gitpullpull's picture
Upload folder using huggingface_hub
bc1260e verified
Raw
History Blame Contribute Delete
327 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.7880220646178093,
"eval_steps": 500,
"global_step": 500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0015760441292356187,
"grad_norm": 0.711133599281311,
"learning_rate": 0.0,
"log_odds_chosen": 0.38395223021507263,
"log_odds_ratio": -0.5362357497215271,
"logits/chosen": -0.6395432353019714,
"logits/rejected": -0.1508907973766327,
"logps/chosen": -1.973930835723877,
"logps/rejected": -2.3123726844787598,
"loss": 2.2406,
"nll_loss": 2.1870110034942627,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19739308953285217,
"rewards/margins": 0.03384416550397873,
"rewards/rejected": -0.2312372624874115,
"step": 1
},
{
"epoch": 0.0031520882584712374,
"grad_norm": 0.6591355204582214,
"learning_rate": 3.1496062992125985e-08,
"log_odds_chosen": 0.39076143503189087,
"log_odds_ratio": -0.5237792730331421,
"logits/chosen": -0.6357220411300659,
"logits/rejected": -0.09711451828479767,
"logps/chosen": -1.8780229091644287,
"logps/rejected": -2.2161149978637695,
"loss": 2.1349,
"nll_loss": 2.082494020462036,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1878022849559784,
"rewards/margins": 0.03380918130278587,
"rewards/rejected": -0.22161146998405457,
"step": 2
},
{
"epoch": 0.004728132387706856,
"grad_norm": 0.74098140001297,
"learning_rate": 6.299212598425197e-08,
"log_odds_chosen": 0.5195883512496948,
"log_odds_ratio": -0.4742986857891083,
"logits/chosen": -0.7715582251548767,
"logits/rejected": -0.26645177602767944,
"logps/chosen": -1.9943277835845947,
"logps/rejected": -2.4578633308410645,
"loss": 2.2471,
"nll_loss": 2.199704647064209,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19943277537822723,
"rewards/margins": 0.046353571116924286,
"rewards/rejected": -0.24578633904457092,
"step": 3
},
{
"epoch": 0.006304176516942475,
"grad_norm": 0.6662443280220032,
"learning_rate": 9.448818897637795e-08,
"log_odds_chosen": 0.55958491563797,
"log_odds_ratio": -0.46615538001060486,
"logits/chosen": -0.5838385820388794,
"logits/rejected": -0.1573001742362976,
"logps/chosen": -2.007845163345337,
"logps/rejected": -2.5115368366241455,
"loss": 2.2568,
"nll_loss": 2.210216760635376,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.20078451931476593,
"rewards/margins": 0.05036917328834534,
"rewards/rejected": -0.2511536777019501,
"step": 4
},
{
"epoch": 0.007880220646178092,
"grad_norm": 0.7787235379219055,
"learning_rate": 1.2598425196850394e-07,
"log_odds_chosen": 0.705410361289978,
"log_odds_ratio": -0.42067474126815796,
"logits/chosen": -0.7229734063148499,
"logits/rejected": -0.27978262305259705,
"logps/chosen": -1.9303616285324097,
"logps/rejected": -2.555299758911133,
"loss": 2.1734,
"nll_loss": 2.1313021183013916,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19303615391254425,
"rewards/margins": 0.06249381601810455,
"rewards/rejected": -0.2555299699306488,
"step": 5
},
{
"epoch": 0.009456264775413711,
"grad_norm": 0.6482278108596802,
"learning_rate": 1.5748031496062992e-07,
"log_odds_chosen": 0.40409255027770996,
"log_odds_ratio": -0.5151315331459045,
"logits/chosen": -0.5276838541030884,
"logits/rejected": -0.05200649052858353,
"logps/chosen": -1.8561073541641235,
"logps/rejected": -2.208005905151367,
"loss": 2.1449,
"nll_loss": 2.0933837890625,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18561072647571564,
"rewards/margins": 0.03518984466791153,
"rewards/rejected": -0.22080056369304657,
"step": 6
},
{
"epoch": 0.01103230890464933,
"grad_norm": 0.6663646697998047,
"learning_rate": 1.889763779527559e-07,
"log_odds_chosen": 0.546273410320282,
"log_odds_ratio": -0.46661460399627686,
"logits/chosen": -0.5808312296867371,
"logits/rejected": -0.19844059646129608,
"logps/chosen": -1.9260658025741577,
"logps/rejected": -2.4119839668273926,
"loss": 2.1774,
"nll_loss": 2.1307828426361084,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1926065981388092,
"rewards/margins": 0.04859180375933647,
"rewards/rejected": -0.24119840562343597,
"step": 7
},
{
"epoch": 0.01260835303388495,
"grad_norm": 0.782015323638916,
"learning_rate": 2.2047244094488187e-07,
"log_odds_chosen": 0.8110038638114929,
"log_odds_ratio": -0.3770079016685486,
"logits/chosen": -0.6198790669441223,
"logits/rejected": -0.24129487574100494,
"logps/chosen": -2.014923572540283,
"logps/rejected": -2.744842290878296,
"loss": 2.2715,
"nll_loss": 2.2338366508483887,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.20149235427379608,
"rewards/margins": 0.07299190014600754,
"rewards/rejected": -0.274484246969223,
"step": 8
},
{
"epoch": 0.014184397163120567,
"grad_norm": 0.6285175085067749,
"learning_rate": 2.519685039370079e-07,
"log_odds_chosen": 0.4325932562351227,
"log_odds_ratio": -0.5066515207290649,
"logits/chosen": -0.5963254570960999,
"logits/rejected": -0.07179142534732819,
"logps/chosen": -1.9378471374511719,
"logps/rejected": -2.31904673576355,
"loss": 2.1726,
"nll_loss": 2.1219170093536377,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1937847137451172,
"rewards/margins": 0.038119956851005554,
"rewards/rejected": -0.23190467059612274,
"step": 9
},
{
"epoch": 0.015760441292356184,
"grad_norm": 0.7205284833908081,
"learning_rate": 2.8346456692913386e-07,
"log_odds_chosen": 0.4166225790977478,
"log_odds_ratio": -0.5110099911689758,
"logits/chosen": -0.6733875274658203,
"logits/rejected": -0.2730729877948761,
"logps/chosen": -1.96082603931427,
"logps/rejected": -2.3293237686157227,
"loss": 2.21,
"nll_loss": 2.158905267715454,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19608259201049805,
"rewards/margins": 0.03684981167316437,
"rewards/rejected": -0.2329324334859848,
"step": 10
},
{
"epoch": 0.017336485421591805,
"grad_norm": 0.7679106593132019,
"learning_rate": 3.1496062992125984e-07,
"log_odds_chosen": 0.48609045147895813,
"log_odds_ratio": -0.48330000042915344,
"logits/chosen": -0.5869827270507812,
"logits/rejected": -0.24110490083694458,
"logps/chosen": -2.035675048828125,
"logps/rejected": -2.4683971405029297,
"loss": 2.2835,
"nll_loss": 2.2351536750793457,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.2035675048828125,
"rewards/margins": 0.043272241950035095,
"rewards/rejected": -0.2468397617340088,
"step": 11
},
{
"epoch": 0.018912529550827423,
"grad_norm": 0.8531121015548706,
"learning_rate": 3.464566929133858e-07,
"log_odds_chosen": 0.5449747443199158,
"log_odds_ratio": -0.461398184299469,
"logits/chosen": -0.7053269147872925,
"logits/rejected": -0.12370388209819794,
"logps/chosen": -2.076099157333374,
"logps/rejected": -2.565109968185425,
"loss": 2.3324,
"nll_loss": 2.286276340484619,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.20760990679264069,
"rewards/margins": 0.04890113323926926,
"rewards/rejected": -0.25651103258132935,
"step": 12
},
{
"epoch": 0.02048857368006304,
"grad_norm": 0.8578523397445679,
"learning_rate": 3.779527559055118e-07,
"log_odds_chosen": 0.5090824365615845,
"log_odds_ratio": -0.47858256101608276,
"logits/chosen": -0.7868019342422485,
"logits/rejected": -0.07148027420043945,
"logps/chosen": -1.9688328504562378,
"logps/rejected": -2.4192914962768555,
"loss": 2.2258,
"nll_loss": 2.1779398918151855,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19688329100608826,
"rewards/margins": 0.045045845210552216,
"rewards/rejected": -0.24192912876605988,
"step": 13
},
{
"epoch": 0.02206461780929866,
"grad_norm": 0.675309956073761,
"learning_rate": 4.0944881889763777e-07,
"log_odds_chosen": 0.3657957911491394,
"log_odds_ratio": -0.5386834144592285,
"logits/chosen": -0.6320536136627197,
"logits/rejected": -0.3780551254749298,
"logps/chosen": -1.9759397506713867,
"logps/rejected": -2.300447463989258,
"loss": 2.2257,
"nll_loss": 2.171861171722412,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19759398698806763,
"rewards/margins": 0.03245077282190323,
"rewards/rejected": -0.23004476726055145,
"step": 14
},
{
"epoch": 0.02364066193853428,
"grad_norm": 0.7733155488967896,
"learning_rate": 4.4094488188976375e-07,
