1128 lines
44 KiB
JSON
1128 lines
44 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.26171159382360637,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.0005234231876472127,
|
|
"grad_norm": 124.9239730834961,
|
|
"learning_rate": 0.0,
|
|
"logits/chosen": -0.4007568359375,
|
|
"logits/rejected": -0.41790771484375,
|
|
"logps/chosen": -297.796875,
|
|
"logps/rejected": -248.078125,
|
|
"logps/weighted_chosen": -4.70263671875,
|
|
"logps/weighted_rejected": -3.35546875,
|
|
"loss": 0.6914,
|
|
"rewards/accuracies": 0.0,
|
|
"rewards/chosen": 0.0,
|
|
"rewards/margins": 0.0,
|
|
"rewards/rejected": 0.0,
|
|
"rewards/weighted_accuracies": 0.0,
|
|
"rewards/weighted_chosen": 0.0,
|
|
"rewards/weighted_margins": 0.0,
|
|
"rewards/weighted_rejected": 0.0,
|
|
"step": 1
|
|
},
|
|
{
|
|
"epoch": 0.005234231876472127,
|
|
"grad_norm": 265.8483581542969,
|
|
"learning_rate": 4.6875e-08,
|
|
"logits/chosen": -0.3176540732383728,
|
|
"logits/rejected": -0.3532511293888092,
|
|
"logps/chosen": -275.5720520019531,
|
|
"logps/rejected": -255.88714599609375,
|
|
"logps/weighted_chosen": -2.5362956523895264,
|
|
"logps/weighted_rejected": -2.774468421936035,
|
|
"loss": 0.6903,
|
|
"rewards/accuracies": 0.2743055522441864,
|
|
"rewards/chosen": 0.0078125,
|
|
"rewards/margins": 0.02973090298473835,
|
|
"rewards/rejected": -0.02191840298473835,
|
|
"rewards/weighted_accuracies": 0.3055555522441864,
|
|
"rewards/weighted_chosen": 0.0025702582206577063,
|
|
"rewards/weighted_margins": 0.0017564562149345875,
|
|
"rewards/weighted_rejected": 0.0008138020639307797,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.010468463752944255,
|
|
"grad_norm": 17.298032760620117,
|
|
"learning_rate": 9.895833333333332e-08,
|
|
"logits/chosen": -0.2979484498500824,
|
|
"logits/rejected": -0.3138488829135895,
|
|
"logps/chosen": -294.40625,
|
|
"logps/rejected": -272.69061279296875,
|
|
"logps/weighted_chosen": -2.3424315452575684,
|
|
"logps/weighted_rejected": -2.307080030441284,
|
|
"loss": 0.6912,
|
|
"rewards/accuracies": 0.296875,
|
|
"rewards/chosen": 0.007617187686264515,
|
|
"rewards/margins": 0.05781250074505806,
|
|
"rewards/rejected": -0.05019531399011612,
|
|
"rewards/weighted_accuracies": 0.3531250059604645,
|
|
"rewards/weighted_chosen": 0.0003913879336323589,
|
|
"rewards/weighted_margins": 0.00116729736328125,
|
|
"rewards/weighted_rejected": -0.0007759094005450606,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.015702695629416383,
|
|
"grad_norm": 79.90486907958984,
|
|
"learning_rate": 1.5104166666666664e-07,
|
|
"logits/chosen": -0.2918685972690582,
|
|
"logits/rejected": -0.3376312255859375,
|
|
"logps/chosen": -298.02032470703125,
|
|
"logps/rejected": -268.0210876464844,
|
|
"logps/weighted_chosen": -2.290698289871216,
|
|
"logps/weighted_rejected": -2.578540086746216,
|
|
"loss": 0.6921,
|
|
"rewards/accuracies": 0.3187499940395355,
|
|
"rewards/chosen": 0.0751953125,
|
|
"rewards/margins": 0.10107421875,
|
|
"rewards/rejected": -0.02587890625,
|
|
"rewards/weighted_accuracies": 0.375,
|
|
"rewards/weighted_chosen": 0.0023864745162427425,
|
|
"rewards/weighted_margins": -9.1552734375e-05,
|
|
"rewards/weighted_rejected": 0.0024780272506177425,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.02093692750588851,
|
|
"grad_norm": 32.818607330322266,
|
|
"learning_rate": 2.03125e-07,
|
|
"logits/chosen": -0.3010299801826477,
|
|
"logits/rejected": -0.3435005247592926,
|
|
"logps/chosen": -278.60235595703125,
|
|
"logps/rejected": -253.9250030517578,
|
|
"logps/weighted_chosen": -2.3670654296875,
|
|
"logps/weighted_rejected": -2.760937452316284,
|
|
"loss": 0.6923,
|
|
"rewards/accuracies": 0.34375,
|
|
"rewards/chosen": 0.12041015923023224,
|
|
"rewards/margins": 0.13857421278953552,
|
|
"rewards/rejected": -0.01816406287252903,
|
|
"rewards/weighted_accuracies": 0.4156250059604645,
|
|
"rewards/weighted_chosen": 0.0007911681896075606,
|
|
"rewards/weighted_margins": -0.00026035308837890625,
|
|
"rewards/weighted_rejected": 0.0010513305896893144,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.02617115938236064,
|
|
"grad_norm": 17.478111267089844,
|
|
"learning_rate": 2.552083333333333e-07,
|
|
"logits/chosen": -0.2825515866279602,
|
|
"logits/rejected": -0.3209354281425476,
|
|
"logps/chosen": -280.23126220703125,
|
|
"logps/rejected": -267.45001220703125,
|
|
"logps/weighted_chosen": -2.329760789871216,
|
|
"logps/weighted_rejected": -2.414306640625,
|
|
"loss": 0.6877,
|
|
"rewards/accuracies": 0.30937498807907104,
|
|
"rewards/chosen": 0.01835937425494194,
|
|
"rewards/margins": -0.06035156175494194,
|
|
"rewards/rejected": 0.07871093600988388,
|
|
"rewards/weighted_accuracies": 0.41874998807907104,
|
|
"rewards/weighted_chosen": 0.0070404051803052425,
|
|
"rewards/weighted_margins": 0.00885009765625,
|
|
"rewards/weighted_rejected": -0.0018066406482830644,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.031405391258832765,
|
|
"grad_norm": 39.04142761230469,
|
|
"learning_rate": 3.0729166666666665e-07,
|
|
"logits/chosen": -0.3157821595668793,
|
|
"logits/rejected": -0.30972975492477417,
|
|
"logps/chosen": -277.58123779296875,
|
|
"logps/rejected": -261.6187438964844,
|
|
"logps/weighted_chosen": -2.753100633621216,
|
|
"logps/weighted_rejected": -2.6282715797424316,
|
|
"loss": 0.6846,
|
|
