qwen2_chat_adamw_iter5 / trainer_state.json
yiran-wang3's picture
End of training
b91f22a verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 100,
"global_step": 42,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"debug/policy_chosen_logits": -1.7926865816116333,
"debug/policy_chosen_logps": -217.38279724121094,
"debug/policy_rejected_logits": -1.8328309059143066,
"debug/policy_rejected_logps": -225.83067321777344,
"debug/reference_chosen_logps": -217.38279724121094,
"debug/reference_rejected_logps": -225.83067321777344,
"epoch": 0.023809523809523808,
"grad_norm": 4.886963528314891,
"learning_rate": 1e-06,
"logits/chosen": -1.7926865816116333,
"logits/rejected": -1.8328309059143066,
"logps/chosen": -217.38279724121094,
"logps/rejected": -225.83067321777344,
"loss": 0.5,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"debug/policy_chosen_logits": -1.8736802339553833,
"debug/policy_chosen_logps": -237.7906036376953,
"debug/policy_rejected_logits": -1.6878823041915894,
"debug/policy_rejected_logps": -252.66542053222656,
"debug/reference_chosen_logps": -237.90771484375,
"debug/reference_rejected_logps": -252.53955078125,
"epoch": 0.047619047619047616,
"grad_norm": 4.425053566972872,
"learning_rate": 1e-06,
"logits/chosen": -1.8736802339553833,
"logits/rejected": -1.6878823041915894,
"logps/chosen": -237.7906036376953,
"logps/rejected": -252.66542053222656,
"loss": 0.4983,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.001171283656731248,
"rewards/margins": 0.0024298285134136677,
"rewards/rejected": -0.001258544740267098,
"step": 2
},
{
"debug/policy_chosen_logits": -1.6082555055618286,
"debug/policy_chosen_logps": -239.86978149414062,
"debug/policy_rejected_logits": -1.6321099996566772,
"debug/policy_rejected_logps": -217.70059204101562,
"debug/reference_chosen_logps": -240.5919189453125,
"debug/reference_rejected_logps": -218.04476928710938,
"epoch": 0.07142857142857142,
"grad_norm": 4.882938644343874,
"learning_rate": 1e-06,
"logits/chosen": -1.6082555055618286,
"logits/rejected": -1.6321099996566772,
"logps/chosen": -239.86978149414062,
"logps/rejected": -217.70059204101562,
"loss": 0.4997,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.007221374660730362,
"rewards/margins": 0.0037795070093125105,
"rewards/rejected": 0.0034418676514178514,
"step": 3
},
{
"debug/policy_chosen_logits": -1.8915444612503052,
"debug/policy_chosen_logps": -215.7807159423828,
"debug/policy_rejected_logits": -1.7285552024841309,
"debug/policy_rejected_logps": -219.4071044921875,
"debug/reference_chosen_logps": -215.05958557128906,
"debug/reference_rejected_logps": -219.14736938476562,
"epoch": 0.09523809523809523,
"grad_norm": 4.417539068018447,
"learning_rate": 1e-06,
"logits/chosen": -1.8915444612503052,
"logits/rejected": -1.7285552024841309,
"logps/chosen": -215.7807159423828,
"logps/rejected": -219.4071044921875,
"loss": 0.5018,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.007211265154182911,
"rewards/margins": -0.004613952711224556,
"rewards/rejected": -0.0025973126757889986,
"step": 4
},
{
"debug/policy_chosen_logits": -1.8312164545059204,
"debug/policy_chosen_logps": -228.4812469482422,
"debug/policy_rejected_logits": -1.8293319940567017,
"debug/policy_rejected_logps": -219.43865966796875,
"debug/reference_chosen_logps": -228.2222137451172,
"debug/reference_rejected_logps": -218.95147705078125,
"epoch": 0.11904761904761904,
"grad_norm": 4.499121781225099,
"learning_rate": 1e-06,
"logits/chosen": -1.8312164545059204,
"logits/rejected": -1.8293319940567017,
"logps/chosen": -228.4812469482422,
"logps/rejected": -219.43865966796875,
"loss": 0.4982,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.0025903128553181887,
"rewards/margins": 0.002281456021592021,
"rewards/rejected": -0.00487176887691021,
"step": 5
},
{
"debug/policy_chosen_logits": -1.7772536277770996,
"debug/policy_chosen_logps": -251.26417541503906,
