gitpullpull's picture
Upload folder using huggingface_hub
d993926 verified
Raw
History Blame Contribute Delete
327 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.3783224024853298,
"eval_steps": 500,
"global_step": 500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.002761477390403866,
"grad_norm": 0.6915842294692993,
"learning_rate": 0.0,
"log_odds_chosen": 0.39300817251205444,
"log_odds_ratio": -0.5194607377052307,
"logits/chosen": -0.6593188047409058,
"logits/rejected": -0.11274649202823639,
"logps/chosen": -1.9585602283477783,
"logps/rejected": -2.3031246662139893,
"loss": 2.2362,
"nll_loss": 2.1842355728149414,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1958560198545456,
"rewards/margins": 0.034456461668014526,
"rewards/rejected": -0.23031246662139893,
"step": 1
},
{
"epoch": 0.005522954780807732,
"grad_norm": 0.8511083722114563,
"learning_rate": 4.587155963302753e-08,
"log_odds_chosen": 0.4412614703178406,
"log_odds_ratio": -0.49888327717781067,
"logits/chosen": -0.6945286393165588,
"logits/rejected": -0.08449751883745193,
"logps/chosen": -1.9912033081054688,
"logps/rejected": -2.382397413253784,
"loss": 2.275,
"nll_loss": 2.2251474857330322,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19912034273147583,
"rewards/margins": 0.0391194187104702,
"rewards/rejected": -0.23823975026607513,
"step": 2
},
{
"epoch": 0.008284432171211598,
"grad_norm": 0.6988061666488647,
"learning_rate": 9.174311926605506e-08,
"log_odds_chosen": 0.49348145723342896,
"log_odds_ratio": -0.4824022948741913,
"logits/chosen": -0.7197601795196533,
"logits/rejected": -0.13051480054855347,
"logps/chosen": -1.9484779834747314,
"logps/rejected": -2.385488271713257,
"loss": 2.1977,
"nll_loss": 2.149470567703247,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19484779238700867,
"rewards/margins": 0.04370103031396866,
"rewards/rejected": -0.23854880034923553,
"step": 3
},
{
"epoch": 0.011045909561615464,
"grad_norm": 0.6778914928436279,
"learning_rate": 1.376146788990826e-07,
"log_odds_chosen": 0.4488891661167145,
"log_odds_ratio": -0.5013620853424072,
"logits/chosen": -0.6518482565879822,
"logits/rejected": -0.08967436850070953,
"logps/chosen": -1.9699251651763916,
"logps/rejected": -2.3685381412506104,
"loss": 2.204,
"nll_loss": 2.1538803577423096,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19699251651763916,
"rewards/margins": 0.0398612916469574,
"rewards/rejected": -0.23685382306575775,
"step": 4
},
{
"epoch": 0.013807386952019331,
"grad_norm": 0.7014802694320679,
"learning_rate": 1.8348623853211012e-07,
"log_odds_chosen": 0.4621143639087677,
"log_odds_ratio": -0.5008962750434875,
"logits/chosen": -0.6857364177703857,
"logits/rejected": -0.15241342782974243,
"logps/chosen": -1.9302603006362915,
"logps/rejected": -2.335421562194824,
"loss": 2.1786,
"nll_loss": 2.128532648086548,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19302603602409363,
"rewards/margins": 0.04051613062620163,
"rewards/rejected": -0.23354215919971466,
"step": 5
},
{
"epoch": 0.016568864342423197,
"grad_norm": 0.6745293140411377,
"learning_rate": 2.2935779816513764e-07,
"log_odds_chosen": 0.37698638439178467,
"log_odds_ratio": -0.5254199504852295,
"logits/chosen": -0.6411757469177246,
"logits/rejected": -0.2817387282848358,
"logps/chosen": -1.9658446311950684,
"logps/rejected": -2.2975409030914307,
"loss": 2.2215,
"nll_loss": 2.168914794921875,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19658446311950684,
"rewards/margins": 0.03316962346434593,
"rewards/rejected": -0.22975412011146545,
"step": 6
},
{
"epoch": 0.019330341732827064,
"grad_norm": 0.6645856499671936,
"learning_rate": 2.752293577981652e-07,
"log_odds_chosen": 0.22311115264892578,
"log_odds_ratio": -0.5942606329917908,
"logits/chosen": -0.5768107175827026,
"logits/rejected": -0.23452866077423096,
"logps/chosen": -2.0238585472106934,
"logps/rejected": -2.223414421081543,
"loss": 2.2796,
"nll_loss": 2.2201786041259766,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.20238585770130157,
"rewards/margins": 0.019955584779381752,
"rewards/rejected": -0.22234144806861877,
"step": 7
},
{
"epoch": 0.022091819123230928,
"grad_norm": 0.7377482652664185,
"learning_rate": 3.211009174311927e-07,
"log_odds_chosen": 0.39164018630981445,
"log_odds_ratio": -0.5184764266014099,
"logits/chosen": -0.566084623336792,
"logits/rejected": -0.12580394744873047,
"logps/chosen": -2.0374059677124023,
"logps/rejected": -2.3862533569335938,
"loss": 2.3026,
"nll_loss": 2.2507545948028564,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.20374059677124023,
"rewards/margins": 0.0348847433924675,
"rewards/rejected": -0.23862534761428833,
"step": 8
},
{
"epoch": 0.024853296513634795,
"grad_norm": 0.7108282446861267,
"learning_rate": 3.6697247706422023e-07,
"log_odds_chosen": 0.6321737170219421,
"log_odds_ratio": -0.4328705072402954,
"logits/chosen": -0.7941892743110657,
"logits/rejected": -0.22760047018527985,
"logps/chosen": -1.9183871746063232,
"logps/rejected": -2.4786453247070312,
"loss": 2.1856,
"nll_loss": 2.142301082611084,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19183871150016785,
"rewards/margins": 0.05602581799030304,
"rewards/rejected": -0.24786454439163208,
"step": 9
},
{
"epoch": 0.027614773904038662,
"grad_norm": 0.8324993252754211,
"learning_rate": 4.128440366972478e-07,
"log_odds_chosen": 0.39953452348709106,
"log_odds_ratio": -0.5179459452629089,
"logits/chosen": -0.7621469497680664,
"logits/rejected": -0.25276613235473633,
"logps/chosen": -2.0076873302459717,
"logps/rejected": -2.3631021976470947,
"loss": 2.2827,
"nll_loss": 2.2309515476226807,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.20076872408390045,
"rewards/margins": 0.03554149717092514,
"rewards/rejected": -0.23631024360656738,
"step": 10
},
{
"epoch": 0.030376251294442526,
"grad_norm": 0.7557547092437744,
"learning_rate": 4.587155963302753e-07,
"log_odds_chosen": 0.5315287113189697,
"log_odds_ratio": -0.4713534414768219,
"logits/chosen": -0.6893935203552246,
"logits/rejected": -0.10655307024717331,
"logps/chosen": -1.9816298484802246,
"logps/rejected": -2.457181930541992,
"loss": 2.257,
"nll_loss": 2.209847927093506,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1981630027294159,
"rewards/margins": 0.04755519703030586,
"rewards/rejected": -0.24571821093559265,
"step": 11
},
{
"epoch": 0.03313772868484639,
"grad_norm": 0.7169145941734314,
"learning_rate": 5.045871559633028e-07,
"log_odds_chosen": 0.38786107301712036,
"log_odds_ratio": -0.5263106822967529,
"logits/chosen": -0.6816898584365845,
"logits/rejected": -0.0623253732919693,
"logps/chosen": -1.9207288026809692,
"logps/rejected": -2.262866973876953,
"loss": 2.1985,
"nll_loss": 2.145916223526001,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19207286834716797,
"rewards/margins": 0.0342138335108757,
"rewards/rejected": -0.22628670930862427,
"step": 12
},
{
"epoch": 0.03589920607525026,
"grad_norm": 0.6504296660423279,
"learning_rate": 5.504587155963304e-07,
"log_odds_chosen": 0.47358548641204834,
"log_odds_ratio": -0.48640355467796326,
"logits/chosen": -0.6333162784576416,
"logits/rejected": -0.1628946214914322,
"logps/chosen": -1.867828369140625,
"logps/rejected": -2.280285596847534,
"loss": 2.1385,
"nll_loss": 2.089879035949707,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1867828369140625,
"rewards/margins": 0.0412457212805748,
"rewards/rejected": -0.2280285507440567,
"step": 13
},
{
"epoch": 0.03866068346565413,
"grad_norm": 0.83363276720047,
"learning_rate": 5.963302752293579e-07,
"log_odds_chosen": 0.4213864207267761,
"log_odds_ratio": -0.521136999130249,
"logits/chosen": -0.7305208444595337,
"logits/rejected": -0.08356200903654099,
"logps/chosen": -2.022002935409546,
"logps/rejected": -2.3972954750061035,
"loss": 2.2795,
"nll_loss": 2.2273383140563965,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.2022002935409546,
"rewards/margins": 0.03752923756837845,
"rewards/rejected": -0.23972955346107483,
"step": 14
},
{
"epoch": 0.04142216085605799,
"grad_norm": 0.8258711099624634,
"learning_rate": 6.422018348623854e-07,
"log_odds_chosen": 0.5449070930480957,
"log_odds_ratio": -0.4581969678401947,
"logits/chosen": -0.8105592131614685,
"logits/rejected": 0.08067083358764648,
"logps/chosen": -1.9474968910217285,
"logps/rejected": -2.428217649459839,
"loss": 2.2093,
"nll_loss": 2.163527488708496,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19474971294403076,
"rewards/margins": 0.04807208105921745,
"rewards/rejected": -0.24282175302505493,
"step": 15
},
{
"epoch": 0.044183638246461855,
"grad_norm": 0.7511261701583862,
"learning_rate": 6.880733944954129e-07,
"log_odds_chosen": 0.4355267882347107,
"log_odds_ratio": -0.5088008046150208,
"logits/chosen": -0.7654774188995361,
"logits/rejected": 0.09955525398254395,
"logps/chosen": -1.873271107673645,
"logps/rejected": -2.249717950820923,
"loss": 2.1545,
"nll_loss": 2.1036317348480225,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18732713162899017,
"rewards/margins": 0.03764466941356659,
"rewards/rejected": -0.22497178614139557,
"step": 16
},
{
"epoch": 0.04694511563686572,
"grad_norm": 0.7405815124511719,
"learning_rate": 7.339449541284405e-07,
"log_odds_chosen": 0.49768519401550293,
"log_odds_ratio": -0.4804110527038574,
"logits/chosen": -0.6771326661109924,
"logits/rejected": 0.09095388650894165,
"logps/chosen": -1.9401291608810425,
"logps/rejected": -2.3801112174987793,
"loss": 2.1836,
"nll_loss": 2.1355254650115967,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19401292502880096,
"rewards/margins": 0.04399820417165756,
"rewards/rejected": -0.23801112174987793,
"step": 17
},
{
"epoch": 0.04970659302726959,
"grad_norm": 0.7064201831817627,
"learning_rate": 7.79816513761468e-07,
"log_odds_chosen": 0.6101190447807312,
"log_odds_ratio": -0.43982943892478943,
"logits/chosen": -0.6823788285255432,
"logits/rejected": -0.1550363153219223,
"logps/chosen": -1.9634783267974854,
"logps/rejected": -2.509146213531494,
"loss": 2.2062,
"nll_loss": 2.162249803543091,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19634784758090973,
"rewards/margins": 0.05456679314374924,
"rewards/rejected": -0.25091463327407837,
"step": 18
},
{
"epoch": 0.05246807041767346,
"grad_norm": 0.709898829460144,
"learning_rate": 8.256880733944956e-07,
"log_odds_chosen": 0.22375424206256866,
"log_odds_ratio": -0.5918958783149719,
"logits/chosen": -0.6519614458084106,
"logits/rejected": -0.030727151781320572,
"logps/chosen": -1.9863125085830688,
"logps/rejected": -2.1811602115631104,
"loss": 2.2528,
"nll_loss": 2.193619966506958,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19863125681877136,
"rewards/margins": 0.01948476769030094,
"rewards/rejected": -0.21811603009700775,
"step": 19
},
{
"epoch": 0.055229547808077324,
"grad_norm": 0.7068823575973511,
"learning_rate": 8.71559633027523e-07,
"log_odds_chosen": 0.4315674602985382,
"log_odds_ratio": -0.5065562129020691,
"logits/chosen": -0.6818419694900513,
"logits/rejected": -0.010788477957248688,
"logps/chosen": -1.9227614402770996,
"logps/rejected": -2.3006279468536377,
"loss": 2.205,
"nll_loss": 2.1543147563934326,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19227616488933563,
"rewards/margins": 0.03778664767742157,
"rewards/rejected": -0.23006278276443481,
"step": 20
},
{
"epoch": 0.057991025198481184,
"grad_norm": 0.7781582474708557,
"learning_rate": 9.174311926605506e-07,
"log_odds_chosen": 0.2822119891643524,
"log_odds_ratio": -0.5701829195022583,
"logits/chosen": -0.5542811155319214,
"logits/rejected": 0.0044986791908741,
"logps/chosen": -2.092810869216919,
"logps/rejected": -2.3459970951080322,
"loss": 2.339,
"nll_loss": 2.2819888591766357,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.2092810869216919,
"rewards/margins": 0.025318622589111328,
"rewards/rejected": -0.23459972441196442,
"step": 21
},
{
"epoch": 0.06075250258888505,
"grad_norm": 0.7080368399620056,
"learning_rate": 9.633027522935782e-07,
"log_odds_chosen": 0.46356773376464844,
"log_odds_ratio": -0.504152774810791,
"logits/chosen": -0.6566605567932129,
"logits/rejected": -0.23016926646232605,
"logps/chosen": -1.9449937343597412,
"logps/rejected": -2.354017972946167,
"loss": 2.1914,
"nll_loss": 2.1410012245178223,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19449937343597412,
"rewards/margins": 0.040902428328990936,
"rewards/rejected": -0.23540180921554565,
"step": 22
},
{
"epoch": 0.06351397997928893,
"grad_norm": 0.7032366394996643,
"learning_rate": 1.0091743119266057e-06,
"log_odds_chosen": 0.41283032298088074,
"log_odds_ratio": -0.5154061913490295,
"logits/chosen": -0.6851422786712646,
"logits/rejected": -0.43942469358444214,
"logps/chosen": -1.9216886758804321,
"logps/rejected": -2.2877392768859863,
"loss": 2.2005,
"nll_loss": 2.1489334106445312,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19216886162757874,
"rewards/margins": 0.03660505264997482,
"rewards/rejected": -0.22877395153045654,
"step": 23
},
{
"epoch": 0.06627545736969279,
"grad_norm": 0.759819746017456,
"learning_rate": 1.055045871559633e-06,
"log_odds_chosen": 0.5156121253967285,
"log_odds_ratio": -0.46936866641044617,
"logits/chosen": -0.6876986622810364,
"logits/rejected": -0.031957462430000305,
"logps/chosen": -1.9351065158843994,
"logps/rejected": -2.390742063522339,
"loss": 2.1928,
"nll_loss": 2.1458399295806885,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19351065158843994,
"rewards/margins": 0.04556357115507126,
"rewards/rejected": -0.2390742301940918,
"step": 24
},
{
"epoch": 0.06903693476009665,
"grad_norm": 0.8110620379447937,
"learning_rate": 1.1009174311926608e-06,
"log_odds_chosen": 0.6373166441917419,
"log_odds_ratio": -0.4303508996963501,
"logits/chosen": -0.6892279982566833,
"logits/rejected": -0.09929540753364563,
"logps/chosen": -1.9359254837036133,
"logps/rejected": -2.5030038356781006,
"loss": 2.2035,
"nll_loss": 2.1604182720184326,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19359253346920013,
"rewards/margins": 0.05670783296227455,
"rewards/rejected": -0.2503003776073456,
"step": 25
},
{
"epoch": 0.07179841215050052,
"grad_norm": 0.7390461564064026,
"learning_rate": 1.1467889908256882e-06,
"log_odds_chosen": 0.36132410168647766,
"log_odds_ratio": -0.5341425538063049,
"logits/chosen": -0.6445504426956177,
"logits/rejected": -0.1332886964082718,
"logps/chosen": -1.9676666259765625,
"logps/rejected": -2.2869884967803955,
"loss": 2.2373,
"nll_loss": 2.1838440895080566,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1967666745185852,
"rewards/margins": 0.03193218633532524,
"rewards/rejected": -0.22869886457920074,
"step": 26
},
{
"epoch": 0.07455988954090438,
"grad_norm": 0.6214093565940857,
"learning_rate": 1.1926605504587159e-06,
"log_odds_chosen": 0.4441767632961273,
"log_odds_ratio": -0.5069965124130249,
"logits/chosen": -0.5817967653274536,
"logits/rejected": -0.287628710269928,
"logps/chosen": -1.8485782146453857,
"logps/rejected": -2.2337605953216553,
"loss": 2.1166,
"nll_loss": 2.065877914428711,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1848578304052353,
"rewards/margins": 0.03851822763681412,
"rewards/rejected": -0.22337606549263,
"step": 27
},
{
"epoch": 0.07732136693130826,
"grad_norm": 0.6619555950164795,
"learning_rate": 1.2385321100917433e-06,
"log_odds_chosen": 0.5049865245819092,
"log_odds_ratio": -0.47638368606567383,
"logits/chosen": -0.6138817071914673,
"logits/rejected": -0.2531018853187561,
"logps/chosen": -1.9172720909118652,
"logps/rejected": -2.361618995666504,
"loss": 2.1756,
"nll_loss": 2.127944231033325,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19172722101211548,
"rewards/margins": 0.04443468153476715,
"rewards/rejected": -0.23616188764572144,
"step": 28
},
{
"epoch": 0.08008284432171212,
"grad_norm": 0.6942340135574341,
"learning_rate": 1.2844036697247707e-06,
"log_odds_chosen": 0.4293629825115204,
"log_odds_ratio": -0.5098441243171692,
"logits/chosen": -0.6483147740364075,
"logits/rejected": -0.05790426582098007,
"logps/chosen": -2.0040318965911865,
"logps/rejected": -2.3848392963409424,
"loss": 2.2622,
"nll_loss": 2.2111713886260986,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.20040319859981537,
"rewards/margins": 0.03808073699474335,
"rewards/rejected": -0.23848393559455872,
"step": 29
},
{
"epoch": 0.08284432171211598,
"grad_norm": 0.7068150043487549,
"learning_rate": 1.3302752293577984e-06,
"log_odds_chosen": 0.6293392181396484,
"log_odds_ratio": -0.43408164381980896,
"logits/chosen": -0.5528810024261475,
"logits/rejected": -0.038134124130010605,
"logps/chosen": -1.9628605842590332,
"logps/rejected": -2.524247407913208,
"loss": 2.2025,
"nll_loss": 2.1590917110443115,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19628606736660004,
"rewards/margins": 0.05613870173692703,
"rewards/rejected": -0.25242477655410767,
"step": 30
},
{
"epoch": 0.08560579910251985,
"grad_norm": 0.7285196185112,
"learning_rate": 1.3761467889908258e-06,
"log_odds_chosen": 0.5635695457458496,
"log_odds_ratio": -0.4573448598384857,
"logits/chosen": -0.6580309271812439,
"logits/rejected": -0.1388925164937973,
"logps/chosen": -1.97215735912323,
"logps/rejected": -2.47501540184021,
"loss": 2.2267,
"nll_loss": 2.1810009479522705,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1972157210111618,
"rewards/margins": 0.050285838544368744,
"rewards/rejected": -0.24750158190727234,
"step": 31
},
{
"epoch": 0.08836727649292371,
"grad_norm": 0.663361132144928,
"learning_rate": 1.4220183486238535e-06,
"log_odds_chosen": 0.4357861876487732,
"log_odds_ratio": -0.5044962763786316,
"logits/chosen": -0.5250373482704163,
"logits/rejected": -0.026727374643087387,
"logps/chosen": -1.9720088243484497,
"logps/rejected": -2.3577969074249268,
"loss": 2.2324,
"nll_loss": 2.1819026470184326,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19720089435577393,
"rewards/margins": 0.0385788157582283,
"rewards/rejected": -0.23577968776226044,
"step": 32
},
{
"epoch": 0.09112875388332758,
"grad_norm": 0.6460317373275757,
"learning_rate": 1.467889908256881e-06,
"log_odds_chosen": 0.5821229219436646,
"log_odds_ratio": -0.45059195160865784,
"logits/chosen": -0.5484606027603149,
"logits/rejected": -0.17182235419750214,
"logps/chosen": -1.8812311887741089,
"logps/rejected": -2.394681453704834,
"loss": 2.1347,
"nll_loss": 2.0896193981170654,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18812312185764313,
"rewards/margins": 0.05134502053260803,
"rewards/rejected": -0.23946812748908997,
"step": 33
},
{
"epoch": 0.09389023127373144,
"grad_norm": 0.6461722254753113,
"learning_rate": 1.5137614678899084e-06,
"log_odds_chosen": 0.5589693784713745,
"log_odds_ratio": -0.466278612613678,
"logits/chosen": -0.5394806861877441,
"logits/rejected": -0.11535287648439407,
"logps/chosen": -1.8948417901992798,
"logps/rejected": -2.3906807899475098,
"loss": 2.1429,
"nll_loss": 2.0962626934051514,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18948417901992798,
"rewards/margins": 0.04958389326930046,
"rewards/rejected": -0.23906809091567993,
"step": 34
},
{
"epoch": 0.09665170866413532,
"grad_norm": 0.6749213933944702,
"learning_rate": 1.559633027522936e-06,
"log_odds_chosen": 0.4189353883266449,
"log_odds_ratio": -0.5096418261528015,
"logits/chosen": -0.8145462274551392,
"logits/rejected": -0.184108167886734,
"logps/chosen": -1.9128409624099731,
"logps/rejected": -2.278416395187378,
"loss": 2.1804,
"nll_loss": 2.12943172454834,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19128410518169403,
"rewards/margins": 0.036557577550411224,
"rewards/rejected": -0.22784166038036346,
"step": 35
},
{
"epoch": 0.09941318605453918,
"grad_norm": 0.7020057439804077,
"learning_rate": 1.6055045871559635e-06,
"log_odds_chosen": 0.5992317199707031,
"log_odds_ratio": -0.4454975128173828,
"logits/chosen": -0.755850613117218,
"logits/rejected": -0.0925610214471817,
"logps/chosen": -1.921108603477478,
"logps/rejected": -2.4531819820404053,
"loss": 2.1777,
"nll_loss": 2.1331448554992676,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19211086630821228,
"rewards/margins": 0.05320734530687332,
"rewards/rejected": -0.245318204164505,
"step": 36
},
{
"epoch": 0.10217466344494304,
"grad_norm": 0.7380937337875366,
"learning_rate": 1.6513761467889911e-06,
"log_odds_chosen": 0.5436604619026184,
"log_odds_ratio": -0.4642760455608368,
"logits/chosen": -0.5516951680183411,
"logits/rejected": -0.17646433413028717,
"logps/chosen": -2.041116952896118,
"logps/rejected": -2.531487226486206,
"loss": 2.3149,
"nll_loss": 2.268436908721924,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.20411169528961182,
"rewards/margins": 0.04903702437877655,
"rewards/rejected": -0.25314873456954956,
"step": 37
},
{
"epoch": 0.10493614083534691,
"grad_norm": 0.697014331817627,
"learning_rate": 1.6972477064220186e-06,
"log_odds_chosen": 0.3754476308822632,
"log_odds_ratio": -0.5259881615638733,
"logits/chosen": -0.6446620225906372,
"logits/rejected": -0.2302635908126831,
"logps/chosen": -1.916767954826355,
"logps/rejected": -2.246654748916626,
"loss": 2.1948,
"nll_loss": 2.142183303833008,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1916767954826355,
"rewards/margins": 0.032988667488098145,
"rewards/rejected": -0.22466546297073364,
"step": 38
},
{
"epoch": 0.10769761822575077,
"grad_norm": 0.6962437033653259,
"learning_rate": 1.743119266055046e-06,
"log_odds_chosen": 0.4513578414916992,
"log_odds_ratio": -0.5085355043411255,
"logits/chosen": -0.5062450170516968,
"logits/rejected": -0.0742938369512558,
"logps/chosen": -2.0077242851257324,
"logps/rejected": -2.4134199619293213,
"loss": 2.2615,
"nll_loss": 2.2106730937957764,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.20077240467071533,
"rewards/margins": 0.04056959226727486,
"rewards/rejected": -0.24134202301502228,
"step": 39
},
{
"epoch": 0.11045909561615465,
"grad_norm": 0.8404136896133423,
"learning_rate": 1.7889908256880737e-06,
"log_odds_chosen": 0.26594164967536926,
"log_odds_ratio": -0.5785322189331055,
"logits/chosen": -0.7206758260726929,
"logits/rejected": -0.2084624469280243,
"logps/chosen": -2.028856039047241,
"logps/rejected": -2.2659192085266113,
"loss": 2.3049,
"nll_loss": 2.2470221519470215,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.20288559794425964,
"rewards/margins": 0.023706313222646713,
"rewards/rejected": -0.22659191489219666,
"step": 40
},
{
"epoch": 0.11322057300655851,
"grad_norm": 0.6305637359619141,
"learning_rate": 1.8348623853211011e-06,
"log_odds_chosen": 0.4789700508117676,
"log_odds_ratio": -0.4894922971725464,
"logits/chosen": -0.6578766107559204,
"logits/rejected": -0.20148660242557526,
"logps/chosen": -1.876691460609436,
"logps/rejected": -2.2974884510040283,
"loss": 2.1332,
"nll_loss": 2.0842790603637695,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18766914308071136,
"rewards/margins": 0.042079709470272064,
"rewards/rejected": -0.22974886000156403,
"step": 41
},
{
"epoch": 0.11598205039696237,
"grad_norm": 0.764166533946991,
"learning_rate": 1.8807339449541288e-06,
"log_odds_chosen": 0.32667985558509827,
"log_odds_ratio": -0.5534327030181885,
"logits/chosen": -0.6471579074859619,
"logits/rejected": -0.24204234778881073,
"logps/chosen": -2.054783582687378,
"logps/rejected": -2.3455071449279785,
"loss": 2.324,
"nll_loss": 2.268658399581909,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.20547834038734436,
"rewards/margins": 0.0290723517537117,
"rewards/rejected": -0.23455071449279785,
"step": 42
},
{
"epoch": 0.11874352778736624,
"grad_norm": 0.6828538775444031,
"learning_rate": 1.9266055045871564e-06,
"log_odds_chosen": 0.39937809109687805,
"log_odds_ratio": -0.5205667018890381,
"logits/chosen": -0.6188192367553711,
"logits/rejected": -0.2733380198478699,
"logps/chosen": -1.9134750366210938,
"logps/rejected": -2.2601161003112793,
"loss": 2.1861,
"nll_loss": 2.134024143218994,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19134750962257385,
"rewards/margins": 0.0346641018986702,
"rewards/rejected": -0.22601160407066345,
"step": 43
},
{
"epoch": 0.1215050051777701,
"grad_norm": 0.8159171938896179,
"learning_rate": 1.9724770642201837e-06,
"log_odds_chosen": 0.32455721497535706,
"log_odds_ratio": -0.5553261041641235,
"logits/chosen": -0.669414758682251,
"logits/rejected": -0.2208831012248993,
"logps/chosen": -2.0293149948120117,
"logps/rejected": -2.317753314971924,
"loss": 2.2895,
"nll_loss": 2.233963966369629,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.20293152332305908,
"rewards/margins": 0.028843821957707405,
"rewards/rejected": -0.23177534341812134,
"step": 44
},
{
"epoch": 0.12426648256817398,
"grad_norm": 0.6607564091682434,
"learning_rate": 2.0183486238532113e-06,
"log_odds_chosen": 0.615048885345459,
"log_odds_ratio": -0.4402877390384674,
"logits/chosen": -0.6855505704879761,
"logits/rejected": -0.38655874133110046,
"logps/chosen": -1.9126825332641602,
"logps/rejected": -2.459097385406494,
"loss": 2.1594,
"nll_loss": 2.115415096282959,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19126826524734497,
"rewards/margins": 0.054641470313072205,
"rewards/rejected": -0.24590973556041718,
"step": 45
},
{
"epoch": 0.12702795995857785,
"grad_norm": 0.6284722685813904,
"learning_rate": 2.064220183486239e-06,
"log_odds_chosen": 0.5434573888778687,
"log_odds_ratio": -0.467115193605423,
"logits/chosen": -0.5798700451850891,
"logits/rejected": -0.13538922369480133,
"logps/chosen": -1.7848091125488281,
"logps/rejected": -2.2502312660217285,
"loss": 2.0394,
"nll_loss": 1.99272620677948,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17848090827465057,
"rewards/margins": 0.046542223542928696,
"rewards/rejected": -0.22502315044403076,
"step": 46
},
{
"epoch": 0.1297894373489817,
"grad_norm": 0.6467039585113525,
"learning_rate": 2.110091743119266e-06,
"log_odds_chosen": 0.3773024082183838,
"log_odds_ratio": -0.5347570180892944,
"logits/chosen": -0.559330403804779,
"logits/rejected": -0.16529709100723267,
"logps/chosen": -2.0019242763519287,
"logps/rejected": -2.339775323867798,
"loss": 2.2632,
"nll_loss": 2.209690809249878,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.2001924067735672,
"rewards/margins": 0.0337851420044899,
"rewards/rejected": -0.2339775562286377,
"step": 47
},
{
"epoch": 0.13255091473938557,
"grad_norm": 0.803294837474823,
"learning_rate": 2.155963302752294e-06,
"log_odds_chosen": 0.16557860374450684,
"log_odds_ratio": -0.6379687190055847,
"logits/chosen": -0.6091009378433228,
"logits/rejected": -0.18343707919120789,
"logps/chosen": -2.083127975463867,
"logps/rejected": -2.241791009902954,
"loss": 2.3479,
"nll_loss": 2.2841379642486572,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.20831279456615448,
"rewards/margins": 0.015866274014115334,
"rewards/rejected": -0.22417910397052765,
"step": 48
},
{
"epoch": 0.13531239212978943,
"grad_norm": 0.7350826263427734,
"learning_rate": 2.2018348623853215e-06,
"log_odds_chosen": 0.663620114326477,
"log_odds_ratio": -0.430046409368515,
"logits/chosen": -0.6330848336219788,
"logits/rejected": -0.12406854331493378,
"logps/chosen": -1.9958436489105225,
"logps/rejected": -2.595897912979126,
"loss": 2.2445,
"nll_loss": 2.201472282409668,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19958436489105225,
"rewards/margins": 0.06000541150569916,
"rewards/rejected": -0.2595897912979126,
"step": 49
},
{
"epoch": 0.1380738695201933,
"grad_norm": 0.6953707337379456,
"learning_rate": 2.2477064220183487e-06,
"log_odds_chosen": 0.558549702167511,
"log_odds_ratio": -0.4727778434753418,
"logits/chosen": -0.6551685333251953,
"logits/rejected": -0.10959107428789139,
"logps/chosen": -1.9449548721313477,
"logps/rejected": -2.4448442459106445,
"loss": 2.1886,
"nll_loss": 2.1412806510925293,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19449549913406372,
"rewards/margins": 0.04998895525932312,
"rewards/rejected": -0.24448445439338684,
"step": 50
},
{
"epoch": 0.14083534691059718,
"grad_norm": 0.7255063056945801,
"learning_rate": 2.2935779816513764e-06,
"log_odds_chosen": 0.439910352230072,
"log_odds_ratio": -0.5028793811798096,
"logits/chosen": -0.5782126784324646,
"logits/rejected": -0.3915577232837677,
"logps/chosen": -1.9679124355316162,
"logps/rejected": -2.3573005199432373,
"loss": 2.2331,
"nll_loss": 2.1828465461730957,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19679124653339386,
"rewards/margins": 0.03893881291151047,
"rewards/rejected": -0.23573008179664612,
"step": 51
},
{
"epoch": 0.14359682430100104,
"grad_norm": 0.6826524138450623,
"learning_rate": 2.339449541284404e-06,
"log_odds_chosen": 0.4098314940929413,
"log_odds_ratio": -0.5124779939651489,
"logits/chosen": -0.5287759304046631,
"logits/rejected": -0.16041654348373413,
"logps/chosen": -1.9367855787277222,
"logps/rejected": -2.296450614929199,
"loss": 2.1899,
"nll_loss": 2.1386733055114746,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19367855787277222,
"rewards/margins": 0.03596651181578636,
"rewards/rejected": -0.22964505851268768,
"step": 52
},
{
"epoch": 0.1463583016914049,
"grad_norm": 0.7680190801620483,
"learning_rate": 2.3853211009174317e-06,
"log_odds_chosen": 0.3706192970275879,
"log_odds_ratio": -0.5296811461448669,
"logits/chosen": -0.707636833190918,
"logits/rejected": -0.3318819999694824,
"logps/chosen": -1.9044169187545776,
"logps/rejected": -2.228182315826416,
"loss": 2.1714,
"nll_loss": 2.118480920791626,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19044168293476105,
"rewards/margins": 0.03237656503915787,
"rewards/rejected": -0.2228182554244995,
"step": 53
},
{
"epoch": 0.14911977908180876,
"grad_norm": 0.6361218690872192,
"learning_rate": 2.431192660550459e-06,
"log_odds_chosen": 0.37140461802482605,
"log_odds_ratio": -0.5330734848976135,
"logits/chosen": -0.5706905126571655,
"logits/rejected": -0.2818271815776825,
"logps/chosen": -1.8717925548553467,
"logps/rejected": -2.1973752975463867,
"loss": 2.1391,
"nll_loss": 2.0858311653137207,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18717925250530243,
"rewards/margins": 0.03255828469991684,
"rewards/rejected": -0.21973752975463867,
"step": 54
},
{
"epoch": 0.15188125647221262,
"grad_norm": 0.6758925914764404,
"learning_rate": 2.4770642201834866e-06,
"log_odds_chosen": 0.351237416267395,
"log_odds_ratio": -0.5363888740539551,
"logits/chosen": -0.6902980804443359,
"logits/rejected": -0.1499267816543579,
"logps/chosen": -1.9404265880584717,
"logps/rejected": -2.248072624206543,
"loss": 2.2093,
"nll_loss": 2.155627965927124,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19404268264770508,
"rewards/margins": 0.030764613300561905,
"rewards/rejected": -0.2248072773218155,
"step": 55
},
{
"epoch": 0.1546427338626165,
"grad_norm": 0.7609654664993286,
"learning_rate": 2.522935779816514e-06,
"log_odds_chosen": 0.6203876733779907,
"log_odds_ratio": -0.4391648769378662,
"logits/chosen": -0.6777594685554504,
"logits/rejected": -0.12744097411632538,
"logps/chosen": -1.920168161392212,
"logps/rejected": -2.470646381378174,
"loss": 2.2005,
"nll_loss": 2.1565370559692383,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1920168101787567,
"rewards/margins": 0.055047836154699326,
"rewards/rejected": -0.24706465005874634,
"step": 56
},
{
"epoch": 0.15740421125302037,
"grad_norm": 0.6103418469429016,
"learning_rate": 2.5688073394495415e-06,
"log_odds_chosen": 0.3935878276824951,
"log_odds_ratio": -0.52540522813797,
"logits/chosen": -0.5779827833175659,
"logits/rejected": -0.07134772837162018,
"logps/chosen": -1.8075436353683472,
"logps/rejected": -2.1494383811950684,
"loss": 2.0685,
"nll_loss": 2.0159270763397217,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18075433373451233,
"rewards/margins": 0.03418949246406555,
"rewards/rejected": -0.21494384109973907,
"step": 57
},
{
"epoch": 0.16016568864342423,
"grad_norm": 0.6918542981147766,
"learning_rate": 2.6146788990825687e-06,
"log_odds_chosen": 0.508127748966217,
"log_odds_ratio": -0.48639971017837524,
"logits/chosen": -0.5526314377784729,
"logits/rejected": -0.17470352351665497,
"logps/chosen": -1.9131710529327393,
"logps/rejected": -2.359304904937744,
"loss": 2.1784,
"nll_loss": 2.1297237873077393,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1913171112537384,
"rewards/margins": 0.044613372534513474,
"rewards/rejected": -0.23593048751354218,
"step": 58
},
{
"epoch": 0.1629271660338281,
"grad_norm": 0.6938620209693909,
"learning_rate": 2.6605504587155968e-06,
"log_odds_chosen": 0.2925473153591156,
"log_odds_ratio": -0.563460111618042,
"logits/chosen": -0.5527081489562988,
"logits/rejected": -0.2188190072774887,
"logps/chosen": -1.9204390048980713,
"logps/rejected": -2.175715684890747,
"loss": 2.1995,
"nll_loss": 2.1431734561920166,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19204390048980713,
"rewards/margins": 0.025527678430080414,
"rewards/rejected": -0.21757157146930695,
"step": 59
},
{
"epoch": 0.16568864342423195,
"grad_norm": 0.6408482193946838,
"learning_rate": 2.706422018348624e-06,
"log_odds_chosen": 0.3286433815956116,
"log_odds_ratio": -0.5471742749214172,
"logits/chosen": -0.5031524300575256,
"logits/rejected": -0.016629882156848907,
"logps/chosen": -1.9514718055725098,
"logps/rejected": -2.240278720855713,
"loss": 2.2124,
"nll_loss": 2.157729387283325,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19514718651771545,
"rewards/margins": 0.028880706056952477,
"rewards/rejected": -0.22402788698673248,
"step": 60
},
{
"epoch": 0.16845012081463584,
"grad_norm": 0.5977794528007507,
"learning_rate": 2.7522935779816517e-06,
"log_odds_chosen": 0.6263030767440796,
"log_odds_ratio": -0.43304184079170227,
"logits/chosen": -0.5808561444282532,
"logits/rejected": -0.16479042172431946,
"logps/chosen": -1.874776840209961,
"logps/rejected": -2.425903797149658,
"loss": 2.1062,
"nll_loss": 2.0628535747528076,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18747767806053162,
"rewards/margins": 0.055112697184085846,
"rewards/rejected": -0.24259036779403687,
"step": 61
},
{
"epoch": 0.1712115982050397,
"grad_norm": 0.7107025980949402,
"learning_rate": 2.798165137614679e-06,
"log_odds_chosen": 0.38438883423805237,
"log_odds_ratio": -0.5274062752723694,
"logits/chosen": -0.5282328724861145,
"logits/rejected": 0.017122909426689148,
"logps/chosen": -2.0265655517578125,
"logps/rejected": -2.3674638271331787,
"loss": 2.2774,
"nll_loss": 2.224700927734375,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.202656552195549,
"rewards/margins": 0.034089840948581696,
"rewards/rejected": -0.2367464005947113,
"step": 62
},
{
"epoch": 0.17397307559544356,
"grad_norm": 0.6914030313491821,
