LLama-8B-Instruct-v0.1-MI-5e-7 / trainer_state.json
tengxiao1
XT
11415f3
{
"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": 7.788778961226321,
"learning_rate": 5.3191489361702123e-08,
"logits/chosen": 0.06188569217920303,
"logits/rejected": 0.03767944127321243,
"logps/chosen": -0.2698679566383362,
"logps/rejected": -0.2681675851345062,
"loss": 1.2746,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.2698679566383362,
"rewards/margins": -0.0017003819812089205,
"rewards/rejected": -0.2681675851345062,
"step": 5
},
{
"epoch": 0.02137608550434202,
"grad_norm": 5.1031447147287015,
"learning_rate": 1.0638297872340425e-07,
"logits/chosen": -0.009919960051774979,
"logits/rejected": -0.011961707845330238,
"logps/chosen": -0.26978033781051636,
"logps/rejected": -0.26762229204177856,
"loss": 1.2723,
"rewards/accuracies": 0.48750001192092896,
"rewards/chosen": -0.26978033781051636,
"rewards/margins": -0.0021580129396170378,
"rewards/rejected": -0.26762229204177856,
"step": 10
},
{
"epoch": 0.03206412825651302,
"grad_norm": 6.478710884493858,
"learning_rate": 1.5957446808510638e-07,
"logits/chosen": 0.0013950124848634005,
"logits/rejected": 0.0043658362701535225,
"logps/chosen": -0.27877503633499146,
"logps/rejected": -0.284164696931839,
"loss": 1.2791,
"rewards/accuracies": 0.44999998807907104,
"rewards/chosen": -0.27877503633499146,
"rewards/margins": 0.005389668978750706,
"rewards/rejected": -0.284164696931839,
"step": 15
},
{
"epoch": 0.04275217100868404,
"grad_norm": 8.960105632559012,
"learning_rate": 2.127659574468085e-07,
"logits/chosen": -0.062316447496414185,
"logits/rejected": -0.06821136176586151,
"logps/chosen": -0.28185540437698364,
"logps/rejected": -0.28997963666915894,
"loss": 1.2761,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.28185540437698364,
"rewards/margins": 0.008124232292175293,
"rewards/rejected": -0.28997963666915894,
"step": 20
},
{
"epoch": 0.053440213760855046,
"grad_norm": 4.829589604518179,
"learning_rate": 2.659574468085106e-07,
"logits/chosen": -0.06830133497714996,
"logits/rejected": -0.047970324754714966,
"logps/chosen": -0.2557140588760376,
"logps/rejected": -0.27477937936782837,
"loss": 1.2567,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.2557140588760376,
"rewards/margins": 0.01906532421708107,
"rewards/rejected": -0.27477937936782837,
"step": 25
},
{
"epoch": 0.06412825651302605,
"grad_norm": 4.836319901094311,
"learning_rate": 3.1914893617021275e-07,
"logits/chosen": -0.014646559953689575,
"logits/rejected": -0.006725949235260487,
"logps/chosen": -0.27856093645095825,
"logps/rejected": -0.29357466101646423,
"loss": 1.2772,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.27856093645095825,
"rewards/margins": 0.015013757161796093,
"rewards/rejected": -0.29357466101646423,
"step": 30
},
{
"epoch": 0.07481629926519706,
"grad_norm": 7.0589927566370365,
"learning_rate": 3.7234042553191484e-07,
"logits/chosen": -0.030227532610297203,
"logits/rejected": -0.009875667281448841,
"logps/chosen": -0.2781219780445099,
"logps/rejected": -0.29566988348960876,
"loss": 1.2752,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": -0.2781219780445099,
"rewards/margins": 0.01754794642329216,
"rewards/rejected": -0.29566988348960876,
"step": 35
},
{
"epoch": 0.08550434201736808,
"grad_norm": 5.508045254637976,
"learning_rate": 4.25531914893617e-07,
"logits/chosen": -0.06681682914495468,
"logits/rejected": -0.07246702909469604,
"logps/chosen": -0.2610124945640564,
"logps/rejected": -0.26930028200149536,
"loss": 1.2684,
"rewards/accuracies": 0.518750011920929,
"rewards/chosen": -0.2610124945640564,
"rewards/margins": 0.008287793025374413,
"rewards/rejected": -0.26930028200149536,
"step": 40
},
{
"epoch": 0.09619238476953908,
"grad_norm": 5.5749249530141105,
"learning_rate": 4.787234042553192e-07,
"logits/chosen": -0.055084120482206345,
"logits/rejected": -0.012284256517887115,
"logps/chosen": -0.28449538350105286,
"logps/rejected": -0.2834285795688629,
"loss": 1.2753,
"rewards/accuracies": 0.53125,
"rewards/chosen": -0.28449538350105286,
"rewards/margins": -0.0010668521281331778,
"rewards/rejected": -0.2834285795688629,
"step": 45
},
{
"epoch": 0.10688042752171009,
"grad_norm": 4.597133841962612,
"learning_rate": 4.999370587356267e-07,
"logits/chosen": -0.019108835607767105,
"logits/rejected": -0.0063239275477826595,
"logps/chosen": -0.2830585837364197,
"logps/rejected": -0.29620057344436646,
"loss": 1.2697,
"rewards/accuracies": 0.53125,
"rewards/chosen": -0.2830585837364197,
"rewards/margins": 0.013141943141818047,
"rewards/rejected": -0.29620057344436646,
"step": 50
},
{
"epoch": 0.11756847027388109,
"grad_norm": 6.058445433480695,
"learning_rate": 4.995525324419337e-07,
"logits/chosen": 0.013618836179375648,
"logits/rejected": -0.007489354815334082,
"logps/chosen": -0.26815882325172424,
"logps/rejected": -0.2764474153518677,
"loss": 1.2635,
"rewards/accuracies": 0.5062500238418579,
"rewards/chosen": -0.26815882325172424,
"rewards/margins": 0.008288596756756306,
"rewards/rejected": -0.2764474153518677,
"step": 55
},
{
"epoch": 0.1282565130260521,
"grad_norm": 6.0182210644235825,
"learning_rate": 4.988189843662815e-07,
"logits/chosen": -0.04398399591445923,
"logits/rejected": -0.07511965185403824,
"logps/chosen": -0.2804185748100281,
