Meta-Llama-3-8B-Instruct-MI-5e-7 / trainer_state.json
tengxiao1
TX
a0ae571
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9982631930527722,
"eval_steps": 400,
"global_step": 467,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01068804275217101,
"grad_norm": 7.670545903309526,
"learning_rate": 5.3191489361702123e-08,
"logits/chosen": 0.029632840305566788,
"logits/rejected": -0.005643837153911591,
"logps/chosen": -0.25496524572372437,
"logps/rejected": -0.2698316276073456,
"loss": 1.2612,
"rewards/accuracies": 0.518750011920929,
"rewards/chosen": -0.25496524572372437,
"rewards/margins": 0.014866364188492298,
"rewards/rejected": -0.2698316276073456,
"step": 5
},
{
"epoch": 0.02137608550434202,
"grad_norm": 6.592494357507575,
"learning_rate": 1.0638297872340425e-07,
"logits/chosen": -0.030216623097658157,
"logits/rejected": -0.021379908546805382,
"logps/chosen": -0.2787334620952606,
"logps/rejected": -0.27201998233795166,
"loss": 1.273,
"rewards/accuracies": 0.44999998807907104,
"rewards/chosen": -0.2787334620952606,
"rewards/margins": -0.00671349698677659,
"rewards/rejected": -0.27201998233795166,
"step": 10
},
{
"epoch": 0.03206412825651302,
"grad_norm": 6.365016282759897,
"learning_rate": 1.5957446808510638e-07,
"logits/chosen": 0.06006206199526787,
"logits/rejected": 0.05343017727136612,
"logps/chosen": -0.29093918204307556,
"logps/rejected": -0.30492520332336426,
"loss": 1.2801,
"rewards/accuracies": 0.4937500059604645,
"rewards/chosen": -0.29093918204307556,
"rewards/margins": 0.013985991477966309,
"rewards/rejected": -0.30492520332336426,
"step": 15
},
{
"epoch": 0.04275217100868404,
"grad_norm": 6.085135360569755,
"learning_rate": 2.127659574468085e-07,
"logits/chosen": 0.036919064819812775,
"logits/rejected": 0.03441498801112175,
"logps/chosen": -0.2613425850868225,
"logps/rejected": -0.27211794257164,
"loss": 1.2665,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.2613425850868225,
"rewards/margins": 0.010775364935398102,
"rewards/rejected": -0.27211794257164,
"step": 20
},
{
"epoch": 0.053440213760855046,
"grad_norm": 5.6906497609423,
"learning_rate": 2.659574468085106e-07,
"logits/chosen": 0.029963787645101547,
"logits/rejected": 0.014045795425772667,
"logps/chosen": -0.2637311816215515,
"logps/rejected": -0.2851249575614929,
"loss": 1.2676,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.2637311816215515,
"rewards/margins": 0.02139372006058693,
"rewards/rejected": -0.2851249575614929,
"step": 25
},
{
"epoch": 0.06412825651302605,
"grad_norm": 6.018351349229231,
"learning_rate": 3.1914893617021275e-07,
"logits/chosen": -0.0884183868765831,
"logits/rejected": -0.06914771348237991,
"logps/chosen": -0.27684831619262695,
"logps/rejected": -0.2800411581993103,
"loss": 1.2677,
"rewards/accuracies": 0.44999998807907104,
"rewards/chosen": -0.27684831619262695,
"rewards/margins": 0.003192795906215906,
"rewards/rejected": -0.2800411581993103,
"step": 30
},
{
"epoch": 0.07481629926519706,
"grad_norm": 4.631909296260356,
"learning_rate": 3.7234042553191484e-07,
"logits/chosen": -0.02534369006752968,
"logits/rejected": -0.05787515640258789,
"logps/chosen": -0.2885574698448181,
"logps/rejected": -0.2887212634086609,
"loss": 1.2839,
"rewards/accuracies": 0.5062500238418579,
"rewards/chosen": -0.2885574698448181,
"rewards/margins": 0.00016381926252506673,
"rewards/rejected": -0.2887212634086609,
"step": 35
},
{
"epoch": 0.08550434201736808,
"grad_norm": 5.991793529016571,
"learning_rate": 4.25531914893617e-07,
"logits/chosen": -0.09816843271255493,
"logits/rejected": -0.10262785106897354,
"logps/chosen": -0.2812947928905487,
"logps/rejected": -0.28836020827293396,
"loss": 1.2799,
"rewards/accuracies": 0.48750001192092896,
"rewards/chosen": -0.2812947928905487,
"rewards/margins": 0.007065430283546448,
"rewards/rejected": -0.28836020827293396,
"step": 40
},
{
"epoch": 0.09619238476953908,
"grad_norm": 5.862607859915412,
"learning_rate": 4.787234042553192e-07,
"logits/chosen": -0.054446518421173096,
"logits/rejected": -0.043239839375019073,
"logps/chosen": -0.2687098979949951,
"logps/rejected": -0.2790478467941284,
"loss": 1.2665,
"rewards/accuracies": 0.53125,
"rewards/chosen": -0.2687098979949951,
"rewards/margins": 0.010338006541132927,
"rewards/rejected": -0.2790478467941284,
"step": 45
},
{
"epoch": 0.10688042752171009,
"grad_norm": 7.216783478332914,
"learning_rate": 4.999370587356267e-07,
"logits/chosen": -0.06357742100954056,
"logits/rejected": -0.08343801647424698,
"logps/chosen": -0.2701963484287262,
"logps/rejected": -0.29289665818214417,
"loss": 1.2612,
"rewards/accuracies": 0.512499988079071,
"rewards/chosen": -0.2701963484287262,
"rewards/margins": 0.022700298577547073,
"rewards/rejected": -0.29289665818214417,
"step": 50
},
{
"epoch": 0.11756847027388109,
"grad_norm": 7.309902512874254,
"learning_rate": 4.995525324419337e-07,
"logits/chosen": -0.027720177546143532,
"logits/rejected": -0.047440748661756516,
"logps/chosen": -0.2775796055793762,
"logps/rejected": -0.3043287396430969,
"loss": 1.2751,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": -0.2775796055793762,
"rewards/margins": 0.026749128475785255,
"rewards/rejected": -0.3043287396430969,
"step": 55
},
{
"epoch": 0.1282565130260521,
"grad_norm": 9.12998569416597,
"learning_rate": 4.988189843662815e-07,
"logits/chosen": -0.037262119352817535,