"log_odds_chosen": 0.41122347116470337,
"log_odds_ratio": -0.5111778378486633,
"logits/chosen": -0.7448755502700806,
"logits/rejected": -0.08966228365898132,
"logps/chosen": -1.9629881381988525,
"logps/rejected": -2.32719087600708,
"loss": 2.2273,
"nll_loss": 2.176145076751709,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19629880785942078,
"rewards/margins": 0.03642028942704201,
"rewards/rejected": -0.23271909356117249,
"step": 15
},
{
"epoch": 0.0252167060677699,
"grad_norm": 0.7707588076591492,
"learning_rate": 4.7244094488188973e-07,
"log_odds_chosen": 0.4248766303062439,
"log_odds_ratio": -0.5078074336051941,
"logits/chosen": -0.6721003651618958,
"logits/rejected": -0.19612114131450653,
"logps/chosen": -1.9606484174728394,
"logps/rejected": -2.336439609527588,
"loss": 2.22,
"nll_loss": 2.1692113876342773,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19606485962867737,
"rewards/margins": 0.03757911175489426,
"rewards/rejected": -0.23364394903182983,
"step": 16
},
{
"epoch": 0.026792750197005517,
"grad_norm": 0.6388130784034729,
"learning_rate": 5.039370078740158e-07,
"log_odds_chosen": 0.6333746314048767,
"log_odds_ratio": -0.43810510635375977,
"logits/chosen": -0.5735284686088562,
"logits/rejected": -0.31118011474609375,
"logps/chosen": -1.9192208051681519,
"logps/rejected": -2.4831156730651855,
"loss": 2.1769,
"nll_loss": 2.133085012435913,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19192209839820862,
"rewards/margins": 0.056389469653367996,
"rewards/rejected": -0.24831153452396393,
"step": 17
},
{
"epoch": 0.028368794326241134,
"grad_norm": 0.7463440299034119,
"learning_rate": 5.354330708661418e-07,
"log_odds_chosen": 0.3546241819858551,
"log_odds_ratio": -0.544340193271637,
"logits/chosen": -0.6276163458824158,
"logits/rejected": 0.05960750952363014,
"logps/chosen": -1.9554523229599,
"logps/rejected": -2.2666845321655273,
"loss": 2.2226,
"nll_loss": 2.168125629425049,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1955452263355255,
"rewards/margins": 0.031123224645853043,
"rewards/rejected": -0.22666846215724945,
"step": 18
},
{
"epoch": 0.029944838455476755,
"grad_norm": 0.6933729648590088,
"learning_rate": 5.669291338582677e-07,
"log_odds_chosen": 0.6329823732376099,
"log_odds_ratio": -0.4341467022895813,
"logits/chosen": -0.753471851348877,
"logits/rejected": -0.27794983983039856,
"logps/chosen": -1.8416026830673218,
"logps/rejected": -2.3966176509857178,
"loss": 2.115,
"nll_loss": 2.07161283493042,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1841602772474289,
"rewards/margins": 0.05550149455666542,
"rewards/rejected": -0.23966176807880402,
"step": 19
},
{
"epoch": 0.03152088258471237,
"grad_norm": 0.7374937534332275,
"learning_rate": 5.984251968503937e-07,
"log_odds_chosen": 0.5668491125106812,
"log_odds_ratio": -0.46694958209991455,
"logits/chosen": -0.562913179397583,
"logits/rejected": -0.25034084916114807,
"logps/chosen": -1.995664119720459,
"logps/rejected": -2.5026793479919434,
"loss": 2.262,
"nll_loss": 2.215284824371338,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19956642389297485,
"rewards/margins": 0.05070152133703232,
"rewards/rejected": -0.25026795268058777,
"step": 20
},
{
"epoch": 0.03309692671394799,
"grad_norm": 0.75541090965271,
"learning_rate": 6.299212598425197e-07,
"log_odds_chosen": 0.35937565565109253,
"log_odds_ratio": -0.5362535715103149,
"logits/chosen": -0.65813809633255,
"logits/rejected": -0.1279284954071045,
"logps/chosen": -2.04636812210083,
"logps/rejected": -2.3680672645568848,
"loss": 2.3058,
"nll_loss": 2.2522225379943848,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.20463679730892181,
"rewards/margins": 0.032169945538043976,
"rewards/rejected": -0.2368067502975464,
"step": 21
},
{
"epoch": 0.03467297084318361,
"grad_norm": 0.7607430219650269,
"learning_rate": 6.614173228346457e-07,
"log_odds_chosen": 0.36671358346939087,
"log_odds_ratio": -0.5375123620033264,
"logits/chosen": -0.7311565279960632,
"logits/rejected": -0.08119023591279984,
"logps/chosen": -1.9978885650634766,
"logps/rejected": -2.3232674598693848,
"loss": 2.2612,
"nll_loss": 2.207477331161499,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1997888684272766,
"rewards/margins": 0.03253789618611336,
"rewards/rejected": -0.23232676088809967,
"step": 22
},
{
"epoch": 0.036249014972419225,
"grad_norm": 0.7234435081481934,
"learning_rate": 6.929133858267716e-07,
"log_odds_chosen": 0.39767685532569885,
"log_odds_ratio": -0.5170127153396606,
"logits/chosen": -0.7005606293678284,
"logits/rejected": -0.19814857840538025,
"logps/chosen": -2.0349388122558594,
"logps/rejected": -2.389338254928589,
"loss": 2.2728,
"nll_loss": 2.221050977706909,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.2034938931465149,
"rewards/margins": 0.03543993830680847,
"rewards/rejected": -0.23893383145332336,
"step": 23
},
{
"epoch": 0.037825059101654845,
"grad_norm": 0.7151917815208435,
"learning_rate": 7.244094488188977e-07,
"log_odds_chosen": 0.4232009947299957,
"log_odds_ratio": -0.5118768215179443,
"logits/chosen": -0.42520439624786377,
"logits/rejected": -0.13911336660385132,
"logps/chosen": -1.974929690361023,
"logps/rejected": -2.349339723587036,
"loss": 2.2336,
"nll_loss": 2.1824424266815186,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19749295711517334,
"rewards/margins": 0.03744099289178848,
"rewards/rejected": -0.23493395745754242,
"step": 24
},
{
"epoch": 0.039401103230890466,
"grad_norm": 0.7771602272987366,
"learning_rate": 7.559055118110236e-07,
"log_odds_chosen": 0.3602335751056671,
"log_odds_ratio": -0.5373751521110535,
"logits/chosen": -0.7194356918334961,
"logits/rejected": -0.2223992645740509,
"logps/chosen": -2.008389711380005,
"logps/rejected": -2.332648277282715,
"loss": 2.2619,
"nll_loss": 2.208117961883545,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.20083898305892944,
"rewards/margins": 0.03242585435509682,
"rewards/rejected": -0.23326483368873596,
"step": 25
},
{
"epoch": 0.04097714736012608,
"grad_norm": 0.7429983615875244,
"learning_rate": 7.874015748031496e-07,
"log_odds_chosen": 0.46550512313842773,
"log_odds_ratio": -0.4939187467098236,
"logits/chosen": -0.8255457282066345,
"logits/rejected": -0.3358853757381439,
"logps/chosen": -1.954911708831787,
"logps/rejected": -2.3684439659118652,
"loss": 2.2146,
"nll_loss": 2.1652259826660156,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19549117982387543,
"rewards/margins": 0.041353195905685425,
"rewards/rejected": -0.23684436082839966,
"step": 26
},
{
"epoch": 0.0425531914893617,
"grad_norm": 0.7892085313796997,
"learning_rate": 8.188976377952755e-07,
"log_odds_chosen": 0.7098885774612427,
"log_odds_ratio": -0.41376522183418274,
"logits/chosen": -0.6670718789100647,
"logits/rejected": -0.18330247700214386,
"logps/chosen": -2.0041563510894775,
"logps/rejected": -2.642397165298462,
"loss": 2.2622,
"nll_loss": 2.2208266258239746,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.20041564106941223,
"rewards/margins": 0.06382407993078232,
"rewards/rejected": -0.26423972845077515,
"step": 27
},
{
"epoch": 0.04412923561859732,
"grad_norm": 0.7302277684211731,
"learning_rate": 8.503937007874016e-07,
"log_odds_chosen": 0.4039226770401001,
"log_odds_ratio": -0.5200110673904419,
"logits/chosen": -0.6893348097801208,
"logits/rejected": -0.09423862397670746,
"logps/chosen": -1.9387571811676025,
"logps/rejected": -2.295835494995117,
"loss": 2.2167,
"nll_loss": 2.1646535396575928,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19387571513652802,
"rewards/margins": 0.03570783883333206,
"rewards/rejected": -0.22958354651927948,
"step": 28
},
{
"epoch": 0.045705279747832936,
"grad_norm": 0.6707211136817932,
"learning_rate": 8.818897637795275e-07,
"log_odds_chosen": 0.35679179430007935,
"log_odds_ratio": -0.5352319478988647,
"logits/chosen": -0.5461763739585876,
"logits/rejected": -0.14010290801525116,
"logps/chosen": -2.050537109375,
"logps/rejected": -2.372122287750244,
"loss": 2.3054,
"nll_loss": 2.2518763542175293,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.20505373179912567,