"rewards/accuracies": 0.34687501192092896,
|
|
"rewards/chosen": 0.13593749701976776,
|
|
"rewards/margins": 0.02968749962747097,
|
|
"rewards/rejected": 0.10625000298023224,
|
|
"rewards/weighted_accuracies": 0.4937500059604645,
|
|
"rewards/weighted_chosen": 0.02093353308737278,
|
|
"rewards/weighted_margins": 0.015999604016542435,
|
|
"rewards/weighted_rejected": 0.0049346922896802425,
|
|
"step": 60
|
|
},
|
|
{
|
|
"epoch": 0.036639623135304895,
|
|
"grad_norm": 73.14978790283203,
|
|
"learning_rate": 3.59375e-07,
|
|
"logits/chosen": -0.32021790742874146,
|
|
"logits/rejected": -0.32699280977249146,
|
|
"logps/chosen": -289.6148376464844,
|
|
"logps/rejected": -244.8367156982422,
|
|
"logps/weighted_chosen": -2.221527099609375,
|
|
"logps/weighted_rejected": -2.581958055496216,
|
|
"loss": 0.6775,
|
|
"rewards/accuracies": 0.49687498807907104,
|
|
"rewards/chosen": 0.43046873807907104,
|
|
"rewards/margins": 0.5244140625,
|
|
"rewards/rejected": -0.09394530951976776,
|
|
"rewards/weighted_accuracies": 0.53125,
|
|
"rewards/weighted_chosen": 0.030600737780332565,
|
|
"rewards/weighted_margins": 0.034238435328006744,
|
|
"rewards/weighted_rejected": -0.0036373138427734375,
|
|
"step": 70
|
|
},
|
|
{
|
|
"epoch": 0.04187385501177702,
|
|
"grad_norm": 65.60453796386719,
|
|
"learning_rate": 4.114583333333333e-07,
|
|
"logits/chosen": -0.28860169649124146,
|
|
"logits/rejected": -0.3099655210971832,
|
|
"logps/chosen": -289.33984375,
|
|
"logps/rejected": -270.3453063964844,
|
|
"logps/weighted_chosen": -2.456127882003784,
|
|
"logps/weighted_rejected": -2.6986327171325684,
|
|
"loss": 0.6733,
|
|
"rewards/accuracies": 0.48124998807907104,
|
|
"rewards/chosen": 0.509765625,
|
|
"rewards/margins": 0.56396484375,
|
|
"rewards/rejected": -0.05419921875,
|
|
"rewards/weighted_accuracies": 0.5625,
|
|
"rewards/weighted_chosen": 0.04326324537396431,
|
|
"rewards/weighted_margins": 0.05183257907629013,
|
|
"rewards/weighted_rejected": -0.008570862002670765,
|
|
"step": 80
|
|
},
|
|
{
|
|
"epoch": 0.04710808688824915,
|
|
"grad_norm": 30.41583824157715,
|
|
"learning_rate": 4.6354166666666664e-07,
|
|
"logits/chosen": -0.3197769224643707,
|
|
"logits/rejected": -0.3126419186592102,
|
|
"logps/chosen": -280.0874938964844,
|
|
"logps/rejected": -257.95782470703125,
|
|
"logps/weighted_chosen": -2.6199097633361816,
|
|
"logps/weighted_rejected": -2.705517530441284,
|
|
"loss": 0.6743,
|
|
"rewards/accuracies": 0.4937500059604645,
|
|
"rewards/chosen": 0.5191406011581421,
|
|
"rewards/margins": 0.6358398199081421,
|
|
"rewards/rejected": -0.11669921875,
|
|
"rewards/weighted_accuracies": 0.5718749761581421,
|
|
"rewards/weighted_chosen": 0.07943420112133026,
|
|
"rewards/weighted_margins": 0.07568969577550888,
|
|
"rewards/weighted_rejected": 0.003753662109375,
|
|
"step": 90
|
|
},
|
|
{
|
|
"epoch": 0.05234231876472128,
|
|
"grad_norm": 61.700382232666016,
|
|
"learning_rate": 5.156249999999999e-07,
|
|
"logits/chosen": -0.2895751893520355,
|
|
"logits/rejected": -0.3621200621128082,
|
|
"logps/chosen": -290.88592529296875,
|
|
"logps/rejected": -286.73126220703125,
|
|
"logps/weighted_chosen": -2.083447217941284,
|
|
"logps/weighted_rejected": -2.6861205101013184,
|
|
"loss": 0.6862,
|
|
"rewards/accuracies": 0.5406249761581421,
|
|
"rewards/chosen": 0.10849609225988388,
|
|
"rewards/margins": 0.94482421875,
|
|
"rewards/rejected": -0.8363281488418579,
|
|
"rewards/weighted_accuracies": 0.574999988079071,
|
|
"rewards/weighted_chosen": 0.05786743015050888,
|
|
"rewards/weighted_margins": 0.058258056640625,
|
|
"rewards/weighted_rejected": -0.0002998351992573589,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.05757655064119341,
|
|
"grad_norm": 33.45664978027344,
|
|
"learning_rate": 5.677083333333333e-07,
|
|
"logits/chosen": -0.3412017822265625,
|
|
"logits/rejected": -0.3312934935092926,
|
|
"logps/chosen": -296.8671875,
|
|
"logps/rejected": -262.3460998535156,
|
|
"logps/weighted_chosen": -2.776904344558716,
|
|
"logps/weighted_rejected": -2.8023438453674316,
|
|
"loss": 0.6835,
|
|
"rewards/accuracies": 0.546875,
|
|
"rewards/chosen": -0.4873046875,
|
|
"rewards/margins": 1.0051758289337158,
|
|
"rewards/rejected": -1.4924805164337158,
|
|
"rewards/weighted_accuracies": 0.5062500238418579,
|
|
"rewards/weighted_chosen": 0.01709899865090847,
|
|
"rewards/weighted_margins": 0.05614929273724556,
|
|
"rewards/weighted_rejected": -0.03909149020910263,
|
|
"step": 110
|
|
},
|
|
{
|
|
"epoch": 0.06281078251766553,
|
|
"grad_norm": 61.74055862426758,
|
|
"learning_rate": 6.197916666666666e-07,
|
|
"logits/chosen": -0.3482359051704407,
|
|
"logits/rejected": -0.3736562728881836,
|
|
"logps/chosen": -295.49530029296875,
|
|
"logps/rejected": -256.08282470703125,
|
|
"logps/weighted_chosen": -2.066760301589966,
|
|
"logps/weighted_rejected": -2.2858643531799316,
|
|
"loss": 0.6833,
|
|
"rewards/accuracies": 0.5687500238418579,
|
|
"rewards/chosen": -0.8194335699081421,
|
|
"rewards/margins": 1.502539038658142,
|
|
"rewards/rejected": -2.321972608566284,
|
|
"rewards/weighted_accuracies": 0.5874999761581421,
|
|
"rewards/weighted_chosen": 0.01521911658346653,
|
|
"rewards/weighted_margins": 0.05682678148150444,
|
|
"rewards/weighted_rejected": -0.04169006273150444,
|
|
"step": 120
|
|
},
|
|
{
|
|
"epoch": 0.06804501439413765,
|
|