"logits/rejected": -0.03407366946339607,
"logps/chosen": -0.2749136984348297,
"logps/rejected": -0.28693363070487976,
"loss": 1.2569,
"rewards/accuracies": 0.53125,
"rewards/chosen": -0.2749136984348297,
"rewards/margins": 0.012019967660307884,
"rewards/rejected": -0.28693363070487976,
"step": 60
},
{
"epoch": 0.13894455577822312,
"grad_norm": 8.45046825310939,
"learning_rate": 4.977374404419837e-07,
"logits/chosen": -0.08383898437023163,
"logits/rejected": -0.058686137199401855,
"logps/chosen": -0.2790556848049164,
"logps/rejected": -0.32426005601882935,
"loss": 1.2561,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.2790556848049164,
"rewards/margins": 0.04520436376333237,
"rewards/rejected": -0.32426005601882935,
"step": 65
},
{
"epoch": 0.14963259853039412,
"grad_norm": 6.213649082386088,
"learning_rate": 4.963094133060148e-07,
"logits/chosen": -0.10034845024347305,
"logits/rejected": -0.10583126544952393,
"logps/chosen": -0.26704102754592896,
"logps/rejected": -0.3004146218299866,
"loss": 1.26,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.26704102754592896,
"rewards/margins": 0.03337355703115463,
"rewards/rejected": -0.3004146218299866,
"step": 70
},
{
"epoch": 0.16032064128256512,
"grad_norm": 5.705188459832667,
"learning_rate": 4.945369001834514e-07,
"logits/chosen": -0.04172532260417938,
"logits/rejected": -0.09454617649316788,
"logps/chosen": -0.3160458207130432,
"logps/rejected": -0.3161882162094116,
"loss": 1.2559,
"rewards/accuracies": 0.42500001192092896,
"rewards/chosen": -0.3160458207130432,
"rewards/margins": 0.00014241260942071676,
"rewards/rejected": -0.3161882162094116,
"step": 75
},
{
"epoch": 0.17100868403473615,
"grad_norm": 8.887537168724824,
"learning_rate": 4.924223800941717e-07,
"logits/chosen": -0.0778362900018692,
"logits/rejected": -0.06695137917995453,
"logps/chosen": -0.29803267121315,
"logps/rejected": -0.36691388487815857,
"loss": 1.2472,
"rewards/accuracies": 0.518750011920929,
"rewards/chosen": -0.29803267121315,
"rewards/margins": 0.06888123601675034,
"rewards/rejected": -0.36691388487815857,
"step": 80
},
{
"epoch": 0.18169672678690715,
"grad_norm": 8.152252527281247,
"learning_rate": 4.899688103857222e-07,
"logits/chosen": -0.1327560842037201,
"logits/rejected": -0.13574589788913727,
"logps/chosen": -0.30018350481987,
"logps/rejected": -0.3354310095310211,
"loss": 1.2511,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.30018350481987,
"rewards/margins": 0.03524748235940933,
"rewards/rejected": -0.3354310095310211,
"step": 85
},
{
"epoch": 0.19238476953907815,
"grad_norm": 5.762681556310691,
"learning_rate": 4.871796225971999e-07,
"logits/chosen": -0.13756980001926422,
"logits/rejected": -0.13764461874961853,
"logps/chosen": -0.30491501092910767,
"logps/rejected": -0.3326208293437958,
"loss": 1.2581,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.30491501092910767,
"rewards/margins": 0.02770586870610714,
"rewards/rejected": -0.3326208293437958,
"step": 90
},
{
"epoch": 0.20307281229124916,
"grad_norm": 9.098632782132587,
"learning_rate": 4.840587176599343e-07,
"logits/chosen": -0.13816510140895844,
"logits/rejected": -0.14487667381763458,
"logps/chosen": -0.31862613558769226,
"logps/rejected": -0.358321875333786,
"loss": 1.257,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": -0.31862613558769226,
"rewards/margins": 0.03969571739435196,
"rewards/rejected": -0.358321875333786,
"step": 95
},
{
"epoch": 0.21376085504342018,
"grad_norm": 10.10450449804474,
"learning_rate": 4.806104604416823e-07,
"logits/chosen": -0.06023276597261429,
"logits/rejected": -0.09339889138936996,
"logps/chosen": -0.3247331976890564,
"logps/rejected": -0.3496086895465851,
"loss": 1.2595,
"rewards/accuracies": 0.53125,
"rewards/chosen": -0.3247331976890564,
"rewards/margins": 0.024875465780496597,
"rewards/rejected": -0.3496086895465851,
"step": 100
},
{
"epoch": 0.22444889779559118,
"grad_norm": 7.472460806707148,
"learning_rate": 4.768396736419662e-07,
"logits/chosen": -0.027901506051421165,
"logits/rejected": -0.022181281819939613,
"logps/chosen": -0.29272037744522095,
"logps/rejected": -0.33167213201522827,
"loss": 1.2473,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.29272037744522095,
"rewards/margins": 0.03895174711942673,
"rewards/rejected": -0.33167213201522827,
"step": 105
},
{
"epoch": 0.23513694054776219,
"grad_norm": 6.945973203940155,
"learning_rate": 4.7275163104709194e-07,
"logits/chosen": -0.08409769833087921,
"logits/rejected": -0.10143758356571198,
"logps/chosen": -0.31818658113479614,
"logps/rejected": -0.36014416813850403,
"loss": 1.2353,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.31818658113479614,
"rewards/margins": 0.041957564651966095,
"rewards/rejected": -0.36014416813850403,
"step": 110
},
{
"epoch": 0.2458249832999332,
"grad_norm": 8.678451264926553,
"learning_rate": 4.683520501542824e-07,
"logits/chosen": -0.04873973876237869,
"logits/rejected": -0.06372807919979095,
"logps/chosen": -0.29218122363090515,
"logps/rejected": -0.36307165026664734,
"loss": 1.2485,
"rewards/accuracies": 0.518750011920929,
"rewards/chosen": -0.29218122363090515,
"rewards/margins": 0.07089041918516159,
"rewards/rejected": -0.36307165026664734,
"step": 115
},
{
"epoch": 0.2565130260521042,
"grad_norm": 11.63534607726989,