"rewards/margins": 0.032158493995666504,
"rewards/rejected": -0.23721222579479218,
"step": 29
},
{
"epoch": 0.04728132387706856,
"grad_norm": 0.8279486894607544,
"learning_rate": 9.133858267716535e-07,
"log_odds_chosen": 0.5706252455711365,
"log_odds_ratio": -0.45796477794647217,
"logits/chosen": -0.6770057678222656,
"logits/rejected": -0.2806178331375122,
"logps/chosen": -2.0110249519348145,
"logps/rejected": -2.521954298019409,
"loss": 2.3011,
"nll_loss": 2.2553153038024902,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.20110251009464264,
"rewards/margins": 0.05109292268753052,
"rewards/rejected": -0.25219541788101196,
"step": 30
},
{
"epoch": 0.04885736800630418,
"grad_norm": 0.7405450344085693,
"learning_rate": 9.448818897637795e-07,
"log_odds_chosen": 0.6274739503860474,
"log_odds_ratio": -0.4411022663116455,
"logits/chosen": -0.6367984414100647,
"logits/rejected": 0.004103410989046097,
"logps/chosen": -2.0546810626983643,
"logps/rejected": -2.6223583221435547,
"loss": 2.3002,
"nll_loss": 2.2560579776763916,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.20546811819076538,
"rewards/margins": 0.056767746806144714,
"rewards/rejected": -0.2622358202934265,
"step": 31
},
{
"epoch": 0.0504334121355398,
"grad_norm": 0.638395369052887,
"learning_rate": 9.763779527559055e-07,
"log_odds_chosen": 0.6788234114646912,
"log_odds_ratio": -0.42549028992652893,
"logits/chosen": -0.570993185043335,
"logits/rejected": -0.3356379568576813,
"logps/chosen": -1.9266096353530884,
"logps/rejected": -2.533726692199707,
"loss": 2.1793,
"nll_loss": 2.136709690093994,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19266097247600555,
"rewards/margins": 0.060711681842803955,
"rewards/rejected": -0.2533726394176483,
"step": 32
},
{
"epoch": 0.05200945626477541,
"grad_norm": 0.6255950331687927,
"learning_rate": 1.0078740157480315e-06,
"log_odds_chosen": 0.5599108338356018,
"log_odds_ratio": -0.4542999565601349,
"logits/chosen": -0.5288444757461548,
"logits/rejected": -0.19333161413669586,
"logps/chosen": -1.9585925340652466,
"logps/rejected": -2.4559030532836914,
"loss": 2.2007,
"nll_loss": 2.155308961868286,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19585925340652466,
"rewards/margins": 0.04973103851079941,
"rewards/rejected": -0.24559029936790466,
"step": 33
},
{
"epoch": 0.05358550039401103,
"grad_norm": 0.6921458840370178,
"learning_rate": 1.0393700787401573e-06,
"log_odds_chosen": 0.3519511818885803,
"log_odds_ratio": -0.5428643226623535,
"logits/chosen": -0.536325216293335,
"logits/rejected": -0.2403833270072937,
"logps/chosen": -1.9013547897338867,
"logps/rejected": -2.205166816711426,
"loss": 2.1703,
"nll_loss": 2.116063117980957,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19013547897338867,
"rewards/margins": 0.030381204560399055,
"rewards/rejected": -0.22051668167114258,
"step": 34
},
{
"epoch": 0.055161544523246654,
"grad_norm": 0.8423165082931519,
"learning_rate": 1.0708661417322836e-06,
"log_odds_chosen": 0.42774710059165955,
"log_odds_ratio": -0.5038249492645264,
"logits/chosen": -0.7048341631889343,
"logits/rejected": 0.0643918514251709,
"logps/chosen": -2.080242872238159,
"logps/rejected": -2.46270489692688,
"loss": 2.3413,
"nll_loss": 2.2909011840820312,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.2080242931842804,
"rewards/margins": 0.03824619948863983,
"rewards/rejected": -0.24627049267292023,
"step": 35
},
{
"epoch": 0.05673758865248227,
"grad_norm": 0.7374855875968933,
"learning_rate": 1.1023622047244094e-06,
"log_odds_chosen": 0.5830976963043213,
"log_odds_ratio": -0.44582459330558777,
"logits/chosen": -0.8039961457252502,
"logits/rejected": -0.06987367570400238,
"logps/chosen": -1.92990243434906,
"logps/rejected": -2.4467267990112305,
"loss": 2.1837,
"nll_loss": 2.139101266860962,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.192990243434906,
"rewards/margins": 0.05168245732784271,
"rewards/rejected": -0.2446727156639099,
"step": 36
},
{
"epoch": 0.05831363278171789,
"grad_norm": 0.6995700001716614,
"learning_rate": 1.1338582677165354e-06,
"log_odds_chosen": 0.4410613477230072,
"log_odds_ratio": -0.5076866149902344,
"logits/chosen": -0.6389314532279968,
"logits/rejected": -0.06876173615455627,
"logps/chosen": -1.9238637685775757,
"logps/rejected": -2.3091113567352295,
"loss": 2.1937,
"nll_loss": 2.1429271697998047,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19238635897636414,
"rewards/margins": 0.03852475434541702,
"rewards/rejected": -0.23091113567352295,
"step": 37
},
{
"epoch": 0.05988967691095351,
"grad_norm": 0.7543565034866333,
"learning_rate": 1.1653543307086612e-06,
"log_odds_chosen": 0.3667663633823395,
"log_odds_ratio": -0.5343418121337891,
"logits/chosen": -0.7570109367370605,
"logits/rejected": -0.23115745186805725,
"logps/chosen": -1.921524167060852,
"logps/rejected": -2.239858865737915,
"loss": 2.1977,
"nll_loss": 2.1442925930023193,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19215241074562073,
"rewards/margins": 0.03183349221944809,
"rewards/rejected": -0.22398591041564941,
"step": 38
},
{
"epoch": 0.061465721040189124,
"grad_norm": 0.7876814007759094,
"learning_rate": 1.1968503937007875e-06,
"log_odds_chosen": 0.2030404657125473,
"log_odds_ratio": -0.6065970659255981,
"logits/chosen": -0.6633272171020508,
"logits/rejected": -0.040516383945941925,
"logps/chosen": -2.038562774658203,
"logps/rejected": -2.220552444458008,
"loss": 2.3176,
"nll_loss": 2.256978988647461,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.20385627448558807,
"rewards/margins": 0.01819896697998047,
"rewards/rejected": -0.22205524146556854,
"step": 39
},
{
"epoch": 0.06304176516942474,
"grad_norm": 0.7621078491210938,
"learning_rate": 1.2283464566929133e-06,
"log_odds_chosen": 0.37610867619514465,
"log_odds_ratio": -0.5284633040428162,
"logits/chosen": -0.7413150668144226,
"logits/rejected": -0.18302536010742188,
"logps/chosen": -1.9204038381576538,
"logps/rejected": -2.2485194206237793,
"loss": 2.1889,
"nll_loss": 2.1360361576080322,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19204038381576538,
"rewards/margins": 0.03281155228614807,
"rewards/rejected": -0.22485193610191345,
"step": 40
},
{
"epoch": 0.06461780929866036,
"grad_norm": 0.6990500688552856,
"learning_rate": 1.2598425196850393e-06,
"log_odds_chosen": 0.27543094754219055,
"log_odds_ratio": -0.5724075436592102,
"logits/chosen": -0.6147856712341309,
"logits/rejected": -0.09826792776584625,
"logps/chosen": -1.9605408906936646,
"logps/rejected": -2.205174446105957,
"loss": 2.2093,
"nll_loss": 2.152068614959717,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1960541009902954,
"rewards/margins": 0.024463361129164696,
"rewards/rejected": -0.22051745653152466,
"step": 41
},
{
"epoch": 0.06619385342789598,
"grad_norm": 0.6998929381370544,
"learning_rate": 1.2913385826771652e-06,
"log_odds_chosen": 0.3089278042316437,
"log_odds_ratio": -0.5570518374443054,
"logits/chosen": -0.6978001594543457,
"logits/rejected": -0.15154145658016205,
"logps/chosen": -1.960097074508667,
"logps/rejected": -2.2320406436920166,
"loss": 2.2221,
"nll_loss": 2.166372537612915,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19600971043109894,
"rewards/margins": 0.02719433605670929,
"rewards/rejected": -0.22320404648780823,
"step": 42
},
{
"epoch": 0.0677698975571316,
"grad_norm": 0.6787785291671753,
"learning_rate": 1.3228346456692914e-06,
"log_odds_chosen": 0.16384947299957275,
"log_odds_ratio": -0.6235592365264893,
"logits/chosen": -0.5829145312309265,
"logits/rejected": -0.14386498928070068,
"logps/chosen": -1.978913426399231,
"logps/rejected": -2.123264789581299,
"loss": 2.2412,
"nll_loss": 2.1788277626037598,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.19789133965969086,
"rewards/margins": 0.014435119926929474,
"rewards/rejected": -0.21232648193836212,
"step": 43
},
{
"epoch": 0.06934594168636722,
"grad_norm": 0.7161246538162231,
"learning_rate": 1.3543307086614172e-06,
"log_odds_chosen": 0.5076345801353455,
"log_odds_ratio": -0.4807063937187195,
