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": 31.966846466064453,
|
||
|
|
"learning_rate": 0.0,
|
||
|
|
"logits/chosen": -2.52783203125,
|
||
|
|
"logits/rejected": -2.498046875,
|
||
|
|
"logps/chosen": -277.71875,
|
||
|
|
"logps/rejected": -223.125,
|
||
|
|
"logps/weighted_chosen": -0.533447265625,
|
||
|
|
"logps/weighted_rejected": -0.5252685546875,
|
||
|
|
"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": 31.4305419921875,
|
||
|
|
"learning_rate": 4.81283422459893e-08,
|
||
|
|
"logits/chosen": -2.42626953125,
|
||
|
|
"logits/rejected": -2.3970539569854736,
|
||
|
|
"logps/chosen": -195.6822967529297,
|
||
|
|
"logps/rejected": -195.890625,
|
||
|
|
"logps/weighted_chosen": -0.5433688759803772,
|
||
|
|
"logps/weighted_rejected": -0.5479600429534912,
|
||
|
|
"loss": 0.6923,
|
||
|
|
"rewards/accuracies": 0.2951388955116272,
|
||
|
|
"rewards/chosen": -0.0735677108168602,
|
||
|
|
"rewards/margins": 0.00434027798473835,
|
||
|
|
"rewards/rejected": -0.0779079869389534,
|
||
|
|
"rewards/weighted_accuracies": 0.34375,
|
||
|
|
"rewards/weighted_chosen": 0.00025431314134038985,
|
||
|
|
"rewards/weighted_margins": 0.0005137125845067203,
|
||
|
|
"rewards/weighted_rejected": -0.0002593994140625,
|
||
|
|
"step": 10
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.010716677829872739,
|
||
|
|
"grad_norm": 17.30971908569336,
|
||
|
|
"learning_rate": 1.0160427807486631e-07,
|
||
|
|
"logits/chosen": -2.400830030441284,
|
||
|
|
"logits/rejected": -2.399658203125,
|
||
|
|
"logps/chosen": -199.91366577148438,
|
||
|
|
"logps/rejected": -202.21328735351562,
|
||
|
|
"logps/weighted_chosen": -0.537811279296875,
|
||
|
|
"logps/weighted_rejected": -0.548992931842804,
|
||
|
|
"loss": 0.6926,
|
||
|
|
"rewards/accuracies": 0.359375,
|
||
|
|
"rewards/chosen": -0.04570312425494194,
|
||
|
|
"rewards/margins": 0.02001953125,
|
||
|
|
"rewards/rejected": -0.06572265923023224,
|
||
|
|
"rewards/weighted_accuracies": 0.3656249940395355,
|
||
|
|
"rewards/weighted_chosen": -0.00011291504051769152,
|
||
|
|
"rewards/weighted_margins": 0.00011787414405262098,
|
||
|
|
"rewards/weighted_rejected": -0.00023076534853316844,
|
||
|
|
"step": 20
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.016075016744809108,
|
||
|
|
"grad_norm": 30.744884490966797,
|
||
|
|
"learning_rate": 1.5508021390374333e-07,
|
||
|
|
"logits/chosen": -2.3249268531799316,
|
||
|
|
"logits/rejected": -2.299609422683716,
|
||
|
|
"logps/chosen": -198.44375610351562,
|
||
|
|
"logps/rejected": -212.16641235351562,
|
||
|
|
"logps/weighted_chosen": -0.555804431438446,
|
||
|
|
"logps/weighted_rejected": -0.547528088092804,
|
||
|
|
"loss": 0.6927,
|
||
|
|
"rewards/accuracies": 0.4000000059604645,
|
||
|
|
"rewards/chosen": 0.0771484375,
|
||
|
|
"rewards/margins": 0.18007811903953552,
|
||
|
|
"rewards/rejected": -0.10292968899011612,
|
||
|
|
"rewards/weighted_accuracies": 0.3656249940395355,
|
||
|
|
"rewards/weighted_chosen": 0.0004203796270303428,
|
||
|
|
"rewards/weighted_margins": -0.0001464843808207661,
|
||
|
|
"rewards/weighted_rejected": 0.0005667686345987022,
|
||
|
|
"step": 30
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.021433355659745478,
|
||
|
|
"grad_norm": 51.52167510986328,
|
||
|
|
"learning_rate": 2.085561497326203e-07,
|
||
|
|
"logits/chosen": -2.3595948219299316,
|
||
|
|
"logits/rejected": -2.375,
|
||
|
|
"logps/chosen": -192.6171875,
|
||
|
|
"logps/rejected": -207.1945343017578,
|
||
|
|
"logps/weighted_chosen": -0.544903576374054,
|
||
|
|
"logps/weighted_rejected": -0.588610827922821,
|
||
|
|
"loss": 0.6926,
|
||
|
|
"rewards/accuracies": 0.31562501192092896,
|
||
|
|
"rewards/chosen": -0.08066406100988388,
|
||
|
|
"rewards/margins": -0.11210937798023224,
|
||
|
|
"rewards/rejected": 0.03144531324505806,
|
||
|
|
"rewards/weighted_accuracies": 0.40937501192092896,
|
||
|
|
"rewards/weighted_chosen": 0.0021041869185864925,
|
||
|
|
"rewards/weighted_margins": 0.0001922607480082661,
|
||
|
|
"rewards/weighted_rejected": 0.0019119263160973787,
|
||
|
|
"step": 40
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.02679169457468185,
|
||
|
|
"grad_norm": 29.50680160522461,
|
||
|
|
"learning_rate": 2.620320855614973e-07,
|
||
|
|
"logits/chosen": -2.3987059593200684,
|
||
|
|
"logits/rejected": -2.3811278343200684,
|
||
|
|
"logps/chosen": -213.0945281982422,
|
||
|
|
"logps/rejected": -228.8328094482422,
|
||
|
|
"logps/weighted_chosen": -0.5421508550643921,
|
||
|
|
"logps/weighted_rejected": -0.557452380657196,
|
||
|
|
"loss": 0.6925,
|
||
|
|
"rewards/accuracies": 0.390625,
|
||
|
|
"rewards/chosen": 0.06367187201976776,
|
||
|
|
"rewards/margins": 0.06484375149011612,
|
||
|
|
"rewards/rejected": -0.0011718750465661287,
|
||
|
|
"rewards/weighted_accuracies": 0.45625001192092896,
|
||
|
|
"rewards/weighted_chosen": 0.0056938170455396175,
|
||
|
|
"rewards/weighted_margins": 0.000339508056640625,
|
||
|
|
"rewards/weighted_rejected": 0.005351257510483265,
|
||
|
|
"step": 50
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.032150033489618215,
|
||
|
|
"grad_norm": 30.643386840820312,
|
||
|
|
"learning_rate": 3.155080213903743e-07,
|
||
|
|
"logits/chosen": -2.347216844558716,
|
||
|
|
"logits/rejected": -2.3227295875549316,
|
||
|
|
"logps/chosen": -211.7421875,
|
||
|
|
"logps/rejected": -218.0234375,
|
||
|
|
"logps/weighted_chosen": -0.536822497844696,
|
||
|
|
"logps/weighted_rejected": -0.568676769733429,
|
||
|
|
"loss": 0.6923,
|
||
|
|
"rewards/accuracies": 0.421875,
|
||
|
|
"rewards/chosen": 0.09257812798023224,
|
||
|
|
"rewards/margins": 0.10820312798023224,
|
||
|
|
"rewards/rejected": -0.015625,
|
||
|
|
"rewards/weighted_accuracies": 0.45625001192092896,
|
||
|
|
"rewards/weighted_chosen": 0.009661102667450905,
|
||
|
|
"rewards/weighted_margins": 0.000804901123046875,
|
||
|
|
"rewards/weighted_rejected": 0.00885620154440403,
|
||
|
|
"step": 60