"grad_norm": 45.78140640258789,
|
|
"learning_rate": 6.718749999999999e-07,
|
|
"logits/chosen": -0.3164505064487457,
|
|
"logits/rejected": -0.3121093809604645,
|
|
"logps/chosen": -306.4761657714844,
|
|
"logps/rejected": -279.87030029296875,
|
|
"logps/weighted_chosen": -2.1032347679138184,
|
|
"logps/weighted_rejected": -2.432788133621216,
|
|
"loss": 0.6718,
|
|
"rewards/accuracies": 0.6312500238418579,
|
|
"rewards/chosen": -0.4637207090854645,
|
|
"rewards/margins": 1.8840820789337158,
|
|
"rewards/rejected": -2.347851514816284,
|
|
"rewards/weighted_accuracies": 0.559374988079071,
|
|
"rewards/weighted_chosen": 0.01874847337603569,
|
|
"rewards/weighted_margins": 0.07195129245519638,
|
|
"rewards/weighted_rejected": -0.05319824069738388,
|
|
"step": 130
|
|
},
|
|
{
|
|
"epoch": 0.07327924627060979,
|
|
"grad_norm": 47.97172546386719,
|
|
"learning_rate": 7.239583333333333e-07,
|
|
"logits/chosen": -0.36130523681640625,
|
|
"logits/rejected": -0.3767889142036438,
|
|
"logps/chosen": -300.2398376464844,
|
|
"logps/rejected": -276.47344970703125,
|
|
"logps/weighted_chosen": -2.402172803878784,
|
|
"logps/weighted_rejected": -2.704345703125,
|
|
"loss": 0.6601,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -1.6025390625,
|
|
"rewards/margins": 2.2777342796325684,
|
|
"rewards/rejected": -3.8802733421325684,
|
|
"rewards/weighted_accuracies": 0.628125011920929,
|
|
"rewards/weighted_chosen": 0.0066894530318677425,
|
|
"rewards/weighted_margins": 0.10413207858800888,
|
|
"rewards/weighted_rejected": -0.0973968505859375,
|
|
"step": 140
|
|
},
|
|
{
|
|
"epoch": 0.07851347814708191,
|
|
"grad_norm": 18.48399543762207,
|
|
"learning_rate": 7.760416666666666e-07,
|
|
"logits/chosen": -0.3260757327079773,
|
|
"logits/rejected": -0.33597564697265625,
|
|
"logps/chosen": -286.6578063964844,
|
|
"logps/rejected": -255.8078155517578,
|
|
"logps/weighted_chosen": -2.939697265625,
|
|
"logps/weighted_rejected": -2.9384522438049316,
|
|
"loss": 0.6693,
|
|
"rewards/accuracies": 0.6312500238418579,
|
|
"rewards/chosen": -2.329296827316284,
|
|
"rewards/margins": 2.532031297683716,
|
|
"rewards/rejected": -4.861328125,
|
|
"rewards/weighted_accuracies": 0.625,
|
|
"rewards/weighted_chosen": 0.01940460130572319,
|
|
"rewards/weighted_margins": 0.12819823622703552,
|
|
"rewards/weighted_rejected": -0.10869445651769638,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 0.08374771002355404,
|
|
"grad_norm": 16.847576141357422,
|
|
"learning_rate": 8.28125e-07,
|
|
"logits/chosen": -0.37299805879592896,
|
|
"logits/rejected": -0.4207763671875,
|
|
"logps/chosen": -306.01873779296875,
|
|
"logps/rejected": -279.0289001464844,
|
|
"logps/weighted_chosen": -2.4432616233825684,
|
|
"logps/weighted_rejected": -2.4922118186950684,
|
|
"loss": 0.631,
|
|
"rewards/accuracies": 0.6468750238418579,
|
|
"rewards/chosen": -2.1587891578674316,
|
|
"rewards/margins": 3.69091796875,
|
|
"rewards/rejected": -5.849707126617432,
|
|
"rewards/weighted_accuracies": 0.637499988079071,
|
|
"rewards/weighted_chosen": 0.04796142503619194,
|
|
"rewards/weighted_margins": 0.21007385849952698,
|
|
"rewards/weighted_rejected": -0.1621856689453125,
|
|
"step": 160
|
|
},
|
|
{
|
|
"epoch": 0.08898194190002617,
|
|
"grad_norm": 237.514404296875,
|
|
"learning_rate": 8.802083333333333e-07,
|
|
"logits/chosen": -0.42098236083984375,
|
|
"logits/rejected": -0.4231323301792145,
|
|
"logps/chosen": -306.859375,
|
|
"logps/rejected": -266.03125,
|
|
"logps/weighted_chosen": -2.6180663108825684,
|
|
"logps/weighted_rejected": -2.9256348609924316,
|
|
"loss": 0.709,
|
|
"rewards/accuracies": 0.6343749761581421,
|
|
"rewards/chosen": -3.73828125,
|
|
"rewards/margins": 4.098242282867432,
|
|
"rewards/rejected": -7.836328029632568,
|
|
"rewards/weighted_accuracies": 0.6031249761581421,
|
|
"rewards/weighted_chosen": 0.06289978325366974,
|
|
"rewards/weighted_margins": 0.186981201171875,
|
|
"rewards/weighted_rejected": -0.12393493950366974,
|
|
"step": 170
|
|
},
|
|
{
|
|
"epoch": 0.0942161737764983,
|
|
"grad_norm": 52.46952819824219,
|
|
"learning_rate": 9.322916666666666e-07,
|
|
"logits/chosen": -0.38860780000686646,
|
|
"logits/rejected": -0.40771445631980896,
|
|
"logps/chosen": -280.2789001464844,
|
|
"logps/rejected": -267.69219970703125,
|
|
"logps/weighted_chosen": -2.579449415206909,
|
|
"logps/weighted_rejected": -3.0037841796875,
|
|
"loss": 0.6853,
|
|
"rewards/accuracies": 0.6343749761581421,
|
|
"rewards/chosen": -6.609179496765137,
|
|
"rewards/margins": 4.819726467132568,
|
|
"rewards/rejected": -11.427831649780273,
|
|
"rewards/weighted_accuracies": 0.6031249761581421,
|
|
"rewards/weighted_chosen": -0.04149780422449112,
|
|
"rewards/weighted_margins": 0.20274047553539276,
|
|
"rewards/weighted_rejected": -0.24451598525047302,
|
|
"step": 180
|
|
},
|
|
{
|
|
"epoch": 0.09945040565297043,
|
|
"grad_norm": 37.741477966308594,
|
|
"learning_rate": 9.84375e-07,
|
|
"logits/chosen": -0.4061637818813324,
|
|
"logits/rejected": -0.4410812258720398,
|
|
"logps/chosen": -314.60626220703125,
|
|
"logps/rejected": -278.87811279296875,
|
|
"logps/weighted_chosen": -2.378369092941284,
|
|
"logps/weighted_rejected": -2.7929444313049316,
|
|
"loss": 0.6517,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -8.988476753234863,
|
|
"rewards/margins": 6.061327934265137,