"logits/chosen": -0.7080238461494446,
"logits/rejected": 0.010522328317165375,
"logps/chosen": -1.9329512119293213,
"logps/rejected": -2.384503126144409,
"loss": 2.1908,
"nll_loss": 2.142735004425049,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19329513609409332,
"rewards/margins": 0.045155204832553864,
"rewards/rejected": -0.2384503185749054,
"step": 44
},
{
"epoch": 0.07092198581560284,
"grad_norm": 0.6544040441513062,
"learning_rate": 1.3858267716535433e-06,
"log_odds_chosen": 0.5975862145423889,
"log_odds_ratio": -0.4529675245285034,
"logits/chosen": -0.5176110863685608,
"logits/rejected": -0.06739248335361481,
"logps/chosen": -1.9081940650939941,
"logps/rejected": -2.438753843307495,
"loss": 2.1874,
"nll_loss": 2.1420602798461914,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1908193975687027,
"rewards/margins": 0.05305597186088562,
"rewards/rejected": -0.24387536942958832,
"step": 45
},
{
"epoch": 0.07249802994483845,
"grad_norm": 0.7946521043777466,
"learning_rate": 1.417322834645669e-06,
"log_odds_chosen": 0.5836816430091858,
"log_odds_ratio": -0.458423912525177,
"logits/chosen": -0.7441750764846802,
"logits/rejected": -0.32784658670425415,
"logps/chosen": -1.9207260608673096,
"logps/rejected": -2.439450740814209,
"loss": 2.1926,
"nll_loss": 2.14674973487854,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19207260012626648,
"rewards/margins": 0.05187246948480606,
"rewards/rejected": -0.24394509196281433,
"step": 46
},
{
"epoch": 0.07407407407407407,
"grad_norm": 0.6708806753158569,
"learning_rate": 1.4488188976377953e-06,
"log_odds_chosen": 0.2689046561717987,
"log_odds_ratio": -0.578476071357727,
"logits/chosen": -0.7736371159553528,
"logits/rejected": -0.19334951043128967,
"logps/chosen": -1.9722037315368652,
"logps/rejected": -2.2119808197021484,
"loss": 2.2323,
"nll_loss": 2.1744863986968994,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19722038507461548,
"rewards/margins": 0.023977704346179962,
"rewards/rejected": -0.22119809687137604,
"step": 47
},
{
"epoch": 0.07565011820330969,
"grad_norm": 0.7817642688751221,
"learning_rate": 1.4803149606299211e-06,
"log_odds_chosen": 0.413425475358963,
"log_odds_ratio": -0.5176951289176941,
"logits/chosen": -0.5976810455322266,
"logits/rejected": -0.1483435034751892,
"logps/chosen": -2.023184061050415,
"logps/rejected": -2.3940744400024414,
"loss": 2.2969,
"nll_loss": 2.2451674938201904,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.20231840014457703,
"rewards/margins": 0.03708904981613159,
"rewards/rejected": -0.23940744996070862,
"step": 48
},
{
"epoch": 0.07722616233254531,
"grad_norm": 0.631397008895874,
"learning_rate": 1.5118110236220472e-06,
"log_odds_chosen": 0.4730543792247772,
"log_odds_ratio": -0.49180155992507935,
"logits/chosen": -0.622490406036377,
"logits/rejected": -0.015296130441129208,
"logps/chosen": -1.9089866876602173,
"logps/rejected": -2.3237972259521484,
"loss": 2.1401,
"nll_loss": 2.0909266471862793,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19089870154857635,
"rewards/margins": 0.04148102179169655,
"rewards/rejected": -0.2323797196149826,
"step": 49
},
{
"epoch": 0.07880220646178093,
"grad_norm": 0.6548082232475281,
"learning_rate": 1.543307086614173e-06,
"log_odds_chosen": 0.2945685088634491,
"log_odds_ratio": -0.5625656843185425,
"logits/chosen": -0.638104259967804,
"logits/rejected": -0.22048690915107727,
"logps/chosen": -1.9748042821884155,
"logps/rejected": -2.235903024673462,
"loss": 2.2419,
"nll_loss": 2.185655355453491,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19748042523860931,
"rewards/margins": 0.026109864935278893,
"rewards/rejected": -0.22359029948711395,
"step": 50
},
{
"epoch": 0.08037825059101655,
"grad_norm": 0.6355348825454712,
"learning_rate": 1.5748031496062992e-06,
"log_odds_chosen": 0.6154743432998657,
"log_odds_ratio": -0.4387228488922119,
"logits/chosen": -0.5254096984863281,
"logits/rejected": -0.08727583289146423,
"logps/chosen": -1.8662408590316772,
"logps/rejected": -2.4071710109710693,
"loss": 2.1184,
"nll_loss": 2.0745527744293213,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18662410974502563,
"rewards/margins": 0.054093025624752045,
"rewards/rejected": -0.24071712791919708,
"step": 51
},
{
"epoch": 0.08195429472025216,
"grad_norm": 0.6476663947105408,
"learning_rate": 1.6062992125984253e-06,
"log_odds_chosen": 0.5350648760795593,
"log_odds_ratio": -0.46677446365356445,
"logits/chosen": -0.4835931360721588,
"logits/rejected": -0.016967706382274628,
"logps/chosen": -2.0277915000915527,
"logps/rejected": -2.507091522216797,
"loss": 2.2718,
"nll_loss": 2.2251675128936768,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.202779158949852,
"rewards/margins": 0.04792997986078262,
"rewards/rejected": -0.2507091462612152,
"step": 52
},
{
"epoch": 0.08353033884948778,
"grad_norm": 0.718711793422699,
"learning_rate": 1.637795275590551e-06,
"log_odds_chosen": 0.3033553957939148,
"log_odds_ratio": -0.5565884709358215,
"logits/chosen": -0.5977045297622681,
"logits/rejected": -0.13718965649604797,
"logps/chosen": -1.9643840789794922,
"logps/rejected": -2.230160713195801,
"loss": 2.2233,
"nll_loss": 2.1675939559936523,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19643841683864594,
"rewards/margins": 0.026577647775411606,
"rewards/rejected": -0.22301605343818665,
"step": 53
},
{
"epoch": 0.0851063829787234,
"grad_norm": 0.6878601908683777,
"learning_rate": 1.6692913385826771e-06,
"log_odds_chosen": 0.30879032611846924,
"log_odds_ratio": -0.5532358288764954,
"logits/chosen": -0.6751337051391602,
"logits/rejected": -0.14412644505500793,
"logps/chosen": -1.903916835784912,
"logps/rejected": -2.1728036403656006,
"loss": 2.1706,
"nll_loss": 2.1152570247650146,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1903916895389557,
"rewards/margins": 0.026888679713010788,
"rewards/rejected": -0.2172803431749344,
"step": 54
},
{
"epoch": 0.08668242710795902,
"grad_norm": 0.7093151211738586,
"learning_rate": 1.7007874015748031e-06,
"log_odds_chosen": 0.42612695693969727,
"log_odds_ratio": -0.5059034824371338,
"logits/chosen": -0.6053857803344727,
"logits/rejected": -0.21064752340316772,
"logps/chosen": -1.9867243766784668,
"logps/rejected": -2.36510968208313,
"loss": 2.2557,
"nll_loss": 2.205127716064453,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19867242872714996,
"rewards/margins": 0.037838518619537354,
"rewards/rejected": -0.23651094734668732,
"step": 55
},
{
"epoch": 0.08825847123719464,
"grad_norm": 0.8093518614768982,
"learning_rate": 1.7322834645669292e-06,
"log_odds_chosen": 0.352740615606308,
"log_odds_ratio": -0.5422862768173218,
"logits/chosen": -0.5116597414016724,
"logits/rejected": -0.0041604433208703995,
"logps/chosen": -2.0882434844970703,
"logps/rejected": -2.4050841331481934,
"loss": 2.3456,
"nll_loss": 2.291337728500366,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.20882436633110046,
"rewards/margins": 0.03168405592441559,
"rewards/rejected": -0.24050842225551605,
"step": 56
},
{
"epoch": 0.08983451536643026,
"grad_norm": 0.6063317656517029,
"learning_rate": 1.763779527559055e-06,
"log_odds_chosen": 0.37979212403297424,
"log_odds_ratio": -0.5244497656822205,
"logits/chosen": -0.5742859840393066,
"logits/rejected": -0.12732906639575958,
"logps/chosen": -1.9634265899658203,
"logps/rejected": -2.2987239360809326,
"loss": 2.2111,
"nll_loss": 2.158637285232544,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19634266197681427,
"rewards/margins": 0.03352972865104675,
"rewards/rejected": -0.22987240552902222,
"step": 57
},
{
"epoch": 0.09141055949566587,
"grad_norm": 0.7049276828765869,
"learning_rate": 1.795275590551181e-06,
"log_odds_chosen": 0.38489818572998047,
"log_odds_ratio": -0.5272811651229858,
"logits/chosen": -0.603106677532196,
"logits/rejected": -0.32046759128570557,
"logps/chosen": -1.931801199913025,
"logps/rejected": -2.274812936782837,
"loss": 2.1909,
"nll_loss": 2.138176918029785,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.193180114030838,
"rewards/margins": 0.034301191568374634,