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03750837240455459,
|
||
|
|
"grad_norm": 27.744924545288086,
|
||
|
|
"learning_rate": 3.689839572192513e-07,
|
||
|
|
"logits/chosen": -2.442822217941284,
|
||
|
|
"logits/rejected": -2.4500975608825684,
|
||
|
|
"logps/chosen": -194.47891235351562,
|
||
|
|
"logps/rejected": -200.30624389648438,
|
||
|
|
"logps/weighted_chosen": -0.5030456781387329,
|
||
|
|
"logps/weighted_rejected": -0.5135132074356079,
|
||
|
|
"loss": 0.6906,
|
||
|
|
"rewards/accuracies": 0.4312500059604645,
|
||
|
|
"rewards/chosen": 0.03203124925494194,
|
||
|
|
"rewards/margins": 0.20937499403953552,
|
||
|
|
"rewards/rejected": -0.17734375596046448,
|
||
|
|
"rewards/weighted_accuracies": 0.5218750238418579,
|
||
|
|
"rewards/weighted_chosen": 0.01783904992043972,
|
||
|
|
"rewards/weighted_margins": 0.0045639039017260075,
|
||
|
|
"rewards/weighted_rejected": 0.013275146484375,
|
||
|
|
"step": 70
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.042866711319490956,
|
||
|
|
"grad_norm": 38.061683654785156,
|
||
|
|
"learning_rate": 4.2245989304812833e-07,
|
||
|
|
"logits/chosen": -2.403271436691284,
|
||
|
|
"logits/rejected": -2.430908203125,
|
||
|
|
"logps/chosen": -219.08438110351562,
|
||
|
|
"logps/rejected": -239.91561889648438,
|
||
|
|
"logps/weighted_chosen": -0.529132068157196,
|
||
|
|
"logps/weighted_rejected": -0.5560547113418579,
|
||
|
|
"loss": 0.6903,
|
||
|
|
"rewards/accuracies": 0.43437498807907104,
|
||
|
|
"rewards/chosen": 0.29023438692092896,
|
||
|
|
"rewards/margins": 0.27910155057907104,
|
||
|
|
"rewards/rejected": 0.01113281212747097,
|
||
|
|
"rewards/weighted_accuracies": 0.518750011920929,
|
||
|
|
"rewards/weighted_chosen": 0.022306060418486595,
|
||
|
|
"rewards/weighted_margins": 0.005640792660415173,
|
||
|
|
"rewards/weighted_rejected": 0.016665268689393997,
|
||
|
|
"step": 80
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04822505023442733,
|
||
|
|
"grad_norm": 28.250593185424805,
|
||
|
|
"learning_rate": 4.7593582887700533e-07,
|
||
|
|
"logits/chosen": -2.392260789871216,
|
||
|
|
"logits/rejected": -2.4065184593200684,
|
||
|
|
"logps/chosen": -195.1953125,
|
||
|
|
"logps/rejected": -207.7804718017578,
|
||
|
|
"logps/weighted_chosen": -0.51409912109375,
|
||
|
|
"logps/weighted_rejected": -0.526226818561554,
|
||
|
|
"loss": 0.6906,
|
||
|
|
"rewards/accuracies": 0.5249999761581421,
|
||
|
|
"rewards/chosen": 0.10097656399011612,
|
||
|
|
"rewards/margins": 0.724414050579071,
|
||
|
|
"rewards/rejected": -0.6234375238418579,
|
||
|
|
"rewards/weighted_accuracies": 0.559374988079071,
|
||
|
|
"rewards/weighted_chosen": 0.025403594598174095,
|
||
|
|
"rewards/weighted_margins": 0.006229400634765625,
|
||
|
|
"rewards/weighted_rejected": 0.01916809007525444,
|
||
|
|
"step": 90
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0535833891493637,
|
||
|
|
"grad_norm": 18.92244529724121,
|
||
|
|
"learning_rate": 5.294117647058823e-07,
|
||
|
|
"logits/chosen": -2.3767333030700684,
|
||
|
|
"logits/rejected": -2.3611817359924316,
|
||
|
|
"logps/chosen": -224.2859344482422,
|
||
|
|
"logps/rejected": -229.6570281982422,
|
||
|
|
"logps/weighted_chosen": -0.534497082233429,
|
||
|
|
"logps/weighted_rejected": -0.5674392580986023,
|
||
|
|
"loss": 0.6903,
|
||
|
|
"rewards/accuracies": 0.5093749761581421,
|
||
|
|
"rewards/chosen": -0.8511718511581421,
|
||
|
|
"rewards/margins": 1.123437523841858,
|
||
|
|
"rewards/rejected": -1.974609375,
|
||
|
|
"rewards/weighted_accuracies": 0.53125,
|
||
|
|
"rewards/weighted_chosen": 0.023969268426299095,
|
||
|
|
"rewards/weighted_margins": 0.00682754535228014,
|
||
|
|
"rewards/weighted_rejected": 0.01714172400534153,
|
||
|
|
"step": 100
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.058941728064300064,
|
||
|
|
"grad_norm": 50.693965911865234,
|
||
|
|
"learning_rate": 5.828877005347593e-07,
|
||
|
|
"logits/chosen": -2.3811888694763184,
|
||
|
|
"logits/rejected": -2.382458448410034,
|
||
|
|
"logps/chosen": -217.30703735351562,
|
||
|
|
"logps/rejected": -225.4343719482422,
|
||
|
|
"logps/weighted_chosen": -0.54705810546875,
|
||
|
|
"logps/weighted_rejected": -0.5594238042831421,
|
||
|
|
"loss": 0.6951,
|
||
|
|
"rewards/accuracies": 0.46562498807907104,
|
||
|
|
"rewards/chosen": -2.3193359375,
|
||
|
|
"rewards/margins": 0.624218761920929,
|
||
|
|
"rewards/rejected": -2.942578077316284,
|
||
|
|
"rewards/weighted_accuracies": 0.4781250059604645,
|
||
|
|
"rewards/weighted_chosen": 0.022613525390625,
|
||
|
|
"rewards/weighted_margins": -0.0009094238048419356,
|
||
|
|
"rewards/weighted_rejected": 0.02351989783346653,
|
||
|
|
"step": 110
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06430006697923643,
|
||
|
|
"grad_norm": 27.526296615600586,
|
||
|
|
"learning_rate": 6.363636363636363e-07,
|
||
|
|
"logits/chosen": -2.459423780441284,
|
||
|
|
"logits/rejected": -2.4720702171325684,
|
||
|
|
"logps/chosen": -187.80313110351562,
|
||
|
|
"logps/rejected": -226.0906219482422,
|
||
|
|
"logps/weighted_chosen": -0.533001720905304,
|
||
|
|
"logps/weighted_rejected": -0.605639636516571,
|
||
|
|
"loss": 0.6903,
|
||
|
|
"rewards/accuracies": 0.49687498807907104,
|
||
|
|
"rewards/chosen": -3.5208983421325684,
|
||
|
|
"rewards/margins": 1.0193359851837158,
|
||
|
|
"rewards/rejected": -4.540820121765137,
|
||
|
|
"rewards/weighted_accuracies": 0.5062500238418579,
|
||
|
|
"rewards/weighted_chosen": 0.005961227230727673,
|
||
|
|
"rewards/weighted_margins": 0.009808349423110485,
|
||
|
|
"rewards/weighted_rejected": -0.0038513182662427425,
|
||
|
|
"step": 120
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06965840589417281,
|
||
|
|
"grad_norm": 26.822662353515625,
|
||
|
|
"learning_rate": 6.898395721925134e-07,
|
||
|
|
"logits/chosen": -2.565673828125,
|
||
|
|
"logits/rejected": -2.564648389816284,
|
||
|
|
"logps/chosen": -231.10000610351562,
|
||
|
|
"logps/rejected": -240.86093139648438,
|
||
|
|