"logps/rejected": -0.30631181597709656,
"loss": 1.2651,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.2804185748100281,
"rewards/margins": 0.025893231853842735,
"rewards/rejected": -0.30631181597709656,
"step": 60
},
{
"epoch": 0.13894455577822312,
"grad_norm": 5.758424974883882,
"learning_rate": 4.977374404419837e-07,
"logits/chosen": -0.029157359153032303,
"logits/rejected": 0.02865619957447052,
"logps/chosen": -0.2859181761741638,
"logps/rejected": -0.30712562799453735,
"loss": 1.2709,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.2859181761741638,
"rewards/margins": 0.021207470446825027,
"rewards/rejected": -0.30712562799453735,
"step": 65
},
{
"epoch": 0.14963259853039412,
"grad_norm": 5.178927137492959,
"learning_rate": 4.963094133060148e-07,
"logits/chosen": -0.020770709961652756,
"logits/rejected": -0.024878982454538345,
"logps/chosen": -0.28995445370674133,
"logps/rejected": -0.33237165212631226,
"loss": 1.2584,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": -0.28995445370674133,
"rewards/margins": 0.04241719841957092,
"rewards/rejected": -0.33237165212631226,
"step": 70
},
{
"epoch": 0.16032064128256512,
"grad_norm": 4.430011398176605,
"learning_rate": 4.945369001834514e-07,
"logits/chosen": 0.060086388140916824,
"logits/rejected": 0.10020087659358978,
"logps/chosen": -0.294359028339386,
"logps/rejected": -0.3206758499145508,
"loss": 1.2689,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.294359028339386,
"rewards/margins": 0.026316797360777855,
"rewards/rejected": -0.3206758499145508,
"step": 75
},
{
"epoch": 0.17100868403473615,
"grad_norm": 6.158496310415975,
"learning_rate": 4.924223800941717e-07,
"logits/chosen": -0.014853144064545631,
"logits/rejected": -0.005943991243839264,
"logps/chosen": -0.2843073010444641,
"logps/rejected": -0.3063741624355316,
"loss": 1.2631,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": -0.2843073010444641,
"rewards/margins": 0.022066866979002953,
"rewards/rejected": -0.3063741624355316,
"step": 80
},
{
"epoch": 0.18169672678690715,
"grad_norm": 4.46789094122989,
"learning_rate": 4.899688103857222e-07,
"logits/chosen": 0.009387167170643806,
"logits/rejected": 0.02460663393139839,
"logps/chosen": -0.28670260310173035,
"logps/rejected": -0.3185740113258362,
"loss": 1.2585,
"rewards/accuracies": 0.48124998807907104,
"rewards/chosen": -0.28670260310173035,
"rewards/margins": 0.031871382147073746,
"rewards/rejected": -0.3185740113258362,
"step": 85
},
{
"epoch": 0.19238476953907815,
"grad_norm": 6.324592677678463,
"learning_rate": 4.871796225971999e-07,
"logits/chosen": 0.02077331207692623,
"logits/rejected": 0.04099290445446968,
"logps/chosen": -0.3005926311016083,
"logps/rejected": -0.3475686311721802,
"loss": 1.2557,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.3005926311016083,
"rewards/margins": 0.04697597771883011,
"rewards/rejected": -0.3475686311721802,
"step": 90
},
{
"epoch": 0.20307281229124916,
"grad_norm": 5.797578925192431,
"learning_rate": 4.840587176599343e-07,
"logits/chosen": 0.013538317754864693,
"logits/rejected": -0.004223898518830538,
"logps/chosen": -0.2979055345058441,
"logps/rejected": -0.308336079120636,
"loss": 1.2608,
"rewards/accuracies": 0.46875,
"rewards/chosen": -0.2979055345058441,
"rewards/margins": 0.01043054647743702,
"rewards/rejected": -0.308336079120636,
"step": 95
},
{
"epoch": 0.21376085504342018,
"grad_norm": 5.582465019704645,
"learning_rate": 4.806104604416823e-07,
"logits/chosen": 0.015301126055419445,
"logits/rejected": 0.018461251631379128,
"logps/chosen": -0.2994353473186493,
"logps/rejected": -0.3412705063819885,
"loss": 1.2598,
"rewards/accuracies": 0.48750001192092896,
"rewards/chosen": -0.2994353473186493,
"rewards/margins": 0.04183514788746834,
"rewards/rejected": -0.3412705063819885,
"step": 100
},
{
"epoch": 0.22444889779559118,
"grad_norm": 9.830901210952655,
"learning_rate": 4.768396736419662e-07,
"logits/chosen": -0.006734785623848438,
"logits/rejected": -0.010221516713500023,
"logps/chosen": -0.2789744436740875,
"logps/rejected": -0.3298511505126953,
"loss": 1.2488,
"rewards/accuracies": 0.5062500238418579,
"rewards/chosen": -0.2789744436740875,
"rewards/margins": 0.050876714289188385,
"rewards/rejected": -0.3298511505126953,
"step": 105
},
{
"epoch": 0.23513694054776219,
"grad_norm": 6.080052683983465,
"learning_rate": 4.7275163104709194e-07,
"logits/chosen": -0.01686781272292137,
"logits/rejected": -0.028651485219597816,
"logps/chosen": -0.28804197907447815,
"logps/rejected": -0.3349292278289795,
"loss": 1.2421,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.28804197907447815,
"rewards/margins": 0.04688727483153343,
"rewards/rejected": -0.3349292278289795,
"step": 110
},
{
"epoch": 0.2458249832999332,
"grad_norm": 5.143799640542211,
"learning_rate": 4.683520501542824e-07,
"logits/chosen": 0.015463327057659626,
"logits/rejected": 0.014460188336670399,
"logps/chosen": -0.2945937514305115,
"logps/rejected": -0.33980080485343933,
"loss": 1.2581,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.2945937514305115,
"rewards/margins": 0.04520703852176666,
"rewards/rejected": -0.33980080485343933,
"step": 115
},
{
"epoch": 0.2565130260521042,
"grad_norm": 6.475084807709587,
"learning_rate": 4.636470841752404e-07,
"logits/chosen": -0.04622779041528702,