"learning_rate": 2.844036697247707e-06,
"log_odds_chosen": 0.4569448232650757,
"log_odds_ratio": -0.4947759211063385,
"logits/chosen": -0.564735472202301,
"logits/rejected": -0.17074617743492126,
"logps/chosen": -1.9550468921661377,
"logps/rejected": -2.357658624649048,
"loss": 2.1959,
"nll_loss": 2.14642596244812,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19550471007823944,
"rewards/margins": 0.040261153131723404,
"rewards/rejected": -0.23576584458351135,
"step": 63
},
{
"epoch": 0.17673455298584742,
"grad_norm": 0.6147691011428833,
"learning_rate": 2.8899082568807342e-06,
"log_odds_chosen": 0.5831525325775146,
"log_odds_ratio": -0.4512834846973419,
"logits/chosen": -0.6652324199676514,
"logits/rejected": -0.30751606822013855,
"logps/chosen": -1.925264835357666,
"logps/rejected": -2.4442431926727295,
"loss": 2.154,
"nll_loss": 2.1088905334472656,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1925264596939087,
"rewards/margins": 0.05189783126115799,
"rewards/rejected": -0.24442431330680847,
"step": 64
},
{
"epoch": 0.17949603037625128,
"grad_norm": 0.6069121360778809,
"learning_rate": 2.935779816513762e-06,
"log_odds_chosen": 0.3449505567550659,
"log_odds_ratio": -0.5405150055885315,
"logits/chosen": -0.46805867552757263,
"logits/rejected": -0.031980015337467194,
"logps/chosen": -1.8449336290359497,
"logps/rejected": -2.141822576522827,
"loss": 2.104,
"nll_loss": 2.049985885620117,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18449336290359497,
"rewards/margins": 0.02968890592455864,
"rewards/rejected": -0.2141822725534439,
"step": 65
},
{
"epoch": 0.18225750776665517,
"grad_norm": 0.5975824594497681,
"learning_rate": 2.981651376146789e-06,
"log_odds_chosen": 0.5348482728004456,
"log_odds_ratio": -0.48265504837036133,
"logits/chosen": -0.6339913606643677,
"logits/rejected": -0.19266335666179657,
"logps/chosen": -1.7976055145263672,
"logps/rejected": -2.2617228031158447,
"loss": 2.0549,
"nll_loss": 2.006657600402832,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17976056039333344,
"rewards/margins": 0.04641173034906387,
"rewards/rejected": -0.2261722981929779,
"step": 66
},
{
"epoch": 0.18501898515705903,
"grad_norm": 0.6583871245384216,
"learning_rate": 3.0275229357798168e-06,
"log_odds_chosen": 0.2934231162071228,
"log_odds_ratio": -0.5610607862472534,
"logits/chosen": -0.49614468216896057,
"logits/rejected": -0.02480134554207325,
"logps/chosen": -1.9567402601242065,
"logps/rejected": -2.21272349357605,
"loss": 2.1884,
"nll_loss": 2.1323015689849854,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19567403197288513,
"rewards/margins": 0.025598343461751938,
"rewards/rejected": -0.22127236425876617,
"step": 67
},
{
"epoch": 0.1877804625474629,
"grad_norm": 0.6593952775001526,
"learning_rate": 3.073394495412844e-06,
"log_odds_chosen": 0.39623624086380005,
"log_odds_ratio": -0.5267960429191589,
"logits/chosen": -0.32998380064964294,
"logits/rejected": 0.019692357629537582,
"logps/chosen": -1.9624451398849487,
"logps/rejected": -2.3119962215423584,
"loss": 2.2478,
"nll_loss": 2.195082902908325,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1962445080280304,
"rewards/margins": 0.03495512157678604,
"rewards/rejected": -0.23119963705539703,
"step": 68
},
{
"epoch": 0.19054193993786675,
"grad_norm": 0.7153931856155396,
"learning_rate": 3.119266055045872e-06,
"log_odds_chosen": 0.5069383382797241,
"log_odds_ratio": -0.4813777804374695,
"logits/chosen": -0.505609393119812,
"logits/rejected": -0.05926704406738281,
"logps/chosen": -2.0001449584960938,
"logps/rejected": -2.453965187072754,
"loss": 2.2604,
"nll_loss": 2.212214231491089,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.20001448690891266,
"rewards/margins": 0.04538201540708542,
"rewards/rejected": -0.24539650976657867,
"step": 69
},
{
"epoch": 0.19330341732827064,
"grad_norm": 0.5841740965843201,
"learning_rate": 3.1651376146788993e-06,
"log_odds_chosen": 0.48365646600723267,
"log_odds_ratio": -0.4906890094280243,
"logits/chosen": -0.4503505527973175,
"logits/rejected": -0.24621139466762543,
"logps/chosen": -1.9794856309890747,
"logps/rejected": -2.4092578887939453,
"loss": 2.2046,
"nll_loss": 2.155548095703125,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19794857501983643,
"rewards/margins": 0.0429772287607193,
"rewards/rejected": -0.24092580378055573,
"step": 70
},
{
"epoch": 0.1960648947186745,
"grad_norm": 0.6693902015686035,
"learning_rate": 3.211009174311927e-06,
"log_odds_chosen": 0.4308607876300812,
"log_odds_ratio": -0.5067814588546753,
"logits/chosen": -0.47226929664611816,
"logits/rejected": -0.22168010473251343,
"logps/chosen": -1.9241232872009277,
"logps/rejected": -2.3047096729278564,
"loss": 2.191,
"nll_loss": 2.14032244682312,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.192412331700325,
"rewards/margins": 0.038058653473854065,
"rewards/rejected": -0.23047097027301788,
"step": 71
},
{
"epoch": 0.19882637210907836,
"grad_norm": 0.6415713429450989,
"learning_rate": 3.256880733944954e-06,
"log_odds_chosen": 0.5713039636611938,
"log_odds_ratio": -0.46779656410217285,
"logits/chosen": -0.4027179479598999,
"logits/rejected": -0.20848074555397034,
"logps/chosen": -1.8328369855880737,
"logps/rejected": -2.313088893890381,
"loss": 2.0868,
"nll_loss": 2.0400490760803223,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18328368663787842,
"rewards/margins": 0.048025187104940414,
"rewards/rejected": -0.23130889236927032,
"step": 72
},
{
"epoch": 0.20158784949948222,
"grad_norm": 0.6714993715286255,
"learning_rate": 3.3027522935779823e-06,
"log_odds_chosen": 0.35765087604522705,
"log_odds_ratio": -0.5315056443214417,
"logits/chosen": -0.387838214635849,
"logits/rejected": -0.10332845896482468,
"logps/chosen": -2.011086940765381,
"logps/rejected": -2.328169107437134,
"loss": 2.2627,
"nll_loss": 2.209596872329712,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.20110869407653809,
"rewards/margins": 0.03170822933316231,
"rewards/rejected": -0.2328169345855713,
"step": 73
},
{
"epoch": 0.20434932688988608,
"grad_norm": 0.6035195589065552,
"learning_rate": 3.3486238532110095e-06,
"log_odds_chosen": 0.37045401334762573,
"log_odds_ratio": -0.528258740901947,
"logits/chosen": -0.4163488447666168,
"logits/rejected": -0.20020049810409546,
"logps/chosen": -1.927206039428711,
"logps/rejected": -2.2507896423339844,
"loss": 2.1911,
"nll_loss": 2.1382455825805664,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19272059202194214,
"rewards/margins": 0.03235836327075958,
"rewards/rejected": -0.22507895529270172,
"step": 74
},
{
"epoch": 0.20711080428028997,
"grad_norm": 0.6681767702102661,
"learning_rate": 3.394495412844037e-06,
"log_odds_chosen": 0.4545160233974457,
"log_odds_ratio": -0.4990374743938446,
"logits/chosen": -0.4695533514022827,
"logits/rejected": -0.14828726649284363,
"logps/chosen": -1.9052660465240479,
"logps/rejected": -2.3053171634674072,
"loss": 2.171,
"nll_loss": 2.121072769165039,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19052661955356598,
"rewards/margins": 0.04000508412718773,
"rewards/rejected": -0.2305317223072052,
"step": 75
},
{
"epoch": 0.20987228167069383,
"grad_norm": 0.5848620533943176,
"learning_rate": 3.4403669724770644e-06,
"log_odds_chosen": 0.4162794351577759,
"log_odds_ratio": -0.5107027292251587,
"logits/chosen": -0.3651241362094879,
"logits/rejected": -0.29924386739730835,
"logps/chosen": -1.9842548370361328,
"logps/rejected": -2.354065418243408,
"loss": 2.2131,
"nll_loss": 2.1620750427246094,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19842548668384552,
"rewards/margins": 0.03698106110095978,
"rewards/rejected": -0.2354065328836441,
"step": 76
},
{
"epoch": 0.2126337590610977,
"grad_norm": 0.6766842007637024,
"learning_rate": 3.486238532110092e-06,
"log_odds_chosen": 0.3890829086303711,
"log_odds_ratio": -0.5331251621246338,
"logits/chosen": -0.5041220784187317,
"logits/rejected": -0.256991982460022,
"logps/chosen": -1.9675860404968262,
"logps/rejected": -2.3138177394866943,
"loss": 2.2335,
"nll_loss": 2.1801676750183105,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19675858318805695,
"rewards/margins": 0.03462318331003189,
"rewards/rejected": -0.23138177394866943,
"step": 77
},
{
"epoch": 0.21539523645150155,
"grad_norm": 0.6400946974754333,
"learning_rate": 3.5321100917431193e-06,
"log_odds_chosen": 0.3764447569847107,
"log_odds_ratio": -0.5406405329704285,
"logits/chosen": -0.30762779712677,
"logits/rejected": -0.010896757245063782,
"logps/chosen": -1.9988197088241577,
"logps/rejected": -2.331996202468872,
"loss": 2.246,
"nll_loss": 2.191932201385498,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19988197088241577,
"rewards/margins": 0.03331765532493591,
"rewards/rejected": -0.23319962620735168,
"step": 78
},
{
"epoch": 0.2181567138419054,
"grad_norm": 0.5379537343978882,
"learning_rate": 3.5779816513761473e-06,
"log_odds_chosen": 0.4429885745048523,
"log_odds_ratio": -0.49778997898101807,
"logits/chosen": -0.32647988200187683,
"logits/rejected": -0.19796019792556763,
"logps/chosen": -1.8951160907745361,
"logps/rejected": -2.282485246658325,
"loss": 2.1565,
"nll_loss": 2.1067562103271484,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18951159715652466,
"rewards/margins": 0.038736920803785324,
"rewards/rejected": -0.22824853658676147,
"step": 79
},
{
"epoch": 0.2209181912323093,
"grad_norm": 0.5976657867431641,
"learning_rate": 3.6238532110091746e-06,
"log_odds_chosen": 0.3772650957107544,
"log_odds_ratio": -0.5323787927627563,
"logits/chosen": -0.396470308303833,
"logits/rejected": -0.1672373265028,
"logps/chosen": -1.8558274507522583,
"logps/rejected": -2.1821842193603516,
"loss": 2.119,
"nll_loss": 2.065802812576294,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1855827420949936,
"rewards/margins": 0.032635681331157684,
"rewards/rejected": -0.21821841597557068,
"step": 80
},
{
"epoch": 0.22367966862271316,
"grad_norm": 0.5696468353271484,
"learning_rate": 3.6697247706422022e-06,
"log_odds_chosen": 0.3498988747596741,
"log_odds_ratio": -0.5373408198356628,
"logits/chosen": -0.32143616676330566,
"logits/rejected": -0.08278146386146545,
"logps/chosen": -1.867413878440857,
"logps/rejected": -2.1695525646209717,
"loss": 2.1158,
"nll_loss": 2.062082052230835,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1867414116859436,
"rewards/margins": 0.0302138552069664,
"rewards/rejected": -0.2169552594423294,
"step": 81
},
{
"epoch": 0.22644114601311702,
"grad_norm": 0.579322874546051,
"learning_rate": 3.7155963302752295e-06,
"log_odds_chosen": 0.3590673804283142,
"log_odds_ratio": -0.5345906019210815,
"logits/chosen": -0.3917956054210663,
"logits/rejected": -0.1658124327659607,
"logps/chosen": -1.9473357200622559,
"logps/rejected": -2.2625551223754883,
"loss": 2.2152,
"nll_loss": 2.1617040634155273,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19473356008529663,
"rewards/margins": 0.03152196481823921,
"rewards/rejected": -0.22625553607940674,
"step": 82
},
{
"epoch": 0.22920262340352088,
"grad_norm": 0.5707481503486633,
"learning_rate": 3.7614678899082575e-06,
"log_odds_chosen": 0.601745069026947,
"log_odds_ratio": -0.44251349568367004,
"logits/chosen": -0.35622307658195496,
"logits/rejected": -0.2330722063779831,
"logps/chosen": -1.8970295190811157,
"logps/rejected": -2.4300365447998047,
"loss": 2.141,
"nll_loss": 2.096763849258423,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18970295786857605,
"rewards/margins": 0.053300708532333374,
"rewards/rejected": -0.24300366640090942,
"step": 83
},
{
"epoch": 0.23196410079392474,
"grad_norm": 0.6338332891464233,
"learning_rate": 3.8073394495412848e-06,
"log_odds_chosen": 0.4194182753562927,
"log_odds_ratio": -0.5094286799430847,
"logits/chosen": -0.38890349864959717,
"logits/rejected": -0.1751769334077835,
"logps/chosen": -1.8893790245056152,
"logps/rejected": -2.2541043758392334,
"loss": 2.1687,
"nll_loss": 2.1177978515625,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18893790245056152,
"rewards/margins": 0.036472536623477936,
"rewards/rejected": -0.22541043162345886,
"step": 84
},
{
"epoch": 0.23472557818432863,
"grad_norm": 0.514668881893158,
"learning_rate": 3.853211009174313e-06,
"log_odds_chosen": 0.5391043424606323,
"log_odds_ratio": -0.46963804960250854,
"logits/chosen": -0.42746701836586,
"logits/rejected": -0.19726645946502686,
"logps/chosen": -1.8130801916122437,
"logps/rejected": -2.285198211669922,
"loss": 2.0648,
"nll_loss": 2.017812490463257,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18130803108215332,
"rewards/margins": 0.047211792320013046,
"rewards/rejected": -0.22851979732513428,
"step": 85
},
{
"epoch": 0.2374870555747325,
"grad_norm": 0.5241694450378418,
"learning_rate": 3.89908256880734e-06,
"log_odds_chosen": 0.5370965003967285,
"log_odds_ratio": -0.47515052556991577,
"logits/chosen": -0.18979063630104065,
"logits/rejected": -0.3587040901184082,
"logps/chosen": -1.9306432008743286,
"logps/rejected": -2.4106080532073975,
"loss": 2.1744,
"nll_loss": 2.1269240379333496,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19306430220603943,
"rewards/margins": 0.04799651354551315,
"rewards/rejected": -0.24106080830097198,
"step": 86
},
{
"epoch": 0.24024853296513635,
"grad_norm": 0.5388301610946655,
"learning_rate": 3.944954128440367e-06,
"log_odds_chosen": 0.47338563203811646,
"log_odds_ratio": -0.4902949929237366,
"logits/chosen": -0.2526980936527252,
"logits/rejected": -0.27716919779777527,
"logps/chosen": -1.8183417320251465,
"logps/rejected": -2.227705955505371,
"loss": 2.0679,
"nll_loss": 2.0188400745391846,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1818341761827469,
"rewards/margins": 0.0409364253282547,
"rewards/rejected": -0.2227705866098404,
"step": 87
},
{
"epoch": 0.2430100103555402,
"grad_norm": 0.5072680115699768,
"learning_rate": 3.9908256880733945e-06,
"log_odds_chosen": 0.5098553895950317,
"log_odds_ratio": -0.4994601309299469,
"logits/chosen": -0.4232753813266754,
"logits/rejected": -0.2069588601589203,
"logps/chosen": -1.7478512525558472,
"logps/rejected": -2.1878132820129395,
"loss": 2.0084,
"nll_loss": 1.9584167003631592,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.17478513717651367,
"rewards/margins": 0.043996214866638184,
"rewards/rejected": -0.21878135204315186,
"step": 88
},
{
"epoch": 0.24577148774594407,
"grad_norm": 0.4936067461967468,
"learning_rate": 4.036697247706423e-06,
"log_odds_chosen": 0.5959604382514954,
"log_odds_ratio": -0.4470658600330353,
"logits/chosen": -0.26887303590774536,
"logits/rejected": -0.2717263400554657,
"logps/chosen": -1.8666045665740967,
"logps/rejected": -2.3904991149902344,
"loss": 2.1004,
"nll_loss": 2.0557374954223633,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18666045367717743,
"rewards/margins": 0.05238945782184601,
"rewards/rejected": -0.23904991149902344,
"step": 89
},
{
"epoch": 0.24853296513634796,
"grad_norm": 0.5457214713096619,
"learning_rate": 4.08256880733945e-06,
"log_odds_chosen": 0.5490537285804749,
"log_odds_ratio": -0.45841851830482483,
"logits/chosen": -0.34721043705940247,
"logits/rejected": -0.10521189868450165,
"logps/chosen": -1.8499623537063599,
"logps/rejected": -2.329256296157837,
"loss": 2.0961,
"nll_loss": 2.050299644470215,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18499624729156494,
"rewards/margins": 0.04792938381433487,
"rewards/rejected": -0.2329256385564804,
"step": 90
},
{
"epoch": 0.2512944425267518,
"grad_norm": 0.5049707293510437,
"learning_rate": 4.128440366972478e-06,
"log_odds_chosen": 0.4017772674560547,
"log_odds_ratio": -0.5216570496559143,
"logits/chosen": -0.26840728521347046,
"logits/rejected": -0.19604754447937012,
"logps/chosen": -1.8395251035690308,
"logps/rejected": -2.1878998279571533,
"loss": 2.0732,
"nll_loss": 2.021005868911743,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18395252525806427,
"rewards/margins": 0.034837473183870316,
"rewards/rejected": -0.2187899798154831,
"step": 91
},
{
"epoch": 0.2540559199171557,
"grad_norm": 0.4878016710281372,
"learning_rate": 4.174311926605505e-06,
"log_odds_chosen": 0.40200310945510864,
"log_odds_ratio": -0.5181869268417358,
"logits/chosen": -0.17307503521442413,
"logits/rejected": -0.5424618721008301,
"logps/chosen": -1.8756885528564453,
"logps/rejected": -2.2274668216705322,
"loss": 2.1098,
"nll_loss": 2.0579941272735596,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18756884336471558,
"rewards/margins": 0.035177819430828094,
"rewards/rejected": -0.22274667024612427,
"step": 92
},
{
"epoch": 0.25681739730755954,
"grad_norm": 0.5148739218711853,
"learning_rate": 4.220183486238532e-06,
"log_odds_chosen": 0.4867369532585144,
"log_odds_ratio": -0.4866551160812378,
"logits/chosen": -0.20501409471035004,
"logits/rejected": -0.3177582919597626,
"logps/chosen": -1.8571135997772217,
"logps/rejected": -2.2809691429138184,
"loss": 2.1117,
"nll_loss": 2.0629868507385254,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1857113391160965,
"rewards/margins": 0.04238557443022728,
"rewards/rejected": -0.22809693217277527,
"step": 93
},
{
"epoch": 0.2595788746979634,
"grad_norm": 0.48695462942123413,
"learning_rate": 4.26605504587156e-06,
"log_odds_chosen": 0.4320299029350281,
"log_odds_ratio": -0.5084698796272278,
"logits/chosen": -0.33521950244903564,
"logits/rejected": -0.22524963319301605,
"logps/chosen": -1.759456753730774,
"logps/rejected": -2.130563974380493,
"loss": 1.9918,
"nll_loss": 1.9409611225128174,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17594566941261292,
"rewards/margins": 0.03711073845624924,
"rewards/rejected": -0.21305640041828156,
"step": 94
},
{
"epoch": 0.26234035208836726,
"grad_norm": 0.4444352984428406,
"learning_rate": 4.311926605504588e-06,
"log_odds_chosen": 0.3173726201057434,
"log_odds_ratio": -0.5506779551506042,
"logits/chosen": -0.15516842901706696,
"logits/rejected": -0.30819153785705566,
"logps/chosen": -1.7852771282196045,
"logps/rejected": -2.0567989349365234,
"loss": 2.0242,
"nll_loss": 1.969139814376831,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17852769792079926,
"rewards/margins": 0.027152204886078835,
"rewards/rejected": -0.20567990839481354,
"step": 95
},
{
"epoch": 0.26510182947877114,
"grad_norm": 0.508222758769989,
"learning_rate": 4.357798165137615e-06,
"log_odds_chosen": 0.26060953736305237,
"log_odds_ratio": -0.5766869187355042,
"logits/chosen": -0.21468104422092438,
"logits/rejected": -0.17443214356899261,
"logps/chosen": -1.9513554573059082,
"logps/rejected": -2.179452419281006,
"loss": 2.1941,
"nll_loss": 2.1364006996154785,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19513554871082306,
"rewards/margins": 0.02280968800187111,
"rewards/rejected": -0.21794524788856506,
"step": 96
},
{
"epoch": 0.26786330686917503,
"grad_norm": 0.4931764602661133,
"learning_rate": 4.403669724770643e-06,
"log_odds_chosen": 0.5352882146835327,
"log_odds_ratio": -0.4726255536079407,
"logits/chosen": -0.10641247034072876,
"logits/rejected": -0.164643332362175,
"logps/chosen": -1.7959346771240234,
"logps/rejected": -2.266484022140503,
"loss": 2.0405,
"nll_loss": 1.9932180643081665,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17959347367286682,
"rewards/margins": 0.047054924070835114,
"rewards/rejected": -0.22664840519428253,
"step": 97
},
{
"epoch": 0.27062478425957887,
"grad_norm": 0.49030837416648865,
"learning_rate": 4.44954128440367e-06,
"log_odds_chosen": 0.38322845101356506,
"log_odds_ratio": -0.5242493748664856,
"logits/chosen": -0.15826918184757233,
"logits/rejected": -0.20204174518585205,
"logps/chosen": -1.8669074773788452,
"logps/rejected": -2.199532985687256,
"loss": 2.1032,
"nll_loss": 2.0507290363311768,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18669073283672333,
"rewards/margins": 0.03326254338026047,
"rewards/rejected": -0.2199532836675644,
"step": 98
},
{
"epoch": 0.27338626164998275,
"grad_norm": 0.49382349848747253,
"learning_rate": 4.4954128440366975e-06,
"log_odds_chosen": 0.5044716000556946,
"log_odds_ratio": -0.4766765236854553,
"logits/chosen": -0.1681555211544037,
"logits/rejected": -0.3671070337295532,
"logps/chosen": -1.9003745317459106,
"logps/rejected": -2.344512939453125,
"loss": 2.1245,
"nll_loss": 2.076801061630249,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19003747403621674,
"rewards/margins": 0.04441382735967636,
"rewards/rejected": -0.2344512790441513,
"step": 99
},
{
"epoch": 0.2761477390403866,
"grad_norm": 0.507435142993927,
"learning_rate": 4.541284403669725e-06,
"log_odds_chosen": 0.2962280511856079,
"log_odds_ratio": -0.5599422454833984,
"logits/chosen": -0.26080605387687683,
"logits/rejected": -0.24701349437236786,
"logps/chosen": -1.798446774482727,
"logps/rejected": -2.050736427307129,
"loss": 2.0576,
"nll_loss": 2.001603841781616,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1798446923494339,
"rewards/margins": 0.0252289529889822,
"rewards/rejected": -0.20507365465164185,
"step": 100
},
{
"epoch": 0.2789092164307905,
"grad_norm": 0.461052268743515,
"learning_rate": 4.587155963302753e-06,
"log_odds_chosen": 0.29842671751976013,
"log_odds_ratio": -0.5633093118667603,
"logits/chosen": -0.11747467517852783,
"logits/rejected": -0.2521146833896637,
"logps/chosen": -1.913146734237671,
"logps/rejected": -2.1769111156463623,
"loss": 2.1524,
"nll_loss": 2.0960795879364014,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1913146674633026,
"rewards/margins": 0.02637643553316593,
"rewards/rejected": -0.21769112348556519,
"step": 101
},
{
"epoch": 0.28167069382119436,
"grad_norm": 0.42991501092910767,
"learning_rate": 4.63302752293578e-06,
"log_odds_chosen": 0.5397889614105225,
"log_odds_ratio": -0.4643154740333557,
"logits/chosen": -0.17212505638599396,
"logits/rejected": -0.3217812776565552,
"logps/chosen": -1.7475972175598145,
"logps/rejected": -2.209057331085205,
"loss": 2.0036,
"nll_loss": 1.9572076797485352,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17475973069667816,
"rewards/margins": 0.046146005392074585,
"rewards/rejected": -0.22090573608875275,
"step": 102
},
{
"epoch": 0.2844321712115982,
"grad_norm": 0.47230181097984314,
"learning_rate": 4.678899082568808e-06,
"log_odds_chosen": 0.29523512721061707,
"log_odds_ratio": -0.5644804239273071,
"logits/chosen": -0.22446005046367645,
"logits/rejected": -0.26140278577804565,
"logps/chosen": -1.7972923517227173,
"logps/rejected": -2.051152229309082,
"loss": 2.0406,
"nll_loss": 1.9841551780700684,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17972922325134277,
"rewards/margins": 0.025385981425642967,
"rewards/rejected": -0.20511522889137268,
"step": 103
},
{
"epoch": 0.2871936486020021,
"grad_norm": 0.4233863651752472,
"learning_rate": 4.724770642201835e-06,
"log_odds_chosen": 0.3453969359397888,
"log_odds_ratio": -0.5395044684410095,
"logits/chosen": -0.18183927237987518,
"logits/rejected": -0.3289150297641754,
"logps/chosen": -1.7190299034118652,
"logps/rejected": -2.0107898712158203,
"loss": 1.9637,
"nll_loss": 1.9097464084625244,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17190299928188324,
"rewards/margins": 0.02917599491775036,
"rewards/rejected": -0.20107899606227875,
"step": 104
},
{
"epoch": 0.2899551259924059,
"grad_norm": 0.4307953417301178,
"learning_rate": 4.770642201834863e-06,
"log_odds_chosen": 0.46213921904563904,
"log_odds_ratio": -0.4912022650241852,
"logits/chosen": -0.1190250962972641,
"logits/rejected": -0.19093452394008636,
"logps/chosen": -1.7502868175506592,
"logps/rejected": -2.1464428901672363,
"loss": 1.9758,
"nll_loss": 1.9266620874404907,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17502868175506592,
"rewards/margins": 0.03961558640003204,
"rewards/rejected": -0.21464428305625916,
"step": 105
},
{
"epoch": 0.2927166033828098,
"grad_norm": 0.43505242466926575,
"learning_rate": 4.816513761467891e-06,
"log_odds_chosen": 0.4836673140525818,
"log_odds_ratio": -0.49270570278167725,
"logits/chosen": -0.08418025076389313,
"logits/rejected": -0.21315333247184753,
"logps/chosen": -1.7327262163162231,
"logps/rejected": -2.145224094390869,
"loss": 1.9644,
"nll_loss": 1.9150831699371338,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17327262461185455,
"rewards/margins": 0.041249774396419525,
"rewards/rejected": -0.21452240645885468,
"step": 106
},
{
"epoch": 0.2954780807732137,
"grad_norm": 0.4454819858074188,
"learning_rate": 4.862385321100918e-06,
"log_odds_chosen": 0.3297927975654602,
"log_odds_ratio": -0.5459839701652527,
"logits/chosen": -0.08551089465618134,
"logits/rejected": -0.3492465913295746,
"logps/chosen": -1.8108294010162354,
"logps/rejected": -2.0928804874420166,
"loss": 2.0393,
"nll_loss": 1.9846614599227905,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18108293414115906,
"rewards/margins": 0.02820511721074581,
"rewards/rejected": -0.20928806066513062,
"step": 107
},
{
"epoch": 0.2982395581636175,
"grad_norm": 0.4228799641132355,
"learning_rate": 4.908256880733945e-06,
"log_odds_chosen": 0.4139629602432251,
"log_odds_ratio": -0.5125847458839417,
"logits/chosen": -0.042439091950654984,
"logits/rejected": -0.3392437696456909,
"logps/chosen": -1.842297077178955,
"logps/rejected": -2.203068971633911,
"loss": 2.0532,
"nll_loss": 2.00195050239563,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18422970175743103,
"rewards/margins": 0.03607717901468277,
"rewards/rejected": -0.2203068882226944,
"step": 108
},
{
"epoch": 0.3010010355540214,
"grad_norm": 0.4132618308067322,
"learning_rate": 4.954128440366973e-06,
"log_odds_chosen": 0.5509217977523804,
"log_odds_ratio": -0.4667437970638275,
"logits/chosen": -0.047298721969127655,
"logits/rejected": -0.492242693901062,
"logps/chosen": -1.8542200326919556,
"logps/rejected": -2.3400068283081055,
"loss": 2.0568,
"nll_loss": 2.0100908279418945,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18542201817035675,
"rewards/margins": 0.04857867211103439,
"rewards/rejected": -0.23400066792964935,
"step": 109
},
{
"epoch": 0.30376251294442524,
"grad_norm": 0.40540555119514465,
"learning_rate": 5e-06,
"log_odds_chosen": 0.26367828249931335,
"log_odds_ratio": -0.5857654213905334,
"logits/chosen": 0.020869266241788864,
"logits/rejected": -0.3348216116428375,
"logps/chosen": -1.8042923212051392,
"logps/rejected": -2.0284535884857178,
"loss": 2.0564,
"nll_loss": 1.997856616973877,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.18042920529842377,
"rewards/margins": 0.02241615764796734,
"rewards/rejected": -0.20284539461135864,
"step": 110
},
{
"epoch": 0.30652399033482913,
"grad_norm": 0.3964520990848541,
"learning_rate": 4.999987154315977e-06,
"log_odds_chosen": 0.39854103326797485,
"log_odds_ratio": -0.5201252698898315,
"logits/chosen": -0.07075213640928268,
"logits/rejected": -0.26321282982826233,
"logps/chosen": -1.7023441791534424,
"logps/rejected": -2.03615665435791,
"loss": 1.9356,
"nll_loss": 1.883634090423584,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17023441195487976,
"rewards/margins": 0.03338123857975006,
"rewards/rejected": -0.20361566543579102,
"step": 111
},
{
"epoch": 0.309285467725233,
"grad_norm": 0.4122212827205658,
"learning_rate": 4.999948617395916e-06,
"log_odds_chosen": 0.3810994029045105,
"log_odds_ratio": -0.528425931930542,
"logits/chosen": 0.04292220622301102,
"logits/rejected": -0.24851810932159424,
"logps/chosen": -1.7573786973953247,
"logps/rejected": -2.084685802459717,
"loss": 1.9844,
"nll_loss": 1.931592583656311,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17573785781860352,
"rewards/margins": 0.03273071348667145,
"rewards/rejected": -0.20846858620643616,
"step": 112
},
{
"epoch": 0.31204694511563685,
"grad_norm": 0.4462045729160309,
"learning_rate": 4.999884389635843e-06,
"log_odds_chosen": 0.392294704914093,
"log_odds_ratio": -0.520103394985199,
"logits/chosen": -0.046707406640052795,
"logits/rejected": -0.22425656020641327,
"logps/chosen": -1.8034018278121948,
"logps/rejected": -2.140183687210083,
"loss": 2.0382,
"nll_loss": 1.9861979484558105,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18034018576145172,
"rewards/margins": 0.03367818892002106,
"rewards/rejected": -0.21401838958263397,
"step": 113
},
{
"epoch": 0.31480842250604074,
"grad_norm": 0.40587544441223145,
"learning_rate": 4.9997944716957985e-06,
"log_odds_chosen": 0.48522624373435974,
"log_odds_ratio": -0.49063414335250854,
"logits/chosen": 0.049725521355867386,
"logits/rejected": -0.3027104139328003,
"logps/chosen": -1.8043849468231201,
"logps/rejected": -2.224425792694092,
"loss": 2.0299,
"nll_loss": 1.9808719158172607,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18043850362300873,
"rewards/margins": 0.04200407862663269,
"rewards/rejected": -0.22244258224964142,
"step": 114
},
{
"epoch": 0.3175698998964446,
"grad_norm": 0.38116294145584106,
"learning_rate": 4.999678864499828e-06,
"log_odds_chosen": 0.35890865325927734,
"log_odds_ratio": -0.5324774384498596,
"logits/chosen": 0.1463293731212616,
"logits/rejected": -0.3165544271469116,
"logps/chosen": -1.83231782913208,
"logps/rejected": -2.1404545307159424,
"loss": 2.0335,
"nll_loss": 1.9802701473236084,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18323180079460144,
"rewards/margins": 0.030813684687018394,
"rewards/rejected": -0.21404549479484558,
"step": 115
},
{
"epoch": 0.32033137728684846,
"grad_norm": 0.3846777081489563,
"learning_rate": 4.999537569235975e-06,
"log_odds_chosen": 0.31865543127059937,
"log_odds_ratio": -0.5562014579772949,
"logits/chosen": 0.06778667122125626,
"logits/rejected": -0.2945130169391632,
"logps/chosen": -1.8204419612884521,
"logps/rejected": -2.0919411182403564,
"loss": 2.0362,
"nll_loss": 1.9805940389633179,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18204417824745178,
"rewards/margins": 0.02714991755783558,
"rewards/rejected": -0.2091941088438034,
"step": 116
},
{
"epoch": 0.32309285467725235,
"grad_norm": 0.3859681189060211,
"learning_rate": 4.999370587356267e-06,
"log_odds_chosen": 0.2864968180656433,
"log_odds_ratio": -0.5685579180717468,
"logits/chosen": 0.021781759336590767,
"logits/rejected": -0.41891583800315857,
"logps/chosen": -1.8915698528289795,
"logps/rejected": -2.1414055824279785,
"loss": 2.1035,
"nll_loss": 2.0466203689575195,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18915697932243347,
"rewards/margins": 0.02498357556760311,
"rewards/rejected": -0.21414057910442352,
"step": 117
},
{
"epoch": 0.3258543320676562,
"grad_norm": 0.4066268801689148,
"learning_rate": 4.9991779205767e-06,
"log_odds_chosen": 0.2647791802883148,
"log_odds_ratio": -0.5728141069412231,
"logits/chosen": -0.04139568656682968,
"logits/rejected": -0.6460073590278625,
"logps/chosen": -1.757622480392456,
"logps/rejected": -1.9820458889007568,
"loss": 1.984,
"nll_loss": 1.9267468452453613,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17576223611831665,
"rewards/margins": 0.022442325949668884,
"rewards/rejected": -0.19820457696914673,
"step": 118
},
{
"epoch": 0.32861580945806007,
"grad_norm": 0.4128899574279785,
"learning_rate": 4.998959570877224e-06,
"log_odds_chosen": 0.3499869406223297,
"log_odds_ratio": -0.5363956093788147,
"logits/chosen": 0.16672371327877045,
"logits/rejected": -0.34432125091552734,
"logps/chosen": -1.9091435670852661,
"logps/rejected": -2.2108755111694336,
"loss": 2.1306,
"nll_loss": 2.0769805908203125,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19091437757015228,
"rewards/margins": 0.0301731638610363,
"rewards/rejected": -0.2210875302553177,
"step": 119
},
{
"epoch": 0.3313772868484639,
"grad_norm": 0.358536958694458,
"learning_rate": 4.99871554050172e-06,
"log_odds_chosen": 0.42555758357048035,
"log_odds_ratio": -0.5050589442253113,
"logits/chosen": 0.02474893629550934,
"logits/rejected": -0.25200581550598145,
"logps/chosen": -1.7013750076293945,
"logps/rejected": -2.060541868209839,
"loss": 1.9062,
"nll_loss": 1.855684757232666,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17013752460479736,
"rewards/margins": 0.03591667488217354,
"rewards/rejected": -0.2060541808605194,
"step": 120
},
{
"epoch": 0.3341387642388678,
"grad_norm": 0.394452840089798,
"learning_rate": 4.9984458319579775e-06,
"log_odds_chosen": 0.423836886882782,
"log_odds_ratio": -0.5092670321464539,
"logits/chosen": -0.0022555014584213495,
"logits/rejected": -0.3431664705276489,
"logps/chosen": -1.7344661951065063,
"logps/rejected": -2.0950186252593994,
"loss": 1.9335,
"nll_loss": 1.8825750350952148,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17344661056995392,
"rewards/margins": 0.03605526313185692,
"rewards/rejected": -0.20950186252593994,
"step": 121
},
{
"epoch": 0.3369002416292717,
"grad_norm": 0.3641405999660492,
"learning_rate": 4.99815044801767e-06,
"log_odds_chosen": 0.31487271189689636,
"log_odds_ratio": -0.5565272569656372,
"logits/chosen": 0.12278047204017639,
"logits/rejected": -0.3144981861114502,
"logps/chosen": -1.7826478481292725,
"logps/rejected": -2.0509307384490967,
"loss": 2.0075,
"nll_loss": 1.951832890510559,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1782647967338562,
"rewards/margins": 0.026828289031982422,
"rewards/rejected": -0.20509308576583862,
"step": 122
},
{
"epoch": 0.3396617190196755,
"grad_norm": 0.4089741110801697,
"learning_rate": 4.9978293917163225e-06,
"log_odds_chosen": 0.1852492243051529,
"log_odds_ratio": -0.6086881756782532,
"logits/chosen": 0.059370554983615875,
"logits/rejected": -0.2047065794467926,
"logps/chosen": -1.9199140071868896,
"logps/rejected": -2.0819344520568848,
"loss": 2.1197,
"nll_loss": 2.0588343143463135,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1919914036989212,
"rewards/margins": 0.016202054917812347,
"rewards/rejected": -0.20819345116615295,
"step": 123
},
{
"epoch": 0.3424231964100794,
"grad_norm": 0.39316678047180176,
"learning_rate": 4.997482666353287e-06,
"log_odds_chosen": 0.3156554102897644,
"log_odds_ratio": -0.5562319755554199,
"logits/chosen": -0.036051761358976364,
"logits/rejected": -0.20177382230758667,
"logps/chosen": -1.8153338432312012,
"logps/rejected": -2.0875864028930664,
"loss": 2.0271,
"nll_loss": 1.9714655876159668,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.18153339624404907,
"rewards/margins": 0.027225244790315628,
"rewards/rejected": -0.2087586373090744,
"step": 124
},
{
"epoch": 0.34518467380048323,
"grad_norm": 0.37397369742393494,
"learning_rate": 4.997110275491702e-06,
"log_odds_chosen": 0.47729045152664185,
"log_odds_ratio": -0.4998089373111725,
"logits/chosen": 0.06823548674583435,
"logits/rejected": -0.6389302015304565,
"logps/chosen": -1.725682258605957,
"logps/rejected": -2.133211612701416,
"loss": 1.9424,
