1128 lines
44 KiB
JSON
1128 lines
44 KiB
JSON
|
{
|
||
|
|
"best_global_step": null,
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 0.2679169457468185,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 500,
|
||
|
|
"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
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"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
|
||
|
|
}
|