"rewards/rejected": -0.22748132050037384,
"step": 58
},
{
"epoch": 0.09298660362490149,
"grad_norm": 0.7724094390869141,
"learning_rate": 1.826771653543307e-06,
"log_odds_chosen": 0.6565301418304443,
"log_odds_ratio": -0.4238013029098511,
"logits/chosen": -0.5493816137313843,
"logits/rejected": -0.04481405392289162,
"logps/chosen": -2.03456711769104,
"logps/rejected": -2.6273677349090576,
"loss": 2.2666,
"nll_loss": 2.2242276668548584,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.20345671474933624,
"rewards/margins": 0.05928007513284683,
"rewards/rejected": -0.2627367675304413,
"step": 59
},
{
"epoch": 0.09456264775413711,
"grad_norm": 0.7019063830375671,
"learning_rate": 1.858267716535433e-06,
"log_odds_chosen": 0.48238804936408997,
"log_odds_ratio": -0.4877777695655823,
"logits/chosen": -0.6007660627365112,
"logits/rejected": -0.25237271189689636,
"logps/chosen": -1.9639641046524048,
"logps/rejected": -2.3943371772766113,
"loss": 2.223,
"nll_loss": 2.174217462539673,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19639641046524048,
"rewards/margins": 0.04303732514381409,
"rewards/rejected": -0.23943373560905457,
"step": 60
},
{
"epoch": 0.09613869188337273,
"grad_norm": 0.7680268883705139,
"learning_rate": 1.889763779527559e-06,
"log_odds_chosen": 0.25518718361854553,
"log_odds_ratio": -0.5833750367164612,
"logits/chosen": -0.5432047247886658,
"logits/rejected": -0.21941693127155304,
"logps/chosen": -2.100229501724243,
"logps/rejected": -2.330411195755005,
"loss": 2.3455,
"nll_loss": 2.2871248722076416,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.2100229412317276,
"rewards/margins": 0.023018185049295425,
"rewards/rejected": -0.23304113745689392,
"step": 61
},
{
"epoch": 0.09771473601260836,
"grad_norm": 0.8415902853012085,
"learning_rate": 1.9212598425196847e-06,
"log_odds_chosen": 0.3082142472267151,
"log_odds_ratio": -0.5612497329711914,
"logits/chosen": -0.6062051653862,
"logits/rejected": -0.11909964680671692,
"logps/chosen": -2.0706098079681396,
"logps/rejected": -2.3465754985809326,
"loss": 2.3287,
"nll_loss": 2.272575616836548,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.20706097781658173,
"rewards/margins": 0.02759658358991146,
"rewards/rejected": -0.23465755581855774,
"step": 62
},
{
"epoch": 0.09929078014184398,
"grad_norm": 0.6543618440628052,
"learning_rate": 1.952755905511811e-06,
"log_odds_chosen": 0.5749139785766602,
"log_odds_ratio": -0.45239800214767456,
"logits/chosen": -0.7079123258590698,
"logits/rejected": -0.2599683403968811,
"logps/chosen": -1.8045680522918701,
"logps/rejected": -2.300361394882202,
"loss": 2.0753,
"nll_loss": 2.030048370361328,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18045681715011597,
"rewards/margins": 0.04957934468984604,
"rewards/rejected": -0.23003613948822021,
"step": 63
},
{
"epoch": 0.1008668242710796,
"grad_norm": 0.6648881435394287,
"learning_rate": 1.9842519685039368e-06,
"log_odds_chosen": 0.6830120086669922,
"log_odds_ratio": -0.4217277765274048,
"logits/chosen": -0.6593011617660522,
"logits/rejected": -0.08247893303632736,
"logps/chosen": -1.8293054103851318,
"logps/rejected": -2.4285809993743896,
"loss": 2.0953,
"nll_loss": 2.0531651973724365,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18293055891990662,
"rewards/margins": 0.059927552938461304,
"rewards/rejected": -0.24285811185836792,
"step": 64
},
{
"epoch": 0.1024428684003152,
"grad_norm": 0.6505405902862549,
"learning_rate": 2.015748031496063e-06,
"log_odds_chosen": 0.6114057898521423,
"log_odds_ratio": -0.4410432279109955,
"logits/chosen": -0.6462138295173645,
"logits/rejected": -0.3533778786659241,
"logps/chosen": -1.9271063804626465,
"logps/rejected": -2.47088360786438,
"loss": 2.1845,
"nll_loss": 2.1404013633728027,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19271063804626465,
"rewards/margins": 0.0543777197599411,
"rewards/rejected": -0.24708837270736694,
"step": 65
},
{
"epoch": 0.10401891252955082,
"grad_norm": 0.6485214829444885,
"learning_rate": 2.047244094488189e-06,
"log_odds_chosen": 0.452458918094635,
"log_odds_ratio": -0.49970927834510803,
"logits/chosen": -0.6230807900428772,
"logits/rejected": -0.13434045016765594,
"logps/chosen": -2.0179097652435303,
"logps/rejected": -2.419290781021118,
"loss": 2.2838,
"nll_loss": 2.233837842941284,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.20179098844528198,
"rewards/margins": 0.04013810679316521,
"rewards/rejected": -0.2419290840625763,
"step": 66
},
{
"epoch": 0.10559495665878645,
"grad_norm": 0.6492588520050049,
"learning_rate": 2.0787401574803147e-06,
"log_odds_chosen": 0.5289919972419739,
"log_odds_ratio": -0.4675235152244568,
"logits/chosen": -0.534376859664917,
"logits/rejected": -0.03799459710717201,
"logps/chosen": -1.955706238746643,
"logps/rejected": -2.4241867065429688,
"loss": 2.1946,
"nll_loss": 2.1478805541992188,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19557063281536102,
"rewards/margins": 0.04684804379940033,
"rewards/rejected": -0.24241869151592255,
"step": 67
},
{
"epoch": 0.10717100078802207,
"grad_norm": 0.774642288684845,
"learning_rate": 2.1102362204724405e-06,
"log_odds_chosen": 0.41134878993034363,
"log_odds_ratio": -0.5170325040817261,
"logits/chosen": -0.6746619939804077,
"logits/rejected": -0.185140922665596,
"logps/chosen": -1.947751760482788,
"logps/rejected": -2.311985969543457,
"loss": 2.2474,
"nll_loss": 2.1957201957702637,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19477517902851105,
"rewards/margins": 0.036423418670892715,
"rewards/rejected": -0.23119859397411346,
"step": 68
},
{
"epoch": 0.10874704491725769,
"grad_norm": 0.8252844214439392,
"learning_rate": 2.141732283464567e-06,
"log_odds_chosen": 0.22346967458724976,
"log_odds_ratio": -0.5949736833572388,
"logits/chosen": -0.41877222061157227,
"logits/rejected": -0.11722514033317566,
"logps/chosen": -2.0319464206695557,
"logps/rejected": -2.2292232513427734,
"loss": 2.3097,
"nll_loss": 2.250192403793335,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.20319463312625885,
"rewards/margins": 0.019727692008018494,
"rewards/rejected": -0.22292232513427734,
"step": 69
},
{
"epoch": 0.11032308904649331,
"grad_norm": 0.6770060658454895,
"learning_rate": 2.173228346456693e-06,
"log_odds_chosen": 0.4207773804664612,
"log_odds_ratio": -0.5108780264854431,
"logits/chosen": -0.5316891670227051,
"logits/rejected": -0.10831936448812485,
"logps/chosen": -1.9700641632080078,
"logps/rejected": -2.342879056930542,
"loss": 2.2395,
"nll_loss": 2.188405752182007,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19700641930103302,
"rewards/margins": 0.03728148713707924,
"rewards/rejected": -0.23428791761398315,
"step": 70
},
{
"epoch": 0.11189913317572892,
"grad_norm": 0.8300355672836304,
"learning_rate": 2.204724409448819e-06,
"log_odds_chosen": 0.5063481330871582,
"log_odds_ratio": -0.48265108466148376,
"logits/chosen": -0.7789384126663208,
"logits/rejected": -0.24630165100097656,
"logps/chosen": -1.9212646484375,
"logps/rejected": -2.370161771774292,
"loss": 2.1808,
"nll_loss": 2.1325840950012207,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19212648272514343,
"rewards/margins": 0.04488971084356308,
"rewards/rejected": -0.23701618611812592,
"step": 71
},
{
"epoch": 0.11347517730496454,
"grad_norm": 0.6843920946121216,
"learning_rate": 2.2362204724409446e-06,
"log_odds_chosen": 0.32779812812805176,
"log_odds_ratio": -0.5560404062271118,
"logits/chosen": -0.5591788291931152,
"logits/rejected": -0.06292789429426193,
"logps/chosen": -1.936835527420044,
"logps/rejected": -2.2265937328338623,
"loss": 2.1968,
"nll_loss": 2.1412172317504883,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19368356466293335,
"rewards/margins": 0.02897578477859497,
"rewards/rejected": -0.22265934944152832,
"step": 72
},
{
"epoch": 0.11505122143420016,
"grad_norm": 0.6691780090332031,
"learning_rate": 2.267716535433071e-06,
"log_odds_chosen": 0.3424806594848633,
"log_odds_ratio": -0.5391549468040466,
"logits/chosen": -0.5084943175315857,
"logits/rejected": -0.18799816071987152,