"learning_rate": 4.636470841752404e-07,
"logits/chosen": -0.06055726855993271,
"logits/rejected": -0.07100611180067062,
"logps/chosen": -0.31277093291282654,
"logps/rejected": -0.3368292450904846,
"loss": 1.2573,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.31277093291282654,
"rewards/margins": 0.024058334529399872,
"rewards/rejected": -0.3368292450904846,
"step": 120
},
{
"epoch": 0.26720106880427524,
"grad_norm": 8.135579287576856,
"learning_rate": 4.5864331343032565e-07,
"logits/chosen": -0.09519238024950027,
"logits/rejected": -0.08622661232948303,
"logps/chosen": -0.3319314420223236,
"logps/rejected": -0.4088417887687683,
"loss": 1.2506,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.3319314420223236,
"rewards/margins": 0.0769103392958641,
"rewards/rejected": -0.4088417887687683,
"step": 125
},
{
"epoch": 0.27788911155644624,
"grad_norm": 7.979535122408603,
"learning_rate": 4.533477361453819e-07,
"logits/chosen": -0.12700992822647095,
"logits/rejected": -0.1654721200466156,
"logps/chosen": -0.31886520981788635,
"logps/rejected": -0.3665064573287964,
"loss": 1.2412,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": -0.31886520981788635,
"rewards/margins": 0.04764125496149063,
"rewards/rejected": -0.3665064573287964,
"step": 130
},
{
"epoch": 0.28857715430861725,
"grad_norm": 6.251297384030462,
"learning_rate": 4.4776775866408533e-07,
"logits/chosen": -0.0819496437907219,
"logits/rejected": -0.07078047096729279,
"logps/chosen": -0.3171434700489044,
"logps/rejected": -0.4331962466239929,
"loss": 1.239,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.3171434700489044,
"rewards/margins": 0.11605282127857208,
"rewards/rejected": -0.4331962466239929,
"step": 135
},
{
"epoch": 0.29926519706078825,
"grad_norm": 5.793187283148429,
"learning_rate": 4.4191118508950277e-07,
"logits/chosen": -0.11466574668884277,
"logits/rejected": -0.13573846220970154,
"logps/chosen": -0.3010789752006531,
"logps/rejected": -0.34237930178642273,
"loss": 1.2512,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": -0.3010789752006531,
"rewards/margins": 0.04130028933286667,
"rewards/rejected": -0.34237930178642273,
"step": 140
},
{
"epoch": 0.30995323981295925,
"grad_norm": 11.065869050969136,
"learning_rate": 4.357862063693485e-07,
"logits/chosen": 0.006140389945358038,
"logits/rejected": -0.06892713904380798,
"logps/chosen": -0.3436379134654999,
"logps/rejected": -0.39155465364456177,
"loss": 1.2445,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": -0.3436379134654999,
"rewards/margins": 0.04791674017906189,
"rewards/rejected": -0.39155465364456177,
"step": 145
},
{
"epoch": 0.32064128256513025,
"grad_norm": 6.596954249277138,
"learning_rate": 4.294013888402029e-07,
"logits/chosen": -0.03650084137916565,
"logits/rejected": -0.05850226432085037,
"logps/chosen": -0.3343154788017273,
"logps/rejected": -0.4158911108970642,
"loss": 1.2408,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.3343154788017273,
"rewards/margins": 0.08157569169998169,
"rewards/rejected": -0.4158911108970642,
"step": 150
},
{
"epoch": 0.33132932531730125,
"grad_norm": 7.905444971631688,
"learning_rate": 4.227656622467162e-07,
"logits/chosen": -0.057654332369565964,
"logits/rejected": -0.08741056174039841,
"logps/chosen": -0.3156217336654663,
"logps/rejected": -0.38537663221359253,
"loss": 1.2467,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.3156217336654663,
"rewards/margins": 0.06975491344928741,
"rewards/rejected": -0.38537663221359253,
"step": 155
},
{
"epoch": 0.3420173680694723,
"grad_norm": 6.977467024402979,
"learning_rate": 4.158883072525528e-07,
"logits/chosen": -0.031004533171653748,
"logits/rejected": -0.07640744745731354,
"logps/chosen": -0.3484176993370056,
"logps/rejected": -0.38878944516181946,
"loss": 1.2482,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.3484176993370056,
"rewards/margins": 0.04037177562713623,
"rewards/rejected": -0.38878944516181946,
"step": 160
},
{
"epoch": 0.3527054108216433,
"grad_norm": 7.010433146556675,
"learning_rate": 4.087789424605447e-07,
"logits/chosen": -0.1136682853102684,
"logits/rejected": -0.11334715038537979,
"logps/chosen": -0.31154853105545044,
"logps/rejected": -0.3925136625766754,
"loss": 1.2296,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.31154853105545044,
"rewards/margins": 0.08096513897180557,
"rewards/rejected": -0.3925136625766754,
"step": 165
},
{
"epoch": 0.3633934535738143,
"grad_norm": 8.49386356506166,
"learning_rate": 4.0144751096020497e-07,
"logits/chosen": -0.0033298402559012175,
"logits/rejected": 0.00819315668195486,
"logps/chosen": -0.3212627172470093,
"logps/rejected": -0.45144709944725037,
"loss": 1.233,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.3212627172470093,
"rewards/margins": 0.13018442690372467,
"rewards/rejected": -0.45144709944725037,
"step": 170
},
{
"epoch": 0.3740814963259853,
"grad_norm": 8.76872679811561,
"learning_rate": 3.939042664214184e-07,
"logits/chosen": -0.05802924558520317,
"logits/rejected": -0.1102098673582077,
"logps/chosen": -0.3099963366985321,
"logps/rejected": -0.4006505608558655,
"loss": 1.2361,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.3099963366985321,
"rewards/margins": 0.0906541496515274,
"rewards/rejected": -0.4006505608558655,
"step": 175
},
{
"epoch": 0.3847695390781563,
"grad_norm": 7.540326639819017,