"nll_loss": 1.8924020528793335,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17256823182106018,
"rewards/margins": 0.04075293242931366,
"rewards/rejected": -0.21332114934921265,
"step": 125
},
{
"epoch": 0.3479461511908871,
"grad_norm": 0.36338478326797485,
"learning_rate": 4.9967122229584614e-06,
"log_odds_chosen": 0.46506431698799133,
"log_odds_ratio": -0.49022355675697327,
"logits/chosen": 0.062216617166996,
"logits/rejected": -0.4744764268398285,
"logps/chosen": -1.7237508296966553,
"logps/rejected": -2.1211488246917725,
"loss": 1.9404,
"nll_loss": 1.891422152519226,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17237509787082672,
"rewards/margins": 0.03973980247974396,
"rewards/rejected": -0.2121148705482483,
"step": 126
},
{
"epoch": 0.350707628581291,
"grad_norm": 0.3949665427207947,
"learning_rate": 4.996288512844169e-06,
"log_odds_chosen": 0.208975687623024,
"log_odds_ratio": -0.603169322013855,
"logits/chosen": 0.15239077806472778,
"logits/rejected": -0.14533445239067078,
"logps/chosen": -1.8812259435653687,
"logps/rejected": -2.058507204055786,
"loss": 2.1066,
"nll_loss": 2.0462498664855957,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.18812260031700134,
"rewards/margins": 0.01772812381386757,
"rewards/rejected": -0.2058507204055786,
"step": 127
},
{
"epoch": 0.35346910597169484,
"grad_norm": 0.3274211585521698,
"learning_rate": 4.995839149503103e-06,
"log_odds_chosen": 0.3612444996833801,
"log_odds_ratio": -0.5432137250900269,
"logits/chosen": 0.15867066383361816,
"logits/rejected": -0.37713369727134705,
"logps/chosen": -1.7504801750183105,
"logps/rejected": -2.060889959335327,
"loss": 1.9636,
"nll_loss": 1.9092310667037964,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17504799365997314,
"rewards/margins": 0.03104100562632084,
"rewards/rejected": -0.20608901977539062,
"step": 128
},
{
"epoch": 0.35623058336209873,
"grad_norm": 0.3613954186439514,
"learning_rate": 4.995364137553166e-06,
"log_odds_chosen": 0.2536008656024933,
"log_odds_ratio": -0.5839525461196899,
"logits/chosen": 0.025492653250694275,
"logits/rejected": -0.21177223324775696,
"logps/chosen": -1.7983472347259521,
"logps/rejected": -2.013514995574951,
"loss": 2.0121,
"nll_loss": 1.953747034072876,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1798347383737564,
"rewards/margins": 0.021516768261790276,
"rewards/rejected": -0.20135147869586945,
"step": 129
},
{
"epoch": 0.35899206075250256,
"grad_norm": 0.33361607789993286,
"learning_rate": 4.994863481875842e-06,
"log_odds_chosen": 0.4904846251010895,
"log_odds_ratio": -0.4827660620212555,
"logits/chosen": 0.06080925092101097,
"logits/rejected": -0.42839962244033813,
"logps/chosen": -1.7444632053375244,
"logps/rejected": -2.1671104431152344,
"loss": 1.9554,
"nll_loss": 1.9071358442306519,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17444632947444916,
"rewards/margins": 0.04226472228765488,
"rewards/rejected": -0.21671104431152344,
"step": 130
},
{
"epoch": 0.36175353814290645,
"grad_norm": 0.38626936078071594,
"learning_rate": 4.99433718761614e-06,
"log_odds_chosen": 0.4231716990470886,
"log_odds_ratio": -0.5083524584770203,
"logits/chosen": 0.15689942240715027,
"logits/rejected": -0.27488887310028076,
"logps/chosen": -1.7832603454589844,
"logps/rejected": -2.1480154991149902,
"loss": 1.9872,
"nll_loss": 1.9363242387771606,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1783260554075241,
"rewards/margins": 0.036475520581007004,
"rewards/rejected": -0.21480156481266022,
"step": 131
},
{
"epoch": 0.36451501553331034,
"grad_norm": 0.3283027410507202,
"learning_rate": 4.993785260182552e-06,
"log_odds_chosen": 0.4390679895877838,
"log_odds_ratio": -0.5016705393791199,
"logits/chosen": 0.13413819670677185,
"logits/rejected": -0.5499831438064575,
"logps/chosen": -1.7239453792572021,
"logps/rejected": -2.099557399749756,
"loss": 1.9329,
"nll_loss": 1.882704734802246,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1723945438861847,
"rewards/margins": 0.037561215460300446,
"rewards/rejected": -0.20995575189590454,
"step": 132
},
{
"epoch": 0.36727649292371417,
"grad_norm": 0.29404011368751526,
"learning_rate": 4.993207705246983e-06,
"log_odds_chosen": 0.33607491850852966,
"log_odds_ratio": -0.5448395609855652,
"logits/chosen": 0.09872237592935562,
"logits/rejected": -0.3414973318576813,
"logps/chosen": -1.7783253192901611,
"logps/rejected": -2.0684056282043457,
"loss": 1.9796,
"nll_loss": 1.925091028213501,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17783252894878387,
"rewards/margins": 0.029008038341999054,
"rewards/rejected": -0.20684055984020233,
"step": 133
},
{
"epoch": 0.37003797031411806,
"grad_norm": 0.303219199180603,
"learning_rate": 4.992604528744705e-06,
"log_odds_chosen": 0.37202000617980957,
"log_odds_ratio": -0.5278716087341309,
"logits/chosen": 0.0883011743426323,
"logits/rejected": -0.6590756177902222,
"logps/chosen": -1.7620971202850342,
"logps/rejected": -2.0787768363952637,
"loss": 1.9649,
"nll_loss": 1.9120779037475586,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1762097179889679,
"rewards/margins": 0.03166797757148743,
"rewards/rejected": -0.20787768065929413,
"step": 134
},
{
"epoch": 0.37279944770452195,
"grad_norm": 0.3061475455760956,
"learning_rate": 4.9919757368742895e-06,
"log_odds_chosen": 0.43486303091049194,
"log_odds_ratio": -0.5079333782196045,
"logits/chosen": 0.13486391305923462,
"logits/rejected": -0.44202011823654175,
"logps/chosen": -1.683558464050293,
"logps/rejected": -2.0501785278320312,
"loss": 1.8894,
"nll_loss": 1.8386157751083374,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16835585236549377,
"rewards/margins": 0.0366620309650898,
"rewards/rejected": -0.20501787960529327,
"step": 135
},
{
"epoch": 0.3755609250949258,
"grad_norm": 0.3027716279029846,
"learning_rate": 4.991321336097546e-06,
"log_odds_chosen": 0.3602009117603302,
"log_odds_ratio": -0.5340627431869507,
"logits/chosen": 0.09272217750549316,
"logits/rejected": -0.37184590101242065,
"logps/chosen": -1.8042678833007812,
"logps/rejected": -2.1151676177978516,
"loss": 1.9799,
"nll_loss": 1.9264534711837769,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18042679131031036,
"rewards/margins": 0.031089982017874718,
"rewards/rejected": -0.21151678264141083,
"step": 136
},
{
"epoch": 0.37832240248532967,
"grad_norm": 0.2968030273914337,
"learning_rate": 4.990641333139455e-06,
"log_odds_chosen": 0.3864700496196747,
"log_odds_ratio": -0.5240495204925537,
"logits/chosen": 0.15802569687366486,
"logits/rejected": -0.40965431928634644,
"logps/chosen": -1.6786091327667236,
"logps/rejected": -2.002851724624634,
"loss": 1.8816,
"nll_loss": 1.8292038440704346,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16786089539527893,
"rewards/margins": 0.032424286007881165,
"rewards/rejected": -0.2002851963043213,
"step": 137
},
{
"epoch": 0.3810838798757335,
"grad_norm": 0.2909601330757141,
"learning_rate": 4.989935734988098e-06,
"log_odds_chosen": 0.2820569574832916,
"log_odds_ratio": -0.5659909844398499,
"logits/chosen": 0.20683330297470093,
"logits/rejected": -0.5098833441734314,
"logps/chosen": -1.7661350965499878,
"logps/rejected": -2.0048882961273193,
"loss": 1.9544,
"nll_loss": 1.8978168964385986,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17661352455615997,
"rewards/margins": 0.023875314742326736,
"rewards/rejected": -0.2004888355731964,
"step": 138
},
{
"epoch": 0.3838453572661374,
"grad_norm": 0.3169582188129425,
"learning_rate": 4.989204548894589e-06,
"log_odds_chosen": 0.27746957540512085,
"log_odds_ratio": -0.5728934407234192,
"logits/chosen": 0.26577499508857727,
"logits/rejected": -0.32127267122268677,
"logps/chosen": -1.846369743347168,
"logps/rejected": -2.0836005210876465,
"loss": 2.0312,
"nll_loss": 1.973919153213501,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.18463698029518127,
"rewards/margins": 0.023723063990473747,
"rewards/rejected": -0.20836003124713898,
"step": 139
},
{
"epoch": 0.3866068346565413,
"grad_norm": 0.2856157720088959,
"learning_rate": 4.988447782372996e-06,
"log_odds_chosen": 0.2677723169326782,
"log_odds_ratio": -0.5744235515594482,
"logits/chosen": 0.08042767643928528,
"logits/rejected": -0.489827036857605,
"logps/chosen": -1.6772124767303467,
"logps/rejected": -1.9024029970169067,
"loss": 1.8706,
"nll_loss": 1.8131682872772217,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1677212417125702,
"rewards/margins": 0.022519057616591454,
"rewards/rejected": -0.1902403086423874,
"step": 140
},
{
"epoch": 0.3893683120469451,
"grad_norm": 0.26832813024520874,
"learning_rate": 4.9876654432002655e-06,
"log_odds_chosen": 0.3760983943939209,
"log_odds_ratio": -0.5314284563064575,
"logits/chosen": 0.22744062542915344,
"logits/rejected": -0.6199472546577454,
"logps/chosen": -1.7124302387237549,
"logps/rejected": -2.035111665725708,
"loss": 1.8915,
"nll_loss": 1.838379979133606,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17124304175376892,
"rewards/margins": 0.032268136739730835,
"rewards/rejected": -0.20351116359233856,
"step": 141
},
{
"epoch": 0.392129789437349,
"grad_norm": 0.2787982225418091,
"learning_rate": 4.986857539416144e-06,
"log_odds_chosen": 0.4650447368621826,
"log_odds_ratio": -0.49311909079551697,
"logits/chosen": 0.13020291924476624,
"logits/rejected": -0.6509877443313599,
"logps/chosen": -1.6808263063430786,
"logps/rejected": -2.0721030235290527,
"loss": 1.8625,
"nll_loss": 1.813225269317627,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16808265447616577,
"rewards/margins": 0.039127662777900696,
"rewards/rejected": -0.20721031725406647,
"step": 142
},
{
"epoch": 0.39489126682775283,
"grad_norm": 0.2971290349960327,
"learning_rate": 4.986024079323092e-06,
"log_odds_chosen": 0.20813080668449402,
"log_odds_ratio": -0.5995200276374817,
"logits/chosen": 0.1712779998779297,
"logits/rejected": -0.2880682647228241,
"logps/chosen": -1.6474640369415283,
"logps/rejected": -1.8199793100357056,
"loss": 1.873,
"nll_loss": 1.8130154609680176,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16474640369415283,
"rewards/margins": 0.017251526936888695,
"rewards/rejected": -0.18199792504310608,
"step": 143
},
{
"epoch": 0.3976527442181567,
"grad_norm": 0.28045693039894104,
"learning_rate": 4.985165071486201e-06,
"log_odds_chosen": 0.3738395869731903,
"log_odds_ratio": -0.5252775549888611,
"logits/chosen": 0.10901468247175217,
"logits/rejected": -0.5548363924026489,
"logps/chosen": -1.7671691179275513,
"logps/rejected": -2.086446762084961,
"loss": 1.9397,
"nll_loss": 1.8872014284133911,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1767169088125229,
"rewards/margins": 0.031927771866321564,
"rewards/rejected": -0.20864468812942505,
"step": 144
},
{
"epoch": 0.4004142216085606,
"grad_norm": 0.28425726294517517,
"learning_rate": 4.984280524733107e-06,
"log_odds_chosen": 0.44026607275009155,
"log_odds_ratio": -0.5015271902084351,
"logits/chosen": 0.08948800712823868,
"logits/rejected": -0.4244663119316101,
"logps/chosen": -1.698014736175537,
"logps/rejected": -2.068047285079956,
"loss": 1.8842,
"nll_loss": 1.8340739011764526,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1698014885187149,
"rewards/margins": 0.037003256380558014,
"rewards/rejected": -0.20680472254753113,
"step": 145
},
{
"epoch": 0.40317569899896444,
"grad_norm": 0.27275997400283813,
"learning_rate": 4.983370448153896e-06,
"log_odds_chosen": 0.34414318203926086,
"log_odds_ratio": -0.53827965259552,
"logits/chosen": 0.12077900767326355,
"logits/rejected": -0.45666831731796265,
"logps/chosen": -1.7140839099884033,
"logps/rejected": -2.004852056503296,
"loss": 1.8944,
"nll_loss": 1.8405368328094482,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17140838503837585,
"rewards/margins": 0.029076814651489258,
"rewards/rejected": -0.2004851996898651,
"step": 146
},
{
"epoch": 0.4059371763893683,
"grad_norm": 0.2800770699977875,
"learning_rate": 4.9824348511010115e-06,
"log_odds_chosen": 0.33458200097084045,
"log_odds_ratio": -0.5480896830558777,
"logits/chosen": 0.24399758875370026,
"logits/rejected": -0.5033762454986572,
"logps/chosen": -1.8108501434326172,
"logps/rejected": -2.099644422531128,
"loss": 1.983,
"nll_loss": 1.9281907081604004,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1810850203037262,
"rewards/margins": 0.02887941710650921,
"rewards/rejected": -0.20996443927288055,
"step": 147
},
{
"epoch": 0.40869865377977216,
"grad_norm": 0.25797146558761597,
"learning_rate": 4.981473743189163e-06,
"log_odds_chosen": 0.5845157504081726,
"log_odds_ratio": -0.45159903168678284,
"logits/chosen": 0.005943242460489273,
"logits/rejected": -0.7690349221229553,
"logps/chosen": -1.5877741575241089,
"logps/rejected": -2.0803065299987793,
"loss": 1.7773,
"nll_loss": 1.73209547996521,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1587774157524109,
"rewards/margins": 0.04925324767827988,
"rewards/rejected": -0.20803067088127136,
"step": 148
},
{
"epoch": 0.41146013117017605,
"grad_norm": 0.27324819564819336,
"learning_rate": 4.98048713429522e-06,
"log_odds_chosen": 0.3932475745677948,
"log_odds_ratio": -0.5176883339881897,
"logits/chosen": 0.20163178443908691,
"logits/rejected": -0.283429354429245,
"logps/chosen": -1.7028628587722778,
"logps/rejected": -2.03279709815979,
"loss": 1.8839,
"nll_loss": 1.8321328163146973,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17028628289699554,
"rewards/margins": 0.032993413507938385,
"rewards/rejected": -0.20327968895435333,
"step": 149
},
{
"epoch": 0.41422160856057993,
"grad_norm": 0.30329883098602295,
"learning_rate": 4.979475034558115e-06,
"log_odds_chosen": 0.3415054678916931,
"log_odds_ratio": -0.5400842428207397,
"logits/chosen": 0.27814820408821106,
"logits/rejected": -0.420722633600235,
"logps/chosen": -1.834177017211914,
"logps/rejected": -2.1293845176696777,
"loss": 2.0005,
"nll_loss": 1.9464527368545532,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18341770768165588,
"rewards/margins": 0.029520753771066666,
"rewards/rejected": -0.21293845772743225,
"step": 150
},
{
"epoch": 0.41698308595098377,
"grad_norm": 0.2631519138813019,
"learning_rate": 4.978437454378741e-06,
"log_odds_chosen": 0.31423458456993103,
"log_odds_ratio": -0.5544165968894958,
"logits/chosen": 0.25030598044395447,
"logits/rejected": -0.45834630727767944,
"logps/chosen": -1.7350115776062012,
"logps/rejected": -1.998989462852478,
"loss": 1.9237,
"nll_loss": 1.8682823181152344,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1735011488199234,
"rewards/margins": 0.026397792622447014,
"rewards/rejected": -0.19989895820617676,
"step": 151
},
{
"epoch": 0.41974456334138766,
"grad_norm": 0.28724023699760437,
"learning_rate": 4.977374404419838e-06,
"log_odds_chosen": 0.34610986709594727,
"log_odds_ratio": -0.5409384965896606,
"logits/chosen": 0.23072689771652222,
"logits/rejected": -0.4912022054195404,
"logps/chosen": -1.7317731380462646,
"logps/rejected": -2.0284037590026855,
"loss": 1.9099,
"nll_loss": 1.8557640314102173,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1731773018836975,
"rewards/margins": 0.029663076624274254,
"rewards/rejected": -0.2028403878211975,
"step": 152
},
{
"epoch": 0.4225060407317915,
"grad_norm": 0.2612636089324951,
"learning_rate": 4.976285895605888e-06,
"log_odds_chosen": 0.4580684304237366,
"log_odds_ratio": -0.4960138499736786,
"logits/chosen": 0.18824972212314606,
"logits/rejected": -0.5351172685623169,
"logps/chosen": -1.6807944774627686,
"logps/rejected": -2.0613059997558594,
"loss": 1.8584,
"nll_loss": 1.8087894916534424,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1680794358253479,
"rewards/margins": 0.038051165640354156,
"rewards/rejected": -0.20613062381744385,
"step": 153
},
{
"epoch": 0.4252675181221954,
"grad_norm": 0.2724829614162445,
"learning_rate": 4.9751719391230055e-06,
"log_odds_chosen": 0.24739423394203186,
"log_odds_ratio": -0.5839247703552246,
"logits/chosen": 0.24238519370555878,
"logits/rejected": -0.4675644040107727,
"logps/chosen": -1.6522984504699707,
"logps/rejected": -1.8578873872756958,
"loss": 1.8598,
"nll_loss": 1.8013783693313599,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16522985696792603,
"rewards/margins": 0.02055889368057251,
"rewards/rejected": -0.18578873574733734,
"step": 154
},
{
"epoch": 0.42802899551259926,
"grad_norm": 0.27581965923309326,
"learning_rate": 4.974032546418816e-06,
"log_odds_chosen": 0.42455926537513733,
"log_odds_ratio": -0.5114243030548096,
"logits/chosen": 0.2685522735118866,
"logits/rejected": -0.49730056524276733,
"logps/chosen": -1.6981408596038818,
"logps/rejected": -2.056694984436035,
"loss": 1.8954,
"nll_loss": 1.8442238569259644,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1698141098022461,
"rewards/margins": 0.035855405032634735,
"rewards/rejected": -0.20566949248313904,
"step": 155
},
{
"epoch": 0.4307904729030031,
"grad_norm": 0.2632681131362915,
"learning_rate": 4.9728677292023405e-06,
"log_odds_chosen": 0.17779600620269775,
"log_odds_ratio": -0.6146007776260376,
"logits/chosen": 0.3470558226108551,
"logits/rejected": -0.47237658500671387,
"logps/chosen": -1.8082042932510376,
"logps/rejected": -1.9588541984558105,
"loss": 1.9785,
"nll_loss": 1.9170793294906616,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18082045018672943,
"rewards/margins": 0.015064971521496773,
"rewards/rejected": -0.19588540494441986,
"step": 156
},
{
"epoch": 0.433551950293407,
"grad_norm": 0.25831714272499084,
"learning_rate": 4.971677499443882e-06,
"log_odds_chosen": 0.30535322427749634,
"log_odds_ratio": -0.5562926530838013,
"logits/chosen": 0.22543753683567047,
"logits/rejected": -0.5224329829216003,
"logps/chosen": -1.7110515832901,
"logps/rejected": -1.9678313732147217,
"loss": 1.8946,
"nll_loss": 1.838951587677002,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17110514640808105,
"rewards/margins": 0.025677980855107307,
"rewards/rejected": -0.1967831254005432,
"step": 157
},
{
"epoch": 0.4363134276838108,
"grad_norm": 0.24926158785820007,
"learning_rate": 4.97046186937489e-06,
"log_odds_chosen": 0.2828327715396881,
"log_odds_ratio": -0.5760643482208252,
"logits/chosen": 0.1486169695854187,
"logits/rejected": -0.5880769491195679,
"logps/chosen": -1.690523386001587,
"logps/rejected": -1.9291006326675415,
"loss": 1.8625,
"nll_loss": 1.804898977279663,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1690523475408554,
"rewards/margins": 0.023857703432440758,
"rewards/rejected": -0.1929100602865219,
"step": 158
},
{
"epoch": 0.4390749050742147,
"grad_norm": 0.26394572854042053,
"learning_rate": 4.9692208514878445e-06,
"log_odds_chosen": 0.21585100889205933,
"log_odds_ratio": -0.593828558921814,
"logits/chosen": 0.3207181692123413,
"logits/rejected": -0.3474537134170532,
"logps/chosen": -1.8141443729400635,
"logps/rejected": -1.9961789846420288,
"loss": 1.9958,
"nll_loss": 1.9364500045776367,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1814144253730774,
"rewards/margins": 0.01820346899330616,
"rewards/rejected": -0.1996179074048996,
"step": 159
},
{
"epoch": 0.4418363824646186,
"grad_norm": 0.26699379086494446,
"learning_rate": 4.967954458536126e-06,
"log_odds_chosen": 0.34597551822662354,
"log_odds_ratio": -0.5411776900291443,
"logits/chosen": 0.29670047760009766,
"logits/rejected": -0.25073421001434326,
"logps/chosen": -1.6443804502487183,
"logps/rejected": -1.9350707530975342,
"loss": 1.833,
"nll_loss": 1.7788795232772827,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16443803906440735,
"rewards/margins": 0.029069023206830025,
"rewards/rejected": -0.19350707530975342,
"step": 160
},
{
"epoch": 0.4445978598550224,
"grad_norm": 0.2796018719673157,
"learning_rate": 4.96666270353388e-06,
"log_odds_chosen": 0.3187919855117798,
"log_odds_ratio": -0.5501160025596619,
"logits/chosen": 0.24211375415325165,
"logits/rejected": -0.5377556681632996,
"logps/chosen": -1.704611897468567,
"logps/rejected": -1.9735724925994873,
"loss": 1.8668,
"nll_loss": 1.8117769956588745,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17046120762825012,
"rewards/margins": 0.02689606510102749,
"rewards/rejected": -0.19735725224018097,
"step": 161
},
{
"epoch": 0.4473593372454263,
"grad_norm": 0.2614901065826416,
"learning_rate": 4.965345599755888e-06,
"log_odds_chosen": 0.3870270848274231,
"log_odds_ratio": -0.5230943560600281,
"logits/chosen": 0.20484450459480286,
"logits/rejected": -0.6116334199905396,
"logps/chosen": -1.7666045427322388,
"logps/rejected": -2.0972108840942383,
"loss": 1.9248,
"nll_loss": 1.8725322484970093,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1766604632139206,
"rewards/margins": 0.03306063264608383,
"rewards/rejected": -0.20972107350826263,
"step": 162
},
{
"epoch": 0.45012081463583015,
"grad_norm": 0.253493994474411,
"learning_rate": 4.964003160737429e-06,
"log_odds_chosen": 0.35205185413360596,
"log_odds_ratio": -0.5439150333404541,
"logits/chosen": 0.24726349115371704,
"logits/rejected": -0.7317219972610474,
"logps/chosen": -1.6719884872436523,
"logps/rejected": -1.9640897512435913,
"loss": 1.8488,
"nll_loss": 1.7943804264068604,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16719885170459747,
"rewards/margins": 0.029210133478045464,
"rewards/rejected": -0.19640898704528809,
"step": 163
},
{
"epoch": 0.45288229202623403,
"grad_norm": 0.23761652410030365,
"learning_rate": 4.9626354002741424e-06,
"log_odds_chosen": 0.38894832134246826,
"log_odds_ratio": -0.5217959880828857,
"logits/chosen": 0.2658887505531311,
"logits/rejected": -0.6289904117584229,
"logps/chosen": -1.6253769397735596,
"logps/rejected": -1.9467687606811523,
"loss": 1.7998,
"nll_loss": 1.747636318206787,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16253769397735596,
"rewards/margins": 0.03213919699192047,
"rewards/rejected": -0.19467687606811523,
"step": 164
},
{
"epoch": 0.4556437694166379,
"grad_norm": 0.2694617509841919,
"learning_rate": 4.9612423324218816e-06,
"log_odds_chosen": 0.383113294839859,
"log_odds_ratio": -0.523147463798523,
"logits/chosen": 0.29493993520736694,
"logits/rejected": -0.8007028102874756,
"logps/chosen": -1.798284649848938,
"logps/rejected": -2.128204584121704,
"loss": 1.9598,
"nll_loss": 1.907509684562683,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1798284649848938,
"rewards/margins": 0.03299199044704437,
"rewards/rejected": -0.21282047033309937,
"step": 165
},
{
"epoch": 0.45840524680704176,
"grad_norm": 0.27424952387809753,
"learning_rate": 4.959823971496575e-06,
"log_odds_chosen": 0.3006168007850647,
"log_odds_ratio": -0.5665806531906128,
"logits/chosen": 0.21390017867088318,
"logits/rejected": -0.4561406373977661,
"logps/chosen": -1.7170822620391846,
"logps/rejected": -1.9759535789489746,
"loss": 1.8963,
"nll_loss": 1.8396110534667969,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.17170822620391846,
"rewards/margins": 0.025887131690979004,
"rewards/rejected": -0.19759535789489746,
"step": 166
},
{
"epoch": 0.46116672419744564,
"grad_norm": 0.28332701325416565,
"learning_rate": 4.958380332074074e-06,
"log_odds_chosen": 0.3598293364048004,
"log_odds_ratio": -0.5315272212028503,
"logits/chosen": 0.3352913558483124,
"logits/rejected": -0.34136950969696045,
"logps/chosen": -1.6895995140075684,
"logps/rejected": -1.9940197467803955,
"loss": 1.8818,
"nll_loss": 1.8286209106445312,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16895994544029236,
"rewards/margins": 0.030442016199231148,
"rewards/rejected": -0.19940195977687836,
"step": 167
},
{
"epoch": 0.4639282015878495,
"grad_norm": 0.2582882344722748,
"learning_rate": 4.95691142899001e-06,
"log_odds_chosen": 0.4415035843849182,
"log_odds_ratio": -0.5072202682495117,
"logits/chosen": 0.23933230340480804,
"logits/rejected": -0.5575604438781738,
"logps/chosen": -1.6840240955352783,
"logps/rejected": -2.0607402324676514,
"loss": 1.8592,
"nll_loss": 1.8084406852722168,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16840240359306335,
"rewards/margins": 0.037671614438295364,
"rewards/rejected": -0.20607402920722961,
"step": 168
},
{
"epoch": 0.46668967897825336,
"grad_norm": 0.2700425982475281,
"learning_rate": 4.955417277339633e-06,
"log_odds_chosen": 0.33803892135620117,
"log_odds_ratio": -0.54216468334198,
"logits/chosen": 0.2614552974700928,
"logits/rejected": -0.5953667163848877,
"logps/chosen": -1.7099006175994873,
"logps/rejected": -1.994602918624878,
"loss": 1.8657,
"nll_loss": 1.8114964962005615,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17099007964134216,
"rewards/margins": 0.028470242395997047,
"rewards/rejected": -0.19946029782295227,
"step": 169
},
{
"epoch": 0.46945115636865725,
"grad_norm": 0.2598535716533661,
"learning_rate": 4.953897892477664e-06,
"log_odds_chosen": 0.37775442004203796,
"log_odds_ratio": -0.5379059910774231,
"logits/chosen": 0.19298075139522552,
"logits/rejected": -0.7334005832672119,
"logps/chosen": -1.6427866220474243,
"logps/rejected": -1.9675416946411133,
"loss": 1.8149,
"nll_loss": 1.7611249685287476,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.16427867114543915,
"rewards/margins": 0.032475508749485016,
"rewards/rejected": -0.19675418734550476,
"step": 170
},
{
"epoch": 0.4722126337590611,
"grad_norm": 0.26744216680526733,
"learning_rate": 4.952353290018132e-06,
"log_odds_chosen": 0.37955784797668457,
"log_odds_ratio": -0.5295799374580383,
"logits/chosen": 0.27365049719810486,
"logits/rejected": -0.16527330875396729,
"logps/chosen": -1.7106455564498901,
"logps/rejected": -2.0328187942504883,
"loss": 1.8951,
"nll_loss": 1.842104196548462,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.171064555644989,
"rewards/margins": 0.032217323780059814,
"rewards/rejected": -0.20328189432621002,
"step": 171
},
{
"epoch": 0.474974111149465,
"grad_norm": 0.298951655626297,
"learning_rate": 4.950783485834218e-06,
"log_odds_chosen": 0.4051695168018341,
"log_odds_ratio": -0.5118554830551147,
"logits/chosen": 0.23584626615047455,
"logits/rejected": -0.5780523419380188,
"logps/chosen": -1.6387077569961548,
"logps/rejected": -1.975178599357605,
"loss": 1.8228,
"nll_loss": 1.7716267108917236,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16387078166007996,
"rewards/margins": 0.0336470901966095,
"rewards/rejected": -0.19751787185668945,
"step": 172
},
{
"epoch": 0.4777355885398688,
"grad_norm": 0.23760944604873657,
"learning_rate": 4.949188496058089e-06,
"log_odds_chosen": 0.308282732963562,
"log_odds_ratio": -0.5617873668670654,
"logits/chosen": 0.2496112585067749,
"logits/rejected": -0.595442533493042,
"logps/chosen": -1.6498539447784424,
"logps/rejected": -1.9084365367889404,
"loss": 1.8249,
"nll_loss": 1.768703818321228,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16498540341854095,
"rewards/margins": 0.02585826814174652,
"rewards/rejected": -0.19084367156028748,
"step": 173
},
{
"epoch": 0.4804970659302727,
"grad_norm": 0.27792298793792725,
"learning_rate": 4.947568337080733e-06,
"log_odds_chosen": 0.12510845065116882,
"log_odds_ratio": -0.6408029198646545,
"logits/chosen": 0.19703012704849243,
"logits/rejected": -0.5250687599182129,
"logps/chosen": -1.7701200246810913,
"logps/rejected": -1.877974033355713,
"loss": 1.9402,
"nll_loss": 1.876126766204834,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.17701202630996704,
"rewards/margins": 0.010785380378365517,
"rewards/rejected": -0.1877973973751068,
"step": 174
},
{
"epoch": 0.4832585433206766,
"grad_norm": 0.26834768056869507,
"learning_rate": 4.945923025551789e-06,
"log_odds_chosen": 0.3476155996322632,
"log_odds_ratio": -0.535990834236145,
"logits/chosen": 0.25689467787742615,
"logits/rejected": -0.616847813129425,
"logps/chosen": -1.7164323329925537,
"logps/rejected": -2.0102245807647705,
"loss": 1.8654,
"nll_loss": 1.8118302822113037,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1716432273387909,
"rewards/margins": 0.029379239305853844,
"rewards/rejected": -0.2010224610567093,
"step": 175
},
{
"epoch": 0.4860200207110804,
"grad_norm": 0.2374398410320282,
"learning_rate": 4.944252578379379e-06,
"log_odds_chosen": 0.2779533267021179,
"log_odds_ratio": -0.5700355172157288,
"logits/chosen": 0.19578704237937927,
"logits/rejected": -0.759753942489624,
"logps/chosen": -1.6441093683242798,
"logps/rejected": -1.8757362365722656,
"loss": 1.8093,
"nll_loss": 1.752274990081787,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16441094875335693,
"rewards/margins": 0.023162685334682465,
"rewards/rejected": -0.1875736117362976,
"step": 176
},
{
"epoch": 0.4887814981014843,
"grad_norm": 0.259815514087677,
"learning_rate": 4.942557012729933e-06,
"log_odds_chosen": 0.3541259169578552,
"log_odds_ratio": -0.5466777086257935,
"logits/chosen": 0.26096147298812866,
"logits/rejected": -0.6699164509773254,
"logps/chosen": -1.7012748718261719,
"logps/rejected": -1.9999008178710938,
"loss": 1.8742,
"nll_loss": 1.819519281387329,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1701274961233139,
"rewards/margins": 0.02986258640885353,
"rewards/rejected": -0.19999009370803833,
"step": 177
},
{
"epoch": 0.49154297549188813,
"grad_norm": 0.2553425133228302,
"learning_rate": 4.940836346028011e-06,
"log_odds_chosen": 0.5353389978408813,
"log_odds_ratio": -0.46747249364852905,
"logits/chosen": 0.19104404747486115,
"logits/rejected": -0.7688809633255005,
"logps/chosen": -1.6563342809677124,
"logps/rejected": -2.104980945587158,
"loss": 1.8241,
"nll_loss": 1.7773433923721313,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.165633425116539,
"rewards/margins": 0.04486468806862831,
"rewards/rejected": -0.21049810945987701,
"step": 178
},
{
"epoch": 0.494304452882292,
"grad_norm": 0.2521730661392212,
"learning_rate": 4.9390905959561254e-06,
"log_odds_chosen": 0.37528765201568604,
"log_odds_ratio": -0.5245440602302551,
"logits/chosen": 0.32407164573669434,
"logits/rejected": -0.7876682281494141,
"logps/chosen": -1.7645964622497559,
"logps/rejected": -2.0821878910064697,
"loss": 1.9004,
"nll_loss": 1.847988247871399,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1764596402645111,
"rewards/margins": 0.03175915777683258,
"rewards/rejected": -0.2082187980413437,
"step": 179
},
{
"epoch": 0.4970659302726959,
"grad_norm": 0.263175904750824,
"learning_rate": 4.937319780454559e-06,
"log_odds_chosen": 0.2128462940454483,
"log_odds_ratio": -0.5969148874282837,
"logits/chosen": 0.281113862991333,
"logits/rejected": -0.6618841290473938,
"logps/chosen": -1.717787504196167,
"logps/rejected": -1.8978548049926758,
"loss": 1.8852,
"nll_loss": 1.8254907131195068,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.17177876830101013,
"rewards/margins": 0.018006734549999237,
"rewards/rejected": -0.18978549540042877,
"step": 180
},
{
"epoch": 0.49982740766309974,
"grad_norm": 0.2627811133861542,
"learning_rate": 4.935523917721182e-06,
"log_odds_chosen": 0.32824423909187317,
"log_odds_ratio": -0.5476853847503662,
"logits/chosen": 0.3165915608406067,
"logits/rejected": -0.5085580945014954,
"logps/chosen": -1.7242554426193237,
"logps/rejected": -1.9997884035110474,
"loss": 1.9077,
"nll_loss": 1.852901816368103,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1724255532026291,
"rewards/margins": 0.027553284540772438,
"rewards/rejected": -0.19997884333133698,
"step": 181
},
{
"epoch": 0.5025888850535036,
"grad_norm": 0.2430126965045929,
"learning_rate": 4.933703026211262e-06,
"log_odds_chosen": 0.37905246019363403,
"log_odds_ratio": -0.528121292591095,
"logits/chosen": 0.3522907495498657,
"logits/rejected": -0.9080434441566467,
"logps/chosen": -1.7351850271224976,
"logps/rejected": -2.059718608856201,
"loss": 1.8866,
"nll_loss": 1.8337947130203247,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17351850867271423,
"rewards/margins": 0.03245338052511215,
"rewards/rejected": -0.2059718817472458,
"step": 182
},
{
"epoch": 0.5053503624439075,
"grad_norm": 0.2511756420135498,
"learning_rate": 4.931857124637276e-06,
"log_odds_chosen": 0.3801102638244629,
"log_odds_ratio": -0.5302396416664124,
"logits/chosen": 0.29899683594703674,
"logits/rejected": -0.5272728204727173,
"logps/chosen": -1.6140385866165161,
"logps/rejected": -1.9330276250839233,
"loss": 1.7807,
"nll_loss": 1.7276661396026611,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16140387952327728,
"rewards/margins": 0.031898900866508484,
"rewards/rejected": -0.19330278038978577,
"step": 183
},
{
"epoch": 0.5081118398343114,
"grad_norm": 0.33855873346328735,
"learning_rate": 4.92998623196872e-06,
"log_odds_chosen": 0.3525073528289795,
"log_odds_ratio": -0.5346016883850098,
"logits/chosen": 0.18561817705631256,
"logits/rejected": -0.9478727579116821,
"logps/chosen": -1.6743947267532349,
"logps/rejected": -1.9694938659667969,
"loss": 1.8278,
"nll_loss": 1.7743027210235596,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16743947565555573,
"rewards/margins": 0.029509922489523888,
"rewards/rejected": -0.19694939255714417,
"step": 184
},
{
"epoch": 0.5108733172247152,
"grad_norm": 0.24746204912662506,
"learning_rate": 4.92809036743191e-06,
"log_odds_chosen": 0.4261741042137146,
"log_odds_ratio": -0.5064104199409485,
"logits/chosen": 0.21586617827415466,
"logits/rejected": -0.7548010349273682,
"logps/chosen": -1.5876227617263794,
"logps/rejected": -1.9392573833465576,
"loss": 1.7364,
"nll_loss": 1.6857661008834839,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15876229107379913,
"rewards/margins": 0.035163454711437225,
"rewards/rejected": -0.19392573833465576,
"step": 185
},
{
"epoch": 0.5136347946151191,
"grad_norm": 0.256515771150589,
"learning_rate": 4.926169550509787e-06,
"log_odds_chosen": 0.2986356019973755,
"log_odds_ratio": -0.5593284964561462,
"logits/chosen": 0.3137073516845703,
"logits/rejected": -0.5803498029708862,
"logps/chosen": -1.7011560201644897,
"logps/rejected": -1.9512310028076172,
"loss": 1.8413,
"nll_loss": 1.7853296995162964,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1701156049966812,
"rewards/margins": 0.025007493793964386,
"rewards/rejected": -0.1951231062412262,
"step": 186
},
{
"epoch": 0.516396272005523,
"grad_norm": 0.26078805327415466,
"learning_rate": 4.924223800941718e-06,
"log_odds_chosen": 0.17883820831775665,
"log_odds_ratio": -0.6130943894386292,
"logits/chosen": 0.2678496241569519,
"logits/rejected": -0.6333581209182739,
"logps/chosen": -1.6809213161468506,
"logps/rejected": -1.829279899597168,
"loss": 1.8416,
"nll_loss": 1.7803112268447876,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.16809213161468506,
"rewards/margins": 0.014835860580205917,
"rewards/rejected": -0.18292801082134247,
"step": 187
},
{
"epoch": 0.5191577493959268,
"grad_norm": 0.3295902609825134,
"learning_rate": 4.9222531387232885e-06,
"log_odds_chosen": 0.2605365216732025,