"logits/rejected": -0.02117704227566719,
"logps/chosen": -0.31878453493118286,
"logps/rejected": -0.3514103889465332,
"loss": 1.259,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.31878453493118286,
"rewards/margins": 0.03262590616941452,
"rewards/rejected": -0.3514103889465332,
"step": 120
},
{
"epoch": 0.26720106880427524,
"grad_norm": 4.989873675244022,
"learning_rate": 4.5864331343032565e-07,
"logits/chosen": -0.0807986855506897,
"logits/rejected": -0.0577215738594532,
"logps/chosen": -0.2865813672542572,
"logps/rejected": -0.3409081995487213,
"loss": 1.2376,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.2865813672542572,
"rewards/margins": 0.05432681366801262,
"rewards/rejected": -0.3409081995487213,
"step": 125
},
{
"epoch": 0.27788911155644624,
"grad_norm": 5.31262070275832,
"learning_rate": 4.533477361453819e-07,
"logits/chosen": -0.03413872793316841,
"logits/rejected": 0.005384595133364201,
"logps/chosen": -0.31652697920799255,
"logps/rejected": -0.35454803705215454,
"loss": 1.2645,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.31652697920799255,
"rewards/margins": 0.03802100569009781,
"rewards/rejected": -0.35454803705215454,
"step": 130
},
{
"epoch": 0.28857715430861725,
"grad_norm": 8.638357586651454,
"learning_rate": 4.4776775866408533e-07,
"logits/chosen": -0.06590205430984497,
"logits/rejected": -0.07048509269952774,
"logps/chosen": -0.30995482206344604,
"logps/rejected": -0.3710452914237976,
"loss": 1.2491,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.30995482206344604,
"rewards/margins": 0.06109045818448067,
"rewards/rejected": -0.3710452914237976,
"step": 135
},
{
"epoch": 0.29926519706078825,
"grad_norm": 7.764947968857639,
"learning_rate": 4.4191118508950277e-07,
"logits/chosen": -0.15944652259349823,
"logits/rejected": -0.15655198693275452,
"logps/chosen": -0.34581616520881653,
"logps/rejected": -0.3991614580154419,
"loss": 1.2569,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": -0.34581616520881653,
"rewards/margins": 0.053345322608947754,
"rewards/rejected": -0.3991614580154419,
"step": 140
},
{
"epoch": 0.30995323981295925,
"grad_norm": 5.8509585456477184,
"learning_rate": 4.357862063693485e-07,
"logits/chosen": -0.020959481596946716,
"logits/rejected": -0.04588431119918823,
"logps/chosen": -0.30446547269821167,
"logps/rejected": -0.35463085770606995,
"loss": 1.2397,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.30446547269821167,
"rewards/margins": 0.05016540735960007,
"rewards/rejected": -0.35463085770606995,
"step": 145
},
{
"epoch": 0.32064128256513025,
"grad_norm": 13.318848211554554,
"learning_rate": 4.294013888402029e-07,
"logits/chosen": -0.06783692538738251,
"logits/rejected": -0.08151838928461075,
"logps/chosen": -0.3216368556022644,
"logps/rejected": -0.41458660364151,
"loss": 1.2473,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.3216368556022644,
"rewards/margins": 0.09294974058866501,
"rewards/rejected": -0.41458660364151,
"step": 150
},
{
"epoch": 0.33132932531730125,
"grad_norm": 7.694806881002591,
"learning_rate": 4.227656622467162e-07,
"logits/chosen": -0.06315052509307861,
"logits/rejected": -0.062115978449583054,
"logps/chosen": -0.3149452805519104,
"logps/rejected": -0.356551855802536,
"loss": 1.2506,
"rewards/accuracies": 0.4625000059604645,
"rewards/chosen": -0.3149452805519104,
"rewards/margins": 0.04160655289888382,
"rewards/rejected": -0.356551855802536,
"step": 155
},
{
"epoch": 0.3420173680694723,
"grad_norm": 6.757643144184794,
"learning_rate": 4.158883072525528e-07,
"logits/chosen": -0.0846010223031044,
"logits/rejected": -0.05645376443862915,
"logps/chosen": -0.31077316403388977,
"logps/rejected": -0.3692784905433655,
"loss": 1.2549,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.31077316403388977,
"rewards/margins": 0.0585053376853466,
"rewards/rejected": -0.3692784905433655,
"step": 160
},
{
"epoch": 0.3527054108216433,
"grad_norm": 8.199115650846204,
"learning_rate": 4.087789424605447e-07,
"logits/chosen": -0.03875167295336723,
"logits/rejected": -0.05049648880958557,
"logps/chosen": -0.29271596670150757,
"logps/rejected": -0.36321207880973816,
"loss": 1.2443,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.29271596670150757,
"rewards/margins": 0.0704960897564888,
"rewards/rejected": -0.36321207880973816,
"step": 165
},
{
"epoch": 0.3633934535738143,
"grad_norm": 5.474896231470313,
"learning_rate": 4.0144751096020497e-07,
"logits/chosen": -0.02952517569065094,
"logits/rejected": -0.003586936742067337,
"logps/chosen": -0.3152143359184265,
"logps/rejected": -0.37791121006011963,
"loss": 1.2466,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.3152143359184265,
"rewards/margins": 0.06269686669111252,
"rewards/rejected": -0.37791121006011963,
"step": 170
},
{
"epoch": 0.3740814963259853,
"grad_norm": 6.674170792303524,
"learning_rate": 3.939042664214184e-07,
"logits/chosen": -0.062463849782943726,
"logits/rejected": -0.05575127527117729,
"logps/chosen": -0.30678054690361023,
"logps/rejected": -0.3601401150226593,
"loss": 1.2515,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.30678054690361023,
"rewards/margins": 0.053359538316726685,
"rewards/rejected": -0.3601401150226593,
"step": 175
},
{
"epoch": 0.3847695390781563,
"grad_norm": 8.128495075307303,
"learning_rate": 3.8615975875375676e-07,