"learning_rate": 3.8615975875375676e-07,
"logits/chosen": -0.009535525925457478,
"logits/rejected": -0.058530379086732864,
"logps/chosen": -0.3339589238166809,
"logps/rejected": -0.40914154052734375,
"loss": 1.2362,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.3339589238166809,
"rewards/margins": 0.07518256455659866,
"rewards/rejected": -0.40914154052734375,
"step": 180
},
{
"epoch": 0.3954575818303273,
"grad_norm": 7.4249629233111305,
"learning_rate": 3.7822481935147655e-07,
"logits/chosen": -0.03331337124109268,
"logits/rejected": -0.07910569757223129,
"logps/chosen": -0.36260131001472473,
"logps/rejected": -0.4173446297645569,
"loss": 1.2293,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.36260131001472473,
"rewards/margins": 0.05474331974983215,
"rewards/rejected": -0.4173446297645569,
"step": 185
},
{
"epoch": 0.4061456245824983,
"grad_norm": 5.280023907320462,
"learning_rate": 3.7011054594483443e-07,
"logits/chosen": -0.055047959089279175,
"logits/rejected": -0.10345073789358139,
"logps/chosen": -0.3115054965019226,
"logps/rejected": -0.3910045027732849,
"loss": 1.2306,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.3115054965019226,
"rewards/margins": 0.0794990137219429,
"rewards/rejected": -0.3910045027732849,
"step": 190
},
{
"epoch": 0.4168336673346693,
"grad_norm": 7.696490999035101,
"learning_rate": 3.618282870789081e-07,
"logits/chosen": -0.030732590705156326,
"logits/rejected": -0.09086582809686661,
"logps/chosen": -0.33225589990615845,
"logps/rejected": -0.4093093276023865,
"loss": 1.238,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.33225589990615845,
"rewards/margins": 0.07705340534448624,
"rewards/rejected": -0.4093093276023865,
"step": 195
},
{
"epoch": 0.42752171008684037,
"grad_norm": 9.865524537565884,
"learning_rate": 3.5338962624163016e-07,
"logits/chosen": -0.04794201999902725,
"logits/rejected": -0.07962854951620102,
"logps/chosen": -0.32879868149757385,
"logps/rejected": -0.3706124424934387,
"loss": 1.2347,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.32879868149757385,
"rewards/margins": 0.04181372746825218,
"rewards/rejected": -0.3706124424934387,
"step": 200
},
{
"epoch": 0.43820975283901137,
"grad_norm": 11.03217764528719,
"learning_rate": 3.448063656632321e-07,
"logits/chosen": 0.025639599189162254,
"logits/rejected": -0.04025055840611458,
"logps/chosen": -0.33458632230758667,
"logps/rejected": -0.4239538311958313,
"loss": 1.2165,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.33458632230758667,
"rewards/margins": 0.08936752378940582,
"rewards/rejected": -0.4239538311958313,
"step": 205
},
{
"epoch": 0.44889779559118237,
"grad_norm": 9.18905664278058,
"learning_rate": 3.360905098097587e-07,
"logits/chosen": -0.15807747840881348,
"logits/rejected": -0.1486450880765915,
"logps/chosen": -0.3539409935474396,
"logps/rejected": -0.4119337201118469,
"loss": 1.2296,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.3539409935474396,
"rewards/margins": 0.057992756366729736,
"rewards/rejected": -0.4119337201118469,
"step": 210
},
{
"epoch": 0.45958583834335337,
"grad_norm": 5.6200397770452435,
"learning_rate": 3.272542485937368e-07,
"logits/chosen": -0.059969425201416016,
"logits/rejected": -0.059418417513370514,
"logps/chosen": -0.33138805627822876,
"logps/rejected": -0.41279467940330505,
"loss": 1.2131,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.33138805627822876,
"rewards/margins": 0.0814066231250763,
"rewards/rejected": -0.41279467940330505,
"step": 215
},
{
"epoch": 0.47027388109552437,
"grad_norm": 11.661689635196453,
"learning_rate": 3.1830994032548e-07,
"logits/chosen": -0.0942499116063118,
"logits/rejected": -0.13178740441799164,
"logps/chosen": -0.3507417142391205,
"logps/rejected": -0.415250301361084,
"loss": 1.2281,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.3507417142391205,
"rewards/margins": 0.0645085945725441,
"rewards/rejected": -0.415250301361084,
"step": 220
},
{
"epoch": 0.48096192384769537,
"grad_norm": 8.913346690203172,
"learning_rate": 3.0927009442887437e-07,
"logits/chosen": -0.09192110598087311,
"logits/rejected": -0.11192705482244492,
"logps/chosen": -0.3596979081630707,
"logps/rejected": -0.4359111785888672,
"loss": 1.2282,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.3596979081630707,
"rewards/margins": 0.07621324807405472,
"rewards/rejected": -0.4359111785888672,
"step": 225
},
{
"epoch": 0.4916499665998664,
"grad_norm": 8.147198569990623,
"learning_rate": 3.001473539458182e-07,
"logits/chosen": -0.06644257158041,
"logits/rejected": -0.06492500752210617,
"logps/chosen": -0.3724405765533447,
"logps/rejected": -0.4937649667263031,
"loss": 1.2288,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.3724405765533447,
"rewards/margins": 0.12132441997528076,
"rewards/rejected": -0.4937649667263031,
"step": 230
},
{
"epoch": 0.5023380093520374,
"grad_norm": 5.622548262622513,
"learning_rate": 2.909544778537844e-07,
"logits/chosen": -0.052472569048404694,
"logits/rejected": -0.08882959187030792,
"logps/chosen": -0.3525467813014984,
"logps/rejected": -0.4478563666343689,
"loss": 1.2195,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.3525467813014984,
"rewards/margins": 0.09530961513519287,
"rewards/rejected": -0.4478563666343689,
"step": 235
},
{
"epoch": 0.5130260521042084,