"log_odds_ratio": -0.5762373208999634,
"logits/chosen": 0.3522469103336334,
"logits/rejected": -0.6649764180183411,
"logps/chosen": -1.7675739526748657,
"logps/rejected": -1.9888498783111572,
"loss": 1.9141,
"nll_loss": 1.8564814329147339,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17675741016864777,
"rewards/margins": 0.022127564996480942,
"rewards/rejected": -0.1988849639892578,
"step": 188
},
{
"epoch": 0.5219192267863307,
"grad_norm": 0.2640776038169861,
"learning_rate": 4.920257584106104e-06,
"log_odds_chosen": 0.26505714654922485,
"log_odds_ratio": -0.5721904039382935,
"logits/chosen": 0.2708177864551544,
"logits/rejected": -0.7849910855293274,
"logps/chosen": -1.7207906246185303,
"logps/rejected": -1.9439318180084229,
"loss": 1.8705,
"nll_loss": 1.8132869005203247,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17207907140254974,
"rewards/margins": 0.022314125671982765,
"rewards/rejected": -0.19439318776130676,
"step": 189
},
{
"epoch": 0.5246807041767345,
"grad_norm": 0.2825080156326294,
"learning_rate": 4.918237157597574e-06,
"log_odds_chosen": 0.343722403049469,
"log_odds_ratio": -0.5388997197151184,
"logits/chosen": 0.17086902260780334,
"logits/rejected": -0.6501764059066772,
"logps/chosen": -1.6524600982666016,
"logps/rejected": -1.9383505582809448,
"loss": 1.8225,
"nll_loss": 1.768616795539856,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16524602472782135,
"rewards/margins": 0.0285890344530344,
"rewards/rejected": -0.1938350349664688,
"step": 190
},
{
"epoch": 0.5274421815671384,
"grad_norm": 0.27104640007019043,
"learning_rate": 4.916191879960708e-06,
"log_odds_chosen": 0.38741055130958557,
"log_odds_ratio": -0.5215877294540405,
"logits/chosen": 0.2510019838809967,
"logits/rejected": -0.7692076563835144,
"logps/chosen": -1.7732622623443604,
"logps/rejected": -2.102111577987671,
"loss": 1.9158,
"nll_loss": 1.8636667728424072,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17732621729373932,
"rewards/margins": 0.03288493677973747,
"rewards/rejected": -0.2102111577987671,
"step": 191
},
{
"epoch": 0.5302036589575423,
"grad_norm": 0.23343509435653687,
"learning_rate": 4.914121772213898e-06,
"log_odds_chosen": 0.36985254287719727,
"log_odds_ratio": -0.5267210006713867,
"logits/chosen": 0.28792741894721985,
"logits/rejected": -0.6279172301292419,
"logps/chosen": -1.6510688066482544,
"logps/rejected": -1.9592763185501099,
"loss": 1.7875,
"nll_loss": 1.7348387241363525,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1651068925857544,
"rewards/margins": 0.03082074038684368,
"rewards/rejected": -0.19592763483524323,
"step": 192
},
{
"epoch": 0.5329651363479462,
"grad_norm": 0.2630840837955475,
"learning_rate": 4.912026855630703e-06,
"log_odds_chosen": 0.4057270884513855,
"log_odds_ratio": -0.5138529539108276,
"logits/chosen": 0.2938230633735657,
"logits/rejected": -0.5338449478149414,
"logps/chosen": -1.6583514213562012,
"logps/rejected": -1.9981603622436523,
"loss": 1.804,
"nll_loss": 1.7526516914367676,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1658351570367813,
"rewards/margins": 0.03398089110851288,
"rewards/rejected": -0.1998160481452942,
"step": 193
},
{
"epoch": 0.5357266137383501,
"grad_norm": 0.32278531789779663,
"learning_rate": 4.909907151739634e-06,
"log_odds_chosen": 0.36290958523750305,
"log_odds_ratio": -0.5369880795478821,
"logits/chosen": 0.2148250937461853,
"logits/rejected": -0.6797182559967041,
"logps/chosen": -1.6431670188903809,
"logps/rejected": -1.9445240497589111,
"loss": 1.7982,
"nll_loss": 1.7444565296173096,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16431671380996704,
"rewards/margins": 0.030135709792375565,
"rewards/rejected": -0.1944524198770523,
"step": 194
},
{
"epoch": 0.5384880911287538,
"grad_norm": 0.27246755361557007,
"learning_rate": 4.907762682323926e-06,
"log_odds_chosen": 0.5421338081359863,
"log_odds_ratio": -0.5049780011177063,
"logits/chosen": 0.2726617455482483,
"logits/rejected": -0.6698200702667236,
"logps/chosen": -1.649804711341858,
"logps/rejected": -2.089289426803589,
"loss": 1.8247,
"nll_loss": 1.774214506149292,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1649804711341858,
"rewards/margins": 0.04394847899675369,
"rewards/rejected": -0.2089289426803589,
"step": 195
},
{
"epoch": 0.5412495685191577,
"grad_norm": 0.25183457136154175,
"learning_rate": 4.905593469421323e-06,
"log_odds_chosen": 0.23532448709011078,
"log_odds_ratio": -0.5886967778205872,
"logits/chosen": 0.26476889848709106,
"logits/rejected": -1.049080729484558,
"logps/chosen": -1.6535186767578125,
"logps/rejected": -1.8511910438537598,
"loss": 1.8103,
"nll_loss": 1.7514796257019043,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.16535183787345886,
"rewards/margins": 0.01976725272834301,
"rewards/rejected": -0.18511910736560822,
"step": 196
},
{
"epoch": 0.5440110459095616,
"grad_norm": 0.4395514130592346,
"learning_rate": 4.90339953532384e-06,
"log_odds_chosen": 0.3145284354686737,
"log_odds_ratio": -0.5530616641044617,
"logits/chosen": 0.2953311502933502,
"logits/rejected": -0.7025068402290344,
"logps/chosen": -1.677915096282959,
"logps/rejected": -1.9411128759384155,
"loss": 1.8327,
"nll_loss": 1.7774040699005127,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16779151558876038,
"rewards/margins": 0.02631976269185543,
"rewards/rejected": -0.19411127269268036,
"step": 197
},
{
"epoch": 0.5467725232999655,
"grad_norm": 0.34831514954566956,
"learning_rate": 4.901180902577549e-06,
"log_odds_chosen": 0.3553253412246704,
"log_odds_ratio": -0.5363969206809998,
"logits/chosen": 0.32768774032592773,
"logits/rejected": -0.8499129414558411,
"logps/chosen": -1.6655187606811523,
"logps/rejected": -1.9629868268966675,
"loss": 1.8077,
"nll_loss": 1.7540751695632935,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1665518581867218,
"rewards/margins": 0.02974681742489338,
"rewards/rejected": -0.19629870355129242,
"step": 198
},
{
"epoch": 0.5495340006903694,
"grad_norm": 0.2638852894306183,
"learning_rate": 4.8989375939823305e-06,
"log_odds_chosen": 0.2459593415260315,
"log_odds_ratio": -0.5942656993865967,
"logits/chosen": 0.28430262207984924,
"logits/rejected": -0.9709011316299438,
"logps/chosen": -1.6272318363189697,
"logps/rejected": -1.8355400562286377,
"loss": 1.7824,
"nll_loss": 1.7229478359222412,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16272318363189697,
"rewards/margins": 0.020830810070037842,
"rewards/rejected": -0.183554008603096,
"step": 199
},
{
"epoch": 0.5522954780807732,
"grad_norm": 0.25321170687675476,
"learning_rate": 4.896669632591652e-06,
"log_odds_chosen": 0.4053685665130615,
"log_odds_ratio": -0.5148746371269226,
"logits/chosen": 0.26351821422576904,
"logits/rejected": -0.6612182259559631,
"logps/chosen": -1.6538773775100708,
"logps/rejected": -1.9941370487213135,
"loss": 1.8067,
"nll_loss": 1.7551857233047485,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16538773477077484,
"rewards/margins": 0.03402596712112427,
"rewards/rejected": -0.1994137167930603,
"step": 200
},
{
"epoch": 0.5550569554711771,
"grad_norm": 0.25087103247642517,
"learning_rate": 4.894377041712327e-06,
"log_odds_chosen": 0.2785293161869049,
"log_odds_ratio": -0.5666120052337646,
"logits/chosen": 0.31340447068214417,
"logits/rejected": -0.8899152278900146,
"logps/chosen": -1.7040454149246216,
"logps/rejected": -1.9381954669952393,
"loss": 1.8502,
"nll_loss": 1.7935274839401245,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17040453851222992,
"rewards/margins": 0.023415017873048782,
"rewards/rejected": -0.1938195526599884,
"step": 201
},
{
"epoch": 0.557818432861581,
"grad_norm": 0.25045764446258545,
"learning_rate": 4.892059844904273e-06,
"log_odds_chosen": 0.3241802752017975,
"log_odds_ratio": -0.5491434335708618,
"logits/chosen": 0.21476463973522186,
"logits/rejected": -0.7644988894462585,
"logps/chosen": -1.6328078508377075,
"logps/rejected": -1.9039433002471924,
"loss": 1.7911,
"nll_loss": 1.7361786365509033,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16328078508377075,
"rewards/margins": 0.027113551273941994,
"rewards/rejected": -0.190394327044487,
"step": 202
},
{
"epoch": 0.5605799102519848,
"grad_norm": 0.23429127037525177,
"learning_rate": 4.889718065980272e-06,
"log_odds_chosen": 0.3891683518886566,
"log_odds_ratio": -0.5242507457733154,
"logits/chosen": 0.25736698508262634,
"logits/rejected": -0.9445868730545044,
"logps/chosen": -1.7043213844299316,
"logps/rejected": -2.0325913429260254,
"loss": 1.8321,
"nll_loss": 1.7796692848205566,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17043213546276093,
"rewards/margins": 0.032826997339725494,
"rewards/rejected": -0.20325914025306702,
"step": 203
},
{
"epoch": 0.5633413876423887,
"grad_norm": 0.23629231750965118,
"learning_rate": 4.8873517290057265e-06,
"log_odds_chosen": 0.5146141052246094,
"log_odds_ratio": -0.470738023519516,
"logits/chosen": 0.21375544369220734,
"logits/rejected": -0.9923663139343262,
"logps/chosen": -1.6708019971847534,
"logps/rejected": -2.108680248260498,
"loss": 1.8157,
"nll_loss": 1.7685816287994385,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16708020865917206,
"rewards/margins": 0.04378781467676163,
"rewards/rejected": -0.2108680158853531,
"step": 204
},
{
"epoch": 0.5661028650327925,
"grad_norm": 0.2618404030799866,
"learning_rate": 4.88496085829841e-06,
"log_odds_chosen": 0.3119449019432068,
"log_odds_ratio": -0.5510252714157104,
"logits/chosen": 0.24412189424037933,
"logits/rejected": -0.44529297947883606,
"logps/chosen": -1.6430044174194336,
"logps/rejected": -1.902493953704834,
"loss": 1.787,
"nll_loss": 1.73191237449646,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16430042684078217,
"rewards/margins": 0.02594897150993347,
"rewards/rejected": -0.19024939835071564,
"step": 205
},
{
"epoch": 0.5688643424231964,
"grad_norm": 0.2525455057621002,
"learning_rate": 4.882545478428219e-06,
"log_odds_chosen": 0.4682312607765198,
"log_odds_ratio": -0.5022274851799011,
"logits/chosen": 0.31566041707992554,
"logits/rejected": -0.9525866508483887,
"logps/chosen": -1.62907075881958,
"logps/rejected": -2.020961046218872,
"loss": 1.7875,
"nll_loss": 1.7372711896896362,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16290709376335144,
"rewards/margins": 0.039189018309116364,
"rewards/rejected": -0.2020961046218872,
"step": 206
},
{
"epoch": 0.5716258198136003,
"grad_norm": 0.29324284195899963,
"learning_rate": 4.880105614216917e-06,
"log_odds_chosen": 0.3941385746002197,
"log_odds_ratio": -0.5201365947723389,
"logits/chosen": 0.3187817335128784,
"logits/rejected": -0.8300870656967163,
"logps/chosen": -1.800098180770874,
"logps/rejected": -2.1378254890441895,
"loss": 1.9198,
"nll_loss": 1.8678245544433594,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18000982701778412,
"rewards/margins": 0.033772725611925125,
"rewards/rejected": -0.21378254890441895,
"step": 207
},
{
"epoch": 0.5743872972040042,
"grad_norm": 0.2698252499103546,
"learning_rate": 4.8776412907378845e-06,
"log_odds_chosen": 0.32584843039512634,
"log_odds_ratio": -0.5465385913848877,
"logits/chosen": 0.34170591831207275,
"logits/rejected": -0.9024736285209656,
"logps/chosen": -1.7152948379516602,
"logps/rejected": -1.9904066324234009,
"loss": 1.8523,
"nll_loss": 1.7976171970367432,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17152947187423706,
"rewards/margins": 0.02751118130981922,
"rewards/rejected": -0.19904065132141113,
"step": 208
},
{
"epoch": 0.577148774594408,
"grad_norm": 0.24523906409740448,
"learning_rate": 4.875152533315859e-06,
"log_odds_chosen": 0.3781871199607849,
"log_odds_ratio": -0.5249034762382507,
"logits/chosen": 0.22711794078350067,
"logits/rejected": -0.8147018551826477,
"logps/chosen": -1.5907987356185913,
"logps/rejected": -1.901723861694336,
"loss": 1.7362,
"nll_loss": 1.6837434768676758,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15907986462116241,
"rewards/margins": 0.031092505902051926,
"rewards/rejected": -0.19017238914966583,
"step": 209
},
{
"epoch": 0.5799102519848118,
"grad_norm": 0.25919055938720703,
"learning_rate": 4.872639367526672e-06,
"log_odds_chosen": 0.39655864238739014,
"log_odds_ratio": -0.5207105875015259,
"logits/chosen": 0.26906242966651917,
"logits/rejected": -0.7278945446014404,
"logps/chosen": -1.6198142766952515,
"logps/rejected": -1.9519634246826172,
"loss": 1.7641,
"nll_loss": 1.7119861841201782,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16198144853115082,
"rewards/margins": 0.03321490436792374,
"rewards/rejected": -0.19519636034965515,
"step": 210
},
{
"epoch": 0.5826717293752157,
"grad_norm": 0.2771226763725281,
"learning_rate": 4.870101819196992e-06,
"log_odds_chosen": 0.33355456590652466,
"log_odds_ratio": -0.5614390969276428,
"logits/chosen": 0.23215994238853455,
"logits/rejected": -0.7148041129112244,
"logps/chosen": -1.6759493350982666,
"logps/rejected": -1.9613184928894043,
"loss": 1.8249,
"nll_loss": 1.7687922716140747,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16759493947029114,
"rewards/margins": 0.028536932542920113,
"rewards/rejected": -0.1961318552494049,
"step": 211
},
{
"epoch": 0.5854332067656196,
"grad_norm": 0.24981357157230377,
"learning_rate": 4.8675399144040535e-06,
"log_odds_chosen": 0.44918757677078247,
"log_odds_ratio": -0.5007237792015076,
"logits/chosen": 0.2401760071516037,
"logits/rejected": -1.0861955881118774,
"logps/chosen": -1.5890363454818726,
"logps/rejected": -1.9614393711090088,
"loss": 1.7223,
"nll_loss": 1.6722639799118042,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15890364348888397,
"rewards/margins": 0.03724028915166855,
"rewards/rejected": -0.19614392518997192,
"step": 212
},
{
"epoch": 0.5881946841560235,
"grad_norm": 0.2642802894115448,
"learning_rate": 4.864953679475392e-06,
"log_odds_chosen": 0.35757988691329956,
"log_odds_ratio": -0.5399623513221741,
"logits/chosen": 0.2965427339076996,
"logits/rejected": -1.087183952331543,
"logps/chosen": -1.645439863204956,
"logps/rejected": -1.9395672082901,
"loss": 1.7654,
"nll_loss": 1.7114168405532837,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1645439863204956,
"rewards/margins": 0.02941274270415306,
"rewards/rejected": -0.19395673274993896,
"step": 213
},
{
"epoch": 0.5909561615464274,
"grad_norm": 0.3119313418865204,
"learning_rate": 4.862343140988573e-06,
"log_odds_chosen": 0.3508715331554413,
"log_odds_ratio": -0.5419243574142456,
"logits/chosen": 0.2440934181213379,
"logits/rejected": -0.9715545773506165,
"logps/chosen": -1.691695213317871,
"logps/rejected": -1.9900875091552734,
"loss": 1.8129,
"nll_loss": 1.75870680809021,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16916951537132263,
"rewards/margins": 0.029839247465133667,
"rewards/rejected": -0.1990087330341339,
"step": 214
},
{
"epoch": 0.5937176389368312,
"grad_norm": 0.2919383943080902,
"learning_rate": 4.859708325770919e-06,
"log_odds_chosen": 0.32950735092163086,
"log_odds_ratio": -0.5485372543334961,
"logits/chosen": 0.314483106136322,
"logits/rejected": -0.6885538697242737,
"logps/chosen": -1.6899092197418213,
"logps/rejected": -1.96530020236969,
"loss": 1.8463,
"nll_loss": 1.791460633277893,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16899092495441437,
"rewards/margins": 0.02753911167383194,
"rewards/rejected": -0.1965300291776657,
"step": 215
},
{
"epoch": 0.596479116327235,
"grad_norm": 0.28394845128059387,
"learning_rate": 4.857049260899233e-06,
"log_odds_chosen": 0.37926068902015686,
"log_odds_ratio": -0.5276373624801636,
"logits/chosen": 0.21555066108703613,
"logits/rejected": -0.8712125420570374,
"logps/chosen": -1.565875768661499,
"logps/rejected": -1.873572826385498,
"loss": 1.7163,
"nll_loss": 1.6635123491287231,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15658757090568542,
"rewards/margins": 0.030769716948270798,
"rewards/rejected": -0.18735727667808533,
"step": 216
},
{
"epoch": 0.5992405937176389,
"grad_norm": 0.2573164105415344,
"learning_rate": 4.854365973699519e-06,
"log_odds_chosen": 0.26477721333503723,
"log_odds_ratio": -0.5851221680641174,
"logits/chosen": 0.2476685345172882,
"logits/rejected": -0.7407121658325195,
"logps/chosen": -1.657826542854309,
"logps/rejected": -1.8721026182174683,
"loss": 1.8129,
"nll_loss": 1.7544105052947998,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.16578267514705658,
"rewards/margins": 0.021427594125270844,
"rewards/rejected": -0.18721026182174683,
"step": 217
},
{
"epoch": 0.6020020711080428,
"grad_norm": 0.2776831388473511,
"learning_rate": 4.851658491746707e-06,
"log_odds_chosen": 0.3577430248260498,
"log_odds_ratio": -0.5342705249786377,
"logits/chosen": 0.17239025235176086,
"logits/rejected": -0.9049993753433228,
"logps/chosen": -1.6406420469284058,
"logps/rejected": -1.9406909942626953,
"loss": 1.773,
"nll_loss": 1.719523549079895,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1640642285346985,
"rewards/margins": 0.030004894360899925,
"rewards/rejected": -0.19406910240650177,
"step": 218
},
{
"epoch": 0.6047635484984467,
"grad_norm": 0.2520454227924347,
"learning_rate": 4.848926842864361e-06,
"log_odds_chosen": 0.4466186463832855,
"log_odds_ratio": -0.4973817467689514,
"logits/chosen": 0.17908504605293274,
"logits/rejected": -1.1680846214294434,
"logps/chosen": -1.6161878108978271,
"logps/rejected": -1.989192247390747,
"loss": 1.7485,
"nll_loss": 1.6987885236740112,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16161878407001495,
"rewards/margins": 0.03730044513940811,
"rewards/rejected": -0.19891922175884247,
"step": 219
},
{
"epoch": 0.6075250258888505,
"grad_norm": 0.2567070722579956,
"learning_rate": 4.846171055124401e-06,
"log_odds_chosen": 0.28004616498947144,
"log_odds_ratio": -0.5727553963661194,
"logits/chosen": 0.2623513638973236,
"logits/rejected": -0.7589821815490723,
"logps/chosen": -1.692448377609253,
"logps/rejected": -1.9293184280395508,
"loss": 1.8107,
"nll_loss": 1.7533810138702393,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16924485564231873,
"rewards/margins": 0.023687003180384636,
"rewards/rejected": -0.19293184578418732,
"step": 220
},
{
"epoch": 0.6102865032792544,
"grad_norm": 0.2756041884422302,
"learning_rate": 4.843391156846811e-06,
"log_odds_chosen": 0.4713705778121948,
"log_odds_ratio": -0.4867982268333435,
"logits/chosen": 0.29524314403533936,
"logits/rejected": -0.8037136793136597,
"logps/chosen": -1.6270312070846558,
"logps/rejected": -2.019744396209717,
"loss": 1.7549,
"nll_loss": 1.7062143087387085,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16270311176776886,
"rewards/margins": 0.039271317422389984,
"rewards/rejected": -0.20197445154190063,
"step": 221
},
{
"epoch": 0.6130479806696583,
"grad_norm": 0.2521502375602722,
"learning_rate": 4.8405871765993435e-06,
"log_odds_chosen": 0.5144533514976501,
"log_odds_ratio": -0.478085458278656,
"logits/chosen": 0.1983458399772644,
"logits/rejected": -0.9259991645812988,
"logps/chosen": -1.681555151939392,
"logps/rejected": -2.111989974975586,
"loss": 1.8047,
"nll_loss": 1.7569189071655273,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1681555211544037,
"rewards/margins": 0.043043479323387146,
"rewards/rejected": -0.21119900047779083,
"step": 222
},
{
"epoch": 0.6158094580600622,
"grad_norm": 0.2711774706840515,
"learning_rate": 4.837759143197237e-06,
"log_odds_chosen": 0.29049259424209595,
"log_odds_ratio": -0.5618168711662292,
"logits/chosen": 0.18880629539489746,
"logits/rejected": -1.0936201810836792,
"logps/chosen": -1.6453745365142822,
"logps/rejected": -1.8913843631744385,
"loss": 1.7765,
"nll_loss": 1.7203593254089355,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1645374447107315,
"rewards/margins": 0.02460099384188652,
"rewards/rejected": -0.18913844227790833,
"step": 223
},
{
"epoch": 0.618570935450466,
"grad_norm": 0.2559034824371338,
"learning_rate": 4.834907085702909e-06,
"log_odds_chosen": 0.39943575859069824,
"log_odds_ratio": -0.516412615776062,
"logits/chosen": 0.22049731016159058,
"logits/rejected": -0.741055428981781,
"logps/chosen": -1.5809171199798584,
"logps/rejected": -1.9103319644927979,
"loss": 1.7174,
"nll_loss": 1.6657910346984863,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1580917090177536,
"rewards/margins": 0.032941486686468124,
"rewards/rejected": -0.19103318452835083,
"step": 224
},
{
"epoch": 0.6213324128408698,
"grad_norm": 0.2551354169845581,
"learning_rate": 4.832031033425663e-06,
"log_odds_chosen": 0.45995259284973145,
"log_odds_ratio": -0.4912871718406677,
"logits/chosen": 0.14968952536582947,
"logits/rejected": -1.1501446962356567,
"logps/chosen": -1.5380923748016357,
"logps/rejected": -1.914685845375061,
"loss": 1.6661,
"nll_loss": 1.6169726848602295,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15380924940109253,
"rewards/margins": 0.03765934333205223,
"rewards/rejected": -0.19146858155727386,
"step": 225
},
{
"epoch": 0.6240938902312737,
"grad_norm": 0.2664376199245453,
"learning_rate": 4.829131015921386e-06,
"log_odds_chosen": 0.3153410255908966,
"log_odds_ratio": -0.5497522950172424,
"logits/chosen": 0.353019654750824,
"logits/rejected": -0.7054316997528076,
"logps/chosen": -1.708608865737915,
"logps/rejected": -1.975217342376709,
"loss": 1.8398,
"nll_loss": 1.784870982170105,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1708608865737915,
"rewards/margins": 0.02666083350777626,
"rewards/rejected": -0.19752173125743866,
"step": 226
},
{
"epoch": 0.6268553676216776,
"grad_norm": 0.25821202993392944,
"learning_rate": 4.826207062992245e-06,
"log_odds_chosen": 0.44117486476898193,
"log_odds_ratio": -0.5039249062538147,
"logits/chosen": 0.22692835330963135,
"logits/rejected": -0.9342468976974487,
"logps/chosen": -1.7786613702774048,
"logps/rejected": -2.1593728065490723,
"loss": 1.8966,
"nll_loss": 1.846253752708435,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1778661459684372,
"rewards/margins": 0.038071148097515106,
"rewards/rejected": -0.2159372866153717,
"step": 227
},
{
"epoch": 0.6296168450120815,
"grad_norm": 0.28877097368240356,
"learning_rate": 4.82325920468638e-06,
"log_odds_chosen": 0.2263374626636505,
"log_odds_ratio": -0.5921541452407837,
"logits/chosen": 0.26566582918167114,
"logits/rejected": -0.8801894187927246,
"logps/chosen": -1.7435269355773926,
"logps/rejected": -1.9335991144180298,
"loss": 1.8735,
"nll_loss": 1.8142577409744263,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17435269057750702,
"rewards/margins": 0.01900722086429596,
"rewards/rejected": -0.19335991144180298,
"step": 228
},
{
"epoch": 0.6323783224024854,
"grad_norm": 0.286994069814682,
"learning_rate": 4.820287471297598e-06,
"log_odds_chosen": 0.34056615829467773,
"log_odds_ratio": -0.5455049872398376,
"logits/chosen": 0.15769648551940918,
"logits/rejected": -0.9189969897270203,
"logps/chosen": -1.612159252166748,
"logps/rejected": -1.8956931829452515,
"loss": 1.7497,
"nll_loss": 1.6951847076416016,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1612159162759781,
"rewards/margins": 0.028353393077850342,
"rewards/rejected": -0.18956929445266724,
"step": 229
},
{
"epoch": 0.6351397997928891,
"grad_norm": 0.25925716757774353,
"learning_rate": 4.817291893365055e-06,
"log_odds_chosen": 0.4459936022758484,
"log_odds_ratio": -0.50335294008255,
"logits/chosen": 0.03054118901491165,
"logits/rejected": -1.0507367849349976,
"logps/chosen": -1.4849401712417603,
"logps/rejected": -1.8426207304000854,
"loss": 1.6416,
"nll_loss": 1.591292142868042,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14849400520324707,
"rewards/margins": 0.035768061876297,
"rewards/rejected": -0.18426206707954407,
"step": 230
},
{
"epoch": 0.637901277183293,
"grad_norm": 0.2912357449531555,
"learning_rate": 4.81427250167295e-06,
"log_odds_chosen": 0.20216652750968933,
"log_odds_ratio": -0.6004579663276672,
"logits/chosen": 0.10540774464607239,
"logits/rejected": -1.1034070253372192,
"logps/chosen": -1.6417537927627563,
"logps/rejected": -1.8084138631820679,
"loss": 1.7758,
"nll_loss": 1.715754508972168,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1641753911972046,
"rewards/margins": 0.01666601374745369,
"rewards/rejected": -0.1808413863182068,
"step": 231
},
{
"epoch": 0.6406627545736969,
"grad_norm": 0.28148511052131653,
"learning_rate": 4.811229327250204e-06,
"log_odds_chosen": 0.40823695063591003,
"log_odds_ratio": -0.5194427967071533,
"logits/chosen": 0.19764533638954163,
"logits/rejected": -1.0543047189712524,
"logps/chosen": -1.68423330783844,
"logps/rejected": -2.0337371826171875,
"loss": 1.8128,
"nll_loss": 1.760822057723999,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1684233546257019,
"rewards/margins": 0.034950368106365204,
"rewards/rejected": -0.20337370038032532,
"step": 232
},
{
"epoch": 0.6434242319641008,
"grad_norm": 0.26084744930267334,
"learning_rate": 4.8081624013701435e-06,
"log_odds_chosen": 0.38514161109924316,
"log_odds_ratio": -0.5289287567138672,
"logits/chosen": 0.16184748709201813,
"logits/rejected": -1.2881039381027222,
"logps/chosen": -1.6542613506317139,
"logps/rejected": -1.9764081239700317,
"loss": 1.7775,
"nll_loss": 1.7246254682540894,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1654261350631714,
"rewards/margins": 0.03221467137336731,
"rewards/rejected": -0.1976408213376999,
"step": 233
},
{
"epoch": 0.6461857093545047,
"grad_norm": 0.26910293102264404,
"learning_rate": 4.805071755550177e-06,
"log_odds_chosen": 0.37266138195991516,
"log_odds_ratio": -0.5266379117965698,
"logits/chosen": 0.2432277947664261,
"logits/rejected": -1.2453206777572632,
"logps/chosen": -1.676849603652954,
"logps/rejected": -1.9903044700622559,
"loss": 1.7904,
"nll_loss": 1.7377511262893677,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16768495738506317,
"rewards/margins": 0.031345486640930176,
"rewards/rejected": -0.19903044402599335,
"step": 234
},
{
"epoch": 0.6489471867449085,
"grad_norm": 0.26596587896347046,
"learning_rate": 4.8019574215514705e-06,
"log_odds_chosen": 0.320311963558197,
"log_odds_ratio": -0.5477744340896606,
"logits/chosen": 0.1768263727426529,
"logits/rejected": -0.8458906412124634,
"logps/chosen": -1.6462191343307495,
"logps/rejected": -1.9113428592681885,
"loss": 1.7877,
"nll_loss": 1.7328777313232422,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16462190449237823,
"rewards/margins": 0.026512393727898598,
"rewards/rejected": -0.19113430380821228,
"step": 235
},
{
"epoch": 0.6517086641353124,
"grad_norm": 0.23039527237415314,
"learning_rate": 4.7988194313786275e-06,
"log_odds_chosen": 0.3400653004646301,
"log_odds_ratio": -0.5446557998657227,
"logits/chosen": 0.17479516565799713,
"logits/rejected": -0.9590482711791992,
"logps/chosen": -1.5938714742660522,
"logps/rejected": -1.8736517429351807,
"loss": 1.706,
"nll_loss": 1.6515262126922607,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15938714146614075,
"rewards/margins": 0.027978049591183662,
"rewards/rejected": -0.18736517429351807,
"step": 236
},
{
"epoch": 0.6544701415257163,
"grad_norm": 0.25597652792930603,
"learning_rate": 4.795657817279349e-06,
"log_odds_chosen": 0.35750117897987366,
"log_odds_ratio": -0.5432247519493103,
"logits/chosen": 0.10542917996644974,
"logits/rejected": -1.0094666481018066,
"logps/chosen": -1.5345871448516846,
"logps/rejected": -1.8278987407684326,
"loss": 1.6735,
"nll_loss": 1.619143009185791,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15345871448516846,
"rewards/margins": 0.02933114767074585,
"rewards/rejected": -0.1827898770570755,
"step": 237
},
{
"epoch": 0.6572316189161201,
"grad_norm": 0.27119508385658264,
"learning_rate": 4.7924726117441135e-06,
"log_odds_chosen": 0.4193371534347534,
"log_odds_ratio": -0.5230008363723755,
"logits/chosen": 0.08888234198093414,
"logits/rejected": -1.0837329626083374,
"logps/chosen": -1.6472194194793701,
"logps/rejected": -1.9990077018737793,
"loss": 1.7734,
"nll_loss": 1.7211425304412842,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16472195088863373,
"rewards/margins": 0.035178832709789276,
"rewards/rejected": -0.1999007910490036,
"step": 238
},
{
"epoch": 0.659993096306524,
"grad_norm": 0.26720085740089417,
"learning_rate": 4.789263847505835e-06,
"log_odds_chosen": 0.4661468267440796,
"log_odds_ratio": -0.49137797951698303,
"logits/chosen": 0.14774960279464722,
"logits/rejected": -0.9653093814849854,
"logps/chosen": -1.634422779083252,
"logps/rejected": -2.0227272510528564,
"loss": 1.7557,
"nll_loss": 1.706547737121582,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16344228386878967,
"rewards/margins": 0.038830459117889404,
"rewards/rejected": -0.20227274298667908,
"step": 239
},
{
"epoch": 0.6627545736969278,
"grad_norm": 0.26344889402389526,
"learning_rate": 4.786031557539532e-06,
"log_odds_chosen": 0.620901882648468,
"log_odds_ratio": -0.43870899081230164,
"logits/chosen": 0.1005333811044693,
"logits/rejected": -1.179240345954895,
"logps/chosen": -1.5813204050064087,
"logps/rejected": -2.10603666305542,
"loss": 1.7039,
"nll_loss": 1.66001558303833,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15813204646110535,
"rewards/margins": 0.05247163027524948,
"rewards/rejected": -0.21060366928577423,
"step": 240
},
{
"epoch": 0.6655160510873317,
"grad_norm": 0.2564525306224823,
"learning_rate": 4.782775775061983e-06,
"log_odds_chosen": 0.3930521011352539,
"log_odds_ratio": -0.5189710855484009,
"logits/chosen": 0.17282657325267792,
"logits/rejected": -0.9231711626052856,
"logps/chosen": -1.5853371620178223,
"logps/rejected": -1.9094042778015137,
"loss": 1.704,
"nll_loss": 1.6520962715148926,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1585337519645691,
"rewards/margins": 0.03240669518709183,
"rewards/rejected": -0.19094042479991913,
"step": 241
},
{
"epoch": 0.6682775284777356,
"grad_norm": 0.23007996380329132,
"learning_rate": 4.779496533531393e-06,
"log_odds_chosen": 0.49815472960472107,
"log_odds_ratio": -0.47664713859558105,
"logits/chosen": 0.16657163202762604,
"logits/rejected": -1.0556104183197021,
"logps/chosen": -1.5395145416259766,
"logps/rejected": -1.9480966329574585,
"loss": 1.6541,
"nll_loss": 1.6063947677612305,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15395145118236542,
"rewards/margins": 0.04085820913314819,
"rewards/rejected": -0.19480964541435242,
"step": 242
},
{
"epoch": 0.6710390058681395,
"grad_norm": 0.2536675035953522,
"learning_rate": 4.7761938666470405e-06,
"log_odds_chosen": 0.3562057316303253,
"log_odds_ratio": -0.5365646481513977,
"logits/chosen": 0.18961578607559204,
"logits/rejected": -1.129898190498352,
"logps/chosen": -1.6646479368209839,
"logps/rejected": -1.964529037475586,
"loss": 1.7902,
"nll_loss": 1.736580729484558,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16646480560302734,
"rewards/margins": 0.029988115653395653,
"rewards/rejected": -0.19645293056964874,
"step": 243
},
{
"epoch": 0.6738004832585434,
"grad_norm": 0.241102397441864,
"learning_rate": 4.7728678083489375e-06,
"log_odds_chosen": 0.287707656621933,
"log_odds_ratio": -0.5653382539749146,
"logits/chosen": 0.18390116095542908,
"logits/rejected": -1.0134061574935913,
"logps/chosen": -1.5824017524719238,
"logps/rejected": -1.8214263916015625,
"loss": 1.707,
"nll_loss": 1.6505118608474731,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1582401841878891,
"rewards/margins": 0.02390245348215103,
"rewards/rejected": -0.18214263021945953,
"step": 244
},
{
"epoch": 0.6765619606489471,
"grad_norm": 0.24698680639266968,
"learning_rate": 4.7695183928174804e-06,
"log_odds_chosen": 0.426491916179657,
"log_odds_ratio": -0.5090312361717224,
"logits/chosen": 0.09436272829771042,
"logits/rejected": -1.081894040107727,
"logps/chosen": -1.614341378211975,
"logps/rejected": -1.9695953130722046,
"loss": 1.7422,
"nll_loss": 1.6912728548049927,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.161434143781662,
"rewards/margins": 0.03552539646625519,
"rewards/rejected": -0.19695955514907837,
"step": 245
},
{
"epoch": 0.679323438039351,
"grad_norm": 0.25812989473342896,
"learning_rate": 4.766145654473096e-06,
"log_odds_chosen": 0.33238697052001953,
"log_odds_ratio": -0.5477334260940552,
"logits/chosen": 0.17383219301700592,
"logits/rejected": -1.1996134519577026,
"logps/chosen": -1.6853795051574707,
"logps/rejected": -1.9657583236694336,
"loss": 1.797,
"nll_loss": 1.7422523498535156,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1685379445552826,
"rewards/margins": 0.028037890791893005,
"rewards/rejected": -0.1965758353471756,
"step": 246
},
{
"epoch": 0.6820849154297549,
"grad_norm": 0.25600266456604004,
"learning_rate": 4.762749627975888e-06,
"log_odds_chosen": 0.3794538974761963,
"log_odds_ratio": -0.5263408422470093,
"logits/chosen": 0.07988805323839188,
"logits/rejected": -1.091023325920105,
"logps/chosen": -1.5263158082962036,
"logps/rejected": -1.8329474925994873,
"loss": 1.6541,
"nll_loss": 1.6014518737792969,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15263158082962036,
"rewards/margins": 0.030663155019283295,
"rewards/rejected": -0.18329472839832306,
"step": 247
},
{
"epoch": 0.6848463928201588,
"grad_norm": 0.24254651367664337,
"learning_rate": 4.7593303482252835e-06,
"log_odds_chosen": 0.34609463810920715,
"log_odds_ratio": -0.541566014289856,
"logits/chosen": 0.14708423614501953,
"logits/rejected": -0.7518667578697205,
"logps/chosen": -1.6179258823394775,
"logps/rejected": -1.905765414237976,
"loss": 1.7327,
"nll_loss": 1.6784985065460205,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1617926061153412,
"rewards/margins": 0.02878393605351448,
"rewards/rejected": -0.19057652354240417,
"step": 248
},
{
"epoch": 0.6876078702105627,
"grad_norm": 0.22665560245513916,
"learning_rate": 4.755887850359673e-06,
"log_odds_chosen": 0.5504661798477173,
"log_odds_ratio": -0.4719165861606598,
"logits/chosen": 0.09460557997226715,
"logits/rejected": -1.225687861442566,
"logps/chosen": -1.4793003797531128,
"logps/rejected": -1.9250984191894531,
"loss": 1.6118,
"nll_loss": 1.5646475553512573,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1479300558567047,
"rewards/margins": 0.04457978904247284,
"rewards/rejected": -0.19250985980033875,
"step": 249
},
{
"epoch": 0.6903693476009665,
"grad_norm": 0.2510414719581604,
"learning_rate": 4.752422169756048e-06,
"log_odds_chosen": 0.3988419771194458,
"log_odds_ratio": -0.5172514915466309,
"logits/chosen": 0.16052858531475067,
"logits/rejected": -1.2513781785964966,
"logps/chosen": -1.6576679944992065,
"logps/rejected": -1.9923875331878662,
"loss": 1.7851,
"nll_loss": 1.7334227561950684,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16576679050922394,
"rewards/margins": 0.033471960574388504,