|
|
"rewards/rejected": -15.050390243530273,
|
|
"rewards/weighted_accuracies": 0.621874988079071,
|
|
"rewards/weighted_chosen": -0.11425475776195526,
|
|
"rewards/weighted_margins": 0.22973021864891052,
|
|
"rewards/weighted_rejected": -0.3441314697265625,
|
|
"step": 190
|
|
},
|
|
{
|
|
"epoch": 0.10468463752944256,
|
|
"grad_norm": 76.11390686035156,
|
|
"learning_rate": 9.99959085414323e-07,
|
|
"logits/chosen": -0.4347488284111023,
|
|
"logits/rejected": -0.46932679414749146,
|
|
"logps/chosen": -319.34686279296875,
|
|
"logps/rejected": -274.1890563964844,
|
|
"logps/weighted_chosen": -2.6759276390075684,
|
|
"logps/weighted_rejected": -3.110583543777466,
|
|
"loss": 0.6539,
|
|
"rewards/accuracies": 0.6312500238418579,
|
|
"rewards/chosen": -10.064160346984863,
|
|
"rewards/margins": 6.332421779632568,
|
|
"rewards/rejected": -16.395898818969727,
|
|
"rewards/weighted_accuracies": 0.612500011920929,
|
|
"rewards/weighted_chosen": -0.08293457329273224,
|
|
"rewards/weighted_margins": 0.25919800996780396,
|
|
"rewards/weighted_rejected": -0.34211426973342896,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.10991886940591468,
|
|
"grad_norm": 19.709537506103516,
|
|
"learning_rate": 9.997587035630105e-07,
|
|
"logits/chosen": -0.4373016357421875,
|
|
"logits/rejected": -0.47842711210250854,
|
|
"logps/chosen": -298.6578063964844,
|
|
"logps/rejected": -305.07733154296875,
|
|
"logps/weighted_chosen": -2.4120116233825684,
|
|
"logps/weighted_rejected": -3.137524366378784,
|
|
"loss": 0.6448,
|
|
"rewards/accuracies": 0.6656249761581421,
|
|
"rewards/chosen": -11.547070503234863,
|
|
"rewards/margins": 8.876855850219727,
|
|
"rewards/rejected": -20.425390243530273,
|
|
"rewards/weighted_accuracies": 0.659375011920929,
|
|
"rewards/weighted_chosen": -0.12418060004711151,
|
|
"rewards/weighted_margins": 0.24833527207374573,
|
|
"rewards/weighted_rejected": -0.37247925996780396,
|
|
"step": 210
|
|
},
|
|
{
|
|
"epoch": 0.11515310128238682,
|
|
"grad_norm": 39.1947135925293,
|
|
"learning_rate": 9.99391406364405e-07,
|
|
"logits/chosen": -0.424050897359848,
|
|
"logits/rejected": -0.4263656735420227,
|
|
"logps/chosen": -306.5453186035156,
|
|
"logps/rejected": -291.21405029296875,
|
|
"logps/weighted_chosen": -2.8793702125549316,
|
|
"logps/weighted_rejected": -3.2673583030700684,
|
|
"loss": 0.6969,
|
|
"rewards/accuracies": 0.612500011920929,
|
|
"rewards/chosen": -14.446874618530273,
|
|
"rewards/margins": 7.897851467132568,
|
|
"rewards/rejected": -22.343358993530273,
|
|
"rewards/weighted_accuracies": 0.606249988079071,
|
|
"rewards/weighted_chosen": -0.2553466856479645,
|
|
"rewards/weighted_margins": 0.21260376274585724,
|
|
"rewards/weighted_rejected": -0.46800535917282104,
|
|
"step": 220
|
|
},
|
|
{
|
|
"epoch": 0.12038733315885894,
|
|
"grad_norm": 437.028564453125,
|
|
"learning_rate": 9.988573164927884e-07,
|
|
"logits/chosen": -0.3824310302734375,
|
|
"logits/rejected": -0.4174056947231293,
|
|
"logps/chosen": -282.2710876464844,
|
|
"logps/rejected": -277.2734375,
|
|
"logps/weighted_chosen": -2.547900438308716,
|
|
"logps/weighted_rejected": -2.85693359375,
|
|
"loss": 0.7367,
|
|
"rewards/accuracies": 0.675000011920929,
|
|
"rewards/chosen": -13.438867568969727,
|
|
"rewards/margins": 12.128320693969727,
|
|
"rewards/rejected": -25.565820693969727,
|
|
"rewards/weighted_accuracies": 0.621874988079071,
|
|
"rewards/weighted_chosen": -0.24434813857078552,
|
|
"rewards/weighted_margins": 0.2164306640625,
|
|
"rewards/weighted_rejected": -0.4607299864292145,
|
|
"step": 230
|
|
},
|
|
{
|
|
"epoch": 0.12562156503533106,
|
|
"grad_norm": 139.88975524902344,
|
|
"learning_rate": 9.98156612329838e-07,
|
|
"logits/chosen": -0.4804244935512543,
|
|
"logits/rejected": -0.527056872844696,
|
|
"logps/chosen": -279.6773376464844,
|
|
"logps/rejected": -310.64532470703125,
|
|
"logps/weighted_chosen": -2.464794874191284,
|
|
"logps/weighted_rejected": -3.058520555496216,
|
|
"loss": 0.6735,
|
|
"rewards/accuracies": 0.637499988079071,
|
|
"rewards/chosen": -14.966699600219727,
|
|
"rewards/margins": 13.450976371765137,
|
|
"rewards/rejected": -28.419336318969727,
|
|
"rewards/weighted_accuracies": 0.6625000238418579,
|
|
"rewards/weighted_chosen": -0.11341552436351776,
|
|
"rewards/weighted_margins": 0.26569825410842896,
|
|
"rewards/weighted_rejected": -0.37919312715530396,
|
|
"step": 240
|
|
},
|
|
{
|
|
"epoch": 0.13085579691180318,
|
|
"grad_norm": 939.5078125,
|
|
"learning_rate": 9.97289527905053e-07,
|
|
"logits/chosen": -0.4899856448173523,
|
|
"logits/rejected": -0.4979141354560852,
|
|
"logps/chosen": -281.9039001464844,
|
|
"logps/rejected": -282.93438720703125,
|
|
"logps/weighted_chosen": -2.8607420921325684,
|
|
"logps/weighted_rejected": -3.0235595703125,
|
|
"loss": 0.7231,
|
|
"rewards/accuracies": 0.6343749761581421,
|
|
"rewards/chosen": -17.384668350219727,
|
|
"rewards/margins": 9.600194931030273,
|
|
"rewards/rejected": -26.986328125,
|
|
"rewards/weighted_accuracies": 0.5843750238418579,
|
|
"rewards/weighted_chosen": -0.20379944145679474,
|
|
"rewards/weighted_margins": 0.18174438178539276,
|
|
"rewards/weighted_rejected": -0.3856750428676605,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 0.1360900287882753,
|
|
"grad_norm": 24.300647735595703,
|
|
"learning_rate": 9.962563528175875e-07,