"logps/weighted_chosen": -0.537158191204071,
|
||
|
|
"logps/weighted_rejected": -0.5712524652481079,
|
||
|
|
"loss": 0.6953,
|
||
|
|
"rewards/accuracies": 0.53125,
|
||
|
|
"rewards/chosen": -2.906054735183716,
|
||
|
|
"rewards/margins": 1.6062500476837158,
|
||
|
|
"rewards/rejected": -4.512109279632568,
|
||
|
|
"rewards/weighted_accuracies": 0.49687498807907104,
|
||
|
|
"rewards/weighted_chosen": 0.00572967529296875,
|
||
|
|
"rewards/weighted_margins": -0.00041961669921875,
|
||
|
|
"rewards/weighted_rejected": 0.0061210631392896175,
|
||
|
|
"step": 130
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07501674480910918,
|
||
|
|
"grad_norm": 19.93072509765625,
|
||
|
|
"learning_rate": 7.433155080213903e-07,
|
||
|
|
"logits/chosen": -2.53515625,
|
||
|
|
"logits/rejected": -2.5311522483825684,
|
||
|
|
"logps/chosen": -198.80313110351562,
|
||
|
|
"logps/rejected": -211.29061889648438,
|
||
|
|
"logps/weighted_chosen": -0.5693206787109375,
|
||
|
|
"logps/weighted_rejected": -0.592010498046875,
|
||
|
|
"loss": 0.694,
|
||
|
|
"rewards/accuracies": 0.543749988079071,
|
||
|
|
"rewards/chosen": -4.0048828125,
|
||
|
|
"rewards/margins": 2.3238282203674316,
|
||
|
|
"rewards/rejected": -6.327929496765137,
|
||
|
|
"rewards/weighted_accuracies": 0.518750011920929,
|
||
|
|
"rewards/weighted_chosen": -0.017972564324736595,
|
||
|
|
"rewards/weighted_margins": 0.005100250244140625,
|
||
|
|
"rewards/weighted_rejected": -0.02306365966796875,
|
||
|
|
"step": 140
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08037508372404555,
|
||
|
|
"grad_norm": 26.39690399169922,
|
||
|
|
"learning_rate": 7.967914438502673e-07,
|
||
|
|
"logits/chosen": -2.509326219558716,
|
||
|
|
"logits/rejected": -2.508056640625,
|
||
|
|
"logps/chosen": -214.25,
|
||
|
|
"logps/rejected": -237.84530639648438,
|
||
|
|
"logps/weighted_chosen": -0.5889526605606079,
|
||
|
|
"logps/weighted_rejected": -0.585980236530304,
|
||
|
|
"loss": 0.7007,
|
||
|
|
"rewards/accuracies": 0.546875,
|
||
|
|
"rewards/chosen": -3.867382764816284,
|
||
|
|
"rewards/margins": 2.546581983566284,
|
||
|
|
"rewards/rejected": -6.4140625,
|
||
|
|
"rewards/weighted_accuracies": 0.550000011920929,
|
||
|
|
"rewards/weighted_chosen": -0.041257478296756744,
|
||
|
|
"rewards/weighted_margins": -0.005215453915297985,
|
||
|
|
"rewards/weighted_rejected": -0.03600044175982475,
|
||
|
|
"step": 150
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08573342263898191,
|
||
|
|
"grad_norm": 25.74460792541504,
|
||
|
|
"learning_rate": 8.502673796791443e-07,
|
||
|
|
"logits/chosen": -2.433154344558716,
|
||
|
|
"logits/rejected": -2.414794921875,
|
||
|
|
"logps/chosen": -200.64999389648438,
|
||
|
|
"logps/rejected": -229.2624969482422,
|
||
|
|
"logps/weighted_chosen": -0.589752197265625,
|
||
|
|
"logps/weighted_rejected": -0.612139880657196,
|
||
|
|
"loss": 0.6933,
|
||
|
|
"rewards/accuracies": 0.5406249761581421,
|
||
|
|
"rewards/chosen": -6.800390720367432,
|
||
|
|
"rewards/margins": 0.5577148199081421,
|
||
|
|
"rewards/rejected": -7.36279296875,
|
||
|
|
"rewards/weighted_accuracies": 0.5218750238418579,
|
||
|
|
"rewards/weighted_chosen": -0.05205688625574112,
|
||
|
|
"rewards/weighted_margins": 0.006626891903579235,
|
||
|
|
"rewards/weighted_rejected": -0.05869255214929581,
|
||
|
|
"step": 160
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09109176155391828,
|
||
|
|
"grad_norm": 24.530479431152344,
|
||
|
|
"learning_rate": 9.037433155080213e-07,
|
||
|
|
"logits/chosen": -2.5284180641174316,
|
||
|
|
"logits/rejected": -2.5411133766174316,
|
||
|
|
"logps/chosen": -204.92031860351562,
|
||
|
|
"logps/rejected": -217.046875,
|
||
|
|
"logps/weighted_chosen": -0.600720226764679,
|
||
|
|
"logps/weighted_rejected": -0.6278625726699829,
|
||
|
|
"loss": 0.6787,
|
||
|
|
"rewards/accuracies": 0.6031249761581421,
|
||
|
|
"rewards/chosen": -7.408789157867432,
|
||
|
|
"rewards/margins": 4.765625,
|
||
|
|
"rewards/rejected": -12.178515434265137,
|
||
|
|
"rewards/weighted_accuracies": 0.612500011920929,
|
||
|
|
"rewards/weighted_chosen": -0.05335540696978569,
|
||
|
|
"rewards/weighted_margins": 0.03420104831457138,
|
||
|
|
"rewards/weighted_rejected": -0.08757324516773224,
|
||
|
|
"step": 170
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09645010046885466,
|
||
|
|
"grad_norm": 17.329570770263672,
|
||
|
|
"learning_rate": 9.572192513368984e-07,
|
||
|
|
"logits/chosen": -2.638134717941284,
|
||
|
|
"logits/rejected": -2.6302247047424316,
|
||
|
|
"logps/chosen": -219.87655639648438,
|
||
|
|
"logps/rejected": -240.1531219482422,
|
||
|
|
"logps/weighted_chosen": -0.637438952922821,
|
||
|
|
"logps/weighted_rejected": -0.700207531452179,
|
||
|
|
"loss": 0.682,
|
||
|
|
"rewards/accuracies": 0.5843750238418579,
|
||
|
|
"rewards/chosen": -10.272656440734863,
|
||
|
|
"rewards/margins": 4.742578029632568,
|
||
|
|
"rewards/rejected": -15.017578125,
|
||
|
|
"rewards/weighted_accuracies": 0.581250011920929,
|
||
|
|
"rewards/weighted_chosen": -0.09931640326976776,
|
||
|
|
"rewards/weighted_margins": 0.0342864990234375,
|
||
|
|
"rewards/weighted_rejected": -0.13359375298023224,
|
||
|
|
"step": 180
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10180843938379103,
|
||
|
|
"grad_norm": 40.06588363647461,
|
||
|
|
"learning_rate": 9.999965031204306e-07,
|
||
|
|
"logits/chosen": -2.677978515625,
|
||
|
|
"logits/rejected": -2.667724609375,
|
||
|
|
"logps/chosen": -217.265625,
|
||
|
|
"logps/rejected": -217.9968719482422,
|
||
|
|
"logps/weighted_chosen": -0.713732898235321,
|
||
|
|
"logps/weighted_rejected": -0.745983898639679,
|
||
|
|
"loss": 0.6972,
|
||
|
|
"rewards/accuracies": 0.559374988079071,
|
||
|
|
"rewards/chosen": -16.422557830810547,
|
||
|
|
"rewards/margins": 3.5362305641174316,
|
||
|
|
"rewards/rejected": -19.955078125,
|
||
|
|
"rewards/weighted_accuracies": 0.578125,
|
||
|
|
"rewards/weighted_chosen": -0.1743011474609375,