"rewards/rejected": -0.19923876225948334,
"step": 250
},
{
"epoch": 0.6931308249913704,
"grad_norm": 0.2423536330461502,
"learning_rate": 4.748933342029639e-06,
"log_odds_chosen": 0.5282813310623169,
"log_odds_ratio": -0.4706994891166687,
"logits/chosen": 0.10225804150104523,
"logits/rejected": -1.2281630039215088,
"logps/chosen": -1.5280183553695679,
"logps/rejected": -1.9655184745788574,
"loss": 1.6654,
"nll_loss": 1.6183708906173706,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15280182659626007,
"rewards/margins": 0.04375001788139343,
"rewards/rejected": -0.1965518444776535,
"step": 251
},
{
"epoch": 0.6958923023817742,
"grad_norm": 0.23662064969539642,
"learning_rate": 4.745421403033548e-06,
"log_odds_chosen": 0.36522558331489563,
"log_odds_ratio": -0.5290880799293518,
"logits/chosen": 0.10924719274044037,
"logits/rejected": -1.0550241470336914,
"logps/chosen": -1.6013975143432617,
"logps/rejected": -1.901705026626587,
"loss": 1.7128,
"nll_loss": 1.6598960161209106,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1601397693157196,
"rewards/margins": 0.0300307534635067,
"rewards/rejected": -0.1901705265045166,
"step": 252
},
{
"epoch": 0.6986537797721781,
"grad_norm": 0.2725629508495331,
"learning_rate": 4.741886388858384e-06,
"log_odds_chosen": 0.31865060329437256,
"log_odds_ratio": -0.5531626343727112,
"logits/chosen": 0.15934640169143677,
"logits/rejected": -1.1095128059387207,
"logps/chosen": -1.6118751764297485,
"logps/rejected": -1.871192455291748,
"loss": 1.7355,
"nll_loss": 1.6801382303237915,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1611875295639038,
"rewards/margins": 0.02593172900378704,
"rewards/rejected": -0.187119260430336,
"step": 253
},
{
"epoch": 0.701415257162582,
"grad_norm": 0.2601236402988434,
"learning_rate": 4.738328335831883e-06,
"log_odds_chosen": 0.3496936559677124,
"log_odds_ratio": -0.5388615131378174,
"logits/chosen": 0.12353205680847168,
"logits/rejected": -1.3011374473571777,
"logps/chosen": -1.5815935134887695,
"logps/rejected": -1.8707093000411987,
"loss": 1.7024,
"nll_loss": 1.6484651565551758,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15815936028957367,
"rewards/margins": 0.028911583125591278,
"rewards/rejected": -0.18707093596458435,
"step": 254
},
{
"epoch": 0.7041767345529858,
"grad_norm": 0.2376219779253006,
"learning_rate": 4.734747280518549e-06,
"log_odds_chosen": 0.4198550879955292,
"log_odds_ratio": -0.5078388452529907,
"logits/chosen": 0.030534474179148674,
"logits/rejected": -1.399457573890686,
"logps/chosen": -1.6249712705612183,
"logps/rejected": -1.9735503196716309,
"loss": 1.7447,
"nll_loss": 1.693869948387146,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1624971330165863,
"rewards/margins": 0.034857891499996185,
"rewards/rejected": -0.19735503196716309,
"step": 255
},
{
"epoch": 0.7069382119433897,
"grad_norm": 0.23082926869392395,
"learning_rate": 4.7311432597192655e-06,
"log_odds_chosen": 0.368667870759964,
"log_odds_ratio": -0.5342724323272705,
"logits/chosen": 0.03557804599404335,
"logits/rejected": -1.3495283126831055,
"logps/chosen": -1.6255284547805786,
"logps/rejected": -1.9313668012619019,
"loss": 1.7347,
"nll_loss": 1.6812456846237183,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1625528484582901,
"rewards/margins": 0.030583834275603294,
"rewards/rejected": -0.19313669204711914,
"step": 256
},
{
"epoch": 0.7096996893337936,
"grad_norm": 0.2494223415851593,
"learning_rate": 4.72751631047092e-06,
"log_odds_chosen": 0.5292459726333618,
"log_odds_ratio": -0.4693402051925659,
"logits/chosen": -0.008031047880649567,
"logits/rejected": -1.0098199844360352,
"logps/chosen": -1.6526001691818237,
"logps/rejected": -2.0972676277160645,
"loss": 1.7807,
"nll_loss": 1.7337857484817505,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16526003181934357,
"rewards/margins": 0.044466737657785416,
"rewards/rejected": -0.2097267508506775,
"step": 257
},
{
"epoch": 0.7124611667241975,
"grad_norm": 0.24077369272708893,
"learning_rate": 4.72386647004603e-06,
"log_odds_chosen": 0.37760159373283386,
"log_odds_ratio": -0.5266671180725098,
"logits/chosen": 0.1133025735616684,
"logits/rejected": -1.0056575536727905,
"logps/chosen": -1.6364243030548096,
"logps/rejected": -1.9491901397705078,
"loss": 1.7621,
"nll_loss": 1.7094427347183228,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16364243626594543,
"rewards/margins": 0.03127656877040863,
"rewards/rejected": -0.19491900503635406,
"step": 258
},
{
"epoch": 0.7152226441146013,
"grad_norm": 0.24284473061561584,
"learning_rate": 4.720193775952352e-06,
"log_odds_chosen": 0.2508072853088379,
"log_odds_ratio": -0.5792805552482605,
"logits/chosen": 0.11287318915128708,
"logits/rejected": -1.0705785751342773,
"logps/chosen": -1.6278916597366333,
"logps/rejected": -1.833707332611084,
"loss": 1.7407,
"nll_loss": 1.6827445030212402,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16278916597366333,
"rewards/margins": 0.02058156207203865,
"rewards/rejected": -0.18337073922157288,
"step": 259
},
{
"epoch": 0.7179841215050051,
"grad_norm": 0.2317313551902771,
"learning_rate": 4.716498265932501e-06,
"log_odds_chosen": 0.5105437636375427,
"log_odds_ratio": -0.4731099009513855,
"logits/chosen": -0.00561100710183382,
"logits/rejected": -1.0285159349441528,
"logps/chosen": -1.4508588314056396,
"logps/rejected": -1.8631662130355835,
"loss": 1.5697,
"nll_loss": 1.5224125385284424,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1450859010219574,
"rewards/margins": 0.04123072326183319,
"rewards/rejected": -0.1863166093826294,
"step": 260
},
{
"epoch": 0.720745598895409,
"grad_norm": 0.25082987546920776,
"learning_rate": 4.712779977963559e-06,
"log_odds_chosen": 0.3830725848674774,
"log_odds_ratio": -0.5252867937088013,
"logits/chosen": 0.09645551443099976,
"logits/rejected": -0.733791172504425,
"logps/chosen": -1.56308114528656,
"logps/rejected": -1.8771851062774658,
"loss": 1.695,
"nll_loss": 1.642515778541565,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.156308114528656,
"rewards/margins": 0.031410399824380875,
"rewards/rejected": -0.18771851062774658,
"step": 261
},
{
"epoch": 0.7235070762858129,
"grad_norm": 0.22368477284908295,
"learning_rate": 4.7090389502566884e-06,
"log_odds_chosen": 0.48179319500923157,
"log_odds_ratio": -0.4859163165092468,
"logits/chosen": 0.042292360216379166,
"logits/rejected": -1.0789254903793335,
"logps/chosen": -1.5697331428527832,
"logps/rejected": -1.9673210382461548,
"loss": 1.6673,
"nll_loss": 1.61872136592865,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15697331726551056,
"rewards/margins": 0.039758771657943726,
"rewards/rejected": -0.19673210382461548,
"step": 262
},
{
"epoch": 0.7262685536762168,
"grad_norm": 0.22603513300418854,
"learning_rate": 4.705275221256738e-06,
"log_odds_chosen": 0.3837694227695465,
"log_odds_ratio": -0.5233887434005737,
"logits/chosen": 0.11829128116369247,
"logits/rejected": -1.0814515352249146,
"logps/chosen": -1.612618088722229,
"logps/rejected": -1.9323992729187012,
"loss": 1.7336,
"nll_loss": 1.6812975406646729,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16126182675361633,
"rewards/margins": 0.03197810798883438,
"rewards/rejected": -0.1932399421930313,
"step": 263
},
{
"epoch": 0.7290300310666207,
"grad_norm": 0.23297946155071259,
"learning_rate": 4.701488829641845e-06,
"log_odds_chosen": 0.3633509874343872,
"log_odds_ratio": -0.5302541255950928,
"logits/chosen": 0.08365479856729507,
"logits/rejected": -1.0365325212478638,
"logps/chosen": -1.5728099346160889,
"logps/rejected": -1.870044231414795,
"loss": 1.6965,
"nll_loss": 1.6434751749038696,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15728099644184113,
"rewards/margins": 0.02972342073917389,
"rewards/rejected": -0.18700441718101501,
"step": 264
},
{
"epoch": 0.7317915084570245,
"grad_norm": 0.215961292386055,
"learning_rate": 4.697679814323044e-06,
"log_odds_chosen": 0.35131096839904785,
"log_odds_ratio": -0.540593147277832,
"logits/chosen": 0.052778296172618866,
"logits/rejected": -1.2138348817825317,
"logps/chosen": -1.6011123657226562,
"logps/rejected": -1.8878912925720215,
"loss": 1.7054,
"nll_loss": 1.6513090133666992,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16011124849319458,
"rewards/margins": 0.028677886351943016,
"rewards/rejected": -0.18878912925720215,
"step": 265
},
{
"epoch": 0.7345529858474283,
"grad_norm": 0.2192918211221695,
"learning_rate": 4.693848214443858e-06,
"log_odds_chosen": 0.3489964008331299,
"log_odds_ratio": -0.538882315158844,
"logits/chosen": 0.0374111570417881,
"logits/rejected": -1.4135199785232544,
"logps/chosen": -1.66756010055542,
"logps/rejected": -1.9600346088409424,
"loss": 1.7678,
"nll_loss": 1.713918924331665,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1667560189962387,
"rewards/margins": 0.029247449710965157,
"rewards/rejected": -0.19600346684455872,
"step": 266
},
{
"epoch": 0.7373144632378322,
"grad_norm": 0.23613475263118744,
"learning_rate": 4.689994069379905e-06,
"log_odds_chosen": 0.6149622797966003,
"log_odds_ratio": -0.43481168150901794,
"logits/chosen": 0.08960134536027908,
"logits/rejected": -1.5427201986312866,
"logps/chosen": -1.5492388010025024,
"logps/rejected": -2.061069965362549,
"loss": 1.6539,
"nll_loss": 1.6104521751403809,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15492388606071472,
"rewards/margins": 0.05118309706449509,
"rewards/rejected": -0.2061069756746292,
"step": 267
},
{
"epoch": 0.7400759406282361,
"grad_norm": 0.22785541415214539,
"learning_rate": 4.686117418738489e-06,
"log_odds_chosen": 0.46340838074684143,
"log_odds_ratio": -0.4908517003059387,
"logits/chosen": 0.09018822759389877,
"logits/rejected": -1.188814640045166,
"logps/chosen": -1.607946753501892,
"logps/rejected": -1.9937546253204346,
"loss": 1.7214,
"nll_loss": 1.6723337173461914,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1607946753501892,
"rewards/margins": 0.03858078643679619,
"rewards/rejected": -0.1993754655122757,
"step": 268
},
{
"epoch": 0.74283741801864,
"grad_norm": 0.20759011805057526,
"learning_rate": 4.6822183023581945e-06,
"log_odds_chosen": 0.39548274874687195,
"log_odds_ratio": -0.5299872756004333,
"logits/chosen": 0.08991736173629761,
"logits/rejected": -1.5129450559616089,
"logps/chosen": -1.5805068016052246,
"logps/rejected": -1.9086334705352783,
"loss": 1.705,
"nll_loss": 1.6519887447357178,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15805068612098694,
"rewards/margins": 0.03281266242265701,
"rewards/rejected": -0.19086334109306335,
"step": 269
},
{
"epoch": 0.7455988954090439,
"grad_norm": 0.21977882087230682,
"learning_rate": 4.678296760308474e-06,
"log_odds_chosen": 0.2966475486755371,
"log_odds_ratio": -0.5605870485305786,
"logits/chosen": -0.026812348514795303,
"logits/rejected": -1.3540985584259033,
"logps/chosen": -1.5566442012786865,
"logps/rejected": -1.7983969449996948,
"loss": 1.6709,
"nll_loss": 1.6148244142532349,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15566441416740417,
"rewards/margins": 0.024175278842449188,
"rewards/rejected": -0.17983968555927277,
"step": 270
},
{
"epoch": 0.7483603727994477,
"grad_norm": 0.24105672538280487,
"learning_rate": 4.674352832889239e-06,
"log_odds_chosen": 0.5675957798957825,
"log_odds_ratio": -0.4524817168712616,
"logits/chosen": 0.017950108274817467,
"logits/rejected": -1.1157424449920654,
"logps/chosen": -1.617455005645752,
"logps/rejected": -2.094006299972534,
"loss": 1.7444,
"nll_loss": 1.6991225481033325,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16174551844596863,
"rewards/margins": 0.0476551279425621,
"rewards/rejected": -0.20940065383911133,
"step": 271
},
{
"epoch": 0.7511218501898516,
"grad_norm": 0.20193441212177277,
"learning_rate": 4.670386560630446e-06,
"log_odds_chosen": 0.4167816638946533,
"log_odds_ratio": -0.5149449110031128,
"logits/chosen": -0.06302893161773682,
"logits/rejected": -1.2852458953857422,
"logps/chosen": -1.538407325744629,
"logps/rejected": -1.8780553340911865,
"loss": 1.6523,
"nll_loss": 1.600759744644165,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15384072065353394,
"rewards/margins": 0.03396480157971382,
"rewards/rejected": -0.18780553340911865,
"step": 272
},
{
"epoch": 0.7538833275802554,
"grad_norm": 0.2135000228881836,
"learning_rate": 4.66639798429168e-06,
"log_odds_chosen": 0.5502100586891174,
"log_odds_ratio": -0.45903506875038147,
"logits/chosen": 0.02270820550620556,
"logits/rejected": -1.4955902099609375,
"logps/chosen": -1.6026809215545654,
"logps/rejected": -2.0662648677825928,
"loss": 1.6999,
"nll_loss": 1.6539511680603027,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1602681279182434,
"rewards/margins": 0.0463583879172802,
"rewards/rejected": -0.2066265046596527,
"step": 273
},
{
"epoch": 0.7566448049706593,
"grad_norm": 0.23999927937984467,
"learning_rate": 4.6623871448617345e-06,
"log_odds_chosen": 0.3460523784160614,
"log_odds_ratio": -0.5406045913696289,
"logits/chosen": -0.06589086353778839,
"logits/rejected": -1.3238555192947388,
"logps/chosen": -1.5511431694030762,
"logps/rejected": -1.8326784372329712,
"loss": 1.6633,
"nll_loss": 1.6092641353607178,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1551143229007721,
"rewards/margins": 0.028153544291853905,
"rewards/rejected": -0.18326784670352936,
"step": 274
},
{
"epoch": 0.7594062823610632,
"grad_norm": 0.2286817878484726,
"learning_rate": 4.6583540835581885e-06,
"log_odds_chosen": 0.4262576103210449,
"log_odds_ratio": -0.5114239454269409,
"logits/chosen": -0.04603245109319687,
"logits/rejected": -1.176690936088562,
"logps/chosen": -1.551270842552185,
"logps/rejected": -1.902989387512207,
"loss": 1.6683,
"nll_loss": 1.6171845197677612,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15512707829475403,
"rewards/margins": 0.03517187386751175,
"rewards/rejected": -0.19029895961284637,
"step": 275
},
{
"epoch": 0.762167759751467,
"grad_norm": 0.20076729357242584,
"learning_rate": 4.654298841826988e-06,
"log_odds_chosen": 0.35074901580810547,
"log_odds_ratio": -0.5370126962661743,
"logits/chosen": -0.04673296585679054,
"logits/rejected": -1.0621964931488037,
"logps/chosen": -1.4891098737716675,
"logps/rejected": -1.7715747356414795,
"loss": 1.5999,
"nll_loss": 1.5461980104446411,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1489109843969345,
"rewards/margins": 0.02824649028480053,
"rewards/rejected": -0.1771574765443802,
"step": 276
},
{
"epoch": 0.7649292371418709,
"grad_norm": 0.2110532820224762,
"learning_rate": 4.6502214613420164e-06,
"log_odds_chosen": 0.6417544484138489,
"log_odds_ratio": -0.4257214069366455,
"logits/chosen": -0.04618730768561363,
"logits/rejected": -1.1318330764770508,
"logps/chosen": -1.457794189453125,
"logps/rejected": -1.9827320575714111,
"loss": 1.5723,
"nll_loss": 1.5297592878341675,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14577943086624146,
"rewards/margins": 0.052493780851364136,
"rewards/rejected": -0.1982732117176056,
"step": 277
},
{
"epoch": 0.7676907145322748,
"grad_norm": 0.23329485952854156,
"learning_rate": 4.646121984004666e-06,
"log_odds_chosen": 0.5122417211532593,
"log_odds_ratio": -0.47840094566345215,
"logits/chosen": -0.014363419264554977,
"logits/rejected": -1.0884288549423218,
"logps/chosen": -1.582539677619934,
"logps/rejected": -2.0093653202056885,
"loss": 1.7056,
"nll_loss": 1.6577521562576294,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15825395286083221,
"rewards/margins": 0.042682573199272156,
"rewards/rejected": -0.20093654096126556,
"step": 278
},
{
"epoch": 0.7704521919226787,
"grad_norm": 0.21924971044063568,
"learning_rate": 4.642000451943409e-06,
"log_odds_chosen": 0.4158315062522888,
"log_odds_ratio": -0.5117031335830688,
"logits/chosen": 0.08326566219329834,
"logits/rejected": -1.0792198181152344,
"logps/chosen": -1.6467266082763672,
"logps/rejected": -1.9971718788146973,
"loss": 1.7532,
"nll_loss": 1.7020772695541382,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16467267274856567,
"rewards/margins": 0.03504452481865883,
"rewards/rejected": -0.199717178940773,
"step": 279
},
{
"epoch": 0.7732136693130826,
"grad_norm": 0.2058788388967514,
"learning_rate": 4.637856907513366e-06,
"log_odds_chosen": 0.5298818945884705,
"log_odds_ratio": -0.4738863706588745,
"logits/chosen": -0.052150044590234756,
"logits/rejected": -1.148370623588562,
"logps/chosen": -1.6041690111160278,
"logps/rejected": -2.045445680618286,
"loss": 1.697,
"nll_loss": 1.6495710611343384,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16041693091392517,
"rewards/margins": 0.04412766546010971,
"rewards/rejected": -0.2045445740222931,
"step": 280
},
{
"epoch": 0.7759751467034863,
"grad_norm": 0.19594821333885193,
"learning_rate": 4.633691393295865e-06,
"log_odds_chosen": 0.33417803049087524,
"log_odds_ratio": -0.5438077449798584,
"logits/chosen": -0.02928408607840538,
"logits/rejected": -1.2465791702270508,
"logps/chosen": -1.6147441864013672,
"logps/rejected": -1.8932818174362183,
"loss": 1.7171,
"nll_loss": 1.6626923084259033,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16147442162036896,
"rewards/margins": 0.027853764593601227,
"rewards/rejected": -0.18932819366455078,
"step": 281
},
{
"epoch": 0.7787366240938902,
"grad_norm": 0.2078380137681961,
"learning_rate": 4.629503952098011e-06,
"log_odds_chosen": 0.6047598123550415,
"log_odds_ratio": -0.4436852037906647,
"logits/chosen": -0.0367959700524807,
"logits/rejected": -1.4540506601333618,
"logps/chosen": -1.5447849035263062,
"logps/rejected": -2.0499391555786133,
"loss": 1.6528,
"nll_loss": 1.6084190607070923,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1544785052537918,
"rewards/margins": 0.050515398383140564,
"rewards/rejected": -0.20499390363693237,
"step": 282
},
{
"epoch": 0.7814981014842941,
"grad_norm": 0.20601390302181244,
"learning_rate": 4.6252946269522406e-06,
"log_odds_chosen": 0.38651254773139954,
"log_odds_ratio": -0.5361747741699219,
"logits/chosen": -0.06736142188310623,
"logits/rejected": -1.498019814491272,
"logps/chosen": -1.5710289478302002,
"logps/rejected": -1.8959977626800537,
"loss": 1.6776,
"nll_loss": 1.6239646673202515,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.15710289776325226,
"rewards/margins": 0.03249688073992729,
"rewards/rejected": -0.18959978222846985,
"step": 283
},
{
"epoch": 0.784259578874698,
"grad_norm": 0.20410317182540894,
"learning_rate": 4.621063461115882e-06,
"log_odds_chosen": 0.36583784222602844,
"log_odds_ratio": -0.5439432263374329,
"logits/chosen": -0.001009071245789528,
"logits/rejected": -1.3512407541275024,
"logps/chosen": -1.6412224769592285,
"logps/rejected": -1.9503740072250366,
"loss": 1.7443,
"nll_loss": 1.6899352073669434,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16412223875522614,
"rewards/margins": 0.030915159732103348,
"rewards/rejected": -0.19503742456436157,
"step": 284
},
{
"epoch": 0.7870210562651019,
"grad_norm": 0.2131362408399582,
"learning_rate": 4.6168104980707105e-06,
"log_odds_chosen": 0.44328027963638306,
"log_odds_ratio": -0.49936914443969727,
"logits/chosen": -0.058265671133995056,
"logits/rejected": -1.3667389154434204,
"logps/chosen": -1.5707849264144897,
"logps/rejected": -1.9372496604919434,
"loss": 1.6771,
"nll_loss": 1.6271687746047974,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15707850456237793,
"rewards/margins": 0.03664647787809372,
"rewards/rejected": -0.19372497498989105,
"step": 285
},
{
"epoch": 0.7897825336555057,
"grad_norm": 0.1902783066034317,
"learning_rate": 4.612535781522504e-06,
"log_odds_chosen": 0.34869980812072754,
"log_odds_ratio": -0.5359938144683838,
"logits/chosen": -0.04153807833790779,
"logits/rejected": -1.3222988843917847,
"logps/chosen": -1.537316918373108,
"logps/rejected": -1.8218798637390137,
"loss": 1.6401,
"nll_loss": 1.5864717960357666,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15373168885707855,
"rewards/margins": 0.02845628745853901,
"rewards/rejected": -0.1821879893541336,
"step": 286
},
{
"epoch": 0.7925440110459095,
"grad_norm": 0.20296384394168854,
"learning_rate": 4.6082393554005855e-06,
"log_odds_chosen": 0.5364274978637695,
"log_odds_ratio": -0.4683885872364044,
"logits/chosen": 0.05323365330696106,
"logits/rejected": -1.2694860696792603,
"logps/chosen": -1.5045408010482788,
"logps/rejected": -1.949967384338379,
"loss": 1.6241,
"nll_loss": 1.5772613286972046,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1504540890455246,
"rewards/margins": 0.04454266279935837,
"rewards/rejected": -0.19499675929546356,
"step": 287
},
{
"epoch": 0.7953054884363134,
"grad_norm": 0.2286815494298935,
"learning_rate": 4.6039212638573835e-06,
"log_odds_chosen": 0.2805465757846832,
"log_odds_ratio": -0.5681173205375671,
"logits/chosen": -0.09262797981500626,
"logits/rejected": -1.4461908340454102,
"logps/chosen": -1.680091381072998,
"logps/rejected": -1.9138922691345215,
"loss": 1.7686,
"nll_loss": 1.7117558717727661,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16800916194915771,
"rewards/margins": 0.02338009513914585,
"rewards/rejected": -0.19138923287391663,
"step": 288
},
{
"epoch": 0.7980669658267173,
"grad_norm": 0.21770860254764557,
"learning_rate": 4.599581551267969e-06,
"log_odds_chosen": 0.5229898691177368,
"log_odds_ratio": -0.47206243872642517,
"logits/chosen": 0.0574706494808197,
"logits/rejected": -1.1304949522018433,
"logps/chosen": -1.6498278379440308,
"logps/rejected": -2.090548038482666,
"loss": 1.7491,
"nll_loss": 1.701880931854248,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16498279571533203,
"rewards/margins": 0.04407202824950218,
"rewards/rejected": -0.20905481278896332,
"step": 289
},
{
"epoch": 0.8008284432171212,
"grad_norm": 0.2001868039369583,
"learning_rate": 4.5952202622296015e-06,
"log_odds_chosen": 0.33771640062332153,
"log_odds_ratio": -0.5422191619873047,
"logits/chosen": -0.08145216107368469,
"logits/rejected": -1.286673903465271,
"logps/chosen": -1.5776478052139282,
"logps/rejected": -1.857193946838379,
"loss": 1.6829,
"nll_loss": 1.6286838054656982,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15776477754116058,
"rewards/margins": 0.027954626828432083,
"rewards/rejected": -0.18571940064430237,
"step": 290
},
{
"epoch": 0.803589920607525,
"grad_norm": 0.19392110407352448,
"learning_rate": 4.590837441561277e-06,
"log_odds_chosen": 0.4482704699039459,
"log_odds_ratio": -0.504833459854126,
"logits/chosen": -0.05314670130610466,
"logits/rejected": -1.3596844673156738,
"logps/chosen": -1.596375584602356,
"logps/rejected": -1.9696097373962402,
"loss": 1.6868,
"nll_loss": 1.6362988948822021,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15963755548000336,
"rewards/margins": 0.037323422729969025,
"rewards/rejected": -0.19696098566055298,
"step": 291
},
{
"epoch": 0.8063513979979289,
"grad_norm": 0.18595081567764282,
"learning_rate": 4.586433134303257e-06,
"log_odds_chosen": 0.5394923090934753,
"log_odds_ratio": -0.4622240662574768,
"logits/chosen": -0.05867187678813934,
"logits/rejected": -1.1816052198410034,
"logps/chosen": -1.4675815105438232,
"logps/rejected": -1.904748797416687,
"loss": 1.5911,
"nll_loss": 1.5448615550994873,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14675816893577576,
"rewards/margins": 0.04371672868728638,
"rewards/rejected": -0.19047488272190094,
"step": 292
},
{
"epoch": 0.8091128753883328,
"grad_norm": 0.2025194615125656,
"learning_rate": 4.582007385716614e-06,
"log_odds_chosen": 0.3834781348705292,
"log_odds_ratio": -0.5252506732940674,
"logits/chosen": -0.03208540380001068,
"logits/rejected": -1.2696198225021362,
"logps/chosen": -1.5880941152572632,
"logps/rejected": -1.9049972295761108,
"loss": 1.7106,
"nll_loss": 1.6580520868301392,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15880942344665527,
"rewards/margins": 0.0316903218626976,
"rewards/rejected": -0.19049973785877228,
"step": 293
},
{
"epoch": 0.8118743527787367,
"grad_norm": 0.21143269538879395,
"learning_rate": 4.57756024128276e-06,
"log_odds_chosen": 0.5385761260986328,
"log_odds_ratio": -0.4631249010562897,
"logits/chosen": -0.12074042856693268,
"logits/rejected": -1.2522529363632202,
"logps/chosen": -1.5002479553222656,
"logps/rejected": -1.9420486688613892,
"loss": 1.6132,
"nll_loss": 1.566838026046753,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15002478659152985,
"rewards/margins": 0.044180065393447876,
"rewards/rejected": -0.19420485198497772,
"step": 294
},
{
"epoch": 0.8146358301691405,
"grad_norm": 0.1943599432706833,
"learning_rate": 4.573091746702988e-06,
"log_odds_chosen": 0.5673821568489075,
"log_odds_ratio": -0.45537808537483215,
"logits/chosen": -0.1322491317987442,
"logits/rejected": -1.209002137184143,
"logps/chosen": -1.5007110834121704,
"logps/rejected": -1.9674400091171265,
"loss": 1.6187,
"nll_loss": 1.573127031326294,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15007109940052032,
"rewards/margins": 0.04667289927601814,
"rewards/rejected": -0.19674400985240936,
"step": 295
},
{
"epoch": 0.8173973075595443,
"grad_norm": 0.17433036863803864,
"learning_rate": 4.5686019478979915e-06,
"log_odds_chosen": 0.6013860702514648,
"log_odds_ratio": -0.4401004910469055,
"logits/chosen": 0.0062029119580984116,
"logits/rejected": -1.6436456441879272,
"logps/chosen": -1.534691572189331,
"logps/rejected": -2.034040927886963,
"loss": 1.6251,
"nll_loss": 1.5810506343841553,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15346917510032654,
"rewards/margins": 0.049934931099414825,
"rewards/rejected": -0.20340411365032196,
"step": 296
},
{
"epoch": 0.8201587849499482,
"grad_norm": 0.20673449337482452,
"learning_rate": 4.564090891007401e-06,
"log_odds_chosen": 0.5072333812713623,
"log_odds_ratio": -0.47898009419441223,
"logits/chosen": -0.12156552821397781,
"logits/rejected": -1.4179987907409668,
"logps/chosen": -1.63102388381958,
"logps/rejected": -2.0585782527923584,
"loss": 1.7285,
"nll_loss": 1.6806358098983765,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1631024032831192,
"rewards/margins": 0.04275544360280037,
"rewards/rejected": -0.20585784316062927,
"step": 297
},
{
"epoch": 0.8229202623403521,
"grad_norm": 0.1835467368364334,
"learning_rate": 4.559558622389304e-06,
"log_odds_chosen": 0.5705875754356384,
"log_odds_ratio": -0.45824331045150757,
"logits/chosen": -0.08186288177967072,
"logits/rejected": -1.502951979637146,
"logps/chosen": -1.5210040807724,
"logps/rejected": -1.987526774406433,
"loss": 1.6191,
"nll_loss": 1.5732712745666504,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15210041403770447,
"rewards/margins": 0.046652257442474365,
"rewards/rejected": -0.19875267148017883,
"step": 298
},
{
"epoch": 0.825681739730756,
"grad_norm": 0.19064722955226898,
"learning_rate": 4.555005188619776e-06,
"log_odds_chosen": 0.5287789106369019,
"log_odds_ratio": -0.473359614610672,
"logits/chosen": -0.1356291025876999,
"logits/rejected": -1.412203073501587,
"logps/chosen": -1.5497990846633911,
"logps/rejected": -1.9918665885925293,
"loss": 1.6437,
"nll_loss": 1.5964010953903198,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1549799144268036,
"rewards/margins": 0.04420673847198486,
"rewards/rejected": -0.19918665289878845,
"step": 299
},
{
"epoch": 0.8284432171211599,
"grad_norm": 0.1883833110332489,
"learning_rate": 4.55043063649239e-06,
"log_odds_chosen": 0.6203457117080688,
"log_odds_ratio": -0.4358068108558655,
"logits/chosen": -0.1520073264837265,
"logits/rejected": -1.6608895063400269,
"logps/chosen": -1.5645240545272827,
"logps/rejected": -2.0832083225250244,
"loss": 1.6559,
"nll_loss": 1.6122933626174927,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15645241737365723,
"rewards/margins": 0.05186842754483223,
"rewards/rejected": -0.20832082629203796,
"step": 300
},
{
"epoch": 0.8312046945115636,
"grad_norm": 0.18228144943714142,
"learning_rate": 4.54583501301775e-06,
"log_odds_chosen": 0.5522174835205078,
"log_odds_ratio": -0.46315470337867737,
"logits/chosen": -0.10763005167245865,
"logits/rejected": -1.706877589225769,
"logps/chosen": -1.5923174619674683,
"logps/rejected": -2.0555925369262695,
"loss": 1.6766,
"nll_loss": 1.630320429801941,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1592317670583725,
"rewards/margins": 0.046327486634254456,
"rewards/rejected": -0.20555922389030457,
"step": 301
},
{
"epoch": 0.8339661719019675,
"grad_norm": 0.18666355311870575,
"learning_rate": 4.541218365422997e-06,
"log_odds_chosen": 0.6547673344612122,
"log_odds_ratio": -0.4269846975803375,
"logits/chosen": -0.1366514414548874,
"logits/rejected": -1.3694720268249512,
"logps/chosen": -1.4531749486923218,
"logps/rejected": -1.9791964292526245,
"loss": 1.5613,
"nll_loss": 1.5185647010803223,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14531749486923218,
"rewards/margins": 0.05260216072201729,
"rewards/rejected": -0.19791966676712036,
"step": 302
},
{
"epoch": 0.8367276492923714,
"grad_norm": 0.19289466738700867,
"learning_rate": 4.536580741151328e-06,
"log_odds_chosen": 0.42363226413726807,
"log_odds_ratio": -0.5060065984725952,
"logits/chosen": -0.1389242559671402,
"logits/rejected": -1.133392095565796,
"logps/chosen": -1.5939542055130005,
"logps/rejected": -1.9430160522460938,
"loss": 1.6864,
"nll_loss": 1.63578200340271,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15939541161060333,
"rewards/margins": 0.034906186163425446,
"rewards/rejected": -0.19430160522460938,
"step": 303
},
{
"epoch": 0.8394891266827753,
"grad_norm": 0.20372274518013,
"learning_rate": 4.531922187861507e-06,
"log_odds_chosen": 0.6614566445350647,
"log_odds_ratio": -0.41916701197624207,
"logits/chosen": -0.11737221479415894,
"logits/rejected": -1.2571710348129272,
"logps/chosen": -1.4728827476501465,
"logps/rejected": -2.0177319049835205,
"loss": 1.5909,
"nll_loss": 1.548986792564392,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1472882777452469,
"rewards/margins": 0.05448490381240845,
"rewards/rejected": -0.20177316665649414,
"step": 304
},
{
"epoch": 0.8422506040731792,
"grad_norm": 0.20864824950695038,
"learning_rate": 4.527242753427378e-06,
"log_odds_chosen": 0.4375610649585724,
"log_odds_ratio": -0.5043913722038269,
"logits/chosen": -0.14712439477443695,
"logits/rejected": -1.1074727773666382,
"logps/chosen": -1.6707913875579834,
"logps/rejected": -2.040156126022339,
"loss": 1.7741,
"nll_loss": 1.723702073097229,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1670791655778885,
"rewards/margins": 0.036936454474925995,
"rewards/rejected": -0.20401562750339508,
"step": 305
},
{
"epoch": 0.845012081463583,
"grad_norm": 0.1923041045665741,
"learning_rate": 4.522542485937369e-06,
"log_odds_chosen": 0.6749688982963562,
"log_odds_ratio": -0.4191080629825592,
"logits/chosen": -0.11244256049394608,
"logits/rejected": -1.4451961517333984,
"logps/chosen": -1.528746485710144,
"logps/rejected": -2.0891406536102295,
"loss": 1.644,
"nll_loss": 1.602088212966919,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1528746634721756,
"rewards/margins": 0.05603940039873123,
"rewards/rejected": -0.20891407132148743,
"step": 306
},
{
"epoch": 0.8477735588539869,
"grad_norm": 0.19692490994930267,
"learning_rate": 4.5178214336940015e-06,
"log_odds_chosen": 0.536999523639679,
"log_odds_ratio": -0.4662824273109436,
"logits/chosen": -0.12849773466587067,
"logits/rejected": -1.3011747598648071,
"logps/chosen": -1.537503957748413,
"logps/rejected": -1.9821370840072632,
"loss": 1.6458,
"nll_loss": 1.5992212295532227,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15375038981437683,
"rewards/margins": 0.044463321566581726,
"rewards/rejected": -0.19821371138095856,
"step": 307
},
{
"epoch": 0.8505350362443908,
"grad_norm": 0.19530101120471954,
"learning_rate": 4.513079645213391e-06,
"log_odds_chosen": 0.5806478261947632,
"log_odds_ratio": -0.45124712586402893,
"logits/chosen": -0.14092004299163818,
"logits/rejected": -1.2499042749404907,
"logps/chosen": -1.447067141532898,
"logps/rejected": -1.9178260564804077,
"loss": 1.543,
"nll_loss": 1.4978904724121094,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14470671117305756,
"rewards/margins": 0.04707589000463486,
"rewards/rejected": -0.19178259372711182,
"step": 308
},
{
"epoch": 0.8532965136347946,
"grad_norm": 0.21278323233127594,
"learning_rate": 4.508317169224752e-06,
"log_odds_chosen": 0.2714424431324005,
"log_odds_ratio": -0.5724983215332031,
"logits/chosen": -0.23690785467624664,
"logits/rejected": -1.2109217643737793,
"logps/chosen": -1.568101406097412,
"logps/rejected": -1.7880043983459473,
"loss": 1.6669,
"nll_loss": 1.6096391677856445,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15681014955043793,
"rewards/margins": 0.021990297362208366,
"rewards/rejected": -0.17880043387413025,
"step": 309
},
{
"epoch": 0.8560579910251985,
"grad_norm": 0.3061252534389496,
"learning_rate": 4.5035340546698915e-06,
"log_odds_chosen": 0.5074098110198975,
"log_odds_ratio": -0.47923699021339417,
"logits/chosen": -0.08015923947095871,
"logits/rejected": -1.4736651182174683,
"logps/chosen": -1.5646302700042725,
"logps/rejected": -1.9794241189956665,
"loss": 1.6599,
"nll_loss": 1.6120116710662842,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15646302700042725,
"rewards/margins": 0.041479405015707016,
"rewards/rejected": -0.19794242084026337,
"step": 310
},
{
"epoch": 0.8588194684156023,
"grad_norm": 0.20862670242786407,
"learning_rate": 4.4987303507027155e-06,
"log_odds_chosen": 0.32739633321762085,
"log_odds_ratio": -0.5497077703475952,
"logits/chosen": -0.057409606873989105,
"logits/rejected": -1.186546802520752,
"logps/chosen": -1.6279197931289673,
"logps/rejected": -1.8978025913238525,
"loss": 1.7296,
"nll_loss": 1.6746082305908203,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.16279199719429016,
"rewards/margins": 0.026988260447978973,
"rewards/rejected": -0.18978025019168854,
"step": 311
},
{
"epoch": 0.8615809458060062,
"grad_norm": 0.1961701512336731,
"learning_rate": 4.493906106688712e-06,
"log_odds_chosen": 0.6683472394943237,
"log_odds_ratio": -0.4163239598274231,
"logits/chosen": -0.08115644752979279,
"logits/rejected": -1.3991774320602417,
"logps/chosen": -1.5248966217041016,
"logps/rejected": -2.0810930728912354,
"loss": 1.633,
"nll_loss": 1.5913276672363281,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15248967707157135,
"rewards/margins": 0.05561964958906174,
"rewards/rejected": -0.20810934901237488,
"step": 312
},
{
"epoch": 0.8643424231964101,
"grad_norm": 0.4846391975879669,
"learning_rate": 4.4890613722044526e-06,
"log_odds_chosen": 0.5416931509971619,
"log_odds_ratio": -0.46186453104019165,
"logits/chosen": -0.13847319781780243,
"logits/rejected": -1.388474702835083,