"logits/chosen": -0.07516147941350937,
"logits/rejected": -0.06471911817789078,
"logps/chosen": -0.31638792157173157,
"logps/rejected": -0.35477930307388306,
"loss": 1.2607,
"rewards/accuracies": 0.4937500059604645,
"rewards/chosen": -0.31638792157173157,
"rewards/margins": 0.038391388952732086,
"rewards/rejected": -0.35477930307388306,
"step": 180
},
{
"epoch": 0.3954575818303273,
"grad_norm": 5.421271435770341,
"learning_rate": 3.7822481935147655e-07,
"logits/chosen": -0.02470918372273445,
"logits/rejected": -0.048990484327077866,
"logps/chosen": -0.3120611310005188,
"logps/rejected": -0.35026127099990845,
"loss": 1.2637,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.3120611310005188,
"rewards/margins": 0.03820016235113144,
"rewards/rejected": -0.35026127099990845,
"step": 185
},
{
"epoch": 0.4061456245824983,
"grad_norm": 7.381230268395833,
"learning_rate": 3.7011054594483443e-07,
"logits/chosen": -0.05781809240579605,
"logits/rejected": -0.03457602858543396,
"logps/chosen": -0.30188003182411194,
"logps/rejected": -0.36022719740867615,
"loss": 1.2467,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.30188003182411194,
"rewards/margins": 0.058347173035144806,
"rewards/rejected": -0.36022719740867615,
"step": 190
},
{
"epoch": 0.4168336673346693,
"grad_norm": 8.206125691472135,
"learning_rate": 3.618282870789081e-07,
"logits/chosen": -0.03373146429657936,
"logits/rejected": -0.008745524100959301,
"logps/chosen": -0.33559438586235046,
"logps/rejected": -0.41109856963157654,
"loss": 1.2425,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.33559438586235046,
"rewards/margins": 0.07550417631864548,
"rewards/rejected": -0.41109856963157654,
"step": 195
},
{
"epoch": 0.42752171008684037,
"grad_norm": 5.62603836496467,
"learning_rate": 3.5338962624163016e-07,
"logits/chosen": -0.010943605564534664,
"logits/rejected": 0.021173829212784767,
"logps/chosen": -0.3039094805717468,
"logps/rejected": -0.3738254904747009,
"loss": 1.2433,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.3039094805717468,
"rewards/margins": 0.0699160099029541,
"rewards/rejected": -0.3738254904747009,
"step": 200
},
{
"epoch": 0.43820975283901137,
"grad_norm": 8.62105429503922,
"learning_rate": 3.448063656632321e-07,
"logits/chosen": -0.05850949138402939,
"logits/rejected": -0.09792699664831161,
"logps/chosen": -0.33592477440834045,
"logps/rejected": -0.39009103178977966,
"loss": 1.2592,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.33592477440834045,
"rewards/margins": 0.05416623875498772,
"rewards/rejected": -0.39009103178977966,
"step": 205
},
{
"epoch": 0.44889779559118237,
"grad_norm": 6.45467133802993,
"learning_rate": 3.360905098097587e-07,
"logits/chosen": -0.06773098558187485,
"logits/rejected": -0.04933014139533043,
"logps/chosen": -0.3153294026851654,
"logps/rejected": -0.37946873903274536,
"loss": 1.2479,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.3153294026851654,
"rewards/margins": 0.06413935124874115,
"rewards/rejected": -0.37946873903274536,
"step": 210
},
{
"epoch": 0.45958583834335337,
"grad_norm": 6.5204274941386755,
"learning_rate": 3.272542485937368e-07,
"logits/chosen": 4.871189594268799e-05,
"logits/rejected": -0.04641323536634445,
"logps/chosen": -0.2955402433872223,
"logps/rejected": -0.3811369240283966,
"loss": 1.2408,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.2955402433872223,
"rewards/margins": 0.08559669554233551,
"rewards/rejected": -0.3811369240283966,
"step": 215
},
{
"epoch": 0.47027388109552437,
"grad_norm": 7.245080500807784,
"learning_rate": 3.1830994032548e-07,
"logits/chosen": -0.08027426153421402,
"logits/rejected": -0.0962793231010437,
"logps/chosen": -0.30348068475723267,
"logps/rejected": -0.4292809069156647,
"loss": 1.2388,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.30348068475723267,
"rewards/margins": 0.1258002072572708,
"rewards/rejected": -0.4292809069156647,
"step": 220
},
{
"epoch": 0.48096192384769537,
"grad_norm": 8.334489989695008,
"learning_rate": 3.0927009442887437e-07,
"logits/chosen": 0.0028877719305455685,
"logits/rejected": -0.0022175461053848267,
"logps/chosen": -0.3067638874053955,
"logps/rejected": -0.38486868143081665,
"loss": 1.2293,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.3067638874053955,
"rewards/margins": 0.07810477167367935,
"rewards/rejected": -0.38486868143081665,
"step": 225
},
{
"epoch": 0.4916499665998664,
"grad_norm": 6.371194399433381,
"learning_rate": 3.001473539458182e-07,
"logits/chosen": -0.005320471711456776,
"logits/rejected": -0.05842217057943344,
"logps/chosen": -0.3131864666938782,
"logps/rejected": -0.3757328391075134,
"loss": 1.2429,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.3131864666938782,
"rewards/margins": 0.06254634261131287,
"rewards/rejected": -0.3757328391075134,
"step": 230
},
{
"epoch": 0.5023380093520374,
"grad_norm": 8.450158822463967,
"learning_rate": 2.909544778537844e-07,
"logits/chosen": -0.03879100829362869,
"logits/rejected": 0.01334688626229763,
"logps/chosen": -0.33406445384025574,
"logps/rejected": -0.4425807595252991,
"loss": 1.2434,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.33406445384025574,
"rewards/margins": 0.10851632058620453,
"rewards/rejected": -0.4425807595252991,
"step": 235
},
{
"epoch": 0.5130260521042084,