|
||
|
|
"rewards/weighted_margins": 0.01756439171731472,
|
||
|
|
"rewards/weighted_rejected": -0.19183501601219177,
|
||
|
|
"step": 190
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1071667782987274,
|
||
|
|
"grad_norm": 26.449254989624023,
|
||
|
|
"learning_rate": 9.998741174712533e-07,
|
||
|
|
"logits/chosen": -2.758007764816284,
|
||
|
|
"logits/rejected": -2.752197265625,
|
||
|
|
"logps/chosen": -251.4640655517578,
|
||
|
|
"logps/rejected": -266.7734375,
|
||
|
|
"logps/weighted_chosen": -0.721697986125946,
|
||
|
|
"logps/weighted_rejected": -0.7427002191543579,
|
||
|
|
"loss": 0.6956,
|
||
|
|
"rewards/accuracies": 0.543749988079071,
|
||
|
|
"rewards/chosen": -23.8759765625,
|
||
|
|
"rewards/margins": 3.2962889671325684,
|
||
|
|
"rewards/rejected": -27.181835174560547,
|
||
|
|
"rewards/weighted_accuracies": 0.5,
|
||
|
|
"rewards/weighted_chosen": -0.14485931396484375,
|
||
|
|
"rewards/weighted_margins": 0.00913848914206028,
|
||
|
|
"rewards/weighted_rejected": -0.153900146484375,
|
||
|
|
"step": 200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11252511721366376,
|
||
|
|
"grad_norm": 25.405826568603516,
|
||
|
|
"learning_rate": 9.995769367531952e-07,
|
||
|
|
"logits/chosen": -2.8843750953674316,
|
||
|
|
"logits/rejected": -2.8621582984924316,
|
||
|
|
"logps/chosen": -249.39999389648438,
|
||
|
|
"logps/rejected": -249.5625,
|
||
|
|
"logps/weighted_chosen": -0.6254638433456421,
|
||
|
|
"logps/weighted_rejected": -0.677978515625,
|
||
|
|
"loss": 0.6845,
|
||
|
|
"rewards/accuracies": 0.5531250238418579,
|
||
|
|
"rewards/chosen": -31.167186737060547,
|
||
|
|
"rewards/margins": 2.1792969703674316,
|
||
|
|
"rewards/rejected": -33.351173400878906,
|
||
|
|
"rewards/weighted_accuracies": 0.528124988079071,
|
||
|
|
"rewards/weighted_chosen": -0.11136627197265625,
|
||
|
|
"rewards/weighted_margins": 0.02726135216653347,
|
||
|
|
"rewards/weighted_rejected": -0.1386566162109375,
|
||
|
|
"step": 210
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11788345612860013,
|
||
|
|
"grad_norm": 22.92214012145996,
|
||
|
|
"learning_rate": 9.991050648838675e-07,
|
||
|
|
"logits/chosen": -3.027539014816284,
|
||
|
|
"logits/rejected": -3.0234375,
|
||
|
|
"logps/chosen": -249.0265655517578,
|
||
|
|
"logps/rejected": -264.1703186035156,
|
||
|
|
"logps/weighted_chosen": -0.72125244140625,
|
||
|
|
"logps/weighted_rejected": -0.7577880620956421,
|
||
|
|
"loss": 0.6887,
|
||
|
|
"rewards/accuracies": 0.5687500238418579,
|
||
|
|
"rewards/chosen": -44.36796951293945,
|
||
|
|
"rewards/margins": 2.092968702316284,
|
||
|
|
"rewards/rejected": -46.47265625,
|
||
|
|
"rewards/weighted_accuracies": 0.5218750238418579,
|
||
|
|
"rewards/weighted_chosen": -0.15605469048023224,
|
||
|
|
"rewards/weighted_margins": 0.022979736328125,
|
||
|
|
"rewards/weighted_rejected": -0.17911987006664276,
|
||
|
|
"step": 220
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12324179504353651,
|
||
|
|
"grad_norm": 18.913394927978516,
|
||
|
|
"learning_rate": 9.98458666866564e-07,
|
||
|
|
"logits/chosen": -2.95361328125,
|
||
|
|
"logits/rejected": -2.9599609375,
|
||
|
|
"logps/chosen": -252.1765594482422,
|
||
|
|
"logps/rejected": -273.97186279296875,
|
||
|
|
"logps/weighted_chosen": -0.7530273199081421,
|
||
|
|
"logps/weighted_rejected": -0.7951415777206421,
|
||
|
|
"loss": 0.6805,
|
||
|
|
"rewards/accuracies": 0.5874999761581421,
|
||
|
|
"rewards/chosen": -55.478126525878906,
|
||
|
|
"rewards/margins": 9.685155868530273,
|
||
|
|
"rewards/rejected": -65.15702819824219,
|
||
|
|
"rewards/weighted_accuracies": 0.546875,
|
||
|
|
"rewards/weighted_chosen": -0.202239990234375,
|
||
|
|
"rewards/weighted_margins": 0.04668273776769638,
|
||
|
|
"rewards/weighted_rejected": -0.248951718211174,
|
||
|
|
"step": 230
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12860013395847286,
|
||
|
|
"grad_norm": 18.727752685546875,
|
||
|
|
"learning_rate": 9.97637968732563e-07,
|
||
|
|
"logits/chosen": -2.9483885765075684,
|
||
|
|
"logits/rejected": -2.934033155441284,
|
||
|
|
"logps/chosen": -264.1937561035156,
|
||
|
|
"logps/rejected": -281.26873779296875,
|
||
|
|
"logps/weighted_chosen": -0.7586425542831421,
|
||
|
|
"logps/weighted_rejected": -0.777172863483429,
|
||
|
|
"loss": 0.7011,
|
||
|
|
"rewards/accuracies": 0.565625011920929,
|
||
|
|
"rewards/chosen": -60.15625,
|
||
|
|
"rewards/margins": 7.514062404632568,
|
||
|
|
"rewards/rejected": -67.6640625,
|
||
|
|
"rewards/weighted_accuracies": 0.528124988079071,
|
||
|
|
"rewards/weighted_chosen": -0.22589111328125,
|
||
|
|
"rewards/weighted_margins": 0.0014251709217205644,
|
||
|
|
"rewards/weighted_rejected": -0.22734984755516052,
|
||
|
|
"step": 240
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13395847287340926,
|
||
|
|
"grad_norm": 17.17424774169922,
|
||
|
|
"learning_rate": 9.966432574620906e-07,
|
||
|
|
"logits/chosen": -2.994140625,
|
||
|
|
"logits/rejected": -2.995654344558716,
|
||
|
|
"logps/chosen": -250.4296875,
|
||
|
|
"logps/rejected": -282.0859375,
|
||
|
|
"logps/weighted_chosen": -0.707995593547821,
|
||
|
|
"logps/weighted_rejected": -0.76885986328125,
|
||
|
|
"loss": 0.6897,
|
||
|
|
"rewards/accuracies": 0.590624988079071,
|
||
|
|
"rewards/chosen": -55.573829650878906,
|
||
|
|
"rewards/margins": 7.142187595367432,
|
||
|
|
"rewards/rejected": -62.73515701293945,
|
||
|
|
"rewards/weighted_accuracies": 0.528124988079071,
|
||
|
|
"rewards/weighted_chosen": -0.18481139838695526,
|
||
|
|
"rewards/weighted_margins": 0.01998596265912056,
|
||
|
|
"rewards/weighted_rejected": -0.20475539565086365,
|
||
|
|
"step": 250
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13931681178834562,
|
||
|
|
"grad_norm": 18.692733764648438,
|
||
|
|
"learning_rate": 9.954748808839674e-07,
|
||
|
|
"logits/chosen": -2.887255907058716,
|
||
|
|
"logits/rejected": -2.898193359375,
|
||
|
|
"logps/chosen": -280.46875,
|
||