"logps/chosen": -1.5097135305404663,
"logps/rejected": -1.9569801092147827,
"loss": 1.59,
"nll_loss": 1.5437886714935303,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15097135305404663,
"rewards/margins": 0.04472666233778,
"rewards/rejected": -0.19569800794124603,
"step": 313
},
{
"epoch": 0.867103900586814,
"grad_norm": 0.2009696215391159,
"learning_rate": 4.484196197037082e-06,
"log_odds_chosen": 0.5546321868896484,
"log_odds_ratio": -0.45888563990592957,
"logits/chosen": -0.07164500653743744,
"logits/rejected": -1.3182318210601807,
"logps/chosen": -1.561537265777588,
"logps/rejected": -2.02632212638855,
"loss": 1.6542,
"nll_loss": 1.6082967519760132,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15615373849868774,
"rewards/margins": 0.046478480100631714,
"rewards/rejected": -0.20263221859931946,
"step": 314
},
{
"epoch": 0.8698653779772179,
"grad_norm": 0.18780747056007385,
"learning_rate": 4.4793106311838e-06,
"log_odds_chosen": 0.5521085262298584,
"log_odds_ratio": -0.46405351161956787,
"logits/chosen": -0.17451435327529907,
"logits/rejected": -1.2368041276931763,
"logps/chosen": -1.547911524772644,
"logps/rejected": -2.007674217224121,
"loss": 1.6276,
"nll_loss": 1.5811690092086792,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15479114651679993,
"rewards/margins": 0.04597627371549606,
"rewards/rejected": -0.2007674127817154,
"step": 315
},
{
"epoch": 0.8726268553676216,
"grad_norm": 0.19249524176120758,
"learning_rate": 4.474404724851356e-06,
"log_odds_chosen": 0.49972984194755554,
"log_odds_ratio": -0.4845547676086426,
"logits/chosen": -0.12106968462467194,
"logits/rejected": -1.4282002449035645,
"logps/chosen": -1.599500060081482,
"logps/rejected": -2.0155510902404785,
"loss": 1.6902,
"nll_loss": 1.641719102859497,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15995000302791595,
"rewards/margins": 0.04160511493682861,
"rewards/rejected": -0.20155511796474457,
"step": 316
},
{
"epoch": 0.8753883327580255,
"grad_norm": 0.2032056301832199,
"learning_rate": 4.469478528455529e-06,
"log_odds_chosen": 0.5046655535697937,
"log_odds_ratio": -0.47710105776786804,
"logits/chosen": -0.27712228894233704,
"logits/rejected": -1.1458078622817993,
"logps/chosen": -1.587306261062622,
"logps/rejected": -2.0054972171783447,
"loss": 1.6787,
"nll_loss": 1.6309565305709839,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1587306261062622,
"rewards/margins": 0.041819095611572266,
"rewards/rejected": -0.20054972171783447,
"step": 317
},
{
"epoch": 0.8781498101484294,
"grad_norm": 0.19460327923297882,
"learning_rate": 4.464532092620607e-06,
"log_odds_chosen": 0.5168122053146362,
"log_odds_ratio": -0.474026620388031,
"logits/chosen": -0.1055205836892128,
"logits/rejected": -1.2003237009048462,
"logps/chosen": -1.6242094039916992,
"logps/rejected": -2.0614237785339355,
"loss": 1.7035,
"nll_loss": 1.6561068296432495,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16242094337940216,
"rewards/margins": 0.04372143745422363,
"rewards/rejected": -0.206142395734787,
"step": 318
},
{
"epoch": 0.8809112875388333,
"grad_norm": 0.17680254578590393,
"learning_rate": 4.4595654681788715e-06,
"log_odds_chosen": 0.6278887987136841,
"log_odds_ratio": -0.43415001034736633,
"logits/chosen": -0.15030619502067566,
"logits/rejected": -1.5502567291259766,
"logps/chosen": -1.5051651000976562,
"logps/rejected": -2.024726629257202,
"loss": 1.5904,
"nll_loss": 1.5469478368759155,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15051651000976562,
"rewards/margins": 0.051956143230199814,
"rewards/rejected": -0.20247265696525574,
"step": 319
},
{
"epoch": 0.8836727649292372,
"grad_norm": 0.20075084269046783,
"learning_rate": 4.454578706170075e-06,
"log_odds_chosen": 0.3344058692455292,
"log_odds_ratio": -0.547435998916626,
"logits/chosen": -0.09762324392795563,
"logits/rejected": -1.3721954822540283,
"logps/chosen": -1.610334873199463,
"logps/rejected": -1.8859809637069702,
"loss": 1.6984,
"nll_loss": 1.6436134576797485,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1610334813594818,
"rewards/margins": 0.02756461501121521,
"rewards/rejected": -0.18859811127185822,
"step": 320
},
{
"epoch": 0.886434242319641,
"grad_norm": 0.1969553381204605,
"learning_rate": 4.449571857840911e-06,
"log_odds_chosen": 0.5939600467681885,
"log_odds_ratio": -0.4538188576698303,
"logits/chosen": -0.02511260285973549,
"logits/rejected": -1.4034056663513184,
"logps/chosen": -1.554402232170105,
"logps/rejected": -2.0581822395324707,
"loss": 1.647,
"nll_loss": 1.6016532182693481,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15544021129608154,
"rewards/margins": 0.05037800967693329,
"rewards/rejected": -0.20581823587417603,
"step": 321
},
{
"epoch": 0.8891957197100449,
"grad_norm": 0.20333482325077057,
"learning_rate": 4.444544974644493e-06,
"log_odds_chosen": 0.4322926700115204,
"log_odds_ratio": -0.5066258311271667,
"logits/chosen": -0.1251221001148224,
"logits/rejected": -1.2058238983154297,
"logps/chosen": -1.5241694450378418,
"logps/rejected": -1.878300666809082,
"loss": 1.6259,
"nll_loss": 1.5751992464065552,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15241695940494537,
"rewards/margins": 0.035413116216659546,
"rewards/rejected": -0.18783007562160492,
"step": 322
},
{
"epoch": 0.8919571971004487,
"grad_norm": 0.19495686888694763,
"learning_rate": 4.4394981082398254e-06,
"log_odds_chosen": 0.4637213945388794,
"log_odds_ratio": -0.5005385875701904,
"logits/chosen": -0.19393548369407654,
"logits/rejected": -1.3736821413040161,
"logps/chosen": -1.5801210403442383,
"logps/rejected": -1.9673550128936768,
"loss": 1.678,
"nll_loss": 1.6279520988464355,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15801210701465607,
"rewards/margins": 0.0387234091758728,
"rewards/rejected": -0.19673550128936768,
"step": 323
},
{
"epoch": 0.8947186744908526,
"grad_norm": 0.21592693030834198,
"learning_rate": 4.434431310491267e-06,
"log_odds_chosen": 0.6085124015808105,
"log_odds_ratio": -0.44234418869018555,
"logits/chosen": -0.20342035591602325,
"logits/rejected": -1.3250751495361328,
"logps/chosen": -1.6450177431106567,
"logps/rejected": -2.15885329246521,
"loss": 1.7148,
"nll_loss": 1.6705245971679688,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16450178623199463,
"rewards/margins": 0.05138356238603592,
"rewards/rejected": -0.21588534116744995,
"step": 324
},
{
"epoch": 0.8974801518812565,
"grad_norm": 0.2041964828968048,
"learning_rate": 4.429344633468005e-06,
"log_odds_chosen": 0.5157475471496582,
"log_odds_ratio": -0.4731763005256653,
"logits/chosen": -0.1373744010925293,
"logits/rejected": -1.3682070970535278,
"logps/chosen": -1.5091853141784668,
"logps/rejected": -1.9341413974761963,
"loss": 1.5962,
"nll_loss": 1.5488858222961426,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15091852843761444,
"rewards/margins": 0.042495615780353546,
"rewards/rejected": -0.19341415166854858,
"step": 325
},
{
"epoch": 0.9002416292716603,
"grad_norm": 0.21775774657726288,
"learning_rate": 4.424238129443515e-06,
"log_odds_chosen": 0.483593225479126,
"log_odds_ratio": -0.48364073038101196,
"logits/chosen": -0.09863223880529404,
"logits/rejected": -1.0222870111465454,
"logps/chosen": -1.6142656803131104,
"logps/rejected": -2.01713228225708,
"loss": 1.7025,
"nll_loss": 1.6541244983673096,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1614265739917755,
"rewards/margins": 0.04028663784265518,
"rewards/rejected": -0.2017132043838501,
"step": 326
},
{
"epoch": 0.9030031066620642,
"grad_norm": 0.20912012457847595,
"learning_rate": 4.4191118508950286e-06,
"log_odds_chosen": 0.5832557678222656,
"log_odds_ratio": -0.44807419180870056,
"logits/chosen": -0.18956468999385834,
"logits/rejected": -1.2079969644546509,
"logps/chosen": -1.5554307699203491,
"logps/rejected": -2.04107666015625,
"loss": 1.659,
"nll_loss": 1.6141505241394043,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15554308891296387,
"rewards/margins": 0.04856458306312561,
"rewards/rejected": -0.20410768687725067,
"step": 327
},
{
"epoch": 0.9057645840524681,
"grad_norm": 0.1959802657365799,
"learning_rate": 4.413965850502987e-06,
"log_odds_chosen": 0.6536615490913391,
"log_odds_ratio": -0.42614060640335083,
"logits/chosen": -0.2154429852962494,
"logits/rejected": -1.3865175247192383,
"logps/chosen": -1.4310425519943237,
"logps/rejected": -1.9644105434417725,
"loss": 1.4983,
"nll_loss": 1.455698013305664,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14310424029827118,
"rewards/margins": 0.05333680659532547,
"rewards/rejected": -0.19644105434417725,
"step": 328
},
{
"epoch": 0.908526061442872,
"grad_norm": 0.19480550289154053,
"learning_rate": 4.408800181150509e-06,
"log_odds_chosen": 0.6978083252906799,
"log_odds_ratio": -0.4107431471347809,
"logits/chosen": -0.20347651839256287,
"logits/rejected": -1.4490666389465332,
"logps/chosen": -1.508465051651001,
"logps/rejected": -2.0856833457946777,
"loss": 1.5878,
"nll_loss": 1.5467225313186646,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15084651112556458,
"rewards/margins": 0.0577218271791935,
"rewards/rejected": -0.20856834948062897,
"step": 329
},
{
"epoch": 0.9112875388332758,
"grad_norm": 0.20806226134300232,
"learning_rate": 4.4036148959228365e-06,
"log_odds_chosen": 0.5531899333000183,
"log_odds_ratio": -0.455844908952713,
"logits/chosen": -0.12704087793827057,
"logits/rejected": -1.144600510597229,
"logps/chosen": -1.5709285736083984,
"logps/rejected": -2.0322890281677246,
"loss": 1.6525,
"nll_loss": 1.6068720817565918,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1570928692817688,
"rewards/margins": 0.046136029064655304,
"rewards/rejected": -0.2032289057970047,
"step": 330
},
{
"epoch": 0.9140490162236796,
"grad_norm": 0.20405174791812897,
"learning_rate": 4.3984100481068e-06,
"log_odds_chosen": 0.5182926654815674,
"log_odds_ratio": -0.4803735911846161,
"logits/chosen": -0.05097893998026848,
"logits/rejected": -1.530269980430603,
"logps/chosen": -1.5223273038864136,
"logps/rejected": -1.9506477117538452,
"loss": 1.602,
"nll_loss": 1.5539740324020386,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15223273634910583,
"rewards/margins": 0.04283204674720764,
"rewards/rejected": -0.19506478309631348,
"step": 331
},
{
"epoch": 0.9168104936140835,
"grad_norm": 0.1879170536994934,
"learning_rate": 4.3931856911902635e-06,
"log_odds_chosen": 0.6412795782089233,
"log_odds_ratio": -0.4325963854789734,
"logits/chosen": -0.23800577223300934,
"logits/rejected": -1.4617422819137573,
"logps/chosen": -1.5373084545135498,
"logps/rejected": -2.0728628635406494,
"loss": 1.6293,
"nll_loss": 1.586013674736023,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1537308394908905,
"rewards/margins": 0.0535554476082325,
"rewards/rejected": -0.2072862982749939,
"step": 332
},
{
"epoch": 0.9195719710044874,
"grad_norm": 0.19721606373786926,
"learning_rate": 4.387941878861578e-06,
"log_odds_chosen": 0.43159598112106323,
"log_odds_ratio": -0.5106831192970276,
"logits/chosen": -0.2015897035598755,
"logits/rejected": -1.1725409030914307,
"logps/chosen": -1.4626684188842773,
"logps/rejected": -1.816298246383667,
"loss": 1.5551,
"nll_loss": 1.5040326118469238,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1462668478488922,
"rewards/margins": 0.03536297380924225,
"rewards/rejected": -0.18162982165813446,
"step": 333
},
{
"epoch": 0.9223334483948913,
"grad_norm": 0.19862483441829681,
"learning_rate": 4.382678665009028e-06,
"log_odds_chosen": 0.5599039196968079,
"log_odds_ratio": -0.45722323656082153,
"logits/chosen": -0.1434212028980255,
"logits/rejected": -1.4838988780975342,
"logps/chosen": -1.5819792747497559,
"logps/rejected": -2.048774003982544,
"loss": 1.6595,
"nll_loss": 1.6138123273849487,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15819790959358215,
"rewards/margins": 0.04667946696281433,
"rewards/rejected": -0.20487739145755768,
"step": 334
},
{
"epoch": 0.9250949257852952,
"grad_norm": 0.20085540413856506,
"learning_rate": 4.3773961037202784e-06,
"log_odds_chosen": 0.5849855542182922,
"log_odds_ratio": -0.4470357596874237,
"logits/chosen": -0.16471579670906067,
"logits/rejected": -1.48972749710083,
"logps/chosen": -1.6449600458145142,
"logps/rejected": -2.1387722492218018,
"loss": 1.7056,
"nll_loss": 1.6609022617340088,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16449600458145142,
"rewards/margins": 0.04938122630119324,
"rewards/rejected": -0.21387723088264465,
"step": 335
},
{
"epoch": 0.927856403175699,
"grad_norm": 0.19542387127876282,
"learning_rate": 4.37209424928182e-06,
"log_odds_chosen": 0.5759137272834778,
"log_odds_ratio": -0.4505816698074341,
"logits/chosen": -0.1124081164598465,
"logits/rejected": -1.1634758710861206,
"logps/chosen": -1.5444155931472778,
"logps/rejected": -2.0244972705841064,
"loss": 1.6261,
"nll_loss": 1.581070065498352,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15444158017635345,
"rewards/margins": 0.048008158802986145,
"rewards/rejected": -0.2024497389793396,
"step": 336
},
{
"epoch": 0.9306178805661028,
"grad_norm": 0.187562957406044,
"learning_rate": 4.366773156178413e-06,
"log_odds_chosen": 0.40636616945266724,
"log_odds_ratio": -0.5145981311798096,
"logits/chosen": -0.23331299424171448,
"logits/rejected": -1.272404432296753,
"logps/chosen": -1.4672964811325073,
"logps/rejected": -1.792708396911621,
"loss": 1.5577,
"nll_loss": 1.5061949491500854,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14672963321208954,
"rewards/margins": 0.03254120051860809,
"rewards/rejected": -0.17927084863185883,
"step": 337
},
{
"epoch": 0.9333793579565067,
"grad_norm": 0.19676648080348969,
"learning_rate": 4.361432879092518e-06,
"log_odds_chosen": 0.6147382259368896,
"log_odds_ratio": -0.444987416267395,
"logits/chosen": -0.2584179937839508,
"logits/rejected": -1.2025710344314575,
"logps/chosen": -1.4777214527130127,
"logps/rejected": -1.9875514507293701,
"loss": 1.5673,
"nll_loss": 1.5228395462036133,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1477721631526947,
"rewards/margins": 0.0509830042719841,
"rewards/rejected": -0.198755145072937,
"step": 338
},
{
"epoch": 0.9361408353469106,
"grad_norm": 0.20113791525363922,
"learning_rate": 4.356073472903747e-06,
"log_odds_chosen": 0.5940407514572144,
"log_odds_ratio": -0.4446149170398712,
"logits/chosen": -0.09678924083709717,
"logits/rejected": -1.2151576280593872,
"logps/chosen": -1.4379996061325073,
"logps/rejected": -1.9237509965896606,
"loss": 1.5294,
"nll_loss": 1.4849097728729248,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14379996061325073,
"rewards/margins": 0.04857514798641205,
"rewards/rejected": -0.19237510859966278,
"step": 339
},
{
"epoch": 0.9389023127373145,
"grad_norm": 0.18967628479003906,
"learning_rate": 4.350694992688289e-06,
"log_odds_chosen": 0.6668899059295654,
"log_odds_ratio": -0.427124559879303,
"logits/chosen": -0.08107070624828339,
"logits/rejected": -1.2518596649169922,
"logps/chosen": -1.471956491470337,
"logps/rejected": -2.021543025970459,
"loss": 1.5567,
"nll_loss": 1.513990879058838,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14719566702842712,
"rewards/margins": 0.05495864897966385,
"rewards/rejected": -0.20215432345867157,
"step": 340
},
{
"epoch": 0.9416637901277183,
"grad_norm": 0.1954769492149353,
"learning_rate": 4.345297493718352e-06,
"log_odds_chosen": 0.5373238325119019,
"log_odds_ratio": -0.46630507707595825,
"logits/chosen": -0.24432893097400665,
"logits/rejected": -1.0526390075683594,
"logps/chosen": -1.472176432609558,
"logps/rejected": -1.9091975688934326,
"loss": 1.5654,
"nll_loss": 1.5187859535217285,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14721764624118805,
"rewards/margins": 0.04370209947228432,
"rewards/rejected": -0.19091975688934326,
"step": 341
},
{
"epoch": 0.9444252675181222,
"grad_norm": 0.18435297906398773,
"learning_rate": 4.339881031461588e-06,
"log_odds_chosen": 0.46578601002693176,
"log_odds_ratio": -0.49112096428871155,
"logits/chosen": -0.18604259192943573,
"logits/rejected": -1.3026976585388184,
"logps/chosen": -1.4346725940704346,
"logps/rejected": -1.7989139556884766,
"loss": 1.5414,
"nll_loss": 1.4923009872436523,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1434672772884369,
"rewards/margins": 0.036424119025468826,
"rewards/rejected": -0.17989139258861542,
"step": 342
},
{
"epoch": 0.9471867449085261,
"grad_norm": 0.19713670015335083,
"learning_rate": 4.334445661580527e-06,
"log_odds_chosen": 0.5911546945571899,
"log_odds_ratio": -0.444367378950119,
"logits/chosen": -0.24173688888549805,
"logits/rejected": -1.4439072608947754,
"logps/chosen": -1.5436309576034546,
"logps/rejected": -2.035305976867676,
"loss": 1.6258,
"nll_loss": 1.5813885927200317,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15436309576034546,
"rewards/margins": 0.049167513847351074,
"rewards/rejected": -0.20353063941001892,
"step": 343
},
{
"epoch": 0.94994822229893,
"grad_norm": 0.19868247210979462,
"learning_rate": 4.328991439932003e-06,
"log_odds_chosen": 0.646172046661377,
"log_odds_ratio": -0.42435139417648315,
"logits/chosen": -0.21409977972507477,
"logits/rejected": -1.1630572080612183,
"logps/chosen": -1.475473165512085,
"logps/rejected": -2.007708787918091,
"loss": 1.5666,
"nll_loss": 1.5241310596466064,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14754730463027954,
"rewards/margins": 0.053223565220832825,
"rewards/rejected": -0.20077086985111237,
"step": 344
},
{
"epoch": 0.9527096996893338,
"grad_norm": 0.2008034884929657,
"learning_rate": 4.323518422566586e-06,
"log_odds_chosen": 0.7171238660812378,
"log_odds_ratio": -0.4020853638648987,
"logits/chosen": -0.14678461849689484,
"logits/rejected": -1.1799049377441406,
"logps/chosen": -1.5431439876556396,
"logps/rejected": -2.141486167907715,
"loss": 1.639,
"nll_loss": 1.5987637042999268,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15431438386440277,
"rewards/margins": 0.05983421206474304,
"rewards/rejected": -0.214148610830307,
"step": 345
},
{
"epoch": 0.9554711770797376,
"grad_norm": 0.18002885580062866,
"learning_rate": 4.318026665727993e-06,
"log_odds_chosen": 0.7151846885681152,
"log_odds_ratio": -0.40609210729599,
"logits/chosen": -0.2377801239490509,
"logits/rejected": -1.4111603498458862,
"logps/chosen": -1.4105151891708374,
"logps/rejected": -1.9907076358795166,
"loss": 1.4953,
"nll_loss": 1.4546504020690918,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1410515308380127,
"rewards/margins": 0.058019235730171204,
"rewards/rejected": -0.1990707665681839,
"step": 346
},
{
"epoch": 0.9582326544701415,
"grad_norm": 0.2088988721370697,
"learning_rate": 4.3125162258525265e-06,
"log_odds_chosen": 0.37778812646865845,
"log_odds_ratio": -0.5246101021766663,
"logits/chosen": -0.26066213846206665,
"logits/rejected": -1.111747145652771,
"logps/chosen": -1.6180295944213867,
"logps/rejected": -1.932243824005127,
"loss": 1.7004,
"nll_loss": 1.6479332447052002,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1618029624223709,
"rewards/margins": 0.03142143785953522,
"rewards/rejected": -0.19322440028190613,
"step": 347
},
{
"epoch": 0.9609941318605454,
"grad_norm": 0.23166809976100922,
"learning_rate": 4.3069871595684795e-06,
"log_odds_chosen": 0.5429641604423523,
"log_odds_ratio": -0.46485579013824463,
"logits/chosen": -0.26122069358825684,
"logits/rejected": -1.2304470539093018,
"logps/chosen": -1.482200026512146,
"logps/rejected": -1.9262490272521973,
"loss": 1.5751,
"nll_loss": 1.5286585092544556,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1482200026512146,
"rewards/margins": 0.044404882937669754,
"rewards/rejected": -0.19262489676475525,
"step": 348
},
{
"epoch": 0.9637556092509493,
"grad_norm": 0.19496473670005798,
"learning_rate": 4.3014395236955635e-06,
"log_odds_chosen": 0.6688944101333618,
"log_odds_ratio": -0.4200526475906372,
"logits/chosen": -0.2004820704460144,
"logits/rejected": -1.322332501411438,
"logps/chosen": -1.4777374267578125,
"logps/rejected": -2.0316760540008545,
"loss": 1.5606,
"nll_loss": 1.5185458660125732,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14777372777462006,
"rewards/margins": 0.055393870919942856,
"rewards/rejected": -0.2031676024198532,
"step": 349
},
{
"epoch": 0.9665170866413532,
"grad_norm": 0.18388359248638153,
"learning_rate": 4.295873375244319e-06,
"log_odds_chosen": 0.5681695938110352,
"log_odds_ratio": -0.4566437602043152,
"logits/chosen": -0.21363496780395508,
"logits/rejected": -1.1769250631332397,
"logps/chosen": -1.439588189125061,
"logps/rejected": -1.9001712799072266,
"loss": 1.5306,
"nll_loss": 1.4849416017532349,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14395882189273834,
"rewards/margins": 0.04605831205844879,
"rewards/rejected": -0.19001714885234833,
"step": 350
},
{
"epoch": 0.9692785640317569,
"grad_norm": 0.20012839138507843,
"learning_rate": 4.290288771415536e-06,
"log_odds_chosen": 0.5067012906074524,
"log_odds_ratio": -0.477801650762558,
"logits/chosen": -0.20678414404392242,
"logits/rejected": -1.340362787246704,
"logps/chosen": -1.5872958898544312,
"logps/rejected": -2.0096263885498047,
"loss": 1.6634,
"nll_loss": 1.6156083345413208,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15872959792613983,
"rewards/margins": 0.04223306477069855,
"rewards/rejected": -0.20096264779567719,
"step": 351
},
{
"epoch": 0.9720400414221608,
"grad_norm": 0.1952664852142334,
"learning_rate": 4.284685769599658e-06,
"log_odds_chosen": 0.5722883343696594,
"log_odds_ratio": -0.45633140206336975,
"logits/chosen": -0.2586257755756378,
"logits/rejected": -1.3633008003234863,
"logps/chosen": -1.495697259902954,
"logps/rejected": -1.9731667041778564,
"loss": 1.5729,
"nll_loss": 1.5272736549377441,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14956970512866974,
"rewards/margins": 0.04774694889783859,
"rewards/rejected": -0.19731666147708893,
"step": 352
},
{
"epoch": 0.9748015188125647,
"grad_norm": 0.194318950176239,
"learning_rate": 4.279064427376199e-06,
"log_odds_chosen": 0.5692598819732666,
"log_odds_ratio": -0.4534481167793274,
"logits/chosen": -0.21069735288619995,
"logits/rejected": -1.2419835329055786,
"logps/chosen": -1.5351362228393555,
"logps/rejected": -2.0084264278411865,
"loss": 1.6002,
"nll_loss": 1.5548111200332642,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15351362526416779,
"rewards/margins": 0.04732901602983475,
"rewards/rejected": -0.20084263384342194,
"step": 353
},
{
"epoch": 0.9775629962029686,
"grad_norm": 0.20712882280349731,
"learning_rate": 4.273424802513145e-06,
"log_odds_chosen": 0.6014809012413025,
"log_odds_ratio": -0.4396442174911499,
"logits/chosen": -0.2209128886461258,
"logits/rejected": -1.3539609909057617,
"logps/chosen": -1.5357590913772583,
"logps/rejected": -2.0328547954559326,
"loss": 1.6122,
"nll_loss": 1.5682051181793213,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15357591211795807,
"rewards/margins": 0.04970954358577728,
"rewards/rejected": -0.20328545570373535,
"step": 354
},
{
"epoch": 0.9803244735933725,
"grad_norm": 0.18310385942459106,
"learning_rate": 4.267766952966369e-06,
"log_odds_chosen": 0.7377204298973083,
"log_odds_ratio": -0.40729206800460815,
"logits/chosen": -0.15849949419498444,
"logits/rejected": -1.2910916805267334,
"logps/chosen": -1.3705458641052246,
"logps/rejected": -1.968411922454834,
"loss": 1.4587,
"nll_loss": 1.4179855585098267,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13705459237098694,
"rewards/margins": 0.05978662520647049,
"rewards/rejected": -0.19684121012687683,
"step": 355
},
{
"epoch": 0.9830859509837763,
"grad_norm": 0.18806028366088867,
"learning_rate": 4.26209093687903e-06,
"log_odds_chosen": 0.4816596806049347,
"log_odds_ratio": -0.48845529556274414,
"logits/chosen": -0.2249145656824112,
"logits/rejected": -1.4047096967697144,
"logps/chosen": -1.5253814458847046,
"logps/rejected": -1.9247775077819824,
"loss": 1.6102,
"nll_loss": 1.5613718032836914,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15253815054893494,
"rewards/margins": 0.03993961960077286,
"rewards/rejected": -0.1924777626991272,
"step": 356
},
{
"epoch": 0.9858474283741802,
"grad_norm": 0.19746609032154083,
"learning_rate": 4.2563968125809734e-06,
"log_odds_chosen": 0.6187906265258789,
"log_odds_ratio": -0.43648919463157654,
"logits/chosen": -0.08207077533006668,
"logits/rejected": -1.3270372152328491,
"logps/chosen": -1.5937108993530273,
"logps/rejected": -2.113786458969116,
"loss": 1.6592,
"nll_loss": 1.615505576133728,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15937110781669617,
"rewards/margins": 0.05200754106044769,
"rewards/rejected": -0.21137863397598267,
"step": 357
},
{
"epoch": 0.988608905764584,
"grad_norm": 0.19708895683288574,
"learning_rate": 4.2506846385881375e-06,
"log_odds_chosen": 0.6972445845603943,
"log_odds_ratio": -0.4114294648170471,
"logits/chosen": -0.37635722756385803,
"logits/rejected": -1.2826493978500366,
"logps/chosen": -1.4048396348953247,
"logps/rejected": -1.9649187326431274,
"loss": 1.4948,
"nll_loss": 1.4536077976226807,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14048396050930023,
"rewards/margins": 0.056007932871580124,
"rewards/rejected": -0.19649189710617065,
"step": 358
},
{
"epoch": 0.9913703831549879,
"grad_norm": 0.19348447024822235,
"learning_rate": 4.2449544736019486e-06,
"log_odds_chosen": 0.5735771656036377,
"log_odds_ratio": -0.4517831802368164,
"logits/chosen": -0.15842154622077942,
"logits/rejected": -1.160691738128662,
"logps/chosen": -1.4741443395614624,
"logps/rejected": -1.9411314725875854,
"loss": 1.559,
"nll_loss": 1.513805866241455,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.147414430975914,
"rewards/margins": 0.04669870436191559,
"rewards/rejected": -0.1941131353378296,
"step": 359
},
{
"epoch": 0.9941318605453918,
"grad_norm": 0.18270692229270935,
"learning_rate": 4.239206376508716e-06,
"log_odds_chosen": 0.7792839407920837,
"log_odds_ratio": -0.38477823138237,
"logits/chosen": -0.15978127717971802,
"logits/rejected": -1.5017080307006836,
"logps/chosen": -1.494106411933899,
"logps/rejected": -2.147690534591675,
"loss": 1.5826,
"nll_loss": 1.5441505908966064,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14941063523292542,
"rewards/margins": 0.06535841524600983,
"rewards/rejected": -0.21476906538009644,
"step": 360
},
{
"epoch": 0.9968933379357956,
"grad_norm": 0.18168719112873077,
"learning_rate": 4.233440406379032e-06,
"log_odds_chosen": 0.5889392495155334,
"log_odds_ratio": -0.4472864866256714,
"logits/chosen": -0.28017136454582214,
"logits/rejected": -1.22226881980896,
"logps/chosen": -1.414659857749939,
"logps/rejected": -1.88670814037323,
"loss": 1.505,
"nll_loss": 1.460268497467041,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.141465961933136,
"rewards/margins": 0.047204844653606415,
"rewards/rejected": -0.1886708289384842,
"step": 361
},
{
"epoch": 0.9996548153261995,
"grad_norm": 0.19425641000270844,
"learning_rate": 4.227656622467162e-06,
"log_odds_chosen": 0.6574209928512573,
"log_odds_ratio": -0.4261106252670288,
"logits/chosen": -0.2706007957458496,
"logits/rejected": -1.1785038709640503,
"logps/chosen": -1.516379475593567,
"logps/rejected": -2.065363883972168,
"loss": 1.5932,
"nll_loss": 1.550638198852539,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1516379415988922,
"rewards/margins": 0.054898452013731,
"rewards/rejected": -0.2065364122390747,
"step": 362
},
{
"epoch": 1.0,
"grad_norm": 0.46724027395248413,
"learning_rate": 4.221855084210433e-06,
"log_odds_chosen": 0.3892417550086975,
"log_odds_ratio": -0.5173466205596924,
"logits/chosen": -0.4694238007068634,
"logits/rejected": -1.6226561069488525,
"logps/chosen": -1.5516700744628906,
"logps/rejected": -1.8701450824737549,
"loss": 1.6319,
"nll_loss": 1.5801714658737183,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15516701340675354,
"rewards/margins": 0.03184749186038971,
"rewards/rejected": -0.18701450526714325,
"step": 363
},
{
"epoch": 1.0027614773904039,
"grad_norm": 0.19933797419071198,
"learning_rate": 4.2160358512286266e-06,
"log_odds_chosen": 0.612895667552948,
"log_odds_ratio": -0.4389383792877197,
"logits/chosen": -0.19393223524093628,
"logits/rejected": -1.2420426607131958,
"logps/chosen": -1.498990535736084,
"logps/rejected": -2.0020782947540283,
"loss": 1.5772,
"nll_loss": 1.5332666635513306,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14989906549453735,
"rewards/margins": 0.05030875653028488,
"rewards/rejected": -0.20020782947540283,
"step": 364
},
{
"epoch": 1.0055229547808078,
"grad_norm": 0.18562446534633636,
"learning_rate": 4.210198983323366e-06,
"log_odds_chosen": 0.6299260258674622,
"log_odds_ratio": -0.4328676760196686,
"logits/chosen": -0.18327558040618896,
"logits/rejected": -1.4133291244506836,
"logps/chosen": -1.5473169088363647,
"logps/rejected": -2.070706605911255,
"loss": 1.6099,
"nll_loss": 1.5666420459747314,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15473169088363647,
"rewards/margins": 0.052338968962430954,
"rewards/rejected": -0.20707066357135773,
"step": 365
},
{
"epoch": 1.0082844321712117,
"grad_norm": 0.19783955812454224,
"learning_rate": 4.204344540477499e-06,
"log_odds_chosen": 0.6581704020500183,
"log_odds_ratio": -0.42357465624809265,
"logits/chosen": -0.21236993372440338,
"logits/rejected": -1.2212865352630615,
"logps/chosen": -1.5292942523956299,
"logps/rejected": -2.0793232917785645,
"loss": 1.6086,
"nll_loss": 1.5662219524383545,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.152929425239563,
"rewards/margins": 0.055002905428409576,
"rewards/rejected": -0.20793233811855316,
"step": 366
},
{
"epoch": 1.0110459095616156,
"grad_norm": 0.18359708786010742,
"learning_rate": 4.1984725828544855e-06,
"log_odds_chosen": 0.7003411054611206,
"log_odds_ratio": -0.40691572427749634,
"logits/chosen": -0.18985068798065186,
"logits/rejected": -1.434674620628357,
"logps/chosen": -1.4541864395141602,
"logps/rejected": -2.0293405055999756,
"loss": 1.531,
"nll_loss": 1.4903078079223633,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1454186588525772,
"rewards/margins": 0.05751540884375572,
"rewards/rejected": -0.20293407142162323,
"step": 367
},
{
"epoch": 1.0138073869520194,
"grad_norm": 0.19799236953258514,
"learning_rate": 4.192583170797775e-06,
"log_odds_chosen": 0.524876594543457,
"log_odds_ratio": -0.4676755368709564,
"logits/chosen": -0.2710600793361664,
"logits/rejected": -1.0796301364898682,
"logps/chosen": -1.492266058921814,
"logps/rejected": -1.9222991466522217,
"loss": 1.5776,
"nll_loss": 1.5308009386062622,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1492266058921814,
"rewards/margins": 0.04300331324338913,
"rewards/rejected": -0.19222994148731232,
"step": 368
},
{
"epoch": 1.016568864342423,
"grad_norm": 0.18693482875823975,
"learning_rate": 4.186676364830187e-06,
"log_odds_chosen": 0.5753119587898254,
"log_odds_ratio": -0.4493744671344757,
"logits/chosen": -0.1405227780342102,
"logits/rejected": -1.2737160921096802,
"logps/chosen": -1.5617775917053223,
"logps/rejected": -2.0400755405426025,
"loss": 1.6285,
"nll_loss": 1.5835627317428589,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15617777407169342,
"rewards/margins": 0.047829799354076385,
"rewards/rejected": -0.2040075659751892,
"step": 369
},
{
"epoch": 1.019330341732827,
"grad_norm": 0.21877062320709229,
"learning_rate": 4.1807522256532925e-06,
"log_odds_chosen": 0.6495308876037598,
"log_odds_ratio": -0.4310915768146515,
"logits/chosen": -0.20181499421596527,
"logits/rejected": -1.3658957481384277,
"logps/chosen": -1.6181979179382324,
"logps/rejected": -2.170933485031128,
"loss": 1.6861,
"nll_loss": 1.6429781913757324,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16181980073451996,
"rewards/margins": 0.05527355521917343,
"rewards/rejected": -0.217093363404274,
"step": 370
},
{
"epoch": 1.0220918191232309,
"grad_norm": 0.18539805710315704,
"learning_rate": 4.174810814146789e-06,
"log_odds_chosen": 0.6834878921508789,
"log_odds_ratio": -0.4169776141643524,
"logits/chosen": -0.08832372725009918,
"logits/rejected": -1.0964831113815308,
"logps/chosen": -1.5441397428512573,
"logps/rejected": -2.117156505584717,
"loss": 1.6166,
"nll_loss": 1.5748991966247559,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15441398322582245,
"rewards/margins": 0.057301655411720276,
"rewards/rejected": -0.21171565353870392,
"step": 371
},
{
"epoch": 1.0248532965136348,
"grad_norm": 0.1914224624633789,
"learning_rate": 4.1688521913678706e-06,
"log_odds_chosen": 0.7822255492210388,
"log_odds_ratio": -0.38675639033317566,
"logits/chosen": -0.16042651236057281,
"logits/rejected": -1.2148573398590088,
"logps/chosen": -1.4989123344421387,
"logps/rejected": -2.1540937423706055,
"loss": 1.5648,
"nll_loss": 1.5261112451553345,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14989124238491058,
"rewards/margins": 0.06551814079284668,
"rewards/rejected": -0.21540936827659607,
"step": 372
},
{
"epoch": 1.0276147739040387,
"grad_norm": 0.1842961460351944,
"learning_rate": 4.162876418550606e-06,
"log_odds_chosen": 0.5339972376823425,
"log_odds_ratio": -0.46532902121543884,
"logits/chosen": -0.2638809084892273,
"logits/rejected": -1.291076421737671,
"logps/chosen": -1.506029486656189,
"logps/rejected": -1.944475769996643,
"loss": 1.5811,
"nll_loss": 1.5345618724822998,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15060293674468994,
"rewards/margins": 0.043844640254974365,
"rewards/rejected": -0.1944475769996643,
"step": 373
},
{
"epoch": 1.0303762512944425,
"grad_norm": 0.19836845993995667,
"learning_rate": 4.156883557105308e-06,
"log_odds_chosen": 0.5632570385932922,
"log_odds_ratio": -0.46161651611328125,
"logits/chosen": -0.23987069725990295,
"logits/rejected": -1.331959843635559,
"logps/chosen": -1.4848788976669312,
"logps/rejected": -1.9496028423309326,
"loss": 1.5517,
"nll_loss": 1.5055512189865112,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1484878808259964,
"rewards/margins": 0.04647241160273552,
"rewards/rejected": -0.19496029615402222,
"step": 374
},
{
"epoch": 1.0331377286848464,
"grad_norm": 0.20077650249004364,
"learning_rate": 4.150873668617899e-06,
"log_odds_chosen": 0.5416385531425476,
"log_odds_ratio": -0.4598068594932556,
"logits/chosen": -0.2892989218235016,
"logits/rejected": -1.227949619293213,
"logps/chosen": -1.5431418418884277,
"logps/rejected": -1.9903440475463867,
"loss": 1.6064,
"nll_loss": 1.5604456663131714,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15431420505046844,
"rewards/margins": 0.04472021758556366,
"rewards/rejected": -0.1990344226360321,
"step": 375
},
{
"epoch": 1.0358992060752503,