"logps/chosen": -1.982399821281433,
"logps/rejected": -2.2857072353363037,
"loss": 2.2447,
"nll_loss": 2.190774917602539,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1982399970293045,
"rewards/margins": 0.0303307194262743,
"rewards/rejected": -0.22857069969177246,
"step": 73
},
{
"epoch": 0.11662726556343578,
"grad_norm": 0.7013347744941711,
"learning_rate": 2.2992125984251967e-06,
"log_odds_chosen": 0.42650213837623596,
"log_odds_ratio": -0.5057096481323242,
"logits/chosen": -0.6094076037406921,
"logits/rejected": -0.11771736294031143,
"logps/chosen": -1.8428946733474731,
"logps/rejected": -2.2101123332977295,
"loss": 2.1276,
"nll_loss": 2.0770435333251953,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18428948521614075,
"rewards/margins": 0.03672178089618683,
"rewards/rejected": -0.22101125121116638,
"step": 74
},
{
"epoch": 0.1182033096926714,
"grad_norm": 0.6637840867042542,
"learning_rate": 2.3307086614173225e-06,
"log_odds_chosen": 0.5096178650856018,
"log_odds_ratio": -0.476134717464447,
"logits/chosen": -0.5619401335716248,
"logits/rejected": -0.2677369713783264,
"logps/chosen": -1.8805265426635742,
"logps/rejected": -2.32961106300354,
"loss": 2.1385,
"nll_loss": 2.0909037590026855,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18805265426635742,
"rewards/margins": 0.044908471405506134,
"rewards/rejected": -0.23296113312244415,
"step": 75
},
{
"epoch": 0.11977935382190702,
"grad_norm": 0.6581621170043945,
"learning_rate": 2.3622047244094483e-06,
"log_odds_chosen": 0.5206456184387207,
"log_odds_ratio": -0.47490978240966797,
"logits/chosen": -0.5504530072212219,
"logits/rejected": -0.03281405568122864,
"logps/chosen": -1.9286243915557861,
"logps/rejected": -2.3910675048828125,
"loss": 2.1717,
"nll_loss": 2.124224901199341,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19286245107650757,
"rewards/margins": 0.04624428227543831,
"rewards/rejected": -0.23910671472549438,
"step": 76
},
{
"epoch": 0.12135539795114263,
"grad_norm": 0.7063953876495361,
"learning_rate": 2.393700787401575e-06,
"log_odds_chosen": 0.4808962941169739,
"log_odds_ratio": -0.49262717366218567,
"logits/chosen": -0.5677655339241028,
"logits/rejected": -0.21073025465011597,
"logps/chosen": -1.9511951208114624,
"logps/rejected": -2.377624988555908,
"loss": 2.2192,
"nll_loss": 2.169985771179199,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19511950016021729,
"rewards/margins": 0.042642995715141296,
"rewards/rejected": -0.23776251077651978,
"step": 77
},
{
"epoch": 0.12293144208037825,
"grad_norm": 0.5751843452453613,
"learning_rate": 2.425196850393701e-06,
"log_odds_chosen": 0.3792150020599365,
"log_odds_ratio": -0.5299723744392395,
"logits/chosen": -0.37227052450180054,
"logits/rejected": -0.4450679123401642,
"logps/chosen": -2.0182228088378906,
"logps/rejected": -2.355861186981201,
"loss": 2.2537,
"nll_loss": 2.200679302215576,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.20182228088378906,
"rewards/margins": 0.03376384079456329,
"rewards/rejected": -0.23558615148067474,
"step": 78
},
{
"epoch": 0.12450748620961387,
"grad_norm": 0.6594578623771667,
"learning_rate": 2.4566929133858266e-06,
"log_odds_chosen": 0.518637478351593,
"log_odds_ratio": -0.47430098056793213,
"logits/chosen": -0.4779280126094818,
"logits/rejected": -0.2910279631614685,
"logps/chosen": -1.9429047107696533,
"logps/rejected": -2.402376174926758,
"loss": 2.1903,
"nll_loss": 2.1428279876708984,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19429044425487518,
"rewards/margins": 0.04594714939594269,
"rewards/rejected": -0.24023759365081787,
"step": 79
},
{
"epoch": 0.12608353033884948,
"grad_norm": 0.6245352625846863,
"learning_rate": 2.488188976377953e-06,
"log_odds_chosen": 0.5585002303123474,
"log_odds_ratio": -0.45803701877593994,
"logits/chosen": -0.5769734978675842,
"logits/rejected": -0.25027596950531006,
"logps/chosen": -1.8669335842132568,
"logps/rejected": -2.356663227081299,
"loss": 2.1157,
"nll_loss": 2.06986403465271,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18669337034225464,
"rewards/margins": 0.048972949385643005,
"rewards/rejected": -0.23566631972789764,
"step": 80
},
{
"epoch": 0.1276595744680851,
"grad_norm": 0.5566908717155457,
"learning_rate": 2.5196850393700787e-06,
"log_odds_chosen": 0.3883778750896454,
"log_odds_ratio": -0.5254943370819092,
"logits/chosen": -0.42693546414375305,
"logits/rejected": -0.2633028030395508,
"logps/chosen": -1.93135666847229,
"logps/rejected": -2.2755541801452637,
"loss": 2.17,
"nll_loss": 2.1174182891845703,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19313567876815796,
"rewards/margins": 0.034419745206832886,
"rewards/rejected": -0.22755542397499084,
"step": 81
},
{
"epoch": 0.12923561859732072,
"grad_norm": 0.5781261324882507,
"learning_rate": 2.5511811023622045e-06,
"log_odds_chosen": 0.524163544178009,
"log_odds_ratio": -0.4806976020336151,
"logits/chosen": -0.4674437940120697,
"logits/rejected": -0.23945724964141846,
"logps/chosen": -1.8828051090240479,
"logps/rejected": -2.3516387939453125,
"loss": 2.1336,
"nll_loss": 2.085569381713867,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18828049302101135,
"rewards/margins": 0.046883389353752136,
"rewards/rejected": -0.23516389727592468,
"step": 82
},
{
"epoch": 0.13081166272655634,
"grad_norm": 0.6088637709617615,
"learning_rate": 2.5826771653543303e-06,
"log_odds_chosen": 0.36324411630630493,
"log_odds_ratio": -0.5395435690879822,
"logits/chosen": -0.39306196570396423,
"logits/rejected": -0.1700359582901001,
"logps/chosen": -1.9157882928848267,
"logps/rejected": -2.237780809402466,
"loss": 2.1604,
"nll_loss": 2.106419563293457,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19157883524894714,
"rewards/margins": 0.032199256122112274,
"rewards/rejected": -0.22377808392047882,
"step": 83
},
{
"epoch": 0.13238770685579196,
"grad_norm": 0.6803274750709534,
"learning_rate": 2.6141732283464566e-06,
"log_odds_chosen": 0.5524423122406006,
"log_odds_ratio": -0.46847474575042725,
"logits/chosen": -0.5545064210891724,
"logits/rejected": -0.24267421662807465,
"logps/chosen": -1.8639458417892456,
"logps/rejected": -2.3503522872924805,
"loss": 2.1458,
"nll_loss": 2.0989749431610107,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1863945871591568,
"rewards/margins": 0.048640646040439606,
"rewards/rejected": -0.235035240650177,
"step": 84
},
{
"epoch": 0.13396375098502758,
"grad_norm": 0.5811371207237244,
"learning_rate": 2.645669291338583e-06,
"log_odds_chosen": 0.5210408568382263,
"log_odds_ratio": -0.46715638041496277,
"logits/chosen": -0.3822883367538452,
"logits/rejected": -0.1786729395389557,
"logps/chosen": -1.9316679239273071,
"logps/rejected": -2.3919014930725098,
"loss": 2.1698,
"nll_loss": 2.1230902671813965,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1931667923927307,
"rewards/margins": 0.04602333903312683,
"rewards/rejected": -0.23919013142585754,
"step": 85
},
{
"epoch": 0.1355397951142632,
"grad_norm": 0.6602110266685486,
"learning_rate": 2.6771653543307086e-06,
"log_odds_chosen": 0.48973286151885986,
"log_odds_ratio": -0.48310309648513794,
"logits/chosen": -0.5846769213676453,
"logits/rejected": -0.20886988937854767,
"logps/chosen": -1.9184874296188354,
"logps/rejected": -2.349517822265625,
"loss": 2.1703,
"nll_loss": 2.1220102310180664,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1918487399816513,
"rewards/margins": 0.043103061616420746,
"rewards/rejected": -0.23495177924633026,
"step": 86
},
{
"epoch": 0.13711583924349882,
"grad_norm": 0.5744991302490234,
"learning_rate": 2.7086614173228344e-06,
"log_odds_chosen": 0.3423335552215576,
"log_odds_ratio": -0.5400257110595703,
"logits/chosen": -0.37106236815452576,
"logits/rejected": -0.360477089881897,
"logps/chosen": -1.922955870628357,
"logps/rejected": -2.2213680744171143,
"loss": 2.1639,
"nll_loss": 2.109863042831421,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19229556620121002,
"rewards/margins": 0.02984124794602394,
"rewards/rejected": -0.22213682532310486,
"step": 87
},
{
"epoch": 0.13869188337273444,