"debug/policy_rejected_logits": -1.7712210416793823,
"debug/policy_rejected_logps": -224.58258056640625,
"debug/reference_chosen_logps": -251.5955047607422,
"debug/reference_rejected_logps": -224.013427734375,
"epoch": 0.14285714285714285,
"grad_norm": 4.746380335819719,
"learning_rate": 1e-06,
"logits/chosen": -1.7772536277770996,
"logits/rejected": -1.7712210416793823,
"logps/chosen": -251.26417541503906,
"logps/rejected": -224.58258056640625,
"loss": 0.5024,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.0033132743556052446,
"rewards/margins": 0.009004650637507439,
"rewards/rejected": -0.005691375583410263,
"step": 6
},
{
"debug/policy_chosen_logits": -1.8587639331817627,
"debug/policy_chosen_logps": -215.83526611328125,
"debug/policy_rejected_logits": -1.830519437789917,
"debug/policy_rejected_logps": -228.4679412841797,
"debug/reference_chosen_logps": -215.88973999023438,
"debug/reference_rejected_logps": -228.85028076171875,
"epoch": 0.16666666666666666,
"grad_norm": 4.703836780587687,
"learning_rate": 1e-06,
"logits/chosen": -1.8587639331817627,
"logits/rejected": -1.830519437789917,
"logps/chosen": -215.83526611328125,
"logps/rejected": -228.4679412841797,
"loss": 0.4979,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.0005447575822472572,
"rewards/margins": -0.0032785222865641117,
"rewards/rejected": 0.0038232803344726562,
"step": 7
},
{
"debug/policy_chosen_logits": -1.713200569152832,
"debug/policy_chosen_logps": -214.9654083251953,
"debug/policy_rejected_logits": -1.6877814531326294,
"debug/policy_rejected_logps": -237.10147094726562,
"debug/reference_chosen_logps": -215.1925811767578,
"debug/reference_rejected_logps": -237.26712036132812,
"epoch": 0.19047619047619047,
"grad_norm": 4.614150814199241,
"learning_rate": 1e-06,
"logits/chosen": -1.713200569152832,
"logits/rejected": -1.6877814531326294,
"logps/chosen": -214.9654083251953,
"logps/rejected": -237.10147094726562,
"loss": 0.4993,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.0022717281244695187,
"rewards/margins": 0.0006152723217383027,
"rewards/rejected": 0.0016564556863158941,
"step": 8
},
{
"debug/policy_chosen_logits": -1.830588459968567,
"debug/policy_chosen_logps": -264.9453430175781,
"debug/policy_rejected_logits": -1.7403862476348877,
"debug/policy_rejected_logps": -236.23504638671875,
"debug/reference_chosen_logps": -265.33489990234375,
"debug/reference_rejected_logps": -236.4605712890625,
"epoch": 0.21428571428571427,
"grad_norm": 4.507439835242966,
"learning_rate": 1e-06,
"logits/chosen": -1.830588459968567,
"logits/rejected": -1.7403862476348877,
"logps/chosen": -264.9453430175781,
"logps/rejected": -236.23504638671875,
"loss": 0.498,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.003895435482263565,
"rewards/margins": 0.0016400336753576994,
"rewards/rejected": 0.002255401574075222,
"step": 9
},
{
"debug/policy_chosen_logits": -1.7635490894317627,
"debug/policy_chosen_logps": -212.54888916015625,
"debug/policy_rejected_logits": -1.7577309608459473,
"debug/policy_rejected_logps": -231.78260803222656,
"debug/reference_chosen_logps": -212.55520629882812,
"debug/reference_rejected_logps": -231.5526885986328,
"epoch": 0.23809523809523808,
"grad_norm": 4.646994116106454,
"learning_rate": 1e-06,
"logits/chosen": -1.7635490894317627,
"logits/rejected": -1.7577309608459473,
"logps/chosen": -212.54888916015625,
"logps/rejected": -231.78260803222656,
"loss": 0.4997,
"rewards/accuracies": 0.625,
"rewards/chosen": 6.330502219498158e-05,
"rewards/margins": 0.0023624992463737726,
"rewards/rejected": -0.002299194224178791,
"step": 10
},
{
"debug/policy_chosen_logits": -1.8275775909423828,
"debug/policy_chosen_logps": -223.30955505371094,
"debug/policy_rejected_logits": -1.7878596782684326,
"debug/policy_rejected_logps": -228.84654235839844,
"debug/reference_chosen_logps": -223.7688751220703,
"debug/reference_rejected_logps": -228.8212890625,
"epoch": 0.2619047619047619,