"grad_norm": 13.487096025706906,
"learning_rate": 2.817043232212371e-07,
"logits/chosen": -0.04630335047841072,
"logits/rejected": -0.09291209280490875,
"logps/chosen": -0.3766464293003082,
"logps/rejected": -0.5016573071479797,
"loss": 1.2266,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.3766464293003082,
"rewards/margins": 0.1250108778476715,
"rewards/rejected": -0.5016573071479797,
"step": 240
},
{
"epoch": 0.5237140948563794,
"grad_norm": 7.58291755925311,
"learning_rate": 2.7240982722585837e-07,
"logits/chosen": -0.06226348876953125,
"logits/rejected": -0.11048658192157745,
"logps/chosen": -0.3789404034614563,
"logps/rejected": -0.4952840805053711,
"loss": 1.2384,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.3789404034614563,
"rewards/margins": 0.1163436621427536,
"rewards/rejected": -0.4952840805053711,
"step": 245
},
{
"epoch": 0.5344021376085505,
"grad_norm": 7.990353520246757,
"learning_rate": 2.63083989060736e-07,
"logits/chosen": -0.012785041704773903,
"logits/rejected": -0.03117755614221096,
"logps/chosen": -0.37290263175964355,
"logps/rejected": -0.4822382926940918,
"loss": 1.2119,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.37290263175964355,
"rewards/margins": 0.10933564603328705,
"rewards/rejected": -0.4822382926940918,
"step": 250
},
{
"epoch": 0.5450901803607214,
"grad_norm": 8.562299251584639,
"learning_rate": 2.537398517538159e-07,
"logits/chosen": -0.1063912883400917,
"logits/rejected": -0.15656878054141998,
"logps/chosen": -0.3503562808036804,
"logps/rejected": -0.47785887122154236,
"loss": 1.2099,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.3503562808036804,
"rewards/margins": 0.12750259041786194,
"rewards/rejected": -0.47785887122154236,
"step": 255
},
{
"epoch": 0.5557782231128925,
"grad_norm": 7.240239076166087,
"learning_rate": 2.4439048392604877e-07,
"logits/chosen": -0.11992867290973663,
"logits/rejected": -0.18333931267261505,
"logps/chosen": -0.37922850251197815,
"logps/rejected": -0.5119472742080688,
"loss": 1.2263,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.37922850251197815,
"rewards/margins": 0.13271880149841309,
"rewards/rejected": -0.5119472742080688,
"step": 260
},
{
"epoch": 0.5664662658650634,
"grad_norm": 10.645153522332302,
"learning_rate": 2.3504896151374144e-07,
"logits/chosen": -0.02706807851791382,
"logits/rejected": -0.0021373010240495205,
"logps/chosen": -0.36091288924217224,
"logps/rejected": -0.492805153131485,
"loss": 1.2117,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.36091288924217224,
"rewards/margins": 0.13189218938350677,
"rewards/rejected": -0.492805153131485,
"step": 265
},
{
"epoch": 0.5771543086172345,
"grad_norm": 10.372879131327405,
"learning_rate": 2.2572834948067795e-07,
"logits/chosen": -0.07857657968997955,
"logits/rejected": -0.10765037685632706,
"logps/chosen": -0.375199556350708,
"logps/rejected": -0.474733829498291,
"loss": 1.2198,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.375199556350708,
"rewards/margins": 0.09953425824642181,
"rewards/rejected": -0.474733829498291,
"step": 270
},
{
"epoch": 0.5878423513694054,
"grad_norm": 10.015449912582644,
"learning_rate": 2.164416835455862e-07,
"logits/chosen": -0.09552633762359619,
"logits/rejected": -0.1291528046131134,
"logps/chosen": -0.3576242923736572,
"logps/rejected": -0.48135852813720703,
"loss": 1.1989,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.3576242923736572,
"rewards/margins": 0.1237342581152916,
"rewards/rejected": -0.48135852813720703,
"step": 275
},
{
"epoch": 0.5985303941215765,
"grad_norm": 10.140058545265708,
"learning_rate": 2.072019519505062e-07,
"logits/chosen": -0.07462620735168457,
"logits/rejected": -0.060869015753269196,
"logps/chosen": -0.35343998670578003,
"logps/rejected": -0.5310408473014832,
"loss": 1.2063,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.35343998670578003,
"rewards/margins": 0.17760084569454193,
"rewards/rejected": -0.5310408473014832,
"step": 280
},
{
"epoch": 0.6092184368737475,
"grad_norm": 9.165516193336138,
"learning_rate": 1.980220772955602e-07,
"logits/chosen": -0.10269057750701904,
"logits/rejected": -0.10618944466114044,
"logps/chosen": -0.37109121680259705,
"logps/rejected": -0.4779582619667053,
"loss": 1.2122,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.37109121680259705,
"rewards/margins": 0.1068669781088829,
"rewards/rejected": -0.4779582619667053,
"step": 285
},
{
"epoch": 0.6199064796259185,
"grad_norm": 7.532797477106183,
"learning_rate": 1.8891489846552644e-07,
"logits/chosen": -0.09100791066884995,
"logits/rejected": -0.11063895374536514,
"logps/chosen": -0.37783291935920715,
"logps/rejected": -0.5191008448600769,
"loss": 1.2027,
"rewards/accuracies": 0.7562500238418579,
"rewards/chosen": -0.37783291935920715,
"rewards/margins": 0.14126797020435333,
"rewards/rejected": -0.5191008448600769,
"step": 290
},
{
"epoch": 0.6305945223780896,
"grad_norm": 11.187112696193623,
"learning_rate": 1.7989315267349933e-07,
"logits/chosen": -0.1488272249698639,
"logits/rejected": -0.1783982813358307,
"logps/chosen": -0.36716532707214355,
"logps/rejected": -0.4786455035209656,
"loss": 1.2272,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.36716532707214355,
"rewards/margins": 0.11148019134998322,
"rewards/rejected": -0.4786455035209656,
"step": 295
},
{
"epoch": 0.6412825651302605,