|
|
"logps/rejected": -280.6640625,
|
||
|
|
"logps/weighted_chosen": -0.7401489019393921,
|
||
|
|
"logps/weighted_rejected": -0.776611328125,
|
||
|
|
"loss": 0.6869,
|
||
|
|
"rewards/accuracies": 0.574999988079071,
|
||
|
|
"rewards/chosen": -53.8828125,
|
||
|
|
"rewards/margins": 5.946484565734863,
|
||
|
|
"rewards/rejected": -59.81757736206055,
|
||
|
|
"rewards/weighted_accuracies": 0.53125,
|
||
|
|
"rewards/weighted_chosen": -0.18103942275047302,
|
||
|
|
"rewards/weighted_margins": 0.02960510179400444,
|
||
|
|
"rewards/weighted_rejected": -0.21054688096046448,
|
||
|
|
"step": 260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.144675150703282,
|
||
|
|
"grad_norm": 16.58523941040039,
|
||
|
|
"learning_rate": 9.941332475539824e-07,
|
||
|
|
"logits/chosen": -2.8628907203674316,
|
||
|
|
"logits/rejected": -2.87939453125,
|
||
|
|
"logps/chosen": -267.26251220703125,
|
||
|
|
"logps/rejected": -306.49688720703125,
|
||
|
|
"logps/weighted_chosen": -0.7835937738418579,
|
||
|
|
"logps/weighted_rejected": -0.8957885503768921,
|
||
|
|
"loss": 0.6712,
|
||
|
|
"rewards/accuracies": 0.625,
|
||
|
|
"rewards/chosen": -57.08281326293945,
|
||
|
|
"rewards/margins": 12.824609756469727,
|
||
|
|
"rewards/rejected": -69.8984375,
|
||
|
|
"rewards/weighted_accuracies": 0.596875011920929,
|
||
|
|
"rewards/weighted_chosen": -0.2109939604997635,
|
||
|
|
"rewards/weighted_margins": 0.07429657131433487,
|
||
|
|
"rewards/weighted_rejected": -0.2853530943393707,
|
||
|
|
"step": 270
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15003348961821836,
|
||
|
|
"grad_norm": 23.606122970581055,
|
||
|
|
"learning_rate": 9.926188266120295e-07,
|
||
|
|
"logits/chosen": -2.9261717796325684,
|
||
|
|
"logits/rejected": -2.9302000999450684,
|
||
|
|
"logps/chosen": -252.35781860351562,
|
||
|
|
"logps/rejected": -270.3921813964844,
|
||
|
|
"logps/weighted_chosen": -0.757946789264679,
|
||
|
|
"logps/weighted_rejected": -0.8466796875,
|
||
|
|
"loss": 0.6669,
|
||
|
|
"rewards/accuracies": 0.5874999761581421,
|
||
|
|
"rewards/chosen": -56.718360900878906,
|
||
|
|
"rewards/margins": 10.275781631469727,
|
||
|
|
"rewards/rejected": -66.990234375,
|
||
|
|
"rewards/weighted_accuracies": 0.606249988079071,
|
||
|
|
"rewards/weighted_chosen": -0.2161209136247635,
|
||
|
|
"rewards/weighted_margins": 0.08074493706226349,
|
||
|
|
"rewards/weighted_rejected": -0.2967941164970398,
|
||
|
|
"step": 280
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15539182853315472,
|
||
|
|
"grad_norm": 25.32463836669922,
|
||
|
|
"learning_rate": 9.909321476180591e-07,
|
||
|
|
"logits/chosen": -2.9480957984924316,
|
||
|
|
"logits/rejected": -2.9312500953674316,
|
||
|
|
"logps/chosen": -281.33123779296875,
|
||
|
|
"logps/rejected": -317.3374938964844,
|
||
|
|
"logps/weighted_chosen": -0.870861828327179,
|
||
|
|
"logps/weighted_rejected": -0.9639892578125,
|
||
|
|
"loss": 0.6868,
|
||
|
|
"rewards/accuracies": 0.6000000238418579,
|
||
|
|
"rewards/chosen": -78.2476577758789,
|
||
|
|
"rewards/margins": 12.921093940734863,
|
||
|
|
"rewards/rejected": -91.15312194824219,
|
||
|
|
"rewards/weighted_accuracies": 0.534375011920929,
|
||
|
|
"rewards/weighted_chosen": -0.32458800077438354,
|
||
|
|
"rewards/weighted_margins": 0.04551086574792862,
|
||
|
|
"rewards/weighted_rejected": -0.37019044160842896,
|
||
|
|
"step": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1607501674480911,
|
||
|
|
"grad_norm": 21.94011116027832,
|
||
|
|
"learning_rate": 9.890738003669027e-07,
|
||
|
|
"logits/chosen": -3.082812547683716,
|
||
|
|
"logits/rejected": -3.089111328125,
|
||
|
|
"logps/chosen": -288.62811279296875,
|
||
|
|
"logps/rejected": -311.38751220703125,
|
||
|
|
"logps/weighted_chosen": -0.955078125,
|
||
|
|
"logps/weighted_rejected": -1.0396239757537842,
|
||
|
|
"loss": 0.687,
|
||
|
|
"rewards/accuracies": 0.5625,
|
||
|
|
"rewards/chosen": -93.6109390258789,
|
||
|
|
"rewards/margins": 9.185546875,
|
||
|
|
"rewards/rejected": -102.78202819824219,
|
||
|
|
"rewards/weighted_accuracies": 0.559374988079071,
|
||
|
|
"rewards/weighted_chosen": -0.40625,
|
||
|
|
"rewards/weighted_margins": 0.07210388034582138,
|
||
|
|
"rewards/weighted_rejected": -0.478240966796875,
|
||
|
|
"step": 300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16610850636302746,
|
||
|
|
"grad_norm": 25.38623046875,
|
||
|
|
"learning_rate": 9.870444346820348e-07,
|
||
|
|
"logits/chosen": -3.173828125,
|
||
|
|
"logits/rejected": -3.147998094558716,
|
||
|
|
"logps/chosen": -297.15936279296875,
|
||
|
|
"logps/rejected": -320.50311279296875,
|
||
|
|
"logps/weighted_chosen": -0.89190673828125,
|
||
|
|
"logps/weighted_rejected": -0.923266589641571,
|
||
|
|
"loss": 0.6987,
|
||
|
|
"rewards/accuracies": 0.6187499761581421,
|
||
|
|
"rewards/chosen": -83.2046890258789,
|
||
|
|
"rewards/margins": 12.489062309265137,
|
||
|
|
"rewards/rejected": -95.6937484741211,
|
||
|
|
"rewards/weighted_accuracies": 0.581250011920929,
|
||
|
|
"rewards/weighted_chosen": -0.33665162324905396,
|
||
|
|
"rewards/weighted_margins": 0.00975952111184597,
|
||
|
|
"rewards/weighted_rejected": -0.34637755155563354,
|
||
|
|
"step": 310
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17146684527796383,
|
||
|
|
"grad_norm": 17.789676666259766,
|
||
|
|
"learning_rate": 9.848447601883433e-07,
|
||
|
|
"logits/chosen": -3.236083984375,
|
||
|
|
"logits/rejected": -3.2168946266174316,
|
||
|
|
"logps/chosen": -267.09844970703125,
|
||
|
|
"logps/rejected": -310.3031311035156,
|
||
|
|
"logps/weighted_chosen": -0.8338623046875,
|
||
|
|
"logps/weighted_rejected": -0.918957531452179,
|
||
|
|
"loss": 0.6805,
|
||
|
|
"rewards/accuracies": 0.612500011920929,
|
||
|
|
"rewards/chosen": -70.2816390991211,
|
||
|
|
"rewards/margins": 19.578907012939453,
|
||
|
|
"rewards/rejected": -89.81953430175781,
|
||
|
|
"rewards/weighted_accuracies": 0.590624988079071,
|
||
|
|