"grad_norm": 4.501453979601088,
"learning_rate": 1e-06,
"logits/chosen": -1.8275775909423828,
"logits/rejected": -1.7878596782684326,
"logps/chosen": -223.30955505371094,
"logps/rejected": -228.84654235839844,
"loss": 0.4975,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.004593219608068466,
"rewards/margins": 0.004845847375690937,
"rewards/rejected": -0.0002526284661144018,
"step": 11
},
{
"debug/policy_chosen_logits": -1.8328152894973755,
"debug/policy_chosen_logps": -229.270751953125,
"debug/policy_rejected_logits": -1.7323297262191772,
"debug/policy_rejected_logps": -245.7848358154297,
"debug/reference_chosen_logps": -231.10006713867188,
"debug/reference_rejected_logps": -244.46746826171875,
"epoch": 0.2857142857142857,
"grad_norm": 4.605283271613626,
"learning_rate": 1e-06,
"logits/chosen": -1.8328152894973755,
"logits/rejected": -1.7323297262191772,
"logps/chosen": -229.270751953125,
"logps/rejected": -245.7848358154297,
"loss": 0.4964,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.018293190747499466,
"rewards/margins": 0.03146692365407944,
"rewards/rejected": -0.013173731975257397,
"step": 12
},
{
"debug/policy_chosen_logits": -1.7267065048217773,
"debug/policy_chosen_logps": -218.44247436523438,
"debug/policy_rejected_logits": -1.6951403617858887,
"debug/policy_rejected_logps": -240.42892456054688,
"debug/reference_chosen_logps": -217.90109252929688,
"debug/reference_rejected_logps": -239.02613830566406,
"epoch": 0.30952380952380953,
"grad_norm": 4.941489045777509,
"learning_rate": 1e-06,
"logits/chosen": -1.7267065048217773,
"logits/rejected": -1.6951403617858887,
"logps/chosen": -218.44247436523438,
"logps/rejected": -240.42892456054688,
"loss": 0.4982,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.005413799546658993,
"rewards/margins": 0.008614081889390945,
"rewards/rejected": -0.014027881436049938,
"step": 13
},
{
"debug/policy_chosen_logits": -1.795722484588623,
"debug/policy_chosen_logps": -235.3023681640625,
"debug/policy_rejected_logits": -1.7265783548355103,
"debug/policy_rejected_logps": -227.51165771484375,
"debug/reference_chosen_logps": -233.77352905273438,
"debug/reference_rejected_logps": -225.99020385742188,
"epoch": 0.3333333333333333,
"grad_norm": 5.251200069837007,
"learning_rate": 1e-06,
"logits/chosen": -1.795722484588623,
"logits/rejected": -1.7265783548355103,
"logps/chosen": -235.3023681640625,
"logps/rejected": -227.51165771484375,
"loss": 0.4978,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.015288218855857849,
"rewards/margins": -7.373839616775513e-05,
"rewards/rejected": -0.015214480459690094,
"step": 14
},
{
"debug/policy_chosen_logits": -1.7187248468399048,
"debug/policy_chosen_logps": -219.02468872070312,
"debug/policy_rejected_logits": -1.710004210472107,
"debug/policy_rejected_logps": -239.28665161132812,
"debug/reference_chosen_logps": -218.7939453125,
"debug/reference_rejected_logps": -236.74258422851562,
"epoch": 0.35714285714285715,
"grad_norm": 5.232970784816434,
"learning_rate": 1e-06,
"logits/chosen": -1.7187248468399048,
"logits/rejected": -1.710004210472107,
"logps/chosen": -219.02468872070312,
"logps/rejected": -239.28665161132812,
"loss": 0.4967,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.002307548187673092,
"rewards/margins": 0.023132991045713425,
"rewards/rejected": -0.025440538302063942,
"step": 15
},
{
"debug/policy_chosen_logits": -1.7510279417037964,
"debug/policy_chosen_logps": -213.79452514648438,
"debug/policy_rejected_logits": -1.6364609003067017,
"debug/policy_rejected_logps": -230.83267211914062,
"debug/reference_chosen_logps": -216.0415802001953,
"debug/reference_rejected_logps": -228.79180908203125,
"epoch": 0.38095238095238093,
"grad_norm": 4.507041423446427,
"learning_rate": 1e-06,
"logits/chosen": -1.7510279417037964,
"logits/rejected": -1.6364609003067017,
"logps/chosen": -213.79452514648438,
"logps/rejected": -230.83267211914062,
"loss": 0.4978,