|
|
"logits/chosen": -0.4584365785121918,
|
|
"logits/rejected": -0.4911254942417145,
|
|
"logps/chosen": -317.06329345703125,
|
|
"logps/rejected": -289.06561279296875,
|
|
"logps/weighted_chosen": -2.4532713890075684,
|
|
"logps/weighted_rejected": -3.330212354660034,
|
|
"loss": 0.6636,
|
|
"rewards/accuracies": 0.590624988079071,
|
|
"rewards/chosen": -18.153711318969727,
|
|
"rewards/margins": 10.150195121765137,
|
|
"rewards/rejected": -28.3046875,
|
|
"rewards/weighted_accuracies": 0.6187499761581421,
|
|
"rewards/weighted_chosen": -0.18357238173484802,
|
|
"rewards/weighted_margins": 0.269927978515625,
|
|
"rewards/weighted_rejected": -0.4535202085971832,
|
|
"step": 260
|
|
},
|
|
{
|
|
"epoch": 0.14132426066474746,
|
|
"grad_norm": 228.46092224121094,
|
|
"learning_rate": 9.950574321395277e-07,
|
|
"logits/chosen": -0.4871993958950043,
|
|
"logits/rejected": -0.5071777105331421,
|
|
"logps/chosen": -310.2124938964844,
|
|
"logps/rejected": -291.5687561035156,
|
|
"logps/weighted_chosen": -2.747607469558716,
|
|
"logps/weighted_rejected": -3.0499024391174316,
|
|
"loss": 0.6525,
|
|
"rewards/accuracies": 0.6156250238418579,
|
|
"rewards/chosen": -20.7822265625,
|
|
"rewards/margins": 7.598730564117432,
|
|
"rewards/rejected": -28.382030487060547,
|
|
"rewards/weighted_accuracies": 0.6499999761581421,
|
|
"rewards/weighted_chosen": -0.2510986328125,
|
|
"rewards/weighted_margins": 0.2596893310546875,
|
|
"rewards/weighted_rejected": -0.5105804204940796,
|
|
"step": 270
|
|
},
|
|
{
|
|
"epoch": 0.14655849254121958,
|
|
"grad_norm": 713.1409912109375,
|
|
"learning_rate": 9.936931663006413e-07,
|
|
"logits/chosen": -0.4856887757778168,
|
|
"logits/rejected": -0.4791274964809418,
|
|
"logps/chosen": -323.875,
|
|
"logps/rejected": -312.1109313964844,
|
|
"logps/weighted_chosen": -2.7660155296325684,
|
|
"logps/weighted_rejected": -3.293591260910034,
|
|
"loss": 0.6155,
|
|
"rewards/accuracies": 0.671875,
|
|
"rewards/chosen": -20.981639862060547,
|
|
"rewards/margins": 11.869140625,
|
|
"rewards/rejected": -32.850975036621094,
|
|
"rewards/weighted_accuracies": 0.675000011920929,
|
|
"rewards/weighted_chosen": -0.13939818739891052,
|
|
"rewards/weighted_margins": 0.34244996309280396,
|
|
"rewards/weighted_rejected": -0.48178404569625854,
|
|
"step": 280
|
|
},
|
|
{
|
|
"epoch": 0.1517927244176917,
|
|
"grad_norm": 10845.4873046875,
|
|
"learning_rate": 9.921640109546357e-07,
|
|
"logits/chosen": -0.433615118265152,
|
|
"logits/rejected": -0.5038871765136719,
|
|
"logps/chosen": -285.6937561035156,
|
|
"logps/rejected": -280.76251220703125,
|
|
"logps/weighted_chosen": -2.63427734375,
|
|
"logps/weighted_rejected": -3.711669921875,
|
|
"loss": 0.65,
|
|
"rewards/accuracies": 0.6625000238418579,
|
|
"rewards/chosen": -18.829492568969727,
|
|
"rewards/margins": 10.345703125,
|
|
"rewards/rejected": -29.1728515625,
|
|
"rewards/weighted_accuracies": 0.6187499761581421,
|
|
"rewards/weighted_chosen": -0.12302856147289276,
|
|
"rewards/weighted_margins": 0.3319152891635895,
|
|
"rewards/weighted_rejected": -0.45494383573532104,
|
|
"step": 290
|
|
},
|
|
{
|
|
"epoch": 0.15702695629416383,
|
|
"grad_norm": 47.70283889770508,
|
|
"learning_rate": 9.90470476826975e-07,
|
|
"logits/chosen": -0.4355606138706207,
|
|
"logits/rejected": -0.4278717041015625,
|
|
"logps/chosen": -288.78045654296875,
|
|
"logps/rejected": -296.9765625,
|
|
"logps/weighted_chosen": -2.532849073410034,
|
|
"logps/weighted_rejected": -3.0221924781799316,
|
|
"loss": 0.6674,
|
|
"rewards/accuracies": 0.6625000238418579,
|
|
"rewards/chosen": -19.346094131469727,
|
|
"rewards/margins": 10.255078315734863,
|
|
"rewards/rejected": -29.59765625,
|
|
"rewards/weighted_accuracies": 0.6187499761581421,
|
|
"rewards/weighted_chosen": -0.21412047743797302,
|
|
"rewards/weighted_margins": 0.2544006407260895,
|
|
"rewards/weighted_rejected": -0.4681999087333679,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.16226118817063595,
|
|
"grad_norm": 644.2051391601562,
|
|
"learning_rate": 9.886131295443002e-07,
|
|
"logits/chosen": -0.462362676858902,
|
|
"logits/rejected": -0.5182906985282898,
|
|
"logps/chosen": -301.2421875,
|
|
"logps/rejected": -280.125,
|
|
"logps/weighted_chosen": -2.5293211936950684,
|
|
"logps/weighted_rejected": -2.9417481422424316,
|
|
"loss": 0.6339,
|
|
"rewards/accuracies": 0.668749988079071,
|
|
"rewards/chosen": -19.940235137939453,
|
|
"rewards/margins": 9.952539443969727,
|
|
"rewards/rejected": -29.892187118530273,
|
|
"rewards/weighted_accuracies": 0.65625,
|
|
"rewards/weighted_chosen": 0.0126800537109375,
|
|
"rewards/weighted_margins": 0.4334716796875,
|
|
"rewards/weighted_rejected": -0.42060548067092896,
|
|
"step": 310
|
|
},
|
|
{
|
|
"epoch": 0.16749542004710807,
|
|
"grad_norm": 42.43592834472656,
|
|
"learning_rate": 9.865925894455166e-07,
|
|
"logits/chosen": -0.538470447063446,
|
|
"logits/rejected": -0.5473419427871704,
|
|
"logps/chosen": -323.84686279296875,
|
|
"logps/rejected": -290.7406311035156,
|
|
"logps/weighted_chosen": -2.8351807594299316,
|
|
"logps/weighted_rejected": -3.2470703125,
|
|
"loss": 0.6762,
|
|
"rewards/accuracies": 0.637499988079071,
|
|
"rewards/chosen": -23.906444549560547,
|
|
"rewards/margins": 9.430078506469727,
|
|
"rewards/rejected": -33.3359375,
|
|
"rewards/weighted_accuracies": 0.6187499761581421,