"grad_norm": 0.1922207772731781,
"learning_rate": 4.144846814849282e-06,
"log_odds_chosen": 0.5037797093391418,
"log_odds_ratio": -0.4755171537399292,
"logits/chosen": -0.10908980667591095,
"logits/rejected": -1.1120904684066772,
"logps/chosen": -1.5532734394073486,
"logps/rejected": -1.9692023992538452,
"loss": 1.6357,
"nll_loss": 1.5881571769714355,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15532734990119934,
"rewards/margins": 0.04159289598464966,
"rewards/rejected": -0.1969202607870102,
"step": 376
},
{
"epoch": 1.0386606834656542,
"grad_norm": 0.19230647385120392,
"learning_rate": 4.138803057734705e-06,
"log_odds_chosen": 0.7433000802993774,
"log_odds_ratio": -0.3926478624343872,
"logits/chosen": -0.16731296479701996,
"logits/rejected": -1.154883623123169,
"logps/chosen": -1.4572123289108276,
"logps/rejected": -2.0705974102020264,
"loss": 1.5333,
"nll_loss": 1.4939861297607422,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14572124183177948,
"rewards/margins": 0.06133852154016495,
"rewards/rejected": -0.20705974102020264,
"step": 377
},
{
"epoch": 1.041422160856058,
"grad_norm": 0.1747986078262329,
"learning_rate": 4.132742459383122e-06,
"log_odds_chosen": 0.8132926821708679,
"log_odds_ratio": -0.37535253167152405,
"logits/chosen": -0.19614934921264648,
"logits/rejected": -1.3223804235458374,
"logps/chosen": -1.3767224550247192,
"logps/rejected": -2.0322914123535156,
"loss": 1.4591,
"nll_loss": 1.4215335845947266,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13767226040363312,
"rewards/margins": 0.06555688381195068,
"rewards/rejected": -0.2032291442155838,
"step": 378
},
{
"epoch": 1.0441836382464618,
"grad_norm": 0.18519026041030884,
"learning_rate": 4.126665082076559e-06,
"log_odds_chosen": 0.5479438900947571,
"log_odds_ratio": -0.46051234006881714,
"logits/chosen": -0.21864745020866394,
"logits/rejected": -0.9870752096176147,
"logps/chosen": -1.5094008445739746,
"logps/rejected": -1.9554524421691895,
"loss": 1.5949,
"nll_loss": 1.5488578081130981,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15094009041786194,
"rewards/margins": 0.044605158269405365,
"rewards/rejected": -0.1955452561378479,
"step": 379
},
{
"epoch": 1.0469451156368657,
"grad_norm": 0.1884365975856781,
"learning_rate": 4.120570988269472e-06,
"log_odds_chosen": 0.6451675891876221,
"log_odds_ratio": -0.4251843988895416,
"logits/chosen": -0.28254202008247375,
"logits/rejected": -1.3985097408294678,
"logps/chosen": -1.5068163871765137,
"logps/rejected": -2.0416600704193115,
"loss": 1.5559,
"nll_loss": 1.5133352279663086,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15068164467811584,
"rewards/margins": 0.053484346717596054,
"rewards/rejected": -0.2041660100221634,
"step": 380
},
{
"epoch": 1.0497065930272695,
"grad_norm": 0.20361238718032837,
"learning_rate": 4.114460240588101e-06,
"log_odds_chosen": 0.8017259836196899,
"log_odds_ratio": -0.3787066638469696,
"logits/chosen": -0.18577685952186584,
"logits/rejected": -1.1138646602630615,
"logps/chosen": -1.5362298488616943,
"logps/rejected": -2.2140719890594482,
"loss": 1.6003,
"nll_loss": 1.5624499320983887,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15362299978733063,
"rewards/margins": 0.06778421998023987,
"rewards/rejected": -0.2214072048664093,
"step": 381
},
{
"epoch": 1.0524680704176734,
"grad_norm": 0.18294233083724976,
"learning_rate": 4.1083329018298356e-06,
"log_odds_chosen": 0.7440272569656372,
"log_odds_ratio": -0.39435288310050964,
"logits/chosen": -0.18778713047504425,
"logits/rejected": -1.1963489055633545,
"logps/chosen": -1.4442341327667236,
"logps/rejected": -2.0572502613067627,
"loss": 1.5068,
"nll_loss": 1.4673796892166138,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14442341029644012,
"rewards/margins": 0.06130162253975868,
"rewards/rejected": -0.2057250291109085,
"step": 382
},
{
"epoch": 1.0552295478080773,
"grad_norm": 0.19120609760284424,
"learning_rate": 4.102189034962561e-06,
"log_odds_chosen": 0.6499487161636353,
"log_odds_ratio": -0.4234417676925659,
"logits/chosen": -0.12440590560436249,
"logits/rejected": -1.0700702667236328,
"logps/chosen": -1.5787014961242676,
"logps/rejected": -2.125746726989746,
"loss": 1.6325,
"nll_loss": 1.5901223421096802,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15787014365196228,
"rewards/margins": 0.05470450222492218,
"rewards/rejected": -0.21257463097572327,
"step": 383
},
{
"epoch": 1.0579910251984812,
"grad_norm": 0.20611584186553955,
"learning_rate": 4.096028703124014e-06,
"log_odds_chosen": 0.6581447720527649,
"log_odds_ratio": -0.427496999502182,
"logits/chosen": -0.243414044380188,
"logits/rejected": -1.0197745561599731,
"logps/chosen": -1.5716708898544312,
"logps/rejected": -2.1266674995422363,
"loss": 1.6172,
"nll_loss": 1.5744341611862183,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15716709196567535,
"rewards/margins": 0.055499687790870667,
"rewards/rejected": -0.21266677975654602,
"step": 384
},
{
"epoch": 1.060752502588885,
"grad_norm": 0.19265590608119965,
"learning_rate": 4.089851969621138e-06,
"log_odds_chosen": 0.7516465783119202,
"log_odds_ratio": -0.4052088260650635,
"logits/chosen": -0.19772981107234955,
"logits/rejected": -1.2006863355636597,
"logps/chosen": -1.4199714660644531,
"logps/rejected": -2.0437204837799072,
"loss": 1.5074,
"nll_loss": 1.466860055923462,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14199714362621307,
"rewards/margins": 0.062374889850616455,
"rewards/rejected": -0.20437204837799072,
"step": 385
},
{
"epoch": 1.063513979979289,
"grad_norm": 0.21174004673957825,
"learning_rate": 4.083658897929425e-06,
"log_odds_chosen": 0.6503540277481079,
"log_odds_ratio": -0.4217776358127594,
"logits/chosen": -0.2460387498140335,
"logits/rejected": -1.1329982280731201,
"logps/chosen": -1.5417426824569702,
"logps/rejected": -2.0824155807495117,
"loss": 1.605,
"nll_loss": 1.5628407001495361,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15417428314685822,
"rewards/margins": 0.05406728759407997,
"rewards/rejected": -0.2082415521144867,
"step": 386
},
{
"epoch": 1.0662754573696929,
"grad_norm": 0.19454148411750793,
"learning_rate": 4.077449551692268e-06,
"log_odds_chosen": 0.5684500336647034,
"log_odds_ratio": -0.45333224534988403,
"logits/chosen": -0.13818205893039703,
"logits/rejected": -1.2238332033157349,
"logps/chosen": -1.5496622323989868,
"logps/rejected": -2.0205225944519043,
"loss": 1.6156,
"nll_loss": 1.570224404335022,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15496623516082764,
"rewards/margins": 0.047086022794246674,
"rewards/rejected": -0.2020522505044937,
"step": 387
},
{
"epoch": 1.0690369347600965,
"grad_norm": 0.19282881915569305,
"learning_rate": 4.071223994720309e-06,
"log_odds_chosen": 0.7608233690261841,
"log_odds_ratio": -0.3879122734069824,
"logits/chosen": -0.26540690660476685,
"logits/rejected": -1.2793084383010864,
"logps/chosen": -1.4896953105926514,
"logps/rejected": -2.125817060470581,
"loss": 1.557,
"nll_loss": 1.518183946609497,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14896953105926514,
"rewards/margins": 0.06361216306686401,
"rewards/rejected": -0.21258167922496796,
"step": 388
},
{
"epoch": 1.0717984121505004,
"grad_norm": 0.18759317696094513,
"learning_rate": 4.064982290990777e-06,
"log_odds_chosen": 0.6020365357398987,
"log_odds_ratio": -0.44241786003112793,
"logits/chosen": -0.2355424463748932,
"logits/rejected": -1.0852099657058716,
"logps/chosen": -1.4403340816497803,
"logps/rejected": -1.9294434785842896,
"loss": 1.5147,
"nll_loss": 1.4704298973083496,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14403343200683594,
"rewards/margins": 0.04891093447804451,
"rewards/rejected": -0.19294434785842896,
"step": 389
},
{
"epoch": 1.0745598895409043,
"grad_norm": 0.19443197548389435,
"learning_rate": 4.058724504646834e-06,
"log_odds_chosen": 0.725110650062561,
"log_odds_ratio": -0.4022667109966278,
"logits/chosen": -0.25374311208724976,
"logits/rejected": -1.207698106765747,
"logps/chosen": -1.4946649074554443,
"logps/rejected": -2.0988166332244873,
"loss": 1.5592,
"nll_loss": 1.518977403640747,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14946648478507996,
"rewards/margins": 0.060415178537368774,
"rewards/rejected": -0.20988167822360992,
"step": 390
},
{
"epoch": 1.0773213669313082,
"grad_norm": 0.19158363342285156,
"learning_rate": 4.0524506999969185e-06,
"log_odds_chosen": 0.5608286261558533,
"log_odds_ratio": -0.4582407772541046,
"logits/chosen": -0.21252039074897766,
"logits/rejected": -1.3294453620910645,
"logps/chosen": -1.54707670211792,
"logps/rejected": -2.0072052478790283,
"loss": 1.606,
"nll_loss": 1.560204029083252,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15470768511295319,
"rewards/margins": 0.04601283743977547,
"rewards/rejected": -0.20072051882743835,
"step": 391
},
{
"epoch": 1.080082844321712,
"grad_norm": 0.18796156346797943,
"learning_rate": 4.046160941514079e-06,
"log_odds_chosen": 0.7382424473762512,
"log_odds_ratio": -0.3966708481311798,
"logits/chosen": -0.1265019178390503,
"logits/rejected": -0.962821900844574,
"logps/chosen": -1.4349395036697388,
"logps/rejected": -2.038311719894409,
"loss": 1.4915,
"nll_loss": 1.4517992734909058,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14349395036697388,
"rewards/margins": 0.06033723056316376,
"rewards/rejected": -0.20383116602897644,
"step": 392
},
{
"epoch": 1.082844321712116,
"grad_norm": 0.18811720609664917,
"learning_rate": 4.039855293835316e-06,
"log_odds_chosen": 0.6325368881225586,
"log_odds_ratio": -0.4284062683582306,
"logits/chosen": -0.2596244513988495,
"logits/rejected": -1.2793852090835571,
"logps/chosen": -1.4744428396224976,
"logps/rejected": -1.99100923538208,
"loss": 1.5311,
"nll_loss": 1.4882373809814453,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14744427800178528,
"rewards/margins": 0.05165664479136467,
"rewards/rejected": -0.19910094141960144,
"step": 393
},
{
"epoch": 1.0856057991025199,
"grad_norm": 0.17588721215724945,
"learning_rate": 4.033533821760917e-06,
"log_odds_chosen": 0.8095629215240479,
"log_odds_ratio": -0.3785027265548706,
"logits/chosen": -0.2389419972896576,
"logits/rejected": -1.3366788625717163,
"logps/chosen": -1.4380571842193604,
"logps/rejected": -2.1090192794799805,
"loss": 1.5041,
"nll_loss": 1.4662492275238037,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14380571246147156,
"rewards/margins": 0.06709621846675873,
"rewards/rejected": -0.21090193092823029,
"step": 394
},
{
"epoch": 1.0883672764929238,
"grad_norm": 0.18705010414123535,
"learning_rate": 4.027196590253786e-06,
"log_odds_chosen": 0.5284130573272705,
"log_odds_ratio": -0.469849556684494,
"logits/chosen": -0.17493271827697754,
"logits/rejected": -1.0048437118530273,
"logps/chosen": -1.4886696338653564,
"logps/rejected": -1.9172314405441284,
"loss": 1.5668,
"nll_loss": 1.5198521614074707,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1488669514656067,
"rewards/margins": 0.04285619407892227,
"rewards/rejected": -0.19172316789627075,
"step": 395
},
{
"epoch": 1.0911287538833276,
"grad_norm": 0.1907300055027008,
"learning_rate": 4.020843664438783e-06,
"log_odds_chosen": 0.6281604766845703,
"log_odds_ratio": -0.43829861283302307,
"logits/chosen": -0.19279785454273224,
"logits/rejected": -1.2376521825790405,
"logps/chosen": -1.4428083896636963,
"logps/rejected": -1.9556207656860352,
"loss": 1.5237,
"nll_loss": 1.4798742532730103,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1442808359861374,
"rewards/margins": 0.05128122866153717,
"rewards/rejected": -0.19556206464767456,
"step": 396
},
{
"epoch": 1.0938902312737315,
"grad_norm": 0.18668295443058014,
"learning_rate": 4.01447510960205e-06,
"log_odds_chosen": 0.7544000148773193,
"log_odds_ratio": -0.39146313071250916,
"logits/chosen": -0.2915095388889313,
"logits/rejected": -1.2768161296844482,
"logps/chosen": -1.5208818912506104,
"logps/rejected": -2.149923801422119,
"loss": 1.5759,
"nll_loss": 1.5367605686187744,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1520881950855255,
"rewards/margins": 0.06290420144796371,
"rewards/rejected": -0.21499237418174744,
"step": 397
},
{
"epoch": 1.0966517086641354,
"grad_norm": 0.1973486691713333,
"learning_rate": 4.008090991190341e-06,
"log_odds_chosen": 0.5326504111289978,
"log_odds_ratio": -0.46492505073547363,
"logits/chosen": -0.09128017723560333,
"logits/rejected": -1.2047038078308105,
"logps/chosen": -1.5613563060760498,
"logps/rejected": -2.0032505989074707,
"loss": 1.6159,
"nll_loss": 1.569425344467163,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15613561868667603,
"rewards/margins": 0.0441894605755806,
"rewards/rejected": -0.20032507181167603,
"step": 398
},
{
"epoch": 1.099413186054539,
"grad_norm": 0.17714130878448486,
"learning_rate": 4.001691374810352e-06,
"log_odds_chosen": 0.815579354763031,
"log_odds_ratio": -0.3770541846752167,
"logits/chosen": -0.1730899065732956,
"logits/rejected": -1.437078833580017,
"logps/chosen": -1.513384222984314,
"logps/rejected": -2.199054718017578,
"loss": 1.5737,
"nll_loss": 1.5360060930252075,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15133842825889587,
"rewards/margins": 0.06856706738471985,
"rewards/rejected": -0.21990549564361572,
"step": 399
},
{
"epoch": 1.102174663444943,
"grad_norm": 0.16967174410820007,
"learning_rate": 3.99527632622804e-06,
"log_odds_chosen": 0.694899320602417,
"log_odds_ratio": -0.4119797945022583,
"logits/chosen": -0.21181365847587585,
"logits/rejected": -1.1213148832321167,
"logps/chosen": -1.4379510879516602,
"logps/rejected": -2.007990837097168,
"loss": 1.5198,
"nll_loss": 1.478610634803772,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14379511773586273,
"rewards/margins": 0.05700398236513138,
"rewards/rejected": -0.2007990926504135,
"step": 400
},
{
"epoch": 1.1049361408353469,
"grad_norm": 0.1991468220949173,
"learning_rate": 3.988845911367957e-06,
"log_odds_chosen": 0.665392279624939,
"log_odds_ratio": -0.423291951417923,
"logits/chosen": -0.19501212239265442,
"logits/rejected": -1.1544811725616455,
"logps/chosen": -1.5978668928146362,
"logps/rejected": -2.159170389175415,
"loss": 1.6409,
"nll_loss": 1.598615050315857,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15978670120239258,
"rewards/margins": 0.05613034963607788,
"rewards/rejected": -0.21591705083847046,
"step": 401
},
{
"epoch": 1.1076976182257507,
"grad_norm": 0.17201201617717743,
"learning_rate": 3.982400196312565e-06,
"log_odds_chosen": 0.7169694900512695,
"log_odds_ratio": -0.39898163080215454,
"logits/chosen": -0.18280558288097382,
"logits/rejected": -1.0723071098327637,
"logps/chosen": -1.4075336456298828,
"logps/rejected": -1.985740303993225,
"loss": 1.4851,
"nll_loss": 1.4452052116394043,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.140753373503685,
"rewards/margins": 0.05782065540552139,
"rewards/rejected": -0.198574036359787,
"step": 402
},
{
"epoch": 1.1104590956161546,
"grad_norm": 0.17941723763942719,
"learning_rate": 3.975939247301558e-06,
"log_odds_chosen": 0.6592923998832703,
"log_odds_ratio": -0.420391321182251,
"logits/chosen": -0.22413836419582367,
"logits/rejected": -1.327505111694336,
"logps/chosen": -1.541399598121643,
"logps/rejected": -2.0909154415130615,
"loss": 1.6127,
"nll_loss": 1.570648193359375,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1541399508714676,
"rewards/margins": 0.05495157092809677,
"rewards/rejected": -0.20909152925014496,
"step": 403
},
{
"epoch": 1.1132205730065585,
"grad_norm": 0.17807722091674805,
"learning_rate": 3.969463130731183e-06,
"log_odds_chosen": 0.6306886076927185,
"log_odds_ratio": -0.43381085991859436,
"logits/chosen": -0.14202114939689636,
"logits/rejected": -1.2107291221618652,
"logps/chosen": -1.4741935729980469,
"logps/rejected": -1.9913864135742188,
"loss": 1.5431,
"nll_loss": 1.4997388124465942,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14741936326026917,
"rewards/margins": 0.051719292998313904,
"rewards/rejected": -0.19913865625858307,
"step": 404
},
{
"epoch": 1.1159820503969624,
"grad_norm": 0.1876784712076187,
"learning_rate": 3.9629719131535595e-06,
"log_odds_chosen": 0.5967572331428528,
"log_odds_ratio": -0.4506050944328308,
"logits/chosen": -0.24889226257801056,
"logits/rejected": -1.0915982723236084,
"logps/chosen": -1.4914859533309937,
"logps/rejected": -1.9845982789993286,
"loss": 1.5603,
"nll_loss": 1.5152877569198608,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1491485983133316,
"rewards/margins": 0.049311213195323944,
"rewards/rejected": -0.19845981895923615,
"step": 405
},
{
"epoch": 1.1187435277873663,
"grad_norm": 0.16931727528572083,
"learning_rate": 3.9564656612759904e-06,
"log_odds_chosen": 0.6932112574577332,
"log_odds_ratio": -0.4113208055496216,
"logits/chosen": -0.1822027862071991,
"logits/rejected": -1.3579068183898926,
"logps/chosen": -1.496368646621704,
"logps/rejected": -2.0719945430755615,
"loss": 1.563,
"nll_loss": 1.521822214126587,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14963684976100922,
"rewards/margins": 0.05756259709596634,
"rewards/rejected": -0.20719945430755615,
"step": 406
},
{
"epoch": 1.1215050051777702,
"grad_norm": 0.18602602183818817,
"learning_rate": 3.94994444196028e-06,
"log_odds_chosen": 0.7077181935310364,
"log_odds_ratio": -0.4127015471458435,
"logits/chosen": -0.15379171073436737,
"logits/rejected": -1.1528825759887695,
"logps/chosen": -1.5301098823547363,
"logps/rejected": -2.1211650371551514,
"loss": 1.5721,
"nll_loss": 1.5308395624160767,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15301097929477692,
"rewards/margins": 0.0591055229306221,
"rewards/rejected": -0.21211649477481842,
"step": 407
},
{
"epoch": 1.124266482568174,
"grad_norm": 0.17830830812454224,
"learning_rate": 3.943408322222049e-06,
"log_odds_chosen": 0.7005403637886047,
"log_odds_ratio": -0.41374269127845764,
"logits/chosen": -0.1289825439453125,
"logits/rejected": -0.9252943396568298,
"logps/chosen": -1.4544087648391724,
"logps/rejected": -2.0350112915039062,
"loss": 1.5331,
"nll_loss": 1.4917351007461548,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14544087648391724,
"rewards/margins": 0.05806024372577667,
"rewards/rejected": -0.2035011202096939,
"step": 408
},
{
"epoch": 1.127027959958578,
"grad_norm": 0.2026163637638092,
"learning_rate": 3.936857369230037e-06,
"log_odds_chosen": 0.5416731834411621,
"log_odds_ratio": -0.4673037528991699,
"logits/chosen": -0.12443944066762924,
"logits/rejected": -1.1075230836868286,
"logps/chosen": -1.5275638103485107,
"logps/rejected": -1.9750971794128418,
"loss": 1.5827,
"nll_loss": 1.5359312295913696,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15275639295578003,
"rewards/margins": 0.04475332424044609,
"rewards/rejected": -0.19750970602035522,
"step": 409
},
{
"epoch": 1.1297894373489816,
"grad_norm": 0.19184917211532593,
"learning_rate": 3.930291650305424e-06,
"log_odds_chosen": 0.741117000579834,
"log_odds_ratio": -0.39208146929740906,
"logits/chosen": -0.14100222289562225,
"logits/rejected": -0.9723483920097351,
"logps/chosen": -1.4884588718414307,
"logps/rejected": -2.1040899753570557,
"loss": 1.5337,
"nll_loss": 1.494455099105835,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14884589612483978,
"rewards/margins": 0.061563119292259216,
"rewards/rejected": -0.2104090005159378,
"step": 410
},
{
"epoch": 1.1325509147393855,
"grad_norm": 0.18398518860340118,
"learning_rate": 3.92371123292113e-06,
"log_odds_chosen": 0.665544867515564,
"log_odds_ratio": -0.41778889298439026,
"logits/chosen": -0.13632512092590332,
"logits/rejected": -1.2031248807907104,
"logps/chosen": -1.5206879377365112,
"logps/rejected": -2.072800636291504,
"loss": 1.5771,
"nll_loss": 1.5352935791015625,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15206880867481232,
"rewards/margins": 0.05521126464009285,
"rewards/rejected": -0.20728005468845367,
"step": 411
},
{
"epoch": 1.1353123921297894,
"grad_norm": 0.18211683630943298,
"learning_rate": 3.917116184701125e-06,
"log_odds_chosen": 0.7499443292617798,
"log_odds_ratio": -0.39630863070487976,
"logits/chosen": -0.17687299847602844,
"logits/rejected": -1.3003441095352173,
"logps/chosen": -1.4918804168701172,
"logps/rejected": -2.1197023391723633,
"loss": 1.5502,
"nll_loss": 1.5105260610580444,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14918804168701172,
"rewards/margins": 0.06278219819068909,
"rewards/rejected": -0.2119702398777008,
"step": 412
},
{
"epoch": 1.1380738695201933,
"grad_norm": 0.20253390073776245,
"learning_rate": 3.910506573419734e-06,
"log_odds_chosen": 0.9193601012229919,
"log_odds_ratio": -0.34398138523101807,
"logits/chosen": -0.14900241792201996,
"logits/rejected": -1.0627330541610718,
"logps/chosen": -1.4866719245910645,
"logps/rejected": -2.259812593460083,
"loss": 1.5385,
"nll_loss": 1.5040783882141113,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14866718649864197,
"rewards/margins": 0.07731407880783081,
"rewards/rejected": -0.22598126530647278,
"step": 413
},
{
"epoch": 1.1408353469105972,
"grad_norm": 0.18668848276138306,
"learning_rate": 3.903882467000938e-06,
"log_odds_chosen": 0.8500868678092957,
"log_odds_ratio": -0.3602939248085022,
"logits/chosen": -0.13838572800159454,
"logits/rejected": -1.1434731483459473,
"logps/chosen": -1.5281840562820435,
"logps/rejected": -2.244469404220581,
"loss": 1.5785,
"nll_loss": 1.542461633682251,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15281839668750763,
"rewards/margins": 0.07162855565547943,
"rewards/rejected": -0.22444695234298706,
"step": 414
},
{
"epoch": 1.143596824301001,
"grad_norm": 0.18311648070812225,
"learning_rate": 3.897243933517679e-06,
"log_odds_chosen": 0.7283087968826294,
"log_odds_ratio": -0.40107017755508423,
"logits/chosen": -0.2569331228733063,
"logits/rejected": -1.2123123407363892,
"logps/chosen": -1.4449797868728638,
"logps/rejected": -2.044313669204712,
"loss": 1.5036,
"nll_loss": 1.463518738746643,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14449797570705414,
"rewards/margins": 0.05993340164422989,
"rewards/rejected": -0.20443139970302582,
"step": 415
},
{
"epoch": 1.146358301691405,
"grad_norm": 0.18268708884716034,
"learning_rate": 3.890591041191162e-06,
"log_odds_chosen": 0.8010820150375366,
"log_odds_ratio": -0.38091176748275757,
"logits/chosen": -0.15429654717445374,
"logits/rejected": -1.0080270767211914,
"logps/chosen": -1.4056251049041748,
"logps/rejected": -2.0608270168304443,
"loss": 1.4656,
"nll_loss": 1.4274916648864746,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1405625194311142,
"rewards/margins": 0.06552018225193024,
"rewards/rejected": -0.20608270168304443,
"step": 416
},
{
"epoch": 1.1491197790818088,
"grad_norm": 0.198701411485672,
"learning_rate": 3.883923858390149e-06,
"log_odds_chosen": 0.818614661693573,
"log_odds_ratio": -0.36802011728286743,
"logits/chosen": -0.189392551779747,
"logits/rejected": -1.2750192880630493,
"logps/chosen": -1.5129953622817993,
"logps/rejected": -2.201220750808716,
"loss": 1.5794,
"nll_loss": 1.5426466464996338,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15129955112934113,
"rewards/margins": 0.06882252544164658,
"rewards/rejected": -0.2201220840215683,
"step": 417
},
{
"epoch": 1.1518812564722127,
"grad_norm": 0.17471423745155334,
"learning_rate": 3.8772424536302565e-06,
"log_odds_chosen": 0.6130560040473938,
"log_odds_ratio": -0.4377739131450653,
"logits/chosen": -0.19274210929870605,
"logits/rejected": -1.2090108394622803,
"logps/chosen": -1.4925827980041504,
"logps/rejected": -1.9969313144683838,
"loss": 1.5448,
"nll_loss": 1.50102698802948,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14925827085971832,
"rewards/margins": 0.05043485015630722,
"rewards/rejected": -0.19969312846660614,
"step": 418
},
{
"epoch": 1.1546427338626164,
"grad_norm": 0.19965821504592896,
"learning_rate": 3.870546895573258e-06,
"log_odds_chosen": 0.7444138526916504,
"log_odds_ratio": -0.39596325159072876,
"logits/chosen": -0.24373914301395416,
"logits/rejected": -0.921118438243866,
"logps/chosen": -1.4715092182159424,
"logps/rejected": -2.086350202560425,
"loss": 1.5252,
"nll_loss": 1.4855574369430542,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.147150918841362,
"rewards/margins": 0.06148412078619003,
"rewards/rejected": -0.20863503217697144,
"step": 419
},
{
"epoch": 1.1574042112530203,
"grad_norm": 0.18526706099510193,
"learning_rate": 3.863837253026372e-06,
"log_odds_chosen": 0.8123873472213745,
"log_odds_ratio": -0.37982794642448425,
"logits/chosen": -0.23291738331317902,
"logits/rejected": -1.1363976001739502,
"logps/chosen": -1.4304343461990356,
"logps/rejected": -2.1056971549987793,
"loss": 1.4989,
"nll_loss": 1.460911512374878,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1430434137582779,
"rewards/margins": 0.06752629578113556,
"rewards/rejected": -0.21056970953941345,
"step": 420
},
{
"epoch": 1.1601656886434242,
"grad_norm": 0.18150660395622253,
"learning_rate": 3.857113594941556e-06,
"log_odds_chosen": 0.6511182188987732,
"log_odds_ratio": -0.4259107708930969,
"logits/chosen": -0.19661128520965576,
"logits/rejected": -1.3280483484268188,
"logps/chosen": -1.527599811553955,
"logps/rejected": -2.070669174194336,
"loss": 1.5857,
"nll_loss": 1.5431334972381592,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15275999903678894,
"rewards/margins": 0.054306939244270325,
"rewards/rejected": -0.20706692337989807,
"step": 421
},
{
"epoch": 1.162927166033828,
"grad_norm": 0.18328295648097992,
"learning_rate": 3.8503759904148005e-06,
"log_odds_chosen": 0.8013736605644226,
"log_odds_ratio": -0.3758004903793335,
"logits/chosen": -0.16402505338191986,
"logits/rejected": -1.2019381523132324,
"logps/chosen": -1.4596202373504639,
"logps/rejected": -2.1256799697875977,
"loss": 1.5176,
"nll_loss": 1.479976773262024,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14596202969551086,
"rewards/margins": 0.06660597026348114,
"rewards/rejected": -0.2125680148601532,
"step": 422
},
{
"epoch": 1.165688643424232,
"grad_norm": 0.18774062395095825,
"learning_rate": 3.843624508685416e-06,
"log_odds_chosen": 0.6741689443588257,
"log_odds_ratio": -0.41605666279792786,
"logits/chosen": -0.20964159071445465,
"logits/rejected": -1.2953227758407593,
"logps/chosen": -1.5373965501785278,
"logps/rejected": -2.098203659057617,
"loss": 1.6025,
"nll_loss": 1.5609112977981567,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15373964607715607,
"rewards/margins": 0.05608072504401207,
"rewards/rejected": -0.20982035994529724,
"step": 423
},
{
"epoch": 1.1684501208146358,
"grad_norm": 0.1771015226840973,
"learning_rate": 3.8368592191353246e-06,
"log_odds_chosen": 0.7563647031784058,
"log_odds_ratio": -0.3966205418109894,
"logits/chosen": -0.19912122189998627,
"logits/rejected": -1.2726441621780396,
"logps/chosen": -1.4663722515106201,
"logps/rejected": -2.0917129516601562,
"loss": 1.5165,
"nll_loss": 1.476802945137024,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1466372162103653,
"rewards/margins": 0.06253407895565033,
"rewards/rejected": -0.20917129516601562,
"step": 424
},
{
"epoch": 1.1712115982050397,
"grad_norm": 0.17853626608848572,
"learning_rate": 3.830080191288342e-06,
"log_odds_chosen": 0.672751784324646,
"log_odds_ratio": -0.41323938965797424,
"logits/chosen": -0.1510731726884842,
"logits/rejected": -1.1135554313659668,
"logps/chosen": -1.4936147928237915,
"logps/rejected": -2.0494277477264404,
"loss": 1.5598,
"nll_loss": 1.518498420715332,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1493614763021469,
"rewards/margins": 0.05558129400014877,
"rewards/rejected": -0.20494277775287628,
"step": 425
},
{
"epoch": 1.1739730755954436,
"grad_norm": 0.19094951450824738,
"learning_rate": 3.823287494809469e-06,
"log_odds_chosen": 0.7997311949729919,
"log_odds_ratio": -0.3781062364578247,
"logits/chosen": -0.25045046210289,
"logits/rejected": -1.166719913482666,
"logps/chosen": -1.472108244895935,
"logps/rejected": -2.1340489387512207,
"loss": 1.5392,
"nll_loss": 1.501369595527649,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14721082150936127,
"rewards/margins": 0.0661940723657608,
"rewards/rejected": -0.21340489387512207,
"step": 426
},
{
"epoch": 1.1767345529858475,
"grad_norm": 0.1843455731868744,
"learning_rate": 3.816481199504171e-06,
"log_odds_chosen": 0.6943097114562988,
"log_odds_ratio": -0.4069713354110718,
"logits/chosen": -0.2397070825099945,
"logits/rejected": -1.1366225481033325,
"logps/chosen": -1.5034641027450562,
"logps/rejected": -2.079597234725952,
"loss": 1.5747,
"nll_loss": 1.5340169668197632,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15034641325473785,
"rewards/margins": 0.057613298296928406,
"rewards/rejected": -0.20795971155166626,
"step": 427
},
{
"epoch": 1.1794960303762512,
"grad_norm": 0.17388851940631866,
"learning_rate": 3.8096613753176635e-06,
"log_odds_chosen": 0.6214280128479004,
"log_odds_ratio": -0.436628133058548,
"logits/chosen": -0.1547197550535202,
"logits/rejected": -1.0357825756072998,
"logps/chosen": -1.3858633041381836,
"logps/rejected": -1.8819453716278076,
"loss": 1.4633,
"nll_loss": 1.4196813106536865,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13858634233474731,
"rewards/margins": 0.04960820823907852,
"rewards/rejected": -0.18819454312324524,
"step": 428
},
{
"epoch": 1.1822575077666553,
"grad_norm": 0.1828448623418808,
"learning_rate": 3.8028280923341927e-06,
"log_odds_chosen": 0.7184998393058777,
"log_odds_ratio": -0.40824440121650696,
"logits/chosen": -0.2581023573875427,
"logits/rejected": -1.0807780027389526,
"logps/chosen": -1.4268940687179565,
"logps/rejected": -2.0194613933563232,
"loss": 1.4935,
"nll_loss": 1.4526987075805664,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14268942177295685,
"rewards/margins": 0.05925672873854637,
"rewards/rejected": -0.20194613933563232,
"step": 429
},
{
"epoch": 1.185018985157059,
"grad_norm": 0.2609389126300812,
"learning_rate": 3.7959814207763134e-06,
"log_odds_chosen": 0.8415360450744629,
"log_odds_ratio": -0.366547167301178,
"logits/chosen": -0.2016637921333313,
"logits/rejected": -1.2747056484222412,
"logps/chosen": -1.539664387702942,
"logps/rejected": -2.2510623931884766,
"loss": 1.5835,
"nll_loss": 1.5468266010284424,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15396642684936523,
"rewards/margins": 0.07113979011774063,
"rewards/rejected": -0.22510623931884766,
"step": 430
},
{
"epoch": 1.1877804625474628,
"grad_norm": 0.18396784365177155,
"learning_rate": 3.789121431004168e-06,
"log_odds_chosen": 0.6930549144744873,
"log_odds_ratio": -0.4097321629524231,
"logits/chosen": -0.24285678565502167,
"logits/rejected": -1.232874870300293,
"logps/chosen": -1.5071865320205688,
"logps/rejected": -2.07889986038208,
"loss": 1.5539,
"nll_loss": 1.5129239559173584,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15071865916252136,
"rewards/margins": 0.05717131495475769,
"rewards/rejected": -0.20788997411727905,
"step": 431
},
{
"epoch": 1.1905419399378667,
"grad_norm": 0.1961180865764618,
"learning_rate": 3.782248193514766e-06,
"log_odds_chosen": 0.6803969144821167,
"log_odds_ratio": -0.41614609956741333,
"logits/chosen": -0.1777564138174057,
"logits/rejected": -1.077903151512146,
"logps/chosen": -1.4115687608718872,
"logps/rejected": -1.9654287099838257,
"loss": 1.4906,
"nll_loss": 1.4489516019821167,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14115691184997559,
"rewards/margins": 0.05538597330451012,
"rewards/rejected": -0.1965428739786148,
"step": 432
},
{
"epoch": 1.1933034173282706,
"grad_norm": 0.18955354392528534,
"learning_rate": 3.775361778941257e-06,
"log_odds_chosen": 0.7743215560913086,
"log_odds_ratio": -0.3833809196949005,
"logits/chosen": -0.19383692741394043,
"logits/rejected": -1.1276659965515137,
"logps/chosen": -1.4961435794830322,
"logps/rejected": -2.144127607345581,
"loss": 1.5554,
"nll_loss": 1.5170167684555054,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1496143639087677,
"rewards/margins": 0.06479839235544205,
"rewards/rejected": -0.21441277861595154,
"step": 433
},
{
"epoch": 1.1960648947186745,
"grad_norm": 0.19281843304634094,
"learning_rate": 3.7684622580522057e-06,
"log_odds_chosen": 0.45453017950057983,
"log_odds_ratio": -0.49613097310066223,
"logits/chosen": -0.1650485396385193,
"logits/rejected": -1.2051373720169067,
"logps/chosen": -1.447697401046753,
"logps/rejected": -1.8144056797027588,
"loss": 1.5077,
"nll_loss": 1.4580379724502563,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14476974308490753,
"rewards/margins": 0.036670833826065063,
"rewards/rejected": -0.1814405769109726,
"step": 434
},
{
"epoch": 1.1988263721090784,
"grad_norm": 0.17417070269584656,
"learning_rate": 3.761549701750865e-06,
"log_odds_chosen": 0.802248477935791,
"log_odds_ratio": -0.38353192806243896,
"logits/chosen": -0.18037042021751404,
"logits/rejected": -1.3770548105239868,
"logps/chosen": -1.4576908349990845,
"logps/rejected": -2.126838445663452,
"loss": 1.5085,
"nll_loss": 1.4701937437057495,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14576907455921173,
"rewards/margins": 0.06691478192806244,
"rewards/rejected": -0.21268387138843536,
"step": 435
},
{
"epoch": 1.2015878494994823,
"grad_norm": 0.17217062413692474,
"learning_rate": 3.7546241810744444e-06,
"log_odds_chosen": 0.6275637149810791,
"log_odds_ratio": -0.4340207278728485,
"logits/chosen": -0.18775464594364166,
"logits/rejected": -0.9219359159469604,
"logps/chosen": -1.4593117237091064,
"logps/rejected": -1.9751626253128052,
"loss": 1.5097,
"nll_loss": 1.4663246870040894,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1459311693906784,
"rewards/margins": 0.05158507823944092,
"rewards/rejected": -0.19751626253128052,
"step": 436
},
{
"epoch": 1.2043493268898862,
"grad_norm": 0.19680339097976685,
"learning_rate": 3.747685767193385e-06,
"log_odds_chosen": 0.6149947047233582,
"log_odds_ratio": -0.4397943317890167,
"logits/chosen": -0.24492347240447998,
"logits/rejected": -1.0724200010299683,
"logps/chosen": -1.5251544713974,
"logps/rejected": -2.0360567569732666,
"loss": 1.5969,
"nll_loss": 1.5528908967971802,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1525154560804367,
"rewards/margins": 0.05109023675322533,
"rewards/rejected": -0.20360569655895233,
"step": 437
},
{
"epoch": 1.20711080428029,
"grad_norm": 0.2228655368089676,
"learning_rate": 3.740734531410626e-06,
"log_odds_chosen": 0.7606922388076782,
"log_odds_ratio": -0.39337587356567383,
"logits/chosen": -0.2465159147977829,
"logits/rejected": -1.2846262454986572,
"logps/chosen": -1.448561191558838,
"logps/rejected": -2.069669008255005,
"loss": 1.5204,