"grad_norm": 8.931102714134676,
"learning_rate": 1.7096945764674398e-07,
"logits/chosen": -0.12898632884025574,
"logits/rejected": -0.16446474194526672,
"logps/chosen": -0.38750600814819336,
"logps/rejected": -0.49209412932395935,
"loss": 1.2176,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.38750600814819336,
"rewards/margins": 0.10458817332983017,
"rewards/rejected": -0.49209412932395935,
"step": 300
},
{
"epoch": 0.6519706078824316,
"grad_norm": 9.395577531406225,
"learning_rate": 1.621562939796643e-07,
"logits/chosen": -0.11750619113445282,
"logits/rejected": -0.16459839046001434,
"logps/chosen": -0.3916749060153961,
"logps/rejected": -0.4877534806728363,
"loss": 1.2037,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.3916749060153961,
"rewards/margins": 0.09607864916324615,
"rewards/rejected": -0.4877534806728363,
"step": 305
},
{
"epoch": 0.6626586506346025,
"grad_norm": 7.24387245035212,
"learning_rate": 1.5346598767856345e-07,
"logits/chosen": -0.04836176708340645,
"logits/rejected": -0.13036546111106873,
"logps/chosen": -0.3734469711780548,
"logps/rejected": -0.47967085242271423,
"loss": 1.2106,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.3734469711780548,
"rewards/margins": 0.10622383654117584,
"rewards/rejected": -0.47967085242271423,
"step": 310
},
{
"epoch": 0.6733466933867736,
"grad_norm": 8.17250962769366,
"learning_rate": 1.4491069292260866e-07,
"logits/chosen": -0.08944126963615417,
"logits/rejected": -0.13419121503829956,
"logps/chosen": -0.3673272132873535,
"logps/rejected": -0.5219535231590271,
"loss": 1.2136,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.3673272132873535,
"rewards/margins": 0.15462630987167358,
"rewards/rejected": -0.5219535231590271,
"step": 315
},
{
"epoch": 0.6840347361389446,
"grad_norm": 9.54861228062624,
"learning_rate": 1.365023750651133e-07,
"logits/chosen": -0.129867285490036,
"logits/rejected": -0.12771280109882355,
"logps/chosen": -0.38212448358535767,
"logps/rejected": -0.5496698617935181,
"loss": 1.1997,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.38212448358535767,
"rewards/margins": 0.16754531860351562,
"rewards/rejected": -0.5496698617935181,
"step": 320
},
{
"epoch": 0.6947227788911156,
"grad_norm": 6.810179057310646,
"learning_rate": 1.2825279389890818e-07,
"logits/chosen": -0.06729793548583984,
"logits/rejected": -0.10314921289682388,
"logps/chosen": -0.37868839502334595,
"logps/rejected": -0.49213042855262756,
"loss": 1.2092,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.37868839502334595,
"rewards/margins": 0.11344204843044281,
"rewards/rejected": -0.49213042855262756,
"step": 325
},
{
"epoch": 0.7054108216432866,
"grad_norm": 6.608583522902243,
"learning_rate": 1.201734872092077e-07,
"logits/chosen": -0.07167549431324005,
"logits/rejected": -0.09838903695344925,
"logps/chosen": -0.37149468064308167,
"logps/rejected": -0.49603986740112305,
"loss": 1.2048,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.37149468064308167,
"rewards/margins": 0.1245451420545578,
"rewards/rejected": -0.49603986740112305,
"step": 330
},
{
"epoch": 0.7160988643954576,
"grad_norm": 13.249759188798002,
"learning_rate": 1.1227575463697439e-07,
"logits/chosen": -0.0337320938706398,
"logits/rejected": -0.06822805106639862,
"logps/chosen": -0.3788215219974518,
"logps/rejected": -0.5212149620056152,
"loss": 1.2014,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.3788215219974518,
"rewards/margins": 0.14239344000816345,
"rewards/rejected": -0.5212149620056152,
"step": 335
},
{
"epoch": 0.7267869071476286,
"grad_norm": 13.767888565392328,
"learning_rate": 1.0457064187534861e-07,
"logits/chosen": -0.09795643389225006,
"logits/rejected": -0.11368696391582489,
"logps/chosen": -0.3989834785461426,
"logps/rejected": -0.5320654511451721,
"loss": 1.2179,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.3989834785461426,
"rewards/margins": 0.13308197259902954,
"rewards/rejected": -0.5320654511451721,
"step": 340
},
{
"epoch": 0.7374749498997996,
"grad_norm": 12.298322164661563,
"learning_rate": 9.706892522124838e-08,
"logits/chosen": -0.1807522028684616,
"logits/rejected": -0.2426176518201828,
"logps/chosen": -0.38689613342285156,
"logps/rejected": -0.5023618340492249,
"loss": 1.2155,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.38689613342285156,
"rewards/margins": 0.11546576023101807,
"rewards/rejected": -0.5023618340492249,
"step": 345
},
{
"epoch": 0.7481629926519706,
"grad_norm": 14.580781339606604,
"learning_rate": 8.978109650374396e-08,
"logits/chosen": -0.12330816686153412,
"logits/rejected": -0.16493481397628784,
"logps/chosen": -0.3923422396183014,
"logps/rejected": -0.512184739112854,
"loss": 1.2063,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.3923422396183014,
"rewards/margins": 0.11984255164861679,
"rewards/rejected": -0.512184739112854,
"step": 350
},
{
"epoch": 0.7588510354041417,
"grad_norm": 9.440132311200133,
"learning_rate": 8.271734841028552e-08,
"logits/chosen": -0.11258058249950409,
"logits/rejected": -0.08831019699573517,
"logps/chosen": -0.3702935576438904,
"logps/rejected": -0.5448261499404907,
"loss": 1.1984,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": -0.3702935576438904,
"rewards/margins": 0.17453262209892273,
"rewards/rejected": -0.5448261499404907,
"step": 355
},
{