|
|
"rewards/weighted_chosen": -0.23797301948070526,
|
|
"rewards/weighted_margins": 0.272064208984375,
|
|
"rewards/weighted_rejected": -0.5102294683456421,
|
|
"step": 320
|
|
},
|
|
{
|
|
"epoch": 0.17272965192358022,
|
|
"grad_norm": 46.229000091552734,
|
|
"learning_rate": 9.84409531374603e-07,
|
|
"logits/chosen": -0.5096954107284546,
|
|
"logits/rejected": -0.4907272458076477,
|
|
"logps/chosen": -332.18280029296875,
|
|
"logps/rejected": -298.5718688964844,
|
|
"logps/weighted_chosen": -2.775341749191284,
|
|
"logps/weighted_rejected": -3.165026903152466,
|
|
"loss": 0.6741,
|
|
"rewards/accuracies": 0.637499988079071,
|
|
"rewards/chosen": -27.408203125,
|
|
"rewards/margins": 8.212109565734863,
|
|
"rewards/rejected": -35.61640548706055,
|
|
"rewards/weighted_accuracies": 0.6312500238418579,
|
|
"rewards/weighted_chosen": -0.17764969170093536,
|
|
"rewards/weighted_margins": 0.241455078125,
|
|
"rewards/weighted_rejected": -0.41897353529930115,
|
|
"step": 330
|
|
},
|
|
{
|
|
"epoch": 0.17796388380005235,
|
|
"grad_norm": 70.134521484375,
|
|
"learning_rate": 9.820646844552219e-07,
|
|
"logits/chosen": -0.5079803466796875,
|
|
"logits/rejected": -0.5622657537460327,
|
|
"logps/chosen": -301.3257751464844,
|
|
"logps/rejected": -303.1468811035156,
|
|
"logps/weighted_chosen": -2.852612257003784,
|
|
"logps/weighted_rejected": -3.0445556640625,
|
|
"loss": 0.6485,
|
|
"rewards/accuracies": 0.6968749761581421,
|
|
"rewards/chosen": -24.6875,
|
|
"rewards/margins": 12.933984756469727,
|
|
"rewards/rejected": -37.61992263793945,
|
|
"rewards/weighted_accuracies": 0.65625,
|
|
"rewards/weighted_chosen": -0.19311293959617615,
|
|
"rewards/weighted_margins": 0.3020080626010895,
|
|
"rewards/weighted_rejected": -0.4951171875,
|
|
"step": 340
|
|
},
|
|
{
|
|
"epoch": 0.18319811567652447,
|
|
"grad_norm": 37.5682373046875,
|
|
"learning_rate": 9.795588318471964e-07,
|
|
"logits/chosen": -0.5787566900253296,
|
|
"logits/rejected": -0.5869846343994141,
|
|
"logps/chosen": -283.0796813964844,
|
|
"logps/rejected": -310.00469970703125,
|
|
"logps/weighted_chosen": -2.5658936500549316,
|
|
"logps/weighted_rejected": -2.95947265625,
|
|
"loss": 0.6307,
|
|
"rewards/accuracies": 0.625,
|
|
"rewards/chosen": -23.71875,
|
|
"rewards/margins": 9.896288871765137,
|
|
"rewards/rejected": -33.61249923706055,
|
|
"rewards/weighted_accuracies": 0.6468750238418579,
|
|
"rewards/weighted_chosen": -0.1531631499528885,
|
|
"rewards/weighted_margins": 0.32664793729782104,
|
|
"rewards/weighted_rejected": -0.479653924703598,
|
|
"step": 350
|
|
},
|
|
{
|
|
"epoch": 0.1884323475529966,
|
|
"grad_norm": 145.69015502929688,
|
|
"learning_rate": 9.768928104849415e-07,
|
|
"logits/chosen": -0.6053832769393921,
|
|
"logits/rejected": -0.6036437749862671,
|
|
"logps/chosen": -306.9140625,
|
|
"logps/rejected": -283.77655029296875,
|
|
"logps/weighted_chosen": -2.835888624191284,
|
|
"logps/weighted_rejected": -2.9885010719299316,
|
|
"loss": 0.6741,
|
|
"rewards/accuracies": 0.612500011920929,
|
|
"rewards/chosen": -24.237499237060547,
|
|
"rewards/margins": 10.624218940734863,
|
|
"rewards/rejected": -34.85859298706055,
|
|
"rewards/weighted_accuracies": 0.640625,
|
|
"rewards/weighted_chosen": -0.1663818359375,
|
|
"rewards/weighted_margins": 0.3080078065395355,
|
|
"rewards/weighted_rejected": -0.4744934141635895,
|
|
"step": 360
|
|
},
|
|
{
|
|
"epoch": 0.19366657942946872,
|
|
"grad_norm": 1347.0242919921875,
|
|
"learning_rate": 9.740675107979355e-07,
|
|
"logits/chosen": -0.5699676275253296,
|
|
"logits/rejected": -0.5878998041152954,
|
|
"logps/chosen": -342.36248779296875,
|
|
"logps/rejected": -311.7203063964844,
|
|
"logps/weighted_chosen": -2.303784132003784,
|
|
"logps/weighted_rejected": -3.164013624191284,
|
|
"loss": 0.6624,
|
|
"rewards/accuracies": 0.6312500238418579,
|
|
"rewards/chosen": -27.253124237060547,
|
|
"rewards/margins": 9.823046684265137,
|
|
"rewards/rejected": -37.07597732543945,
|
|
"rewards/weighted_accuracies": 0.643750011920929,
|
|
"rewards/weighted_chosen": -0.24745483696460724,
|
|
"rewards/weighted_margins": 0.27247315645217896,
|
|
"rewards/weighted_rejected": -0.5200088620185852,
|
|
"step": 370
|
|
},
|
|
{
|
|
"epoch": 0.19890081130594087,
|
|
"grad_norm": 1152.541748046875,
|
|
"learning_rate": 9.71083876413323e-07,
|
|
"logits/chosen": -0.5426742434501648,
|
|
"logits/rejected": -0.5510879755020142,
|
|
"logps/chosen": -331.25390625,
|
|
"logps/rejected": -310.33123779296875,
|
|
"logps/weighted_chosen": -2.4026856422424316,
|
|
"logps/weighted_rejected": -3.172229051589966,
|
|
"loss": 0.6676,
|
|
"rewards/accuracies": 0.621874988079071,
|
|
"rewards/chosen": -28.869922637939453,
|
|
"rewards/margins": 11.865234375,
|
|
"rewards/rejected": -40.73320388793945,
|
|
"rewards/weighted_accuracies": 0.6187499761581421,
|
|
"rewards/weighted_chosen": -0.24112549424171448,
|
|
"rewards/weighted_margins": 0.2662292420864105,
|
|
"rewards/weighted_rejected": -0.5070465207099915,
|
|
"step": 380
|
|
},
|
|
{
|
|
"epoch": 0.204135043182413,
|
|
"grad_norm": 69.06751251220703,
|
|
"learning_rate": 9.67942903840751e-07,
|
|
"logits/chosen": -0.5844222903251648,
|
|
"logits/rejected": -0.6357940435409546,
|
|
"logps/chosen": -331.53436279296875,
|
|
"logps/rejected": -316.50469970703125,
|
|