"rewards/weighted_chosen": -0.2871032655239105,
|
||
|
|
"rewards/weighted_margins": 0.06427612155675888,
|
||
|
|
"rewards/weighted_rejected": -0.3513946533203125,
|
||
|
|
"step": 320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1768251841929002,
|
||
|
|
"grad_norm": 24.225061416625977,
|
||
|
|
"learning_rate": 9.824755460639899e-07,
|
||
|
|
"logits/chosen": -3.1796875,
|
||
|
|
"logits/rejected": -3.179150342941284,
|
||
|
|
"logps/chosen": -265.12030029296875,
|
||
|
|
"logps/rejected": -308.19842529296875,
|
||
|
|
"logps/weighted_chosen": -0.833850085735321,
|
||
|
|
"logps/weighted_rejected": -0.8948730230331421,
|
||
|
|
"loss": 0.6836,
|
||
|
|
"rewards/accuracies": 0.6156250238418579,
|
||
|
|
"rewards/chosen": -73.27030944824219,
|
||
|
|
"rewards/margins": 16.651952743530273,
|
||
|
|
"rewards/rejected": -89.87968444824219,
|
||
|
|
"rewards/weighted_accuracies": 0.5531250238418579,
|
||
|
|
"rewards/weighted_chosen": -0.31196290254592896,
|
||
|
|
"rewards/weighted_margins": 0.04750366136431694,
|
||
|
|
"rewards/weighted_rejected": -0.35954588651657104,
|
||
|
|
"step": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18218352310783656,
|
||
|
|
"grad_norm": 19.119462966918945,
|
||
|
|
"learning_rate": 9.799376207714444e-07,
|
||
|
|
"logits/chosen": -3.1602540016174316,
|
||
|
|
"logits/rejected": -3.154492139816284,
|
||
|
|
"logps/chosen": -277.9624938964844,
|
||
|
|
"logps/rejected": -296.7093811035156,
|
||
|
|
"logps/weighted_chosen": -0.902087390422821,
|
||
|
|
"logps/weighted_rejected": -0.974658191204071,
|
||
|
|
"loss": 0.6727,
|
||
|
|
"rewards/accuracies": 0.5406249761581421,
|
||
|
|
"rewards/chosen": -77.0367202758789,
|
||
|
|
"rewards/margins": 9.926562309265137,
|
||
|
|
"rewards/rejected": -86.9476547241211,
|
||
|
|
"rewards/weighted_accuracies": 0.559374988079071,
|
||
|
|
"rewards/weighted_chosen": -0.334219366312027,
|
||
|
|
"rewards/weighted_margins": 0.06939239799976349,
|
||
|
|
"rewards/weighted_rejected": -0.4036361575126648,
|
||
|
|
"step": 340
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18754186202277295,
|
||
|
|
"grad_norm": 22.155990600585938,
|
||
|
|
"learning_rate": 9.772318717677903e-07,
|
||
|
|
"logits/chosen": -3.130908250808716,
|
||
|
|
"logits/rejected": -3.132128953933716,
|
||
|
|
"logps/chosen": -288.41094970703125,
|
||
|
|
"logps/rejected": -313.0843811035156,
|
||
|
|
"logps/weighted_chosen": -0.9267333745956421,
|
||
|
|
"logps/weighted_rejected": -1.040771484375,
|
||
|
|
"loss": 0.6683,
|
||
|
|
"rewards/accuracies": 0.625,
|
||
|
|
"rewards/chosen": -88.24609375,
|
||
|
|
"rewards/margins": 18.800390243530273,
|
||
|
|
"rewards/rejected": -107.046875,
|
||
|
|
"rewards/weighted_accuracies": 0.6156250238418579,
|
||
|
|
"rewards/weighted_chosen": -0.39622801542282104,
|
||
|
|
"rewards/weighted_margins": 0.08176574856042862,
|
||
|
|
"rewards/weighted_rejected": -0.47796326875686646,
|
||
|
|
"step": 350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19290020093770932,
|
||
|
|
"grad_norm": 26.421598434448242,
|
||
|
|
"learning_rate": 9.743592451943998e-07,
|
||
|
|
"logits/chosen": -3.229687452316284,
|
||
|
|
"logits/rejected": -3.228222608566284,
|
||
|
|
"logps/chosen": -327.7328186035156,
|
||
|
|
"logps/rejected": -361.0015563964844,
|
||
|
|
"logps/weighted_chosen": -0.990966796875,
|
||
|
|
"logps/weighted_rejected": -1.057519555091858,
|
||
|
|
"loss": 0.6726,
|
||
|
|
"rewards/accuracies": 0.590624988079071,
|
||
|
|
"rewards/chosen": -116.12812805175781,
|
||
|
|
"rewards/margins": 18.506641387939453,
|
||
|
|
"rewards/rejected": -134.654296875,
|
||
|
|
"rewards/weighted_accuracies": 0.543749988079071,
|
||
|
|
"rewards/weighted_chosen": -0.4190429747104645,
|
||
|
|
"rewards/weighted_margins": 0.08066101372241974,
|
||
|
|
"rewards/weighted_rejected": -0.499612420797348,
|
||
|
|
"step": 360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1982585398526457,
|
||
|
|
"grad_norm": 19.50035858154297,
|
||
|
|
"learning_rate": 9.713207455460892e-07,
|
||
|
|
"logits/chosen": -3.184375047683716,
|
||
|
|
"logits/rejected": -3.209912061691284,
|
||
|
|
"logps/chosen": -326.4937438964844,
|
||
|
|
"logps/rejected": -343.4593811035156,
|
||
|
|
"logps/weighted_chosen": -1.00506591796875,
|
||
|
|
"logps/weighted_rejected": -1.092797875404358,
|
||
|
|
"loss": 0.6762,
|
||
|
|
"rewards/accuracies": 0.5687500238418579,
|
||
|
|
"rewards/chosen": -120.6937484741211,
|
||
|
|
"rewards/margins": 6.901953220367432,
|
||
|
|
"rewards/rejected": -127.60469055175781,
|
||
|
|
"rewards/weighted_accuracies": 0.5531250238418579,
|
||
|
|
"rewards/weighted_chosen": -0.4098266661167145,
|
||
|
|
"rewards/weighted_margins": 0.08511047065258026,
|
||
|
|
"rewards/weighted_rejected": -0.495004266500473,
|
||
|
|
"step": 370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20361687876758205,
|
||
|
|
"grad_norm": 17.131114959716797,
|
||
|
|
"learning_rate": 9.681174353198686e-07,
|
||
|
|
"logits/chosen": -3.2510743141174316,
|
||
|
|
"logits/rejected": -3.23828125,
|
||
|
|
"logps/chosen": -330.140625,
|
||
|
|
"logps/rejected": -368.10467529296875,
|
||
|
|
"logps/weighted_chosen": -0.9594482183456421,
|
||
|
|
"logps/weighted_rejected": -1.056616187095642,
|
||
|
|
"loss": 0.6719,
|
||
|
|
"rewards/accuracies": 0.6156250238418579,
|
||
|
|
"rewards/chosen": -119.54219055175781,
|
||
|
|
"rewards/margins": 20.350391387939453,
|
||
|
|
"rewards/rejected": -139.82461547851562,
|
||
|
|
"rewards/weighted_accuracies": 0.574999988079071,
|
||
|
|
"rewards/weighted_chosen": -0.415090948343277,
|
||
|
|
"rewards/weighted_margins": 0.086700439453125,
|
||
|
|
"rewards/weighted_rejected": -0.501788318157196,
|
||
|
|
"step": 380
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20897521768251842,
|
||
|
|
"grad_norm": 38.318756103515625,
|
||
|
|
"learning_rate": 9.647504346434103e-07,
|
||
|
|
"logits/chosen": -3.445019483566284,
|
||
|
|