"rewards/accuracies": 0.875,
"rewards/chosen": 0.022470567375421524,
"rewards/margins": 0.04287933558225632,
"rewards/rejected": -0.020408764481544495,
"step": 16
},
{
"debug/policy_chosen_logits": -1.7066298723220825,
"debug/policy_chosen_logps": -234.27340698242188,
"debug/policy_rejected_logits": -1.7131836414337158,
"debug/policy_rejected_logps": -234.25454711914062,
"debug/reference_chosen_logps": -235.27389526367188,
"debug/reference_rejected_logps": -234.8538818359375,
"epoch": 0.40476190476190477,
"grad_norm": 4.798628310057689,
"learning_rate": 1e-06,
"logits/chosen": -1.7066298723220825,
"logits/rejected": -1.7131836414337158,
"logps/chosen": -234.27340698242188,
"logps/rejected": -234.25454711914062,
"loss": 0.5,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.010004977695643902,
"rewards/margins": 0.004011764191091061,
"rewards/rejected": 0.005993213504552841,
"step": 17
},
{
"debug/policy_chosen_logits": -1.7528653144836426,
"debug/policy_chosen_logps": -251.7496337890625,
"debug/policy_rejected_logits": -1.561893343925476,
"debug/policy_rejected_logps": -227.67825317382812,
"debug/reference_chosen_logps": -250.92552185058594,
"debug/reference_rejected_logps": -227.5825958251953,
"epoch": 0.42857142857142855,
"grad_norm": 4.585902807542465,
"learning_rate": 1e-06,
"logits/chosen": -1.7528653144836426,
"logits/rejected": -1.561893343925476,
"logps/chosen": -251.7496337890625,
"logps/rejected": -227.67825317382812,
"loss": 0.4927,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.008241119794547558,
"rewards/margins": -0.007284717168658972,
"rewards/rejected": -0.0009564019273966551,
"step": 18
},
{
"debug/policy_chosen_logits": -1.7094969749450684,
"debug/policy_chosen_logps": -218.6276397705078,
"debug/policy_rejected_logits": -1.6478347778320312,
"debug/policy_rejected_logps": -222.29495239257812,
"debug/reference_chosen_logps": -218.80328369140625,
"debug/reference_rejected_logps": -223.06387329101562,
"epoch": 0.4523809523809524,
"grad_norm": 4.499072587904488,
"learning_rate": 1e-06,
"logits/chosen": -1.7094969749450684,
"logits/rejected": -1.6478347778320312,
"logps/chosen": -218.6276397705078,
"logps/rejected": -222.29495239257812,
"loss": 0.4999,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.001756363082677126,
"rewards/margins": -0.0059328461065888405,
"rewards/rejected": 0.007689208723604679,
"step": 19
},
{
"debug/policy_chosen_logits": -1.7157065868377686,
"debug/policy_chosen_logps": -233.549560546875,
"debug/policy_rejected_logits": -1.6739444732666016,
"debug/policy_rejected_logps": -228.99595642089844,
"debug/reference_chosen_logps": -234.72760009765625,
"debug/reference_rejected_logps": -227.83563232421875,
"epoch": 0.47619047619047616,
"grad_norm": 4.575115483844653,
"learning_rate": 1e-06,
"logits/chosen": -1.7157065868377686,
"logits/rejected": -1.6739444732666016,
"logps/chosen": -233.549560546875,
"logps/rejected": -228.99595642089844,
"loss": 0.4987,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.011780375614762306,
"rewards/margins": 0.023383673280477524,
"rewards/rejected": -0.011603298597037792,
"step": 20
},
{
"debug/policy_chosen_logits": -1.6785526275634766,
"debug/policy_chosen_logps": -229.71510314941406,
"debug/policy_rejected_logits": -1.5240322351455688,
"debug/policy_rejected_logps": -219.5631103515625,
"debug/reference_chosen_logps": -229.66180419921875,
"debug/reference_rejected_logps": -219.04263305664062,
"epoch": 0.5,
"grad_norm": 4.692408631629601,
"learning_rate": 1e-06,
"logits/chosen": -1.6785526275634766,
"logits/rejected": -1.5240322351455688,
"logps/chosen": -229.71510314941406,
"logps/rejected": -219.5631103515625,
"loss": 0.504,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.0005329509731382132,
"rewards/margins": 0.004671857692301273,
"rewards/rejected": -0.00520480889827013,
"step": 21
},
{
"debug/policy_chosen_logits": -1.7138959169387817,
"debug/policy_chosen_logps": -258.2476806640625,