"logps/weighted_chosen": -2.523181200027466,
|
|
"logps/weighted_rejected": -3.170581102371216,
|
|
"loss": 0.6281,
|
|
"rewards/accuracies": 0.659375011920929,
|
|
"rewards/chosen": -27.312108993530273,
|
|
"rewards/margins": 14.989062309265137,
|
|
"rewards/rejected": -42.309181213378906,
|
|
"rewards/weighted_accuracies": 0.675000011920929,
|
|
"rewards/weighted_chosen": -0.17550353705883026,
|
|
"rewards/weighted_margins": 0.357177734375,
|
|
"rewards/weighted_rejected": -0.532818615436554,
|
|
"step": 390
|
|
},
|
|
{
|
|
"epoch": 0.2093692750588851,
|
|
"grad_norm": 64.58200073242188,
|
|
"learning_rate": 9.646456421395447e-07,
|
|
"logits/chosen": -0.573455810546875,
|
|
"logits/rejected": -0.5838836431503296,
|
|
"logps/chosen": -348.1484375,
|
|
"logps/rejected": -349.8687438964844,
|
|
"logps/weighted_chosen": -2.479809522628784,
|
|
"logps/weighted_rejected": -3.226489305496216,
|
|
"loss": 0.6551,
|
|
"rewards/accuracies": 0.65625,
|
|
"rewards/chosen": -29.957422256469727,
|
|
"rewards/margins": 16.478321075439453,
|
|
"rewards/rejected": -46.43476486206055,
|
|
"rewards/weighted_accuracies": 0.640625,
|
|
"rewards/weighted_chosen": -0.203724667429924,
|
|
"rewards/weighted_margins": 0.254974365234375,
|
|
"rewards/weighted_rejected": -0.4586502015590668,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 0.21460350693535724,
|
|
"grad_norm": 19.509164810180664,
|
|
"learning_rate": 9.611931925683266e-07,
|
|
"logits/chosen": -0.5067977905273438,
|
|
"logits/rejected": -0.5407257080078125,
|
|
"logps/chosen": -337.97967529296875,
|
|
"logps/rejected": -311.12811279296875,
|
|
"logps/weighted_chosen": -2.369091749191284,
|
|
"logps/weighted_rejected": -2.994799852371216,
|
|
"loss": 0.5919,
|
|
"rewards/accuracies": 0.6656249761581421,
|
|
"rewards/chosen": -31.741016387939453,
|
|
"rewards/margins": 15.541796684265137,
|
|
"rewards/rejected": -47.293357849121094,
|
|
"rewards/weighted_accuracies": 0.6937500238418579,
|
|
"rewards/weighted_chosen": -0.17744140326976776,
|
|
"rewards/weighted_margins": 0.386505126953125,
|
|
"rewards/weighted_rejected": -0.563507080078125,
|
|
"step": 410
|
|
},
|
|
{
|
|
"epoch": 0.21983773881182936,
|
|
"grad_norm": 29.50520133972168,
|
|
"learning_rate": 9.575867082172085e-07,
|
|
"logits/chosen": -0.5286194086074829,
|
|
"logits/rejected": -0.5688308477401733,
|
|
"logps/chosen": -344.96563720703125,
|
|
"logps/rejected": -328.64532470703125,
|
|
"logps/weighted_chosen": -2.8730225563049316,
|
|
"logps/weighted_rejected": -2.930432081222534,
|
|
"loss": 0.6442,
|
|
"rewards/accuracies": 0.675000011920929,
|
|
"rewards/chosen": -40.091796875,
|
|
"rewards/margins": 18.596485137939453,
|
|
"rewards/rejected": -58.685935974121094,
|
|
"rewards/weighted_accuracies": 0.628125011920929,
|
|
"rewards/weighted_chosen": -0.38405150175094604,
|
|
"rewards/weighted_margins": 0.39207762479782104,
|
|
"rewards/weighted_rejected": -0.7763046026229858,
|
|
"step": 420
|
|
},
|
|
{
|
|
"epoch": 0.22507197068830148,
|
|
"grad_norm": 15.081392288208008,
|
|
"learning_rate": 9.538273936226673e-07,
|
|
"logits/chosen": -0.5923309326171875,
|
|
"logits/rejected": -0.625653088092804,
|
|
"logps/chosen": -301.640625,
|
|
"logps/rejected": -313.77032470703125,
|
|
"logps/weighted_chosen": -3.0785889625549316,
|
|
"logps/weighted_rejected": -3.5872559547424316,
|
|
"loss": 0.668,
|
|
"rewards/accuracies": 0.628125011920929,
|
|
"rewards/chosen": -36.12675857543945,
|
|
"rewards/margins": 13.111328125,
|
|
"rewards/rejected": -49.236328125,
|
|
"rewards/weighted_accuracies": 0.6312500238418579,
|
|
"rewards/weighted_chosen": -0.28452759981155396,
|
|
"rewards/weighted_margins": 0.3635620176792145,
|
|
"rewards/weighted_rejected": -0.64788818359375,
|
|
"step": 430
|
|
},
|
|
{
|
|
"epoch": 0.23030620256477363,
|
|
"grad_norm": 23.0412540435791,
|
|
"learning_rate": 9.499165043652391e-07,
|
|
"logits/chosen": -0.5760772824287415,
|
|
"logits/rejected": -0.5774017572402954,
|
|
"logps/chosen": -328.1015625,
|
|
"logps/rejected": -318.84454345703125,
|
|
"logps/weighted_chosen": -3.0999755859375,
|
|
"logps/weighted_rejected": -3.371142625808716,
|
|
"loss": 0.6452,
|
|
"rewards/accuracies": 0.640625,
|
|
"rewards/chosen": -36.97431564331055,
|
|
"rewards/margins": 15.057812690734863,
|
|
"rewards/rejected": -52.0390625,
|
|
"rewards/weighted_accuracies": 0.628125011920929,
|
|
"rewards/weighted_chosen": -0.39061737060546875,
|
|
"rewards/weighted_margins": 0.3595825135707855,
|
|
"rewards/weighted_rejected": -0.7500244379043579,
|
|
"step": 440
|
|
},
|
|
{
|
|
"epoch": 0.23554043444124576,
|
|
"grad_norm": 22.85993003845215,
|
|
"learning_rate": 9.458553466501665e-07,
|
|
"logits/chosen": -0.6301513910293579,
|
|
"logits/rejected": -0.6610687375068665,
|
|
"logps/chosen": -324.77032470703125,
|
|
"logps/rejected": -298.57501220703125,
|
|
"logps/weighted_chosen": -3.1078124046325684,
|
|
"logps/weighted_rejected": -3.38037109375,
|
|
"loss": 0.6501,
|
|
"rewards/accuracies": 0.640625,
|
|
"rewards/chosen": -38.617774963378906,
|
|
"rewards/margins": 16.141407012939453,
|
|
"rewards/rejected": -54.763671875,
|
|
"rewards/weighted_accuracies": 0.653124988079071,
|
|
"rewards/weighted_chosen": -0.4693542420864105,
|
|
"rewards/weighted_margins": 0.4438629150390625,
|
|
"rewards/weighted_rejected": -0.9127838015556335,
