697.0875244140625,
|
||
|
|
"rewards/weighted_accuracies": 0.684374988079071,
|
||
|
|
"rewards/weighted_chosen": -3.447314500808716,
|
||
|
|
"rewards/weighted_margins": 0.4397216737270355,
|
||
|
|
"rewards/weighted_rejected": -3.887988328933716,
|
||
|
|
"step": 970
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5251172136637642,
|
||
|
|
"grad_norm": 67.22262573242188,
|
||
|
|
"learning_rate": 5.448196544517167e-07,
|
||
|
|
"logits/chosen": -3.939257860183716,
|
||
|
|
"logits/rejected": -3.950000047683716,
|
||
|
|
"logps/chosen": -1683.112548828125,
|
||
|
|
"logps/rejected": -2141.637451171875,
|
||
|
|
"logps/weighted_chosen": -4.157031059265137,
|
||
|
|
"logps/weighted_rejected": -4.748242378234863,
|
||
|
|
"loss": 0.6054,
|
||
|
|
"rewards/accuracies": 0.6343749761581421,
|
||
|
|
"rewards/chosen": -1484.3062744140625,
|
||
|
|
"rewards/margins": 448.54998779296875,
|
||
|
|
"rewards/rejected": -1932.387451171875,
|
||
|
|
"rewards/weighted_accuracies": 0.6156250238418579,
|
||
|
|
"rewards/weighted_chosen": -3.728564500808716,
|
||
|
|
"rewards/weighted_margins": 0.5644286870956421,
|
||
|
|
"rewards/weighted_rejected": -4.292675971984863,
|
||
|
|
"step": 980
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5304755525787006,
|
||
|
|
"grad_norm": 62.878604888916016,
|
||
|
|
"learning_rate": 5.355000230782267e-07,
|
||
|
|
"logits/chosen": -3.724902391433716,
|
||
|
|
"logits/rejected": -3.7308592796325684,
|
||
|
|
"logps/chosen": -1693.425048828125,
|
||
|
|
"logps/rejected": -2090.09375,
|
||
|
|
"logps/weighted_chosen": -4.06298828125,
|
||
|
|
"logps/weighted_rejected": -4.559179782867432,
|
||
|
|
"loss": 0.6163,
|
||
|
|
"rewards/accuracies": 0.606249988079071,
|
||
|
|
"rewards/chosen": -1488.6500244140625,
|
||
|
|
"rewards/margins": 378.28125,
|
||
|
|
"rewards/rejected": -1866.75,
|
||
|
|
"rewards/weighted_accuracies": 0.671875,
|
||
|
|
"rewards/weighted_chosen": -3.603222608566284,
|
||
|
|
"rewards/weighted_margins": 0.4825805723667145,
|
||
|
|
"rewards/weighted_rejected": -4.086816310882568,
|
||
|
|
"step": 990
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.535833891493637,
|
||
|
|
"grad_norm": 132.68304443359375,
|
||
|
|
"learning_rate": 5.26167978121472e-07,
|
||
|
|
"logits/chosen": -3.9014649391174316,
|
||
|
|
"logits/rejected": -3.8974609375,
|
||
|
|
"logps/chosen": -1748.393798828125,
|
||
|
|
"logps/rejected": -2370.02490234375,
|
||
|
|
"logps/weighted_chosen": -3.965136766433716,
|
||
|
|
"logps/weighted_rejected": -4.65234375,
|
||
|
|
"loss": 0.5721,
|
||
|
|
"rewards/accuracies": 0.6625000238418579,
|
||
|
|
"rewards/chosen": -1539.40625,
|
||
|
|
"rewards/margins": 598.4609375,
|
||
|
|
"rewards/rejected": -2137.625,
|
||
|
|
"rewards/weighted_accuracies": 0.690625011920929,
|
||
|
|
"rewards/weighted_chosen": -3.532958984375,
|
||
|
|
"rewards/weighted_margins": 0.640454113483429,
|
||
|
|
"rewards/weighted_rejected": -4.173095703125,
|
||
|
|
"step": 1000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.535833891493637,
|
||
|
|
"eval_logits/chosen": -4.246370315551758,
|
||
|
|
"eval_logits/rejected": -4.254964828491211,
|
||
|
|
"eval_logps/chosen": -2158.052001953125,
|
||
|
|
"eval_logps/rejected": -2603.586181640625,
|
||
|
|
"eval_logps/weighted_chosen": -4.515809059143066,
|
||
|
|
"eval_logps/weighted_rejected": -5.166130542755127,
|
||
|
|
"eval_loss": 0.6049003601074219,
|
||
|
|
"eval_rewards/accuracies": 0.6298397779464722,
|
||
|
|
"eval_rewards/chosen": -1955.2657470703125,
|
||
|
|
"eval_rewards/margins": 431.65203857421875,
|
||
|
|
"eval_rewards/rejected": -2386.94921875,
|
||
|
|
"eval_rewards/weighted_accuracies": 0.6779038906097412,
|
||
|
|
"eval_rewards/weighted_chosen": -4.057965278625488,
|
||
|
|
"eval_rewards/weighted_margins": 0.6343361735343933,
|
||
|
|
"eval_rewards/weighted_rejected": -4.692301273345947,
|
||
|
|
"eval_runtime": 2385.1831,
|
||
|
|
"eval_samples_per_second": 1.255,
|
||
|
|
"eval_steps_per_second": 0.314,
|
||
|
|
"step": 1000
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 10,
|
||
|
|
"max_steps": 1867,
|
||
|
|
"num_input_tokens_seen": 0,
|
||
|
|
"num_train_epochs": 1,
|
||
|
|
"save_steps": 500,
|
||
|
|
"stateful_callbacks": {
|
||
|
|
"TrainerControl": {
|
||
|
|
"args": {
|
||
|
|
"should_epoch_stop": false,
|
||
|
|
"should_evaluate": false,
|
||
|
|
"should_log": false,
|
||
|
|
"should_save": true,
|
||
|
|
"should_training_stop": false
|
||
|
|
},
|
||
|
|
"attributes": {}
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"total_flos": 0.0,
|
||
|
|
"train_batch_size": 1,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|