"logits/rejected": -3.4564452171325684,
|
||
|
|
"logps/chosen": -371.9156188964844,
|
||
|
|
"logps/rejected": -389.43438720703125,
|
||
|
|
"logps/weighted_chosen": -1.062597632408142,
|
||
|
|
"logps/weighted_rejected": -1.165747046470642,
|
||
|
|
"loss": 0.6656,
|
||
|
|
"rewards/accuracies": 0.578125,
|
||
|
|
"rewards/chosen": -150.7859344482422,
|
||
|
|
"rewards/margins": 19.3828125,
|
||
|
|
"rewards/rejected": -170.19686889648438,
|
||
|
|
"rewards/weighted_accuracies": 0.578125,
|
||
|
|
"rewards/weighted_chosen": -0.49369508028030396,
|
||
|
|
"rewards/weighted_margins": 0.09916381537914276,
|
||
|
|
"rewards/weighted_rejected": -0.5928589105606079,
|
||
|
|
"step": 390
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2143335565974548,
|
||
|
|
"grad_norm": 37.75257873535156,
|
||
|
|
"learning_rate": 9.612209208833646e-07,
|
||
|
|
"logits/chosen": -3.4891600608825684,
|
||
|
|
"logits/rejected": -3.4913086891174316,
|
||
|
|
"logps/chosen": -373.2046813964844,
|
||
|
|
"logps/rejected": -387.3968811035156,
|
||
|
|
"logps/weighted_chosen": -1.159387230873108,
|
||
|
|
"logps/weighted_rejected": -1.2549560070037842,
|
||
|
|
"loss": 0.6802,
|
||
|
|
"rewards/accuracies": 0.6031249761581421,
|
||
|
|
"rewards/chosen": -173.24063110351562,
|
||
|
|
"rewards/margins": 12.66796875,
|
||
|
|
"rewards/rejected": -185.9562530517578,
|
||
|
|
"rewards/weighted_accuracies": 0.5531250238418579,
|
||
|
|
"rewards/weighted_chosen": -0.619030773639679,
|
||
|
|
"rewards/weighted_margins": 0.07501526176929474,
|
||
|
|
"rewards/weighted_rejected": -0.694287121295929,
|
||
|
|
"step": 400
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21969189551239116,
|
||
|
|
"grad_norm": 62.04045486450195,
|
||
|
|
"learning_rate": 9.5753012823366e-07,
|
||
|
|
"logits/chosen": -3.5625,
|
||
|
|
"logits/rejected": -3.555224657058716,
|
||
|
|
"logps/chosen": -392.5843811035156,
|
||
|
|
"logps/rejected": -429.76873779296875,
|
||
|
|
"logps/weighted_chosen": -1.109643578529358,
|
||
|
|
"logps/weighted_rejected": -1.2004883289337158,
|
||
|
|
"loss": 0.6903,
|
||
|
|
"rewards/accuracies": 0.6156250238418579,
|
||
|
|
"rewards/chosen": -180.3562469482422,
|
||
|
|
"rewards/margins": 26.413280487060547,
|
||
|
|
"rewards/rejected": -206.75936889648438,
|
||
|
|
"rewards/weighted_accuracies": 0.5375000238418579,
|
||
|
|
"rewards/weighted_chosen": -0.5794036984443665,
|
||
|
|
"rewards/weighted_margins": 0.08088378608226776,
|
||
|
|
"rewards/weighted_rejected": -0.660137951374054,
|
||
|
|
"step": 410
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22505023442732752,
|
||
|
|
"grad_norm": 29.432497024536133,
|
||
|
|
"learning_rate": 9.536793472839324e-07,
|
||
|
|
"logits/chosen": -3.7582030296325684,
|
||
|
|
"logits/rejected": -3.77783203125,
|
||
|
|
"logps/chosen": -413.109375,
|
||
|
|
"logps/rejected": -459.015625,
|
||
|
|
"logps/weighted_chosen": -1.0903198719024658,
|
||
|
|
"logps/weighted_rejected": -1.1765625476837158,
|
||
|
|
"loss": 0.6833,
|
||
|
|
"rewards/accuracies": 0.6031249761581421,
|
||
|
|
"rewards/chosen": -182.9140625,
|
||
|
|
"rewards/margins": 33.59453201293945,
|
||
|
|
"rewards/rejected": -216.5343780517578,
|
||
|
|
"rewards/weighted_accuracies": 0.590624988079071,
|
||
|
|
"rewards/weighted_chosen": -0.528045654296875,
|
||
|
|
"rewards/weighted_margins": 0.08157958835363388,
|
||
|
|
"rewards/weighted_rejected": -0.609661877155304,
|
||
|
|
"step": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2304085733422639,
|
||
|
|
"grad_norm": 34.61969757080078,
|
||
|
|
"learning_rate": 9.496699245682351e-07,
|
||
|
|
"logits/chosen": -3.876757860183716,
|
||
|
|
"logits/rejected": -3.873339891433716,
|
||
|
|
"logps/chosen": -411.23748779296875,
|
||
|
|
"logps/rejected": -474.3687438964844,
|
||
|
|
"logps/weighted_chosen": -1.104589819908142,
|
||
|
|
"logps/weighted_rejected": -1.18817138671875,
|
||
|
|
"loss": 0.6878,
|
||
|
|
"rewards/accuracies": 0.6156250238418579,
|
||
|
|
"rewards/chosen": -190.3468780517578,
|
||
|
|
"rewards/margins": 43.072265625,
|
||
|
|
"rewards/rejected": -233.43124389648438,
|
||
|
|
"rewards/weighted_accuracies": 0.5531250238418579,
|
||
|
|
"rewards/weighted_chosen": -0.543225109577179,
|
||
|
|
"rewards/weighted_margins": 0.05898132175207138,
|
||
|
|
"rewards/weighted_rejected": -0.6022491455078125,
|
||
|
|
"step": 430
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23576691225720026,
|
||
|
|
"grad_norm": 25.689876556396484,
|
||
|
|
"learning_rate": 9.455032620941839e-07,
|
||
|
|
"logits/chosen": -3.893261671066284,
|
||
|
|
"logits/rejected": -3.866406202316284,
|
||
|
|
"logps/chosen": -390.3531188964844,
|
||
|
|
"logps/rejected": -474.80780029296875,
|
||
|
|
"logps/weighted_chosen": -1.0908081531524658,
|
||
|
|
"logps/weighted_rejected": -1.19012451171875,
|
||
|
|
"loss": 0.6673,
|
||
|
|
"rewards/accuracies": 0.621874988079071,
|
||
|
|
"rewards/chosen": -196.52188110351562,
|
||
|
|
"rewards/margins": 52.310546875,
|
||
|
|
"rewards/rejected": -248.7218780517578,
|
||
|
|
"rewards/weighted_accuracies": 0.5718749761581421,
|
||
|
|
"rewards/weighted_chosen": -0.555419921875,
|
||
|
|
"rewards/weighted_margins": 0.09650878608226776,
|
||
|
|
"rewards/weighted_rejected": -0.651898205280304,
|
||
|
|
"step": 440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24112525117213665,
|
||
|
|
"grad_norm": 54.604705810546875,
|
||
|
|
"learning_rate": 9.411808168527066e-07,
|
||
|
|
"logits/chosen": -3.88916015625,
|
||
|
|
"logits/rejected": -3.8941407203674316,
|
||
|
|
"logps/chosen": -411.3031311035156,
|
||
|
|
"logps/rejected": -472.203125,
|
||
|
|
"logps/weighted_chosen": -1.082067847251892,
|
||
|
|
"logps/weighted_rejected": -1.1839110851287842,
|
||
|
|
"loss": 0.6816,
|
||
|
|
"rewards/accuracies": 0.6000000238418579,
|
||
|
|
"rewards/chosen": -205.703125,
|
||
|
|
"rewards/margins": 46.81132888793945,
|
||
|
|
"rewards/rejected": -252.5343780517578,
|
||