"debug/policy_rejected_logits": -1.6827083826065063,
"debug/policy_rejected_logps": -217.47686767578125,
"debug/reference_chosen_logps": -258.50408935546875,
"debug/reference_rejected_logps": -218.35800170898438,
"epoch": 0.5238095238095238,
"grad_norm": 5.349249118587083,
"learning_rate": 1e-06,
"logits/chosen": -1.7138959169387817,
"logits/rejected": -1.6827083826065063,
"logps/chosen": -258.2476806640625,
"logps/rejected": -217.47686767578125,
"loss": 0.5027,
"rewards/accuracies": 0.375,
"rewards/chosen": 0.0025640674866735935,
"rewards/margins": -0.00624719588086009,
"rewards/rejected": 0.008811263367533684,
"step": 22
},
{
"debug/policy_chosen_logits": -1.7912077903747559,
"debug/policy_chosen_logps": -216.38299560546875,
"debug/policy_rejected_logits": -1.8135955333709717,
"debug/policy_rejected_logps": -212.30813598632812,
"debug/reference_chosen_logps": -217.41909790039062,
"debug/reference_rejected_logps": -213.4788055419922,
"epoch": 0.5476190476190477,
"grad_norm": 5.700609487442719,
"learning_rate": 1e-06,
"logits/chosen": -1.7912077903747559,
"logits/rejected": -1.8135955333709717,
"logps/chosen": -216.38299560546875,
"logps/rejected": -212.30813598632812,
"loss": 0.5012,
"rewards/accuracies": 0.375,
"rewards/chosen": 0.010360946878790855,
"rewards/margins": -0.0013457299210131168,
"rewards/rejected": 0.011706676334142685,
"step": 23
},
{
"debug/policy_chosen_logits": -1.7215830087661743,
"debug/policy_chosen_logps": -224.13088989257812,
"debug/policy_rejected_logits": -1.7539652585983276,
"debug/policy_rejected_logps": -252.66017150878906,
"debug/reference_chosen_logps": -223.94528198242188,
"debug/reference_rejected_logps": -251.1186065673828,
"epoch": 0.5714285714285714,
"grad_norm": 4.945978414250695,
"learning_rate": 1e-06,
"logits/chosen": -1.7215830087661743,
"logits/rejected": -1.7539652585983276,
"logps/chosen": -224.13088989257812,
"logps/rejected": -252.66017150878906,
"loss": 0.4913,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.0018562317127361894,
"rewards/margins": 0.013559339568018913,
"rewards/rejected": -0.015415573492646217,
"step": 24
},
{
"debug/policy_chosen_logits": -1.7326081991195679,
"debug/policy_chosen_logps": -226.6582794189453,
"debug/policy_rejected_logits": -1.701865553855896,
"debug/policy_rejected_logps": -225.5404052734375,
"debug/reference_chosen_logps": -227.42691040039062,
"debug/reference_rejected_logps": -224.72911071777344,
"epoch": 0.5952380952380952,
"grad_norm": 5.05266825293041,
"learning_rate": 1e-06,
"logits/chosen": -1.7326081991195679,
"logits/rejected": -1.701865553855896,
"logps/chosen": -226.6582794189453,
"logps/rejected": -225.5404052734375,
"loss": 0.4988,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.007686404511332512,
"rewards/margins": 0.01579955965280533,
"rewards/rejected": -0.008113155141472816,
"step": 25
},
{
"debug/policy_chosen_logits": -1.797877550125122,
"debug/policy_chosen_logps": -232.49713134765625,
"debug/policy_rejected_logits": -1.7819045782089233,
"debug/policy_rejected_logps": -239.3779296875,
"debug/reference_chosen_logps": -231.29766845703125,
"debug/reference_rejected_logps": -239.3263702392578,
"epoch": 0.6190476190476191,
"grad_norm": 5.6239784950743745,
"learning_rate": 1e-06,
"logits/chosen": -1.797877550125122,
"logits/rejected": -1.7819045782089233,
"logps/chosen": -232.49713134765625,
"logps/rejected": -239.3779296875,
"loss": 0.4983,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.011994647793471813,
"rewards/margins": -0.011478938162326813,
"rewards/rejected": -0.0005157091654837132,
"step": 26
},
{
"debug/policy_chosen_logits": -1.6310824155807495,
"debug/policy_chosen_logps": -223.48495483398438,
"debug/policy_rejected_logits": -1.5401698350906372,
"debug/policy_rejected_logps": -237.32968139648438,
"debug/reference_chosen_logps": -225.53250122070312,
"debug/reference_rejected_logps": -237.51010131835938,