"nll_loss": 1.4810588359832764,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14485612511634827,
"rewards/margins": 0.062110789120197296,
"rewards/rejected": -0.20696690678596497,
"step": 438
},
{
"epoch": 1.2098722816706937,
"grad_norm": 0.18277090787887573,
"learning_rate": 3.7337705451608676e-06,
"log_odds_chosen": 0.9290235638618469,
"log_odds_ratio": -0.3463304042816162,
"logits/chosen": -0.3210224509239197,
"logits/rejected": -1.3052295446395874,
"logps/chosen": -1.3797541856765747,
"logps/rejected": -2.1371095180511475,
"loss": 1.4436,
"nll_loss": 1.4089810848236084,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13797542452812195,
"rewards/margins": 0.07573550194501877,
"rewards/rejected": -0.2137109339237213,
"step": 439
},
{
"epoch": 1.2126337590610976,
"grad_norm": 0.19237011671066284,
"learning_rate": 3.7267938800098454e-06,
"log_odds_chosen": 0.9403778910636902,
"log_odds_ratio": -0.3412759602069855,
"logits/chosen": -0.4038398861885071,
"logits/rejected": -1.273897409439087,
"logps/chosen": -1.3479474782943726,
"logps/rejected": -2.1216483116149902,
"loss": 1.4158,
"nll_loss": 1.3816239833831787,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13479475677013397,
"rewards/margins": 0.0773700550198555,
"rewards/rejected": -0.21216480433940887,
"step": 440
},
{
"epoch": 1.2153952364515015,
"grad_norm": 0.2002696543931961,
"learning_rate": 3.7198046076535865e-06,
"log_odds_chosen": 0.7807128429412842,
"log_odds_ratio": -0.3827586770057678,
"logits/chosen": -0.2198580801486969,
"logits/rejected": -1.0754787921905518,
"logps/chosen": -1.4956470727920532,
"logps/rejected": -2.1461679935455322,
"loss": 1.5459,
"nll_loss": 1.507658839225769,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1495647132396698,
"rewards/margins": 0.0650520920753479,
"rewards/rejected": -0.2146168202161789,
"step": 441
},
{
"epoch": 1.2181567138419054,
"grad_norm": 0.18718883395195007,
"learning_rate": 3.71280279991768e-06,
"log_odds_chosen": 0.8355327248573303,
"log_odds_ratio": -0.36628708243370056,
"logits/chosen": -0.27813613414764404,
"logits/rejected": -1.077792763710022,
"logps/chosen": -1.4404205083847046,
"logps/rejected": -2.134765625,
"loss": 1.5121,
"nll_loss": 1.4754693508148193,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1440420299768448,
"rewards/margins": 0.0694345235824585,
"rewards/rejected": -0.21347656846046448,
"step": 442
},
{
"epoch": 1.2209181912323093,
"grad_norm": 0.17261095345020294,
"learning_rate": 3.705788528756533e-06,
"log_odds_chosen": 0.8777885437011719,
"log_odds_ratio": -0.3522123098373413,
"logits/chosen": -0.30880072712898254,
"logits/rejected": -1.328255534172058,
"logps/chosen": -1.4321343898773193,
"logps/rejected": -2.1607189178466797,
"loss": 1.4899,
"nll_loss": 1.4546407461166382,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1432134509086609,
"rewards/margins": 0.07285845279693604,
"rewards/rejected": -0.21607188880443573,
"step": 443
},
{
"epoch": 1.2236796686227132,
"grad_norm": 0.2517964541912079,
"learning_rate": 3.698761866252635e-06,
"log_odds_chosen": 0.6931225061416626,
"log_odds_ratio": -0.42155617475509644,
"logits/chosen": -0.262115478515625,
"logits/rejected": -1.257841944694519,
"logps/chosen": -1.5295952558517456,
"logps/rejected": -2.1139965057373047,
"loss": 1.5875,
"nll_loss": 1.5453757047653198,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15295952558517456,
"rewards/margins": 0.05844012275338173,
"rewards/rejected": -0.211399644613266,
"step": 444
},
{
"epoch": 1.226441146013117,
"grad_norm": 0.19526347517967224,
"learning_rate": 3.691722884615814e-06,
"log_odds_chosen": 0.6913182735443115,
"log_odds_ratio": -0.41308411955833435,
"logits/chosen": -0.17629948258399963,
"logits/rejected": -0.7438918352127075,
"logps/chosen": -1.4464210271835327,
"logps/rejected": -2.0083394050598145,
"loss": 1.5191,
"nll_loss": 1.4777559041976929,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14464209973812103,
"rewards/margins": 0.05619185417890549,
"rewards/rejected": -0.20083396136760712,
"step": 445
},
{
"epoch": 1.229202623403521,
"grad_norm": 0.1902356892824173,
"learning_rate": 3.684671656182497e-06,
"log_odds_chosen": 0.8285303115844727,
"log_odds_ratio": -0.36952173709869385,
"logits/chosen": -0.2879321277141571,
"logits/rejected": -1.2174959182739258,
"logps/chosen": -1.4798123836517334,
"logps/rejected": -2.1687145233154297,
"loss": 1.5442,
"nll_loss": 1.507254719734192,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14798125624656677,
"rewards/margins": 0.06889019906520844,
"rewards/rejected": -0.216871440410614,
"step": 446
},
{
"epoch": 1.2319641007939248,
"grad_norm": 0.19624045491218567,
"learning_rate": 3.6776082534149664e-06,
"log_odds_chosen": 0.7684917449951172,
"log_odds_ratio": -0.3862813413143158,
"logits/chosen": -0.272920161485672,
"logits/rejected": -1.2850197553634644,
"logps/chosen": -1.5005940198898315,
"logps/rejected": -2.1409354209899902,
"loss": 1.5377,
"nll_loss": 1.4990581274032593,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15005940198898315,
"rewards/margins": 0.06403413414955139,
"rewards/rejected": -0.21409353613853455,
"step": 447
},
{
"epoch": 1.2347255781843287,
"grad_norm": 0.1784961074590683,
"learning_rate": 3.670532748900615e-06,
"log_odds_chosen": 0.8203001618385315,
"log_odds_ratio": -0.37937578558921814,
"logits/chosen": -0.2516147494316101,
"logits/rejected": -1.1510305404663086,
"logps/chosen": -1.4173564910888672,
"logps/rejected": -2.08891224861145,
"loss": 1.4755,
"nll_loss": 1.4375797510147095,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14173565804958344,
"rewards/margins": 0.06715556979179382,
"rewards/rejected": -0.20889122784137726,
"step": 448
},
{
"epoch": 1.2374870555747326,
"grad_norm": 0.1783873736858368,
"learning_rate": 3.663445215351198e-06,
"log_odds_chosen": 0.9026142954826355,
"log_odds_ratio": -0.35365813970565796,
"logits/chosen": -0.3238006830215454,
"logits/rejected": -1.0488125085830688,
"logps/chosen": -1.4057360887527466,
"logps/rejected": -2.1597814559936523,
"loss": 1.4666,
"nll_loss": 1.4311884641647339,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14057360589504242,
"rewards/margins": 0.07540452480316162,
"rewards/rejected": -0.21597814559936523,
"step": 449
},
{
"epoch": 1.2402485329651363,
"grad_norm": 0.1859315186738968,
"learning_rate": 3.656345725602089e-06,
"log_odds_chosen": 0.9581824541091919,
"log_odds_ratio": -0.3327292799949646,
"logits/chosen": -0.22859053313732147,
"logits/rejected": -1.3163652420043945,
"logps/chosen": -1.5037806034088135,
"logps/rejected": -2.316061019897461,
"loss": 1.5517,
"nll_loss": 1.5183771848678589,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1503780633211136,
"rewards/margins": 0.08122803270816803,
"rewards/rejected": -0.23160609602928162,
"step": 450
},
{
"epoch": 1.2430100103555402,
"grad_norm": 0.19541142880916595,
"learning_rate": 3.6492343526115292e-06,
"log_odds_chosen": 0.7478545308113098,
"log_odds_ratio": -0.3891213536262512,
"logits/chosen": -0.1598832607269287,
"logits/rejected": -1.259619116783142,
"logps/chosen": -1.5255135297775269,
"logps/rejected": -2.1506118774414062,
"loss": 1.5591,
"nll_loss": 1.5201823711395264,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15255135297775269,
"rewards/margins": 0.06250984221696854,
"rewards/rejected": -0.21506118774414062,
"step": 451
},
{
"epoch": 1.245771487745944,
"grad_norm": 0.18374916911125183,
"learning_rate": 3.642111169459879e-06,
"log_odds_chosen": 0.5907194018363953,
"log_odds_ratio": -0.4446646273136139,
"logits/chosen": -0.2301802933216095,
"logits/rejected": -1.1353992223739624,
"logps/chosen": -1.4880855083465576,
"logps/rejected": -1.973372459411621,
"loss": 1.5312,
"nll_loss": 1.4867122173309326,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.148808553814888,
"rewards/margins": 0.04852868244051933,
"rewards/rejected": -0.19733723998069763,
"step": 452
},
{
"epoch": 1.248532965136348,
"grad_norm": 0.17803093791007996,
"learning_rate": 3.634976249348867e-06,
"log_odds_chosen": 0.7315319776535034,
"log_odds_ratio": -0.4015371799468994,
"logits/chosen": -0.2719256281852722,
"logits/rejected": -1.3226613998413086,
"logps/chosen": -1.4791927337646484,
"logps/rejected": -2.084644317626953,
"loss": 1.54,
"nll_loss": 1.4997994899749756,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14791928231716156,
"rewards/margins": 0.06054516136646271,
"rewards/rejected": -0.20846444368362427,
"step": 453
},
{
"epoch": 1.2512944425267518,
"grad_norm": 0.2094297856092453,
"learning_rate": 3.6278296656008366e-06,
"log_odds_chosen": 0.9680303931236267,
"log_odds_ratio": -0.3453172743320465,
"logits/chosen": -0.2988622188568115,
"logits/rejected": -1.1323243379592896,
"logps/chosen": -1.350219964981079,
"logps/rejected": -2.144009590148926,
"loss": 1.4208,
"nll_loss": 1.3862643241882324,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13502199947834015,
"rewards/margins": 0.07937898486852646,
"rewards/rejected": -0.21440096199512482,
"step": 454
},
{
"epoch": 1.2540559199171557,
"grad_norm": 0.16929712891578674,
"learning_rate": 3.6206714916579925e-06,
"log_odds_chosen": 0.8621344566345215,
"log_odds_ratio": -0.36102256178855896,
"logits/chosen": -0.259186714887619,
"logits/rejected": -1.2395236492156982,
"logps/chosen": -1.3943541049957275,
"logps/rejected": -2.1026394367218018,
"loss": 1.4532,
"nll_loss": 1.4170664548873901,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13943539559841156,
"rewards/margins": 0.07082855701446533,
"rewards/rejected": -0.2102639526128769,
"step": 455
},
{
"epoch": 1.2568173973075596,
"grad_norm": 0.18482358753681183,
"learning_rate": 3.613501801081648e-06,
"log_odds_chosen": 0.6896051168441772,
"log_odds_ratio": -0.4120646119117737,
"logits/chosen": -0.24805930256843567,
"logits/rejected": -1.0325236320495605,
"logps/chosen": -1.4436662197113037,
"logps/rejected": -2.008178472518921,
"loss": 1.5044,
"nll_loss": 1.463235855102539,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14436662197113037,
"rewards/margins": 0.056451231241226196,
"rewards/rejected": -0.20081785321235657,
"step": 456
},
{
"epoch": 1.2595788746979635,
"grad_norm": 0.19444166123867035,
"learning_rate": 3.606320667551466e-06,
"log_odds_chosen": 0.815831184387207,
"log_odds_ratio": -0.3756023645401001,
"logits/chosen": -0.2379026561975479,
"logits/rejected": -1.2224661111831665,
"logps/chosen": -1.4957644939422607,
"logps/rejected": -2.179534673690796,
"loss": 1.5552,
"nll_loss": 1.5175931453704834,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14957645535469055,
"rewards/margins": 0.06837702542543411,
"rewards/rejected": -0.21795348823070526,
"step": 457
},
{
"epoch": 1.2623403520883674,
"grad_norm": 0.18903037905693054,
"learning_rate": 3.599128164864706e-06,
"log_odds_chosen": 0.8830035924911499,
"log_odds_ratio": -0.35656800866127014,
"logits/chosen": -0.3331121802330017,
"logits/rejected": -1.2537503242492676,
"logps/chosen": -1.4486404657363892,
"logps/rejected": -2.1861586570739746,
"loss": 1.4942,
"nll_loss": 1.458591341972351,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1448640525341034,
"rewards/margins": 0.07375183701515198,
"rewards/rejected": -0.21861587464809418,
"step": 458
},
{
"epoch": 1.265101829478771,
"grad_norm": 0.19099265336990356,
"learning_rate": 3.5919243669354585e-06,
"log_odds_chosen": 0.653681218624115,
"log_odds_ratio": -0.4278097450733185,
"logits/chosen": -0.3065020442008972,
"logits/rejected": -0.8541854023933411,
"logps/chosen": -1.5121026039123535,
"logps/rejected": -2.0558552742004395,
"loss": 1.5649,
"nll_loss": 1.5221350193023682,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1512102633714676,
"rewards/margins": 0.054375264793634415,
"rewards/rejected": -0.2055855393409729,
"step": 459
},
{
"epoch": 1.2678633068691751,
"grad_norm": 0.2050788551568985,
"learning_rate": 3.5847093477938955e-06,
"log_odds_chosen": 0.719269871711731,
"log_odds_ratio": -0.403872549533844,
"logits/chosen": -0.3556126356124878,
"logits/rejected": -1.1764421463012695,
"logps/chosen": -1.5112565755844116,
"logps/rejected": -2.108696460723877,
"loss": 1.5556,
"nll_loss": 1.5151852369308472,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15112565457820892,
"rewards/margins": 0.059744007885456085,
"rewards/rejected": -0.2108696550130844,
"step": 460
},
{
"epoch": 1.2706247842595788,
"grad_norm": 0.18415868282318115,
"learning_rate": 3.5774831815855017e-06,
"log_odds_chosen": 0.9078155755996704,
"log_odds_ratio": -0.34306374192237854,
"logits/chosen": -0.34916579723358154,
"logits/rejected": -1.1399445533752441,
"logps/chosen": -1.4003783464431763,
"logps/rejected": -2.1475114822387695,
"loss": 1.4682,
"nll_loss": 1.4339377880096436,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14003783464431763,
"rewards/margins": 0.0747133195400238,
"rewards/rejected": -0.21475116908550262,
"step": 461
},
{
"epoch": 1.2733862616499827,
"grad_norm": 0.1681346744298935,
"learning_rate": 3.5702459425703146e-06,
"log_odds_chosen": 0.9646233320236206,
"log_odds_ratio": -0.3330087959766388,
"logits/chosen": -0.3065449595451355,
"logits/rejected": -1.3354847431182861,
"logps/chosen": -1.4999949932098389,
"logps/rejected": -2.319182872772217,
"loss": 1.5538,
"nll_loss": 1.5204527378082275,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14999951422214508,
"rewards/margins": 0.08191878348588943,
"rewards/rejected": -0.23191829025745392,
"step": 462
},
{
"epoch": 1.2761477390403866,
"grad_norm": 0.18663005530834198,
"learning_rate": 3.562997705122162e-06,
"log_odds_chosen": 0.8870431184768677,
"log_odds_ratio": -0.34840965270996094,
"logits/chosen": -0.2869050204753876,
"logits/rejected": -1.1682400703430176,
"logps/chosen": -1.4178146123886108,
"logps/rejected": -2.1509153842926025,
"loss": 1.4721,
"nll_loss": 1.4372905492782593,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14178146421909332,
"rewards/margins": 0.07331006228923798,
"rewards/rejected": -0.2150915116071701,
"step": 463
},
{
"epoch": 1.2789092164307905,
"grad_norm": 0.17033839225769043,
"learning_rate": 3.5557385437279e-06,
"log_odds_chosen": 0.923761785030365,
"log_odds_ratio": -0.3432496190071106,
"logits/chosen": -0.2502524256706238,
"logits/rejected": -0.997616708278656,
"logps/chosen": -1.4954802989959717,
"logps/rejected": -2.2727675437927246,
"loss": 1.5311,
"nll_loss": 1.4967255592346191,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1495480239391327,
"rewards/margins": 0.07772872596979141,
"rewards/rejected": -0.2272767573595047,
"step": 464
},
{
"epoch": 1.2816706938211944,
"grad_norm": 0.195389986038208,
"learning_rate": 3.5484685329866424e-06,
"log_odds_chosen": 0.6268041133880615,
"log_odds_ratio": -0.4326058626174927,
"logits/chosen": -0.2863166928291321,
"logits/rejected": -1.0226547718048096,
"logps/chosen": -1.5096933841705322,
"logps/rejected": -2.028193950653076,
"loss": 1.5479,
"nll_loss": 1.5046144723892212,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15096935629844666,
"rewards/margins": 0.05185003951191902,
"rewards/rejected": -0.20281939208507538,
"step": 465
},
{
"epoch": 1.2844321712115983,
"grad_norm": 0.17330817878246307,
"learning_rate": 3.541187747608998e-06,
"log_odds_chosen": 1.0745373964309692,
"log_odds_ratio": -0.31021982431411743,
"logits/chosen": -0.37704816460609436,
"logits/rejected": -1.22407865524292,
"logps/chosen": -1.5697916746139526,
"logps/rejected": -2.4969704151153564,
"loss": 1.5937,
"nll_loss": 1.562636375427246,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15697917342185974,
"rewards/margins": 0.09271789342164993,
"rewards/rejected": -0.24969705939292908,
"step": 466
},
{
"epoch": 1.2871936486020021,
"grad_norm": 0.20620717108249664,
"learning_rate": 3.533896262416302e-06,
"log_odds_chosen": 0.8361949920654297,
"log_odds_ratio": -0.36747199296951294,
"logits/chosen": -0.31524112820625305,
"logits/rejected": -1.093992829322815,
"logps/chosen": -1.4830838441848755,
"logps/rejected": -2.1828489303588867,
"loss": 1.5323,
"nll_loss": 1.4955836534500122,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1483083963394165,
"rewards/margins": 0.06997650861740112,
"rewards/rejected": -0.21828490495681763,
"step": 467
},
{
"epoch": 1.2899551259924058,
"grad_norm": 0.20733030140399933,
"learning_rate": 3.5265941523398455e-06,
"log_odds_chosen": 0.9227453470230103,
"log_odds_ratio": -0.3383994400501251,
"logits/chosen": -0.3382907211780548,
"logits/rejected": -1.0753273963928223,
"logps/chosen": -1.5061396360397339,
"logps/rejected": -2.2841649055480957,
"loss": 1.5394,
"nll_loss": 1.5055965185165405,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1506139636039734,
"rewards/margins": 0.07780253887176514,
"rewards/rejected": -0.22841650247573853,
"step": 468
},
{
"epoch": 1.29271660338281,
"grad_norm": 0.1836915910243988,
"learning_rate": 3.519281492420108e-06,
"log_odds_chosen": 0.5979352593421936,
"log_odds_ratio": -0.446799635887146,
"logits/chosen": -0.29712000489234924,
"logits/rejected": -1.041902780532837,
"logps/chosen": -1.5359019041061401,
"logps/rejected": -2.0312960147857666,
"loss": 1.5753,
"nll_loss": 1.5306230783462524,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15359018743038177,
"rewards/margins": 0.04953942447900772,
"rewards/rejected": -0.2031296193599701,
"step": 469
},
{
"epoch": 1.2954780807732136,
"grad_norm": 0.19433459639549255,
"learning_rate": 3.5119583578059845e-06,
"log_odds_chosen": 0.7800890803337097,
"log_odds_ratio": -0.3985646963119507,
"logits/chosen": -0.2822642922401428,
"logits/rejected": -1.1507655382156372,
"logps/chosen": -1.6121432781219482,
"logps/rejected": -2.281339406967163,
"loss": 1.6286,
"nll_loss": 1.588757038116455,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16121432185173035,
"rewards/margins": 0.06691960990428925,
"rewards/rejected": -0.2281339317560196,
"step": 470
},
{
"epoch": 1.2982395581636175,
"grad_norm": 0.1745569258928299,
"learning_rate": 3.504624823754014e-06,
"log_odds_chosen": 0.9689381122589111,
"log_odds_ratio": -0.32553234696388245,
"logits/chosen": -0.3070724904537201,
"logits/rejected": -1.4188494682312012,
"logps/chosen": -1.4123187065124512,
"logps/rejected": -2.2118630409240723,
"loss": 1.4628,
"nll_loss": 1.4302698373794556,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14123189449310303,
"rewards/margins": 0.07995443046092987,
"rewards/rejected": -0.2211862951517105,
"step": 471
},
{
"epoch": 1.3010010355540214,
"grad_norm": 0.18898555636405945,
"learning_rate": 3.4972809656276047e-06,
"log_odds_chosen": 0.8960251212120056,
"log_odds_ratio": -0.3498467206954956,
"logits/chosen": -0.273061066865921,
"logits/rejected": -1.3158522844314575,
"logps/chosen": -1.4974440336227417,
"logps/rejected": -2.2527313232421875,
"loss": 1.5449,
"nll_loss": 1.5098881721496582,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1497444063425064,
"rewards/margins": 0.07552873343229294,
"rewards/rejected": -0.22527314722537994,
"step": 472
},
{
"epoch": 1.3037625129444252,
"grad_norm": 0.1694575846195221,
"learning_rate": 3.4899268588962613e-06,
"log_odds_chosen": 0.9621706604957581,
"log_odds_ratio": -0.32910794019699097,
"logits/chosen": -0.3364591598510742,
"logits/rejected": -1.1894400119781494,
"logps/chosen": -1.363714575767517,
"logps/rejected": -2.1495935916900635,
"loss": 1.4326,
"nll_loss": 1.3997074365615845,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1363714635372162,
"rewards/margins": 0.07858789712190628,
"rewards/rejected": -0.21495933830738068,
"step": 473
},
{
"epoch": 1.3065239903348291,
"grad_norm": 0.18671829998493195,
"learning_rate": 3.4825625791348093e-06,
"log_odds_chosen": 0.9318122863769531,
"log_odds_ratio": -0.3445737957954407,
"logits/chosen": -0.29330548644065857,
"logits/rejected": -1.1094152927398682,
"logps/chosen": -1.4132261276245117,
"logps/rejected": -2.188481569290161,
"loss": 1.4629,
"nll_loss": 1.4284807443618774,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14132261276245117,
"rewards/margins": 0.0775255486369133,
"rewards/rejected": -0.21884815394878387,
"step": 474
},
{
"epoch": 1.309285467725233,
"grad_norm": 0.18166309595108032,
"learning_rate": 3.4751882020226174e-06,
"log_odds_chosen": 0.8953260779380798,
"log_odds_ratio": -0.35324275493621826,
"logits/chosen": -0.30813199281692505,
"logits/rejected": -1.1583383083343506,
"logps/chosen": -1.5182853937149048,
"logps/rejected": -2.2794241905212402,
"loss": 1.5593,
"nll_loss": 1.5240164995193481,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15182854235172272,
"rewards/margins": 0.07611385732889175,
"rewards/rejected": -0.22794242203235626,
"step": 475
},
{
"epoch": 1.312046945115637,
"grad_norm": 0.1913100630044937,
"learning_rate": 3.467803803342821e-06,
"log_odds_chosen": 0.8503288626670837,
"log_odds_ratio": -0.36000919342041016,
"logits/chosen": -0.2365928590297699,
"logits/rejected": -1.175534725189209,
"logps/chosen": -1.4916030168533325,
"logps/rejected": -2.2044079303741455,
"loss": 1.5322,
"nll_loss": 1.4961673021316528,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14916031062602997,
"rewards/margins": 0.07128050923347473,
"rewards/rejected": -0.2204408198595047,
"step": 476
},
{
"epoch": 1.3148084225060408,
"grad_norm": 0.1990288347005844,
"learning_rate": 3.4604094589815402e-06,
"log_odds_chosen": 0.7814106345176697,
"log_odds_ratio": -0.3809904456138611,
"logits/chosen": -0.33347588777542114,
"logits/rejected": -1.1712278127670288,
"logps/chosen": -1.4147124290466309,
"logps/rejected": -2.0545654296875,
"loss": 1.4683,
"nll_loss": 1.4301787614822388,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1414712369441986,
"rewards/margins": 0.06398531049489975,
"rewards/rejected": -0.20545653998851776,
"step": 477
},
{
"epoch": 1.3175698998964447,
"grad_norm": 0.18192388117313385,
"learning_rate": 3.4530052449271044e-06,
"log_odds_chosen": 0.9670149087905884,
"log_odds_ratio": -0.33280444145202637,
"logits/chosen": -0.372698038816452,
"logits/rejected": -1.1448473930358887,
"logps/chosen": -1.4152264595031738,
"logps/rejected": -2.2173709869384766,
"loss": 1.4648,
"nll_loss": 1.4315613508224487,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14152264595031738,
"rewards/margins": 0.08021444082260132,
"rewards/rejected": -0.2217371016740799,
"step": 478
},
{
"epoch": 1.3203313772868484,
"grad_norm": 0.19329139590263367,
"learning_rate": 3.4455912372692696e-06,
"log_odds_chosen": 0.6430727243423462,
"log_odds_ratio": -0.430493026971817,
"logits/chosen": -0.28841686248779297,
"logits/rejected": -1.1759401559829712,
"logps/chosen": -1.4796922206878662,
"logps/rejected": -2.0091733932495117,
"loss": 1.5189,
"nll_loss": 1.475854516029358,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14796923100948334,
"rewards/margins": 0.05294811725616455,
"rewards/rejected": -0.2009173333644867,
"step": 479
},
{
"epoch": 1.3230928546772525,
"grad_norm": 0.18555399775505066,
"learning_rate": 3.438167512198436e-06,
"log_odds_chosen": 0.6809258460998535,
"log_odds_ratio": -0.4141601622104645,
"logits/chosen": -0.31838759779930115,
"logits/rejected": -1.10935640335083,
"logps/chosen": -1.522485613822937,
"logps/rejected": -2.0886290073394775,
"loss": 1.5706,
"nll_loss": 1.529231309890747,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1522485464811325,
"rewards/margins": 0.05661435052752495,
"rewards/rejected": -0.20886291563510895,
"step": 480
},
{
"epoch": 1.3258543320676561,
"grad_norm": 0.185867041349411,
"learning_rate": 3.4307341460048633e-06,
"log_odds_chosen": 0.8023340106010437,
"log_odds_ratio": -0.37456732988357544,
"logits/chosen": -0.29839807748794556,
"logits/rejected": -1.1352858543395996,
"logps/chosen": -1.5517854690551758,
"logps/rejected": -2.226045846939087,
"loss": 1.5881,
"nll_loss": 1.5506727695465088,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15517854690551758,
"rewards/margins": 0.06742605566978455,
"rewards/rejected": -0.22260460257530212,
"step": 481
},
{
"epoch": 1.32861580945806,
"grad_norm": 0.19534382224082947,
"learning_rate": 3.4232912150778914e-06,
"log_odds_chosen": 0.8740422129631042,
"log_odds_ratio": -0.35599130392074585,
"logits/chosen": -0.3661443293094635,
"logits/rejected": -1.2379000186920166,
"logps/chosen": -1.4489246606826782,
"logps/rejected": -2.17354416847229,
"loss": 1.5169,
"nll_loss": 1.4813332557678223,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14489248394966125,
"rewards/margins": 0.07246193289756775,
"rewards/rejected": -0.2173544317483902,
"step": 482
},
{
"epoch": 1.331377286848464,
"grad_norm": 0.18097054958343506,
"learning_rate": 3.415838795905151e-06,
"log_odds_chosen": 1.0503449440002441,
"log_odds_ratio": -0.3068162798881531,
"logits/chosen": -0.4407689571380615,
"logits/rejected": -1.4052069187164307,
"logps/chosen": -1.4251679182052612,
"logps/rejected": -2.2970073223114014,
"loss": 1.4703,
"nll_loss": 1.4396311044692993,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14251679182052612,
"rewards/margins": 0.08718395978212357,
"rewards/rejected": -0.2297007441520691,
"step": 483
},
{
"epoch": 1.3341387642388678,
"grad_norm": 0.19162198901176453,
"learning_rate": 3.408376965071779e-06,
"log_odds_chosen": 0.7587519884109497,
"log_odds_ratio": -0.3879585862159729,
"logits/chosen": -0.28846538066864014,
"logits/rejected": -1.1753207445144653,
"logps/chosen": -1.5185155868530273,
"logps/rejected": -2.149603843688965,
"loss": 1.5523,
"nll_loss": 1.5134819746017456,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1518515795469284,
"rewards/margins": 0.06310880929231644,
"rewards/rejected": -0.21496038138866425,
"step": 484
},
{
"epoch": 1.3369002416292717,
"grad_norm": 0.1803843080997467,
"learning_rate": 3.400905799259634e-06,
"log_odds_chosen": 0.7836498022079468,
"log_odds_ratio": -0.3856756389141083,
"logits/chosen": -0.3168273866176605,
"logits/rejected": -1.1095941066741943,
"logps/chosen": -1.4380414485931396,
"logps/rejected": -2.0824549198150635,
"loss": 1.4834,
"nll_loss": 1.444858431816101,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.143804132938385,
"rewards/margins": 0.06444136798381805,
"rewards/rejected": -0.20824551582336426,
"step": 485
},
{
"epoch": 1.3396617190196756,
"grad_norm": 0.18349698185920715,
"learning_rate": 3.393425375246503e-06,
"log_odds_chosen": 0.7499913573265076,
"log_odds_ratio": -0.3889506757259369,
"logits/chosen": -0.34318801760673523,
"logits/rejected": -1.1768746376037598,
"logps/chosen": -1.4634371995925903,
"logps/rejected": -2.080674648284912,
"loss": 1.5059,
"nll_loss": 1.4669642448425293,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14634370803833008,
"rewards/margins": 0.0617237389087677,
"rewards/rejected": -0.20806746184825897,
"step": 486
},
{
"epoch": 1.3424231964100795,
"grad_norm": 0.19035322964191437,
"learning_rate": 3.3859357699053165e-06,
"log_odds_chosen": 0.8554298877716064,
"log_odds_ratio": -0.36611807346343994,
"logits/chosen": -0.31382206082344055,
"logits/rejected": -1.058935523033142,
"logps/chosen": -1.3999180793762207,
"logps/rejected": -2.1072771549224854,
"loss": 1.4434,
"nll_loss": 1.4067771434783936,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13999181985855103,
"rewards/margins": 0.0707358866930008,
"rewards/rejected": -0.210727721452713,
"step": 487
},
{
"epoch": 1.3451846738004831,
"grad_norm": 0.1878138780593872,
"learning_rate": 3.3784370602033572e-06,
"log_odds_chosen": 0.992766797542572,
"log_odds_ratio": -0.3239111304283142,
"logits/chosen": -0.40365132689476013,
"logits/rejected": -1.291216254234314,
"logps/chosen": -1.4312440156936646,
"logps/rejected": -2.263004779815674,
"loss": 1.4795,
"nll_loss": 1.4471262693405151,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14312440156936646,
"rewards/margins": 0.08317607641220093,
"rewards/rejected": -0.22630049288272858,
"step": 488
},
{
"epoch": 1.3479461511908872,
"grad_norm": 0.19307951629161835,
"learning_rate": 3.3709293232014705e-06,
"log_odds_chosen": 0.8466112613677979,
"log_odds_ratio": -0.36574095487594604,
"logits/chosen": -0.36816614866256714,
"logits/rejected": -1.4561880826950073,
"logps/chosen": -1.4958292245864868,
"logps/rejected": -2.210787534713745,
"loss": 1.5377,
"nll_loss": 1.501145839691162,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14958293735980988,
"rewards/margins": 0.07149583101272583,
"rewards/rejected": -0.2210787534713745,
"step": 489
},
{
"epoch": 1.350707628581291,
"grad_norm": 0.1835506707429886,
"learning_rate": 3.3634126360532694e-06,
"log_odds_chosen": 0.9759219288825989,
"log_odds_ratio": -0.3324146568775177,
"logits/chosen": -0.24411487579345703,
"logits/rejected": -1.3185425996780396,
"logps/chosen": -1.5735536813735962,
"logps/rejected": -2.4141926765441895,
"loss": 1.6083,
"nll_loss": 1.5750340223312378,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15735535323619843,
"rewards/margins": 0.08406390994787216,
"rewards/rejected": -0.24141928553581238,
"step": 490
},
{
"epoch": 1.3534691059716948,
"grad_norm": 0.20382438600063324,
"learning_rate": 3.355887076004345e-06,
"log_odds_chosen": 0.8967198133468628,
"log_odds_ratio": -0.3447577953338623,
"logits/chosen": -0.45378029346466064,
"logits/rejected": -1.0346810817718506,
"logps/chosen": -1.4815733432769775,
"logps/rejected": -2.233039140701294,
"loss": 1.5336,
"nll_loss": 1.4991440773010254,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14815731346607208,
"rewards/margins": 0.07514660060405731,
"rewards/rejected": -0.2233039289712906,
"step": 491
},
{
"epoch": 1.3562305833620987,
"grad_norm": 0.1867835968732834,
"learning_rate": 3.3483527203914694e-06,
"log_odds_chosen": 0.9929035902023315,
"log_odds_ratio": -0.3237488567829132,
"logits/chosen": -0.3180537819862366,
"logits/rejected": -1.2090908288955688,
"logps/chosen": -1.517293095588684,
"logps/rejected": -2.360647678375244,
"loss": 1.5624,
"nll_loss": 1.5299937725067139,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15172931551933289,
"rewards/margins": 0.08433545380830765,
"rewards/rejected": -0.23606477677822113,
"step": 492
},
{
"epoch": 1.3589920607525026,
"grad_norm": 0.18780100345611572,
"learning_rate": 3.340809646641805e-06,
"log_odds_chosen": 0.8353948593139648,
"log_odds_ratio": -0.36322087049484253,
"logits/chosen": -0.23974277079105377,
"logits/rejected": -1.23904550075531,
"logps/chosen": -1.5875341892242432,
"logps/rejected": -2.3006584644317627,
"loss": 1.6096,
"nll_loss": 1.5732414722442627,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1587534248828888,
"rewards/margins": 0.07131239771842957,
"rewards/rejected": -0.23006582260131836,
"step": 493
},
{
"epoch": 1.3617535381429065,
"grad_norm": 0.18200430274009705,
"learning_rate": 3.333257932272105e-06,
"log_odds_chosen": 0.9004358053207397,
"log_odds_ratio": -0.3532771170139313,
"logits/chosen": -0.2815527021884918,
"logits/rejected": -1.1550835371017456,
"logps/chosen": -1.4390370845794678,
"logps/rejected": -2.194286346435547,
"loss": 1.4934,
"nll_loss": 1.4581207036972046,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1439037024974823,
"rewards/margins": 0.07552491873502731,
"rewards/rejected": -0.21942861378192902,
"step": 494
},
{
"epoch": 1.3645150155333103,
"grad_norm": 0.19926373660564423,
"learning_rate": 3.3256976548879183e-06,
"log_odds_chosen": 0.8690962195396423,
"log_odds_ratio": -0.3561793863773346,
"logits/chosen": -0.3800399601459503,
"logits/rejected": -1.047258973121643,
"logps/chosen": -1.4909493923187256,
"logps/rejected": -2.219834327697754,
"loss": 1.5367,
"nll_loss": 1.5010950565338135,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14909493923187256,
"rewards/margins": 0.07288849353790283,
"rewards/rejected": -0.2219834327697754,
"step": 495
},
{
"epoch": 1.3672764929237142,
"grad_norm": 0.18317997455596924,
"learning_rate": 3.3181288921827925e-06,
"log_odds_chosen": 0.9415011405944824,
"log_odds_ratio": -0.3410661220550537,
"logits/chosen": -0.27754640579223633,
"logits/rejected": -1.0399134159088135,
"logps/chosen": -1.4013381004333496,
"logps/rejected": -2.179905414581299,
"loss": 1.465,
"nll_loss": 1.4309097528457642,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1401338279247284,
"rewards/margins": 0.07785671204328537,
"rewards/rejected": -0.21799054741859436,
"step": 496
},
{
"epoch": 1.3700379703141181,
"grad_norm": 0.1992860585451126,
"learning_rate": 3.310551721937475e-06,
"log_odds_chosen": 0.9155603647232056,
"log_odds_ratio": -0.340352326631546,
"logits/chosen": -0.30399274826049805,
"logits/rejected": -1.2022323608398438,
"logps/chosen": -1.4278628826141357,
"logps/rejected": -2.1877853870391846,
"loss": 1.4555,
"nll_loss": 1.4214258193969727,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14278629422187805,
"rewards/margins": 0.07599223405122757,
"rewards/rejected": -0.21877853572368622,
"step": 497
},
{
"epoch": 1.372799447704522,
"grad_norm": 0.2043248564004898,
"learning_rate": 3.3029662220191146e-06,
"log_odds_chosen": 0.9598441123962402,
"log_odds_ratio": -0.3286186754703522,
"logits/chosen": -0.38024866580963135,
"logits/rejected": -1.2354223728179932,
"logps/chosen": -1.33193838596344,
"logps/rejected": -2.1157238483428955,
"loss": 1.388,
"nll_loss": 1.3551721572875977,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13319383561611176,
"rewards/margins": 0.07837854325771332,
"rewards/rejected": -0.21157239377498627,
"step": 498
},
{
"epoch": 1.3755609250949257,
"grad_norm": 0.17761792242527008,
"learning_rate": 3.2953724703804572e-06,
"log_odds_chosen": 1.0203739404678345,
"log_odds_ratio": -0.3161318898200989,
"logits/chosen": -0.31488168239593506,
"logits/rejected": -1.2814425230026245,
"logps/chosen": -1.4082773923873901,
"logps/rejected": -2.2629306316375732,
"loss": 1.4532,
"nll_loss": 1.4215905666351318,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1408277451992035,
"rewards/margins": 0.08546529710292816,
"rewards/rejected": -0.22629307210445404,
"step": 499
},
{
"epoch": 1.3783224024853298,
"grad_norm": 0.19437268376350403,
"learning_rate": 3.2877705450590525e-06,
"log_odds_chosen": 0.8638174533843994,
"log_odds_ratio": -0.35613882541656494,
"logits/chosen": -0.3166804015636444,
"logits/rejected": -1.1927976608276367,
"logps/chosen": -1.4706673622131348,
"logps/rejected": -2.1903886795043945,
"loss": 1.5439,
"nll_loss": 1.5082558393478394,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14706675708293915,
"rewards/margins": 0.07197214663028717,
"rewards/rejected": -0.21903888881206512,
"step": 500
}
],
"logging_steps": 1,
"max_steps": 1089,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"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
}