|
|
"step": 450
|
|
},
|
|
{
|
|
"epoch": 0.24077466631771788,
|
|
"grad_norm": 11.338366508483887,
|
|
"learning_rate": 9.416452768711366e-07,
|
|
"logits/chosen": -0.6437957882881165,
|
|
"logits/rejected": -0.674633800983429,
|
|
"logps/chosen": -333.87188720703125,
|
|
"logps/rejected": -318.10467529296875,
|
|
"logps/weighted_chosen": -2.8603515625,
|
|
"logps/weighted_rejected": -3.472119092941284,
|
|
"loss": 0.6881,
|
|
"rewards/accuracies": 0.6468750238418579,
|
|
"rewards/chosen": -41.862693786621094,
|
|
"rewards/margins": 16.437694549560547,
|
|
"rewards/rejected": -58.305076599121094,
|
|
"rewards/weighted_accuracies": 0.653124988079071,
|
|
"rewards/weighted_chosen": -0.43403321504592896,
|
|
"rewards/weighted_margins": 0.4345336854457855,
|
|
"rewards/weighted_rejected": -0.8688415288925171,
|
|
"step": 460
|
|
},
|
|
{
|
|
"epoch": 0.24600889819419,
|
|
"grad_norm": 35.14228057861328,
|
|
"learning_rate": 9.372877011572557e-07,
|
|
"logits/chosen": -0.5738860964775085,
|
|
"logits/rejected": -0.5953949093818665,
|
|
"logps/chosen": -350.23046875,
|
|
"logps/rejected": -329.484375,
|
|
"logps/weighted_chosen": -2.7630372047424316,
|
|
"logps/weighted_rejected": -3.2092041969299316,
|
|
"loss": 0.6424,
|
|
"rewards/accuracies": 0.6187499761581421,
|
|
"rewards/chosen": -39.640037536621094,
|
|
"rewards/margins": 15.155468940734863,
|
|
"rewards/rejected": -54.796485900878906,
|
|
"rewards/weighted_accuracies": 0.671875,
|
|
"rewards/weighted_chosen": -0.338470458984375,
|
|
"rewards/weighted_margins": 0.4430099427700043,
|
|
"rewards/weighted_rejected": -0.781768798828125,
|
|
"step": 470
|
|
},
|
|
{
|
|
"epoch": 0.2512431300706621,
|
|
"grad_norm": 55.78861618041992,
|
|
"learning_rate": 9.327840749034141e-07,
|
|
"logits/chosen": -0.6301849484443665,
|
|
"logits/rejected": -0.6710245013237,
|
|
"logps/chosen": -324.8578186035156,
|
|
"logps/rejected": -336.734375,
|
|
"logps/weighted_chosen": -2.5625243186950684,
|
|
"logps/weighted_rejected": -3.774975538253784,
|
|
"loss": 0.636,
|
|
"rewards/accuracies": 0.668749988079071,
|
|
"rewards/chosen": -37.658203125,
|
|
"rewards/margins": 22.687694549560547,
|
|
"rewards/rejected": -60.357421875,
|
|
"rewards/weighted_accuracies": 0.6812499761581421,
|
|
"rewards/weighted_chosen": -0.2594970762729645,
|
|
"rewards/weighted_margins": 0.48378294706344604,
|
|
"rewards/weighted_rejected": -0.7435302734375,
|
|
"step": 480
|
|
},
|
|
{
|
|
"epoch": 0.2564773619471343,
|
|
"grad_norm": 38.52030563354492,
|
|
"learning_rate": 9.281359022841965e-07,
|
|
"logits/chosen": -0.5341835021972656,
|
|
"logits/rejected": -0.550585925579071,
|
|
"logps/chosen": -320.76171875,
|
|
"logps/rejected": -314.70391845703125,
|
|
"logps/weighted_chosen": -2.91943359375,
|
|
"logps/weighted_rejected": -4.169653415679932,
|
|
"loss": 0.6375,
|
|
"rewards/accuracies": 0.653124988079071,
|
|
"rewards/chosen": -45.831443786621094,
|
|
"rewards/margins": 24.026952743530273,
|
|
"rewards/rejected": -69.88359069824219,
|
|
"rewards/weighted_accuracies": 0.668749988079071,
|
|
"rewards/weighted_chosen": -0.516644299030304,
|
|
"rewards/weighted_margins": 0.518634021282196,
|
|
"rewards/weighted_rejected": -1.0354125499725342,
|
|
"step": 490
|
|
},
|
|
{
|
|
"epoch": 0.26171159382360637,
|
|
"grad_norm": 15.318764686584473,
|
|
"learning_rate": 9.233447357514989e-07,
|
|
"logits/chosen": -0.5592590570449829,
|
|
"logits/rejected": -0.6022475957870483,
|
|
"logps/chosen": -347.77423095703125,
|
|
"logps/rejected": -342.21875,
|
|
"logps/weighted_chosen": -3.2605223655700684,
|
|
"logps/weighted_rejected": -3.667773485183716,
|
|
"loss": 0.6494,
|
|
"rewards/accuracies": 0.659375011920929,
|
|
"rewards/chosen": -49.72968673706055,
|
|
"rewards/margins": 22.665430068969727,
|
|
"rewards/rejected": -72.4097671508789,
|
|
"rewards/weighted_accuracies": 0.65625,
|
|
"rewards/weighted_chosen": -0.4633804261684418,
|
|
"rewards/weighted_margins": 0.527294933795929,
|
|
"rewards/weighted_rejected": -0.990936279296875,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.26171159382360637,
|
|
"eval_logits/chosen": -0.6649160385131836,
|
|
"eval_logits/rejected": -0.682785153388977,
|
|
"eval_logps/chosen": -345.551513671875,
|
|
"eval_logps/rejected": -345.9989929199219,
|
|
"eval_logps/weighted_chosen": -2.992426872253418,
|
|
"eval_logps/weighted_rejected": -3.637266159057617,
|
|
"eval_loss": 0.6663314700126648,
|
|
"eval_rewards/accuracies": 0.6305000185966492,
|
|
"eval_rewards/chosen": -56.88743591308594,
|
|
"eval_rewards/margins": 21.448062896728516,
|
|
"eval_rewards/rejected": -78.32637786865234,
|
|
"eval_rewards/weighted_accuracies": 0.6480000019073486,
|
|
"eval_rewards/weighted_chosen": -0.5315031409263611,
|
|
"eval_rewards/weighted_margins": 0.43671876192092896,
|
|
"eval_rewards/weighted_rejected": -0.9682218432426453,
|
|
"eval_runtime": 1455.1048,
|
|
"eval_samples_per_second": 1.374,
|
|
"eval_steps_per_second": 0.344,
|
|
"step": 500
|
|
}
|
|
],
|
|
"logging_steps": 10,
|
|
"max_steps": 1911,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 1,
|
|
"save_steps": 500,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": false
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 0.0,
|
|
"train_batch_size": 1,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|