"grad_norm": 6.072988469243308,
"learning_rate": 2.817043232212371e-07,
"logits/chosen": -0.06821739673614502,
"logits/rejected": -0.04843694344162941,
"logps/chosen": -0.30397796630859375,
"logps/rejected": -0.4096347391605377,
"loss": 1.2477,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.30397796630859375,
"rewards/margins": 0.10565674304962158,
"rewards/rejected": -0.4096347391605377,
"step": 240
},
{
"epoch": 0.5237140948563794,
"grad_norm": 5.092574721635333,
"learning_rate": 2.7240982722585837e-07,
"logits/chosen": -0.1030288115143776,
"logits/rejected": -0.09166723489761353,
"logps/chosen": -0.3165212571620941,
"logps/rejected": -0.3529934883117676,
"loss": 1.2312,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.3165212571620941,
"rewards/margins": 0.03647221624851227,
"rewards/rejected": -0.3529934883117676,
"step": 245
},
{
"epoch": 0.5344021376085505,
"grad_norm": 8.48565741064265,
"learning_rate": 2.63083989060736e-07,
"logits/chosen": -0.03879087418317795,
"logits/rejected": -0.043008435517549515,
"logps/chosen": -0.33534130454063416,
"logps/rejected": -0.44996342062950134,
"loss": 1.2295,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": -0.33534130454063416,
"rewards/margins": 0.1146220937371254,
"rewards/rejected": -0.44996342062950134,
"step": 250
},
{
"epoch": 0.5450901803607214,
"grad_norm": 9.606596247391161,
"learning_rate": 2.537398517538159e-07,
"logits/chosen": -0.04518505185842514,
"logits/rejected": -0.03536449745297432,
"logps/chosen": -0.36244145035743713,
"logps/rejected": -0.432295560836792,
"loss": 1.2516,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.36244145035743713,
"rewards/margins": 0.06985411792993546,
"rewards/rejected": -0.432295560836792,
"step": 255
},
{
"epoch": 0.5557782231128925,
"grad_norm": 10.8283575979542,
"learning_rate": 2.4439048392604877e-07,
"logits/chosen": -0.10192953050136566,
"logits/rejected": -0.09843325614929199,
"logps/chosen": -0.32212352752685547,
"logps/rejected": -0.40814656019210815,
"loss": 1.2399,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.32212352752685547,
"rewards/margins": 0.08602304756641388,
"rewards/rejected": -0.40814656019210815,
"step": 260
},
{
"epoch": 0.5664662658650634,
"grad_norm": 16.490660464227766,
"learning_rate": 2.3504896151374144e-07,
"logits/chosen": -0.07496851682662964,
"logits/rejected": -0.07402872294187546,
"logps/chosen": -0.3090594410896301,
"logps/rejected": -0.3913823962211609,
"loss": 1.2455,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.3090594410896301,
"rewards/margins": 0.08232297748327255,
"rewards/rejected": -0.3913823962211609,
"step": 265
},
{
"epoch": 0.5771543086172345,
"grad_norm": 6.869735430783954,
"learning_rate": 2.2572834948067795e-07,
"logits/chosen": -0.04213152453303337,
"logits/rejected": -0.03718320652842522,
"logps/chosen": -0.3201075792312622,
"logps/rejected": -0.48250871896743774,
"loss": 1.2286,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.3201075792312622,
"rewards/margins": 0.16240113973617554,
"rewards/rejected": -0.48250871896743774,
"step": 270
},
{
"epoch": 0.5878423513694054,
"grad_norm": 6.123315952988323,
"learning_rate": 2.164416835455862e-07,
"logits/chosen": -0.028377573937177658,
"logits/rejected": -0.050650130957365036,
"logps/chosen": -0.318294495344162,
"logps/rejected": -0.35608866810798645,
"loss": 1.2416,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.318294495344162,
"rewards/margins": 0.037794191390275955,
"rewards/rejected": -0.35608866810798645,
"step": 275
},
{
"epoch": 0.5985303941215765,
"grad_norm": 6.592928863606633,
"learning_rate": 2.072019519505062e-07,
"logits/chosen": -0.0866287425160408,
"logits/rejected": -0.06696648895740509,
"logps/chosen": -0.31993359327316284,
"logps/rejected": -0.3936074376106262,
"loss": 1.2363,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.31993359327316284,
"rewards/margins": 0.07367385178804398,
"rewards/rejected": -0.3936074376106262,
"step": 280
},
{
"epoch": 0.6092184368737475,
"grad_norm": 9.598921674964332,
"learning_rate": 1.980220772955602e-07,
"logits/chosen": -0.08209933340549469,
"logits/rejected": -0.09453925490379333,
"logps/chosen": -0.3198639750480652,
"logps/rejected": -0.3884400725364685,
"loss": 1.2506,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.3198639750480652,
"rewards/margins": 0.06857611238956451,
"rewards/rejected": -0.3884400725364685,
"step": 285
},
{
"epoch": 0.6199064796259185,
"grad_norm": 10.59331357545079,
"learning_rate": 1.8891489846552644e-07,
"logits/chosen": 0.006256720516830683,
"logits/rejected": -0.015608558431267738,
"logps/chosen": -0.31744810938835144,
"logps/rejected": -0.4133140444755554,
"loss": 1.2544,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.31744810938835144,
"rewards/margins": 0.09586595743894577,
"rewards/rejected": -0.4133140444755554,
"step": 290
},
{
"epoch": 0.6305945223780896,
"grad_norm": 5.292198128419299,
"learning_rate": 1.7989315267349933e-07,
"logits/chosen": -0.023702971637248993,
"logits/rejected": 0.03860974311828613,
"logps/chosen": -0.2653047740459442,
"logps/rejected": -0.3553427457809448,
"loss": 1.2377,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.2653047740459442,
"rewards/margins": 0.0900380089879036,
"rewards/rejected": -0.3553427457809448,