"grad_norm": 0.6963973045349121,
"learning_rate": 2.7401574803149607e-06,
"log_odds_chosen": 0.5418673753738403,
"log_odds_ratio": -0.475824773311615,
"logits/chosen": -0.6239266991615295,
"logits/rejected": -0.2058069109916687,
"logps/chosen": -1.848305583000183,
"logps/rejected": -2.326037645339966,
"loss": 2.1293,
"nll_loss": 2.0817408561706543,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18483057618141174,
"rewards/margins": 0.04777318611741066,
"rewards/rejected": -0.2326037585735321,
"step": 88
},
{
"epoch": 0.14026792750197006,
"grad_norm": 0.6552391648292542,
"learning_rate": 2.7716535433070865e-06,
"log_odds_chosen": 0.500076174736023,
"log_odds_ratio": -0.4754161834716797,
"logits/chosen": -0.44256362318992615,
"logits/rejected": -0.20719635486602783,
"logps/chosen": -1.9562758207321167,
"logps/rejected": -2.3987882137298584,
"loss": 2.2135,
"nll_loss": 2.165970802307129,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1956275850534439,
"rewards/margins": 0.04425125569105148,
"rewards/rejected": -0.2398788183927536,
"step": 89
},
{
"epoch": 0.14184397163120568,
"grad_norm": 0.6141228079795837,
"learning_rate": 2.8031496062992123e-06,
"log_odds_chosen": 0.428195595741272,
"log_odds_ratio": -0.5067068934440613,
"logits/chosen": -0.3649379312992096,
"logits/rejected": -0.1602931022644043,
"logps/chosen": -2.002263307571411,
"logps/rejected": -2.3845441341400146,
"loss": 2.2267,
"nll_loss": 2.1760218143463135,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.20022635161876678,
"rewards/margins": 0.038228072226047516,
"rewards/rejected": -0.2384544163942337,
"step": 90
},
{
"epoch": 0.1434200157604413,
"grad_norm": 0.5524324774742126,
"learning_rate": 2.834645669291338e-06,
"log_odds_chosen": 0.4524117112159729,
"log_odds_ratio": -0.49862101674079895,
"logits/chosen": -0.36407744884490967,
"logits/rejected": -0.2811731696128845,
"logps/chosen": -1.8061617612838745,
"logps/rejected": -2.198903799057007,
"loss": 2.0471,
"nll_loss": 1.997222900390625,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18061619997024536,
"rewards/margins": 0.0392741933465004,
"rewards/rejected": -0.21989038586616516,
"step": 91
},
{
"epoch": 0.1449960598896769,
"grad_norm": 0.5921797752380371,
"learning_rate": 2.8661417322834644e-06,
"log_odds_chosen": 0.5024532675743103,
"log_odds_ratio": -0.47816595435142517,
"logits/chosen": -0.44370609521865845,
"logits/rejected": -0.18724730610847473,
"logps/chosen": -1.829555869102478,
"logps/rejected": -2.2681233882904053,
"loss": 2.0879,
"nll_loss": 2.040083169937134,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18295560777187347,
"rewards/margins": 0.04385674372315407,
"rewards/rejected": -0.22681234776973724,
"step": 92
},
{
"epoch": 0.14657210401891252,
"grad_norm": 0.628095805644989,
"learning_rate": 2.8976377952755906e-06,
"log_odds_chosen": 0.43953385949134827,
"log_odds_ratio": -0.5222434401512146,
"logits/chosen": -0.4026286005973816,
"logits/rejected": -0.23625504970550537,
"logps/chosen": -1.9340780973434448,
"logps/rejected": -2.3274731636047363,
"loss": 2.18,
"nll_loss": 2.127760410308838,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.19340780377388,
"rewards/margins": 0.03933952748775482,
"rewards/rejected": -0.23274733126163483,
"step": 93
},
{
"epoch": 0.14814814814814814,
"grad_norm": 0.5658537149429321,
"learning_rate": 2.9291338582677165e-06,
"log_odds_chosen": 0.45617368817329407,
"log_odds_ratio": -0.4992842674255371,
"logits/chosen": -0.36675694584846497,
"logits/rejected": -0.3160392642021179,
"logps/chosen": -1.9161105155944824,
"logps/rejected": -2.315382480621338,
"loss": 2.1513,
"nll_loss": 2.1013565063476562,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19161105155944824,
"rewards/margins": 0.03992719575762749,
"rewards/rejected": -0.23153826594352722,
"step": 94
},
{
"epoch": 0.14972419227738376,
"grad_norm": 0.639642596244812,
"learning_rate": 2.9606299212598423e-06,
"log_odds_chosen": 0.5549046993255615,
"log_odds_ratio": -0.463879257440567,
"logits/chosen": -0.43291109800338745,
"logits/rejected": -0.15716485679149628,
"logps/chosen": -1.9124903678894043,
"logps/rejected": -2.4041366577148438,
"loss": 2.1648,
"nll_loss": 2.1183886528015137,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1912490427494049,
"rewards/margins": 0.049164604395627975,
"rewards/rejected": -0.240413635969162,
"step": 95
},
{
"epoch": 0.15130023640661938,
"grad_norm": 0.569778323173523,
"learning_rate": 2.9921259842519685e-06,
"log_odds_chosen": 0.5220546722412109,
"log_odds_ratio": -0.46969637274742126,
"logits/chosen": -0.42379647493362427,
"logits/rejected": -0.3440685570240021,
"logps/chosen": -1.9186266660690308,
"logps/rejected": -2.3788628578186035,
"loss": 2.141,
"nll_loss": 2.0940771102905273,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19186267256736755,
"rewards/margins": 0.04602360725402832,
"rewards/rejected": -0.23788626492023468,
"step": 96
},
{
"epoch": 0.152876280535855,
"grad_norm": 0.56700599193573,
"learning_rate": 3.0236220472440943e-06,
"log_odds_chosen": 0.350454717874527,
"log_odds_ratio": -0.5345020294189453,
"logits/chosen": -0.47989675402641296,
"logits/rejected": -0.17767037451267242,
"logps/chosen": -1.8971889019012451,
"logps/rejected": -2.200305461883545,
"loss": 2.1227,
"nll_loss": 2.069289445877075,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18971890211105347,
"rewards/margins": 0.03031165711581707,
"rewards/rejected": -0.22003056108951569,
"step": 97
},
{
"epoch": 0.15445232466509062,
"grad_norm": 0.5734896063804626,
"learning_rate": 3.05511811023622e-06,
"log_odds_chosen": 0.4143810570240021,
"log_odds_ratio": -0.512945294380188,
"logits/chosen": -0.31117746233940125,
"logits/rejected": -0.24713104963302612,
"logps/chosen": -2.0033698081970215,
"logps/rejected": -2.3717055320739746,
"loss": 2.2459,
"nll_loss": 2.1946518421173096,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.2003369927406311,
"rewards/margins": 0.03683357313275337,
"rewards/rejected": -0.23717054724693298,
"step": 98
},
{
"epoch": 0.15602836879432624,
"grad_norm": 0.6178714036941528,
"learning_rate": 3.086614173228346e-06,
"log_odds_chosen": 0.24748294055461884,
"log_odds_ratio": -0.5794985890388489,
"logits/chosen": -0.5359123945236206,
"logits/rejected": -0.3172164559364319,
"logps/chosen": -1.911259651184082,
"logps/rejected": -2.12546443939209,
"loss": 2.1682,
"nll_loss": 2.1102778911590576,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19112597405910492,
"rewards/margins": 0.021420463919639587,
"rewards/rejected": -0.2125464230775833,
"step": 99
},
{
"epoch": 0.15760441292356187,
"grad_norm": 0.590636670589447,
"learning_rate": 3.1181102362204722e-06,
"log_odds_chosen": 0.36373111605644226,
"log_odds_ratio": -0.533814549446106,
"logits/chosen": -0.4497278034687042,
"logits/rejected": -0.2689790427684784,
"logps/chosen": -1.8333114385604858,
"logps/rejected": -2.1512179374694824,
"loss": 2.0894,
"nll_loss": 2.036029815673828,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18333116173744202,
"rewards/margins": 0.031790636479854584,
"rewards/rejected": -0.2151218056678772,
"step": 100
},
{
"epoch": 0.15918045705279749,
"grad_norm": 0.5497896075248718,
"learning_rate": 3.1496062992125985e-06,
"log_odds_chosen": 0.42619574069976807,
"log_odds_ratio": -0.5083091855049133,
"logits/chosen": -0.3605027198791504,
"logits/rejected": -0.24103917181491852,
"logps/chosen": -1.8285539150238037,
"logps/rejected": -2.200834035873413,
"loss": 2.0895,
"nll_loss": 2.038670301437378,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18285538256168365,
"rewards/margins": 0.03722800686955452,
"rewards/rejected": -0.22008340060710907,
"step": 101
},
{
"epoch": 0.1607565011820331,
"grad_norm": 0.5178012251853943,
"learning_rate": 3.1811023622047243e-06,
"log_odds_chosen": 0.43138545751571655,
"log_odds_ratio": -0.5122407078742981,
"logits/chosen": -0.3726266324520111,
"logits/rejected": -0.37165510654449463,
"logps/chosen": -1.7784022092819214,
"logps/rejected": -2.1532044410705566,
"loss": 2.0323,