"epoch": 0.6428571428571429,
"grad_norm": 5.775503509441982,
"learning_rate": 1e-06,
"logits/chosen": -1.6310824155807495,
"logits/rejected": -1.5401698350906372,
"logps/chosen": -223.48495483398438,
"logps/rejected": -237.32968139648438,
"loss": 0.4974,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.020475387573242188,
"rewards/margins": 0.018671227619051933,
"rewards/rejected": 0.0018041613511741161,
"step": 27
},
{
"debug/policy_chosen_logits": -1.8868080377578735,
"debug/policy_chosen_logps": -228.5826416015625,
"debug/policy_rejected_logits": -1.640351414680481,
"debug/policy_rejected_logps": -232.02215576171875,
"debug/reference_chosen_logps": -227.31100463867188,
"debug/reference_rejected_logps": -231.01402282714844,
"epoch": 0.6666666666666666,
"grad_norm": 4.615984568922464,
"learning_rate": 1e-06,
"logits/chosen": -1.8868080377578735,
"logits/rejected": -1.640351414680481,
"logps/chosen": -228.5826416015625,
"logps/rejected": -232.02215576171875,
"loss": 0.5044,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.012716369703412056,
"rewards/margins": -0.0026350021362304688,
"rewards/rejected": -0.010081367567181587,
"step": 28
},
{
"debug/policy_chosen_logits": -1.6878303289413452,
"debug/policy_chosen_logps": -207.95928955078125,
"debug/policy_rejected_logits": -1.6597003936767578,
"debug/policy_rejected_logps": -203.81204223632812,
"debug/reference_chosen_logps": -209.3488006591797,
"debug/reference_rejected_logps": -204.5792694091797,
"epoch": 0.6904761904761905,
"grad_norm": 5.141749172246753,
"learning_rate": 1e-06,
"logits/chosen": -1.6878303289413452,
"logits/rejected": -1.6597003936767578,
"logps/chosen": -207.95928955078125,
"logps/rejected": -203.81204223632812,
"loss": 0.5038,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.013895034790039062,
"rewards/margins": 0.006222839932888746,
"rewards/rejected": 0.0076721953228116035,
"step": 29
},
{
"debug/policy_chosen_logits": -1.654941201210022,
"debug/policy_chosen_logps": -214.6881866455078,
"debug/policy_rejected_logits": -1.7197308540344238,
"debug/policy_rejected_logps": -230.4603271484375,
"debug/reference_chosen_logps": -213.96749877929688,
"debug/reference_rejected_logps": -229.0462188720703,
"epoch": 0.7142857142857143,
"grad_norm": 4.802743085723465,
"learning_rate": 1e-06,
"logits/chosen": -1.654941201210022,
"logits/rejected": -1.7197308540344238,
"logps/chosen": -214.6881866455078,
"logps/rejected": -230.4603271484375,
"loss": 0.4929,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.0072069549933075905,
"rewards/margins": 0.006934070959687233,
"rewards/rejected": -0.014141025952994823,
"step": 30
},
{
"debug/policy_chosen_logits": -1.5828349590301514,
"debug/policy_chosen_logps": -237.3705596923828,
"debug/policy_rejected_logits": -1.5794568061828613,
"debug/policy_rejected_logps": -254.6802978515625,
"debug/reference_chosen_logps": -235.51742553710938,
"debug/reference_rejected_logps": -251.95379638671875,
"epoch": 0.7380952380952381,
"grad_norm": 5.710393511185945,
"learning_rate": 1e-06,
"logits/chosen": -1.5828349590301514,
"logits/rejected": -1.5794568061828613,
"logps/chosen": -237.3705596923828,
"logps/rejected": -254.6802978515625,
"loss": 0.4965,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.0185314379632473,
"rewards/margins": 0.00873336661607027,
"rewards/rejected": -0.027264803647994995,
"step": 31
},
{
"debug/policy_chosen_logits": -1.8172823190689087,
"debug/policy_chosen_logps": -233.02899169921875,
"debug/policy_rejected_logits": -1.8409276008605957,
"debug/policy_rejected_logps": -226.48519897460938,
"debug/reference_chosen_logps": -232.89686584472656,
"debug/reference_rejected_logps": -226.06564331054688,
"epoch": 0.7619047619047619,
"grad_norm": 4.735509214480415,
"learning_rate": 1e-06,
"logits/chosen": -1.8172823190689087,
"logits/rejected": -1.8409276008605957,
"logps/chosen": -233.02899169921875,
"logps/rejected": -226.48519897460938,