"step": 295
},
{
"epoch": 0.6412825651302605,
"grad_norm": 12.156996229810368,
"learning_rate": 1.7096945764674398e-07,
"logits/chosen": -0.06409218162298203,
"logits/rejected": -0.008139371871948242,
"logps/chosen": -0.3294127881526947,
"logps/rejected": -0.4992917478084564,
"loss": 1.2352,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.3294127881526947,
"rewards/margins": 0.1698789894580841,
"rewards/rejected": -0.4992917478084564,
"step": 300
},
{
"epoch": 0.6519706078824316,
"grad_norm": 7.623030069962012,
"learning_rate": 1.621562939796643e-07,
"logits/chosen": -0.0799182876944542,
"logits/rejected": -0.0722065195441246,
"logps/chosen": -0.3097606599330902,
"logps/rejected": -0.3971176743507385,
"loss": 1.2431,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.3097606599330902,
"rewards/margins": 0.08735699951648712,
"rewards/rejected": -0.3971176743507385,
"step": 305
},
{
"epoch": 0.6626586506346025,
"grad_norm": 5.936446487789816,
"learning_rate": 1.5346598767856345e-07,
"logits/chosen": -0.14804649353027344,
"logits/rejected": -0.1550634354352951,
"logps/chosen": -0.3042837977409363,
"logps/rejected": -0.3682674467563629,
"loss": 1.2481,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.3042837977409363,
"rewards/margins": 0.06398364901542664,
"rewards/rejected": -0.3682674467563629,
"step": 310
},
{
"epoch": 0.6733466933867736,
"grad_norm": 10.616574151746162,
"learning_rate": 1.4491069292260866e-07,
"logits/chosen": -0.08723266422748566,
"logits/rejected": -0.07947458326816559,
"logps/chosen": -0.35345691442489624,
"logps/rejected": -0.41427382826805115,
"loss": 1.2571,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.35345691442489624,
"rewards/margins": 0.06081696227192879,
"rewards/rejected": -0.41427382826805115,
"step": 315
},
{
"epoch": 0.6840347361389446,
"grad_norm": 8.526821542838308,
"learning_rate": 1.365023750651133e-07,
"logits/chosen": -0.07338618487119675,
"logits/rejected": -0.0861336812376976,
"logps/chosen": -0.3063948154449463,
"logps/rejected": -0.3334597945213318,
"loss": 1.261,
"rewards/accuracies": 0.518750011920929,
"rewards/chosen": -0.3063948154449463,
"rewards/margins": 0.027065003290772438,
"rewards/rejected": -0.3334597945213318,
"step": 320
},
{
"epoch": 0.6947227788911156,
"grad_norm": 9.444848804844204,
"learning_rate": 1.2825279389890818e-07,
"logits/chosen": -0.10052672773599625,
"logits/rejected": -0.07519718259572983,
"logps/chosen": -0.31606850028038025,
"logps/rejected": -0.45330485701560974,
"loss": 1.2442,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.31606850028038025,
"rewards/margins": 0.1372363269329071,
"rewards/rejected": -0.45330485701560974,
"step": 325
},
{
"epoch": 0.7054108216432866,
"grad_norm": 5.315121636991915,
"learning_rate": 1.201734872092077e-07,
"logits/chosen": -0.143130362033844,
"logits/rejected": -0.1398681104183197,
"logps/chosen": -0.30278146266937256,
"logps/rejected": -0.3507356643676758,
"loss": 1.2357,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.30278146266937256,
"rewards/margins": 0.047954194247722626,
"rewards/rejected": -0.3507356643676758,
"step": 330
},
{
"epoch": 0.7160988643954576,
"grad_norm": 6.664051479903874,
"learning_rate": 1.1227575463697439e-07,
"logits/chosen": -0.06139817088842392,
"logits/rejected": -0.07857407629489899,
"logps/chosen": -0.31398746371269226,
"logps/rejected": -0.4074636995792389,
"loss": 1.2491,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.31398746371269226,
"rewards/margins": 0.09347621351480484,
"rewards/rejected": -0.4074636995792389,
"step": 335
},
{
"epoch": 0.7267869071476286,
"grad_norm": 6.466017993804309,
"learning_rate": 1.0457064187534861e-07,
"logits/chosen": -0.13862331211566925,
"logits/rejected": -0.12122078239917755,
"logps/chosen": -0.3089817464351654,
"logps/rejected": -0.4134499430656433,
"loss": 1.2489,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.3089817464351654,
"rewards/margins": 0.1044682040810585,
"rewards/rejected": -0.4134499430656433,
"step": 340
},
{
"epoch": 0.7374749498997996,
"grad_norm": 7.883921432886144,
"learning_rate": 9.706892522124838e-08,
"logits/chosen": -0.1288374960422516,
"logits/rejected": -0.13176000118255615,
"logps/chosen": -0.2964598536491394,
"logps/rejected": -0.4006675183773041,
"loss": 1.2362,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.2964598536491394,
"rewards/margins": 0.10420763492584229,
"rewards/rejected": -0.4006675183773041,
"step": 345
},
{
"epoch": 0.7481629926519706,
"grad_norm": 8.13826941223703,
"learning_rate": 8.978109650374396e-08,
"logits/chosen": -0.050217024981975555,
"logits/rejected": -0.06327749788761139,
"logps/chosen": -0.32183343172073364,
"logps/rejected": -0.4269295334815979,
"loss": 1.2292,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.32183343172073364,
"rewards/margins": 0.10509608685970306,
"rewards/rejected": -0.4269295334815979,
"step": 350
},
{
"epoch": 0.7588510354041417,
"grad_norm": 5.382196991571276,
"learning_rate": 8.271734841028552e-08,
"logits/chosen": -0.11176713556051254,
"logits/rejected": -0.10792241245508194,
"logps/chosen": -0.3432878851890564,
"logps/rejected": -0.4941299557685852,
"loss": 1.2361,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.3432878851890564,
"rewards/margins": 0.15084199607372284,