"epoch": 0.7695390781563126,
"grad_norm": 7.623549767902818,
"learning_rate": 7.588756023130833e-08,
"logits/chosen": -0.13030676543712616,
"logits/rejected": -0.14040207862854004,
"logps/chosen": -0.40232768654823303,
"logps/rejected": -0.5432428121566772,
"loss": 1.2027,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.40232768654823303,
"rewards/margins": 0.1409151703119278,
"rewards/rejected": -0.5432428121566772,
"step": 360
},
{
"epoch": 0.7802271209084837,
"grad_norm": 9.392417635076814,
"learning_rate": 6.930128404315214e-08,
"logits/chosen": -0.08242303133010864,
"logits/rejected": -0.14365339279174805,
"logps/chosen": -0.3742930591106415,
"logps/rejected": -0.48327428102493286,
"loss": 1.2155,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.3742930591106415,
"rewards/margins": 0.10898121446371078,
"rewards/rejected": -0.48327428102493286,
"step": 365
},
{
"epoch": 0.7909151636606546,
"grad_norm": 7.694121650797251,
"learning_rate": 6.296773134861824e-08,
"logits/chosen": -0.178551584482193,
"logits/rejected": -0.19603314995765686,
"logps/chosen": -0.41331925988197327,
"logps/rejected": -0.5020118951797485,
"loss": 1.2167,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.41331925988197327,
"rewards/margins": 0.08869270235300064,
"rewards/rejected": -0.5020118951797485,
"step": 370
},
{
"epoch": 0.8016032064128257,
"grad_norm": 8.81810478954552,
"learning_rate": 5.6895760193850145e-08,
"logits/chosen": -0.09992343187332153,
"logits/rejected": -0.10953284800052643,
"logps/chosen": -0.4108389914035797,
"logps/rejected": -0.5310851335525513,
"loss": 1.2019,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.4108389914035797,
"rewards/margins": 0.12024614959955215,
"rewards/rejected": -0.5310851335525513,
"step": 375
},
{
"epoch": 0.8122912491649966,
"grad_norm": 8.345914437235688,
"learning_rate": 5.109386277955477e-08,
"logits/chosen": -0.07882420718669891,
"logits/rejected": -0.12227978557348251,
"logps/chosen": -0.404563844203949,
"logps/rejected": -0.47655385732650757,
"loss": 1.2129,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.404563844203949,
"rewards/margins": 0.07199002057313919,
"rewards/rejected": -0.47655385732650757,
"step": 380
},
{
"epoch": 0.8229792919171677,
"grad_norm": 15.681686443247571,
"learning_rate": 4.557015358389216e-08,
"logits/chosen": -0.06385800987482071,
"logits/rejected": -0.03621825575828552,
"logps/chosen": -0.3926911950111389,
"logps/rejected": -0.5792160034179688,
"loss": 1.2111,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": -0.3926911950111389,
"rewards/margins": 0.18652474880218506,
"rewards/rejected": -0.5792160034179688,
"step": 385
},
{
"epoch": 0.8336673346693386,
"grad_norm": 9.223256642330082,
"learning_rate": 4.0332358013644015e-08,
"logits/chosen": -0.03278634697198868,
"logits/rejected": -0.13274672627449036,
"logps/chosen": -0.3848794102668762,
"logps/rejected": -0.5192353129386902,
"loss": 1.2095,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.3848794102668762,
"rewards/margins": 0.134355828166008,
"rewards/rejected": -0.5192353129386902,
"step": 390
},
{
"epoch": 0.8443553774215097,
"grad_norm": 8.533321895055286,
"learning_rate": 3.538780159953347e-08,
"logits/chosen": -0.10506091266870499,
"logits/rejected": -0.11381878703832626,
"logps/chosen": -0.3806256651878357,
"logps/rejected": -0.5088571906089783,
"loss": 1.1919,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.3806256651878357,
"rewards/margins": 0.12823154032230377,
"rewards/rejected": -0.5088571906089783,
"step": 395
},
{
"epoch": 0.8550434201736807,
"grad_norm": 6.692568306598245,
"learning_rate": 3.074339975080836e-08,
"logits/chosen": -0.10104751586914062,
"logits/rejected": -0.08141423761844635,
"logps/chosen": -0.38590914011001587,
"logps/rejected": -0.491323858499527,
"loss": 1.2128,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.38590914011001587,
"rewards/margins": 0.10541467368602753,
"rewards/rejected": -0.491323858499527,
"step": 400
},
{
"epoch": 0.8550434201736807,
"eval_logits/chosen": 0.102480448782444,
"eval_logits/rejected": 0.08468419313430786,
"eval_logps/chosen": -0.3920242488384247,
"eval_logps/rejected": -0.5305575132369995,
"eval_loss": 1.2118037939071655,
"eval_rewards/accuracies": 0.7174796462059021,
"eval_rewards/chosen": -0.3920242488384247,
"eval_rewards/margins": 0.13853329420089722,
"eval_rewards/rejected": -0.5305575132369995,
"eval_runtime": 426.9778,
"eval_samples_per_second": 4.593,
"eval_steps_per_second": 0.288,
"step": 400
},
{
"epoch": 0.8657314629258517,
"grad_norm": 15.478975607455368,
"learning_rate": 2.6405648083415833e-08,
"logits/chosen": -0.10661216825246811,
"logits/rejected": -0.15089339017868042,
"logps/chosen": -0.40621575713157654,
"logps/rejected": -0.5219891667366028,
"loss": 1.2137,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.40621575713157654,
"rewards/margins": 0.11577335745096207,
"rewards/rejected": -0.5219891667366028,
"step": 405
},
{
"epoch": 0.8764195056780227,
"grad_norm": 10.163686409803713,
"learning_rate": 2.2380613335296033e-08,
"logits/chosen": -0.1509750932455063,
"logits/rejected": -0.1919194757938385,
"logps/chosen": -0.3878978490829468,
"logps/rejected": -0.49552789330482483,
"loss": 1.2191,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.3878978490829468,