|
|
"rewards/weighted_accuracies": 0.590624988079071,
|
||
|
|
"rewards/weighted_chosen": -0.554046630859375,
|
||
|
|
"rewards/weighted_margins": 0.07362671196460724,
|
||
|
|
"rewards/weighted_rejected": -0.6277831792831421,
|
||
|
|
"step": 450
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24648359008707302,
|
||
|
|
"grad_norm": 26.44339942932129,
|
||
|
|
"learning_rate": 9.367041003085648e-07,
|
||
|
|
"logits/chosen": -3.8785157203674316,
|
||
|
|
"logits/rejected": -3.866894483566284,
|
||
|
|
"logps/chosen": -523.2593994140625,
|
||
|
|
"logps/rejected": -523.2750244140625,
|
||
|
|
"logps/weighted_chosen": -1.2444579601287842,
|
||
|
|
"logps/weighted_rejected": -1.3485107421875,
|
||
|
|
"loss": 0.6775,
|
||
|
|
"rewards/accuracies": 0.5625,
|
||
|
|
"rewards/chosen": -304.52813720703125,
|
||
|
|
"rewards/margins": 4.764062404632568,
|
||
|
|
"rewards/rejected": -309.3812561035156,
|
||
|
|
"rewards/weighted_accuracies": 0.590624988079071,
|
||
|
|
"rewards/weighted_chosen": -0.6716858148574829,
|
||
|
|
"rewards/weighted_margins": 0.08544921875,
|
||
|
|
"rewards/weighted_rejected": -0.757153332233429,
|
||
|
|
"step": 460
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25184192900200936,
|
||
|
|
"grad_norm": 28.781457901000977,
|
||
|
|
"learning_rate": 9.320746778718274e-07,
|
||
|
|
"logits/chosen": -4.042578220367432,
|
||
|
|
"logits/rejected": -4.053515434265137,
|
||
|
|
"logps/chosen": -444.046875,
|
||
|
|
"logps/rejected": -514.0531005859375,
|
||
|
|
"logps/weighted_chosen": -1.281274437904358,
|
||
|
|
"logps/weighted_rejected": -1.4283936023712158,
|
||
|
|
"loss": 0.6526,
|
||
|
|
"rewards/accuracies": 0.6187499761581421,
|
||
|
|
"rewards/chosen": -256.4828186035156,
|
||
|
|
"rewards/margins": 53.646095275878906,
|
||
|
|
"rewards/rejected": -310.15313720703125,
|
||
|
|
"rewards/weighted_accuracies": 0.596875011920929,
|
||
|
|
"rewards/weighted_chosen": -0.714306652545929,
|
||
|
|
"rewards/weighted_margins": 0.13612060248851776,
|
||
|
|
"rewards/weighted_rejected": -0.8504638671875,
|
||
|
|
"step": 470
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2572002679169457,
|
||
|
|
"grad_norm": 58.31265640258789,
|
||
|
|
"learning_rate": 9.272941683504808e-07,
|
||
|
|
"logits/chosen": -4.060449123382568,
|
||
|
|
"logits/rejected": -4.083691596984863,
|
||
|
|
"logps/chosen": -564.7312622070312,
|
||
|
|
"logps/rejected": -582.9124755859375,
|
||
|
|
"logps/weighted_chosen": -1.305688500404358,
|
||
|
|
"logps/weighted_rejected": -1.470678687095642,
|
||
|
|
"loss": 0.6627,
|
||
|
|
"rewards/accuracies": 0.5687500238418579,
|
||
|
|
"rewards/chosen": -346.97344970703125,
|
||
|
|
"rewards/margins": 15.486719131469727,
|
||
|
|
"rewards/rejected": -362.28594970703125,
|
||
|
|
"rewards/weighted_accuracies": 0.596875011920929,
|
||
|
|
"rewards/weighted_chosen": -0.7694152593612671,
|
||
|
|
"rewards/weighted_margins": 0.14488831162452698,
|
||
|
|
"rewards/weighted_rejected": -0.9143127202987671,
|
||
|
|
"step": 480
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2625586068318821,
|
||
|
|
"grad_norm": 28.284282684326172,
|
||
|
|
"learning_rate": 9.223642433843679e-07,
|
||
|
|
"logits/chosen": -3.8697266578674316,
|
||
|
|
"logits/rejected": -3.857128858566284,
|
||
|
|
"logps/chosen": -500.0718688964844,
|
||
|
|
"logps/rejected": -534.6281127929688,
|
||
|
|
"logps/weighted_chosen": -1.1939697265625,
|
||
|
|
"logps/weighted_rejected": -1.323388695716858,
|
||
|
|
"loss": 0.6567,
|
||
|
|
"rewards/accuracies": 0.621874988079071,
|
||
|
|
"rewards/chosen": -298.09063720703125,
|
||
|
|
"rewards/margins": 29.775781631469727,
|
||
|
|
"rewards/rejected": -328.0640563964844,
|
||
|
|
"rewards/weighted_accuracies": 0.59375,
|
||
|
|
"rewards/weighted_chosen": -0.631103515625,
|
||
|
|
"rewards/weighted_margins": 0.13239136338233948,
|
||
|
|
"rewards/weighted_rejected": -0.763378918170929,
|
||
|
|
"step": 490
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2679169457468185,
|
||
|
|
"grad_norm": 32.619327545166016,
|
||
|
|
"learning_rate": 9.172866268606513e-07,
|
||
|
|
"logits/chosen": -3.828320264816284,
|
||
|
|
"logits/rejected": -3.82373046875,
|
||
|
|
"logps/chosen": -527.2687377929688,
|
||
|
|
"logps/rejected": -588.8343505859375,
|
||
|
|
"logps/weighted_chosen": -1.291723608970642,
|
||
|
|
"logps/weighted_rejected": -1.4711182117462158,
|
||
|
|
"loss": 0.6559,
|
||
|
|
"rewards/accuracies": 0.625,
|
||
|
|
"rewards/chosen": -312.36248779296875,
|
||
|
|
"rewards/margins": 48.47734451293945,
|
||
|
|
"rewards/rejected": -360.8890686035156,
|
||
|
|
"rewards/weighted_accuracies": 0.59375,
|
||
|
|
"rewards/weighted_chosen": -0.727709949016571,
|
||
|
|
"rewards/weighted_margins": 0.145904541015625,
|
||
|
|
"rewards/weighted_rejected": -0.8731445074081421,
|
||
|
|
"step": 500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2679169457468185,
|
||
|
|
"eval_logits/chosen": -3.994471788406372,
|
||
|
|
"eval_logits/rejected": -4.005757808685303,
|
||
|
|
"eval_logps/chosen": -530.9920043945312,
|
||
|
|
"eval_logps/rejected": -592.9425659179688,
|
||
|
|
"eval_logps/weighted_chosen": -1.3731945753097534,
|
||
|
|
"eval_logps/weighted_rejected": -1.5277278423309326,
|
||
|
|
"eval_loss": 0.6660141348838806,
|
||
|
|
"eval_rewards/accuracies": 0.5874499082565308,
|
||
|
|
"eval_rewards/chosen": -328.0513916015625,
|
||
|
|
"eval_rewards/margins": 48.42765426635742,
|
||
|
|
"eval_rewards/rejected": -376.4005432128906,
|
||
|
|
"eval_rewards/weighted_accuracies": 0.5901201367378235,
|
||
|
|
"eval_rewards/weighted_chosen": -0.8177139759063721,
|
||
|
|
"eval_rewards/weighted_margins": 0.13399113714694977,
|
||
|
|
"eval_rewards/weighted_rejected": -0.9517050385475159,
|
||
|
|
"eval_runtime": 1889.7494,
|
||
|
|
"eval_samples_per_second": 1.584,
|
||
|
|
"eval_steps_per_second": 0.396,
|
||
|
|
"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
|
||
|
|
}
|