"rewards/rejected": -0.4941299557685852,
"step": 355
},
{
"epoch": 0.7695390781563126,
"grad_norm": 19.58123389841092,
"learning_rate": 7.588756023130833e-08,
"logits/chosen": -0.06944067776203156,
"logits/rejected": -0.07753388583660126,
"logps/chosen": -0.34678810834884644,
"logps/rejected": -0.3968583941459656,
"loss": 1.2575,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": -0.34678810834884644,
"rewards/margins": 0.05007026717066765,
"rewards/rejected": -0.3968583941459656,
"step": 360
},
{
"epoch": 0.7802271209084837,
"grad_norm": 9.51840066626221,
"learning_rate": 6.930128404315214e-08,
"logits/chosen": -0.12009277194738388,
"logits/rejected": -0.12382064759731293,
"logps/chosen": -0.3429993987083435,
"logps/rejected": -0.4403919279575348,
"loss": 1.2425,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.3429993987083435,
"rewards/margins": 0.0973924994468689,
"rewards/rejected": -0.4403919279575348,
"step": 365
},
{
"epoch": 0.7909151636606546,
"grad_norm": 8.472250659850234,
"learning_rate": 6.296773134861824e-08,
"logits/chosen": -0.06690394878387451,
"logits/rejected": -0.050654083490371704,
"logps/chosen": -0.33747923374176025,
"logps/rejected": -0.39239710569381714,
"loss": 1.2363,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.33747923374176025,
"rewards/margins": 0.05491787940263748,
"rewards/rejected": -0.39239710569381714,
"step": 370
},
{
"epoch": 0.8016032064128257,
"grad_norm": 8.100035385536074,
"learning_rate": 5.6895760193850145e-08,
"logits/chosen": -0.10875822603702545,
"logits/rejected": -0.09210009127855301,
"logps/chosen": -0.32545924186706543,
"logps/rejected": -0.3734380006790161,
"loss": 1.2345,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.32545924186706543,
"rewards/margins": 0.04797876626253128,
"rewards/rejected": -0.3734380006790161,
"step": 375
},
{
"epoch": 0.8122912491649966,
"grad_norm": 7.4124570883149135,
"learning_rate": 5.109386277955477e-08,
"logits/chosen": -0.09277497231960297,
"logits/rejected": -0.09483953565359116,
"logps/chosen": -0.31688663363456726,
"logps/rejected": -0.4329262375831604,
"loss": 1.2342,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.31688663363456726,
"rewards/margins": 0.11603957414627075,
"rewards/rejected": -0.4329262375831604,
"step": 380
},
{
"epoch": 0.8229792919171677,
"grad_norm": 9.699621065575215,
"learning_rate": 4.557015358389216e-08,
"logits/chosen": -0.0970136821269989,
"logits/rejected": -0.07916976511478424,
"logps/chosen": -0.33148500323295593,
"logps/rejected": -0.3951139450073242,
"loss": 1.2473,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.33148500323295593,
"rewards/margins": 0.0636288970708847,
"rewards/rejected": -0.3951139450073242,
"step": 385
},
{
"epoch": 0.8336673346693386,
"grad_norm": 6.48805841042973,
"learning_rate": 4.0332358013644015e-08,
"logits/chosen": -0.10146977752447128,
"logits/rejected": -0.07709420472383499,
"logps/chosen": -0.31857064366340637,
"logps/rejected": -0.3821341395378113,
"loss": 1.2466,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.31857064366340637,
"rewards/margins": 0.06356349587440491,
"rewards/rejected": -0.3821341395378113,
"step": 390
},
{
"epoch": 0.8443553774215097,
"grad_norm": 7.587194466968942,
"learning_rate": 3.538780159953347e-08,
"logits/chosen": -0.13655772805213928,
"logits/rejected": -0.12785808742046356,
"logps/chosen": -0.3200170397758484,
"logps/rejected": -0.35742777585983276,
"loss": 1.2443,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.3200170397758484,
"rewards/margins": 0.03741071745753288,
"rewards/rejected": -0.35742777585983276,
"step": 395
},
{
"epoch": 0.8550434201736807,
"grad_norm": 6.278551070521003,
"learning_rate": 3.074339975080836e-08,
"logits/chosen": -0.15586154162883759,
"logits/rejected": -0.1442781686782837,
"logps/chosen": -0.3342689573764801,
"logps/rejected": -0.3648763597011566,
"loss": 1.2474,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.3342689573764801,
"rewards/margins": 0.03060741163790226,
"rewards/rejected": -0.3648763597011566,
"step": 400
},
{
"epoch": 0.8550434201736807,
"eval_logits/chosen": 0.05778730660676956,
"eval_logits/rejected": 0.048719972372055054,
"eval_logps/chosen": -0.32926085591316223,
"eval_logps/rejected": -0.3879503607749939,
"eval_loss": 1.244562029838562,
"eval_rewards/accuracies": 0.5813007950782776,
"eval_rewards/chosen": -0.32926085591316223,
"eval_rewards/margins": 0.05868948996067047,
"eval_rewards/rejected": -0.3879503607749939,
"eval_runtime": 422.4499,
"eval_samples_per_second": 4.642,
"eval_steps_per_second": 0.291,
"step": 400
},
{
"epoch": 0.8657314629258517,
"grad_norm": 8.840635048074553,
"learning_rate": 2.6405648083415833e-08,
"logits/chosen": -0.07431348413228989,
"logits/rejected": -0.12591402232646942,
"logps/chosen": -0.33717232942581177,
"logps/rejected": -0.36238372325897217,
"loss": 1.247,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.33717232942581177,
"rewards/margins": 0.025211378931999207,
"rewards/rejected": -0.36238372325897217,
"step": 405
},
{
"epoch": 0.8764195056780227,
"grad_norm": 8.461073615162963,
"learning_rate": 2.2380613335296033e-08,
"logits/chosen": -0.065019391477108,
"logits/rejected": -0.09228098392486572,
"logps/chosen": -0.3428507447242737,
"logps/rejected": -0.3787713050842285,
"loss": 1.2627,