"loss": 0.5039,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.0013212203048169613,
"rewards/margins": 0.0028741275891661644,
"rewards/rejected": -0.0041953460313379765,
"step": 32
},
{
"debug/policy_chosen_logits": -1.7416476011276245,
"debug/policy_chosen_logps": -210.79525756835938,
"debug/policy_rejected_logits": -1.65560781955719,
"debug/policy_rejected_logps": -239.12855529785156,
"debug/reference_chosen_logps": -209.15957641601562,
"debug/reference_rejected_logps": -239.18234252929688,
"epoch": 0.7857142857142857,
"grad_norm": 4.895068163607816,
"learning_rate": 1e-06,
"logits/chosen": -1.7416476011276245,
"logits/rejected": -1.65560781955719,
"logps/chosen": -210.79525756835938,
"logps/rejected": -239.12855529785156,
"loss": 0.5056,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.016356829553842545,
"rewards/margins": -0.016894912347197533,
"rewards/rejected": 0.0005380823276937008,
"step": 33
},
{
"debug/policy_chosen_logits": -1.7726426124572754,
"debug/policy_chosen_logps": -225.79507446289062,
"debug/policy_rejected_logits": -1.7965084314346313,
"debug/policy_rejected_logps": -220.8616943359375,
"debug/reference_chosen_logps": -225.74952697753906,
"debug/reference_rejected_logps": -221.5814666748047,
"epoch": 0.8095238095238095,
"grad_norm": 5.792452647884142,
"learning_rate": 1e-06,
"logits/chosen": -1.7726426124572754,
"logits/rejected": -1.7965084314346313,
"logps/chosen": -225.79507446289062,
"logps/rejected": -220.8616943359375,
"loss": 0.4985,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.00045545492321252823,
"rewards/margins": -0.007653160020709038,
"rewards/rejected": 0.007197704166173935,
"step": 34
},
{
"debug/policy_chosen_logits": -1.6315754652023315,
"debug/policy_chosen_logps": -234.38040161132812,
"debug/policy_rejected_logits": -1.613837718963623,
"debug/policy_rejected_logps": -246.2030029296875,
"debug/reference_chosen_logps": -236.5584716796875,
"debug/reference_rejected_logps": -244.6172332763672,
"epoch": 0.8333333333333334,
"grad_norm": 5.023497019038571,
"learning_rate": 1e-06,
"logits/chosen": -1.6315754652023315,
"logits/rejected": -1.613837718963623,
"logps/chosen": -234.38040161132812,
"logps/rejected": -246.2030029296875,
"loss": 0.4951,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.021780777722597122,
"rewards/margins": 0.0376385897397995,
"rewards/rejected": -0.015857810154557228,
"step": 35
},
{
"debug/policy_chosen_logits": -1.733331561088562,
"debug/policy_chosen_logps": -232.724609375,
"debug/policy_rejected_logits": -1.7223798036575317,
"debug/policy_rejected_logps": -222.6837158203125,
"debug/reference_chosen_logps": -233.48643493652344,
"debug/reference_rejected_logps": -222.51834106445312,
"epoch": 0.8571428571428571,
"grad_norm": 4.953158777901106,
"learning_rate": 1e-06,
"logits/chosen": -1.733331561088562,
"logits/rejected": -1.7223798036575317,
"logps/chosen": -232.724609375,
"logps/rejected": -222.6837158203125,
"loss": 0.4898,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.00761817954480648,
"rewards/margins": 0.009272022172808647,
"rewards/rejected": -0.0016538426280021667,
"step": 36
},
{
"debug/policy_chosen_logits": -1.859521508216858,
"debug/policy_chosen_logps": -225.0269775390625,
"debug/policy_rejected_logits": -1.8794121742248535,
"debug/policy_rejected_logps": -238.612548828125,
"debug/reference_chosen_logps": -226.9108428955078,
"debug/reference_rejected_logps": -235.47463989257812,
"epoch": 0.8809523809523809,
"grad_norm": 4.654227939974065,
"learning_rate": 1e-06,
"logits/chosen": -1.859521508216858,
"logits/rejected": -1.8794121742248535,
"logps/chosen": -225.0269775390625,
"logps/rejected": -238.612548828125,
"loss": 0.4792,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.01883869245648384,
"rewards/margins": 0.05021755024790764,
"rewards/rejected": -0.031378861516714096,
"step": 37
},
{
"debug/policy_chosen_logits": -1.8789831399917603,