"rewards/margins": 0.10763005912303925,
"rewards/rejected": -0.49552789330482483,
"step": 410
},
{
"epoch": 0.8871075484301937,
"grad_norm": 10.299286899820958,
"learning_rate": 1.8673924881500823e-08,
"logits/chosen": -0.055319927632808685,
"logits/rejected": -0.10443171113729477,
"logps/chosen": -0.41571861505508423,
"logps/rejected": -0.5507162809371948,
"loss": 1.2175,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.41571861505508423,
"rewards/margins": 0.13499762117862701,
"rewards/rejected": -0.5507162809371948,
"step": 415
},
{
"epoch": 0.8977955911823647,
"grad_norm": 11.270224781986835,
"learning_rate": 1.5290766861003475e-08,
"logits/chosen": -0.08639408648014069,
"logits/rejected": -0.15111452341079712,
"logps/chosen": -0.36737141013145447,
"logps/rejected": -0.5103423595428467,
"loss": 1.2052,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.36737141013145447,
"rewards/margins": 0.14297088980674744,
"rewards/rejected": -0.5103423595428467,
"step": 420
},
{
"epoch": 0.9084836339345357,
"grad_norm": 12.926708544168248,
"learning_rate": 1.2235870926211616e-08,
"logits/chosen": 0.01661798171699047,
"logits/rejected": -0.014898231253027916,
"logps/chosen": -0.3715210556983948,
"logps/rejected": -0.47712254524230957,
"loss": 1.2047,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.3715210556983948,
"rewards/margins": 0.105601467192173,
"rewards/rejected": -0.47712254524230957,
"step": 425
},
{
"epoch": 0.9191716766867067,
"grad_norm": 7.163643566777999,
"learning_rate": 9.513509625323518e-09,
"logits/chosen": -0.02985338866710663,
"logits/rejected": -0.11251659691333771,
"logps/chosen": -0.3954943120479584,
"logps/rejected": -0.508020281791687,
"loss": 1.2119,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.3954943120479584,
"rewards/margins": 0.1125260591506958,
"rewards/rejected": -0.508020281791687,
"step": 430
},
{
"epoch": 0.9298597194388778,
"grad_norm": 6.770678170385186,
"learning_rate": 7.127490426783123e-09,
"logits/chosen": -0.12328235805034637,
"logits/rejected": -0.15495580434799194,
"logps/chosen": -0.3658512830734253,
"logps/rejected": -0.5462064146995544,
"loss": 1.2004,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.3658512830734253,
"rewards/margins": 0.1803552210330963,
"rewards/rejected": -0.5462064146995544,
"step": 435
},
{
"epoch": 0.9405477621910487,
"grad_norm": 8.399767120686823,
"learning_rate": 5.08115039419113e-09,
"logits/chosen": -0.11677885055541992,
"logits/rejected": -0.14224281907081604,
"logps/chosen": -0.39421266317367554,
"logps/rejected": -0.5207501649856567,
"loss": 1.2155,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": -0.39421266317367554,
"rewards/margins": 0.12653745710849762,
"rewards/rejected": -0.5207501649856567,
"step": 440
},
{
"epoch": 0.9512358049432198,
"grad_norm": 8.32203352730185,
"learning_rate": 3.3773515191196646e-09,
"logits/chosen": -0.10022030025720596,
"logits/rejected": -0.09951629489660263,
"logps/chosen": -0.3814024329185486,
"logps/rejected": -0.5090724229812622,
"loss": 1.2051,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.3814024329185486,
"rewards/margins": 0.12766999006271362,
"rewards/rejected": -0.5090724229812622,
"step": 445
},
{
"epoch": 0.9619238476953907,
"grad_norm": 9.446955616928552,
"learning_rate": 2.0184767183584474e-09,
"logits/chosen": -0.05858677625656128,
"logits/rejected": -0.07191181182861328,
"logps/chosen": -0.41846340894699097,
"logps/rejected": -0.5389441251754761,
"loss": 1.2009,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.41846340894699097,
"rewards/margins": 0.12048067897558212,
"rewards/rejected": -0.5389441251754761,
"step": 450
},
{
"epoch": 0.9726118904475618,
"grad_norm": 16.2892173576613,
"learning_rate": 1.0064265011902328e-09,
"logits/chosen": -0.019699424505233765,
"logits/rejected": -0.08497850596904755,
"logps/chosen": -0.3675594925880432,
"logps/rejected": -0.4840230941772461,
"loss": 1.1891,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.3675594925880432,
"rewards/margins": 0.11646362394094467,
"rewards/rejected": -0.4840230941772461,
"step": 455
},
{
"epoch": 0.9832999331997327,
"grad_norm": 11.507314738344544,
"learning_rate": 3.4261631135654167e-10,
"logits/chosen": 0.04134285822510719,
"logits/rejected": -0.051879990845918655,
"logps/chosen": -0.373710960149765,
"logps/rejected": -0.507372260093689,
"loss": 1.2095,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.373710960149765,
"rewards/margins": 0.13366124033927917,
"rewards/rejected": -0.507372260093689,
"step": 460
},
{
"epoch": 0.9939879759519038,
"grad_norm": 8.049283368745556,
"learning_rate": 2.797454743164174e-11,
"logits/chosen": -0.040327288210392,
"logits/rejected": -0.08210764825344086,
"logps/chosen": -0.3866434097290039,
"logps/rejected": -0.4750938415527344,
"loss": 1.2088,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.3866434097290039,
"rewards/margins": 0.08845046907663345,
"rewards/rejected": -0.4750938415527344,
"step": 465
},
{
"epoch": 0.9982631930527722,
"step": 467,
"total_flos": 0.0,
"train_loss": 1.2289682925385632,
"train_runtime": 21112.5317,
"train_samples_per_second": 2.836,
"train_steps_per_second": 0.022
}
],
"logging_steps": 5,
"max_steps": 467,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 1000000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}