"rewards/accuracies": 0.518750011920929,
"rewards/chosen": -0.3428507447242737,
"rewards/margins": 0.03592059016227722,
"rewards/rejected": -0.3787713050842285,
"step": 410
},
{
"epoch": 0.8871075484301937,
"grad_norm": 9.178492500949973,
"learning_rate": 1.8673924881500823e-08,
"logits/chosen": -0.08692122995853424,
"logits/rejected": -0.06387011706829071,
"logps/chosen": -0.27576693892478943,
"logps/rejected": -0.377163827419281,
"loss": 1.2333,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.27576693892478943,
"rewards/margins": 0.10139693319797516,
"rewards/rejected": -0.377163827419281,
"step": 415
},
{
"epoch": 0.8977955911823647,
"grad_norm": 5.901298908278412,
"learning_rate": 1.5290766861003475e-08,
"logits/chosen": -0.042976394295692444,
"logits/rejected": -0.024144111201167107,
"logps/chosen": -0.3224348723888397,
"logps/rejected": -0.45535239577293396,
"loss": 1.243,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.3224348723888397,
"rewards/margins": 0.13291755318641663,
"rewards/rejected": -0.45535239577293396,
"step": 420
},
{
"epoch": 0.9084836339345357,
"grad_norm": 5.019524890791478,
"learning_rate": 1.2235870926211616e-08,
"logits/chosen": 0.01839056983590126,
"logits/rejected": -0.026639455929398537,
"logps/chosen": -0.30540984869003296,
"logps/rejected": -0.41048941016197205,
"loss": 1.2396,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.30540984869003296,
"rewards/margins": 0.10507960617542267,
"rewards/rejected": -0.41048941016197205,
"step": 425
},
{
"epoch": 0.9191716766867067,
"grad_norm": 5.654732068658596,
"learning_rate": 9.513509625323518e-09,
"logits/chosen": -0.0712607279419899,
"logits/rejected": -0.05221161991357803,
"logps/chosen": -0.28964343667030334,
"logps/rejected": -0.34897297620773315,
"loss": 1.2369,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.28964343667030334,
"rewards/margins": 0.059329528361558914,
"rewards/rejected": -0.34897297620773315,
"step": 430
},
{
"epoch": 0.9298597194388778,
"grad_norm": 7.192592801186788,
"learning_rate": 7.127490426783123e-09,
"logits/chosen": -0.05717062950134277,
"logits/rejected": -0.07238337397575378,
"logps/chosen": -0.3144196569919586,
"logps/rejected": -0.4318385720252991,
"loss": 1.2379,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.3144196569919586,
"rewards/margins": 0.11741892248392105,
"rewards/rejected": -0.4318385720252991,
"step": 435
},
{
"epoch": 0.9405477621910487,
"grad_norm": 25.472141832477323,
"learning_rate": 5.08115039419113e-09,
"logits/chosen": -0.09659489244222641,
"logits/rejected": -0.10936260223388672,
"logps/chosen": -0.32775628566741943,
"logps/rejected": -0.37708064913749695,
"loss": 1.2516,
"rewards/accuracies": 0.53125,
"rewards/chosen": -0.32775628566741943,
"rewards/margins": 0.049324359744787216,
"rewards/rejected": -0.37708064913749695,
"step": 440
},
{
"epoch": 0.9512358049432198,
"grad_norm": 7.24892892824306,
"learning_rate": 3.3773515191196646e-09,
"logits/chosen": -0.008545724675059319,
"logits/rejected": -0.007621192838996649,
"logps/chosen": -0.3069409132003784,
"logps/rejected": -0.37095940113067627,
"loss": 1.2593,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.3069409132003784,
"rewards/margins": 0.06401850283145905,
"rewards/rejected": -0.37095940113067627,
"step": 445
},
{
"epoch": 0.9619238476953907,
"grad_norm": 7.087915891602145,
"learning_rate": 2.0184767183584474e-09,
"logits/chosen": -0.06611478328704834,
"logits/rejected": -0.0581117644906044,
"logps/chosen": -0.31249338388442993,
"logps/rejected": -0.39341428875923157,
"loss": 1.2302,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.31249338388442993,
"rewards/margins": 0.08092090487480164,
"rewards/rejected": -0.39341428875923157,
"step": 450
},
{
"epoch": 0.9726118904475618,
"grad_norm": 6.957525491196618,
"learning_rate": 1.0064265011902328e-09,
"logits/chosen": -0.02204318158328533,
"logits/rejected": 0.01280857902020216,
"logps/chosen": -0.3316726088523865,
"logps/rejected": -0.3979867994785309,
"loss": 1.252,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.3316726088523865,
"rewards/margins": 0.0663142055273056,
"rewards/rejected": -0.3979867994785309,
"step": 455
},
{
"epoch": 0.9832999331997327,
"grad_norm": 6.51993431880911,
"learning_rate": 3.4261631135654167e-10,
"logits/chosen": -0.019794732332229614,
"logits/rejected": -0.043111272156238556,
"logps/chosen": -0.321623295545578,
"logps/rejected": -0.362581729888916,
"loss": 1.2517,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": -0.321623295545578,
"rewards/margins": 0.04095841199159622,
"rewards/rejected": -0.362581729888916,
"step": 460
},
{
"epoch": 0.9939879759519038,
"grad_norm": 7.2911344813134455,
"learning_rate": 2.797454743164174e-11,
"logits/chosen": 0.01124805398285389,
"logits/rejected": -0.010005452670156956,
"logps/chosen": -0.3086443543434143,
"logps/rejected": -0.398124635219574,
"loss": 1.2441,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.3086443543434143,
"rewards/margins": 0.08948027342557907,
"rewards/rejected": -0.398124635219574,
"step": 465
},
{
"epoch": 0.9982631930527722,
"step": 467,
"total_flos": 0.0,
"train_loss": 1.2499028640845093,
"train_runtime": 20893.3787,
"train_samples_per_second": 2.866,
"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
}