"debug/policy_chosen_logps": -208.9774169921875,
"debug/policy_rejected_logits": -1.8391132354736328,
"debug/policy_rejected_logps": -209.525390625,
"debug/reference_chosen_logps": -207.11997985839844,
"debug/reference_rejected_logps": -208.88546752929688,
"epoch": 0.9047619047619048,
"grad_norm": 5.750158702312609,
"learning_rate": 1e-06,
"logits/chosen": -1.8789831399917603,
"logits/rejected": -1.8391132354736328,
"logps/chosen": -208.9774169921875,
"logps/rejected": -209.525390625,
"loss": 0.4916,
"rewards/accuracies": 0.25,
"rewards/chosen": -0.018574314191937447,
"rewards/margins": -0.012175025418400764,
"rewards/rejected": -0.0063992878422141075,
"step": 38
},
{
"debug/policy_chosen_logits": -1.8669378757476807,
"debug/policy_chosen_logps": -256.5040283203125,
"debug/policy_rejected_logits": -1.771195411682129,
"debug/policy_rejected_logps": -240.91348266601562,
"debug/reference_chosen_logps": -256.64874267578125,
"debug/reference_rejected_logps": -240.50926208496094,
"epoch": 0.9285714285714286,
"grad_norm": 5.01330786712291,
"learning_rate": 1e-06,
"logits/chosen": -1.8669378757476807,
"logits/rejected": -1.771195411682129,
"logps/chosen": -256.5040283203125,
"logps/rejected": -240.91348266601562,
"loss": 0.4963,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.0014472389593720436,
"rewards/margins": 0.005489482544362545,
"rewards/rejected": -0.004042243584990501,
"step": 39
},
{
"debug/policy_chosen_logits": -1.8057416677474976,
"debug/policy_chosen_logps": -199.05746459960938,
"debug/policy_rejected_logits": -1.72264564037323,
"debug/policy_rejected_logps": -230.5078125,
"debug/reference_chosen_logps": -198.6903533935547,
"debug/reference_rejected_logps": -230.05313110351562,
"epoch": 0.9523809523809523,
"grad_norm": 4.6267949338716425,
"learning_rate": 1e-06,
"logits/chosen": -1.8057416677474976,
"logits/rejected": -1.72264564037323,
"logps/chosen": -199.05746459960938,
"logps/rejected": -230.5078125,
"loss": 0.5011,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.0036712647415697575,
"rewards/margins": 0.0008756828028708696,
"rewards/rejected": -0.004546947777271271,
"step": 40
},
{
"debug/policy_chosen_logits": -1.7298365831375122,
"debug/policy_chosen_logps": -215.87039184570312,
"debug/policy_rejected_logits": -1.7618889808654785,
"debug/policy_rejected_logps": -218.1046600341797,
"debug/reference_chosen_logps": -216.2220001220703,
"debug/reference_rejected_logps": -218.1280059814453,
"epoch": 0.9761904761904762,
"grad_norm": 5.043714508829569,
"learning_rate": 1e-06,
"logits/chosen": -1.7298365831375122,
"logits/rejected": -1.7618889808654785,
"logps/chosen": -215.87039184570312,
"logps/rejected": -218.1046600341797,
"loss": 0.4946,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.003516044234856963,
"rewards/margins": 0.003282546764239669,
"rewards/rejected": 0.00023349723778665066,
"step": 41
},
{
"debug/policy_chosen_logits": -1.9528135061264038,
"debug/policy_chosen_logps": -210.9334259033203,
"debug/policy_rejected_logits": -1.8600839376449585,
"debug/policy_rejected_logps": -225.0123291015625,
"debug/reference_chosen_logps": -212.72988891601562,
"debug/reference_rejected_logps": -225.8553924560547,
"epoch": 1.0,
"grad_norm": 5.608640172933804,
"learning_rate": 1e-06,
"logits/chosen": -1.9528135061264038,
"logits/rejected": -1.8600839376449585,
"logps/chosen": -210.9334259033203,
"logps/rejected": -225.0123291015625,
"loss": 0.468,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.017964612692594528,
"rewards/margins": 0.009534033946692944,
"rewards/rejected": 0.00843057781457901,
"step": 42
},
{
"epoch": 1.0,
"step": 42,
"total_flos": 0.0,
"train_loss": 0.49727832632405417,
"train_runtime": 148.4721,
"train_samples_per_second": 17.822,
"train_steps_per_second": 0.283
}
],
"logging_steps": 1,
"max_steps": 42,
"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": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}