"nll_loss": 1.9810512065887451,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1778402179479599,
"rewards/margins": 0.03748023882508278,
"rewards/rejected": -0.21532045304775238,
"step": 102
},
{
"epoch": 0.16233254531126873,
"grad_norm": 0.5599470734596252,
"learning_rate": 3.2125984251968505e-06,
"log_odds_chosen": 0.4818297326564789,
"log_odds_ratio": -0.49377134442329407,
"logits/chosen": -0.2876685857772827,
"logits/rejected": -0.038474664092063904,
"logps/chosen": -1.7715296745300293,
"logps/rejected": -2.1862025260925293,
"loss": 2.0313,
"nll_loss": 1.9819062948226929,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17715296149253845,
"rewards/margins": 0.04146728664636612,
"rewards/rejected": -0.21862025558948517,
"step": 103
},
{
"epoch": 0.16390858944050432,
"grad_norm": 0.5964917540550232,
"learning_rate": 3.2440944881889763e-06,
"log_odds_chosen": 0.4325031638145447,
"log_odds_ratio": -0.5093069672584534,
"logits/chosen": -0.23736125230789185,
"logits/rejected": -0.06159596145153046,
"logps/chosen": -1.9128376245498657,
"logps/rejected": -2.294654369354248,
"loss": 2.1641,
"nll_loss": 2.113180637359619,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19128376245498657,
"rewards/margins": 0.03818168863654137,
"rewards/rejected": -0.22946545481681824,
"step": 104
},
{
"epoch": 0.16548463356973994,
"grad_norm": 0.4927206039428711,
"learning_rate": 3.275590551181102e-06,
"log_odds_chosen": 0.5299558043479919,
"log_odds_ratio": -0.4695979058742523,
"logits/chosen": -0.377352774143219,
"logits/rejected": -0.17640215158462524,
"logps/chosen": -1.7906301021575928,
"logps/rejected": -2.250798225402832,
"loss": 2.0519,
"nll_loss": 2.0049400329589844,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17906302213668823,
"rewards/margins": 0.04601679742336273,
"rewards/rejected": -0.22507980465888977,
"step": 105
},
{
"epoch": 0.16706067769897556,
"grad_norm": 0.5032868385314941,
"learning_rate": 3.307086614173228e-06,
"log_odds_chosen": 0.4509121775627136,
"log_odds_ratio": -0.5058165788650513,
"logits/chosen": -0.35679227113723755,
"logits/rejected": -0.1107356995344162,
"logps/chosen": -1.8084826469421387,
"logps/rejected": -2.190898895263672,
"loss": 2.0546,
"nll_loss": 2.003988742828369,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18084825575351715,
"rewards/margins": 0.03824164718389511,
"rewards/rejected": -0.21908989548683167,
"step": 106
},
{
"epoch": 0.16863672182821118,
"grad_norm": 0.5300337672233582,
"learning_rate": 3.3385826771653542e-06,
"log_odds_chosen": 0.43685051798820496,
"log_odds_ratio": -0.5093865394592285,
"logits/chosen": -0.2651398479938507,
"logits/rejected": -0.28625091910362244,
"logps/chosen": -1.8772473335266113,
"logps/rejected": -2.2612218856811523,
"loss": 2.118,
"nll_loss": 2.0670440196990967,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1877247393131256,
"rewards/margins": 0.03839743137359619,
"rewards/rejected": -0.226122185587883,
"step": 107
},
{
"epoch": 0.1702127659574468,
"grad_norm": 0.49163365364074707,
"learning_rate": 3.37007874015748e-06,
"log_odds_chosen": 0.3183455765247345,
"log_odds_ratio": -0.5563209056854248,
"logits/chosen": -0.23882922530174255,
"logits/rejected": -0.3078491985797882,
"logps/chosen": -1.85860013961792,
"logps/rejected": -2.1333258152008057,
"loss": 2.102,
"nll_loss": 2.0463998317718506,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1858600229024887,
"rewards/margins": 0.02747257985174656,
"rewards/rejected": -0.21333259344100952,
"step": 108
},
{
"epoch": 0.17178881008668243,
"grad_norm": 0.5118012428283691,
"learning_rate": 3.4015748031496063e-06,
"log_odds_chosen": 0.4563853442668915,
"log_odds_ratio": -0.49264228343963623,
"logits/chosen": -0.16680516302585602,
"logits/rejected": -0.2830784320831299,
"logps/chosen": -1.9145252704620361,
"logps/rejected": -2.3134219646453857,
"loss": 2.1528,
"nll_loss": 2.1035213470458984,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19145254790782928,
"rewards/margins": 0.03988967090845108,
"rewards/rejected": -0.23134221136569977,
"step": 109
},
{
"epoch": 0.17336485421591805,
"grad_norm": 0.505682110786438,
"learning_rate": 3.433070866141732e-06,
"log_odds_chosen": 0.49830734729766846,
"log_odds_ratio": -0.48064208030700684,
"logits/chosen": -0.27441835403442383,
"logits/rejected": -0.24226327240467072,
"logps/chosen": -1.8005558252334595,
"logps/rejected": -2.2267792224884033,
"loss": 2.0182,
"nll_loss": 1.9701097011566162,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18005558848381042,
"rewards/margins": 0.04262235015630722,
"rewards/rejected": -0.22267794609069824,
"step": 110
},
{
"epoch": 0.17494089834515367,
"grad_norm": 0.5464332103729248,
"learning_rate": 3.4645669291338583e-06,
"log_odds_chosen": 0.5071738362312317,
"log_odds_ratio": -0.4801686406135559,
"logits/chosen": -0.23766781389713287,
"logits/rejected": -0.30691930651664734,
"logps/chosen": -1.902522087097168,
"logps/rejected": -2.3450286388397217,
"loss": 2.1483,
"nll_loss": 2.10026478767395,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19025221467018127,
"rewards/margins": 0.044250644743442535,
"rewards/rejected": -0.2345028668642044,
"step": 111
},
{
"epoch": 0.1765169424743893,
"grad_norm": 0.5223987102508545,
"learning_rate": 3.496062992125984e-06,
"log_odds_chosen": 0.3665584325790405,
"log_odds_ratio": -0.5315839052200317,
"logits/chosen": -0.20885254442691803,
"logits/rejected": -0.12856575846672058,
"logps/chosen": -1.9509872198104858,
"logps/rejected": -2.271317481994629,
"loss": 2.1848,
"nll_loss": 2.131622314453125,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19509869813919067,
"rewards/margins": 0.0320330373942852,
"rewards/rejected": -0.22713173925876617,
"step": 112
},
{
"epoch": 0.1780929866036249,
"grad_norm": 0.5140707492828369,
"learning_rate": 3.52755905511811e-06,
"log_odds_chosen": 0.38427114486694336,
"log_odds_ratio": -0.526176929473877,
"logits/chosen": -0.3687783181667328,
"logits/rejected": -0.2154863178730011,
"logps/chosen": -1.7950341701507568,
"logps/rejected": -2.127782106399536,
"loss": 2.0463,
"nll_loss": 1.9936522245407104,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1795034110546112,
"rewards/margins": 0.03327479213476181,
"rewards/rejected": -0.21277819573879242,
"step": 113
},
{
"epoch": 0.17966903073286053,
"grad_norm": 0.48041167855262756,
"learning_rate": 3.559055118110236e-06,
"log_odds_chosen": 0.2929834723472595,
"log_odds_ratio": -0.5727202892303467,
"logits/chosen": -0.11503149569034576,
"logits/rejected": -0.22482499480247498,
"logps/chosen": -1.921325922012329,
"logps/rejected": -2.181408405303955,
"loss": 2.167,
"nll_loss": 2.1096832752227783,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1921325922012329,
"rewards/margins": 0.026008253917098045,
"rewards/rejected": -0.2181408405303955,
"step": 114
},
{
"epoch": 0.18124507486209615,
"grad_norm": 0.483804315328598,
"learning_rate": 3.590551181102362e-06,
"log_odds_chosen": 0.46048107743263245,
"log_odds_ratio": -0.4994431436061859,
"logits/chosen": -0.21135865151882172,
"logits/rejected": -0.4032437205314636,
"logps/chosen": -1.8342138528823853,
"logps/rejected": -2.2355825901031494,
"loss": 2.0671,
"nll_loss": 2.017176389694214,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18342137336730957,
"rewards/margins": 0.040136873722076416,
"rewards/rejected": -0.2235582321882248,
"step": 115
},
{
"epoch": 0.18282111899133174,
"grad_norm": 0.4813881814479828,
"learning_rate": 3.622047244094488e-06,
"log_odds_chosen": 0.3711977005004883,
"log_odds_ratio": -0.5412157773971558,
"logits/chosen": -0.22255532443523407,
"logits/rejected": -0.2175833135843277,
"logps/chosen": -1.8901299238204956,
"logps/rejected": -2.2157669067382812,
"loss": 2.1143,
"nll_loss": 2.060180187225342,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18901298940181732,
"rewards/margins": 0.032563693821430206,
"rewards/rejected": -0.22157667577266693,
"step": 116
},
{
"epoch": 0.18439716312056736,
"grad_norm": 0.4870263934135437,
"learning_rate": 3.653543307086614e-06,
"log_odds_chosen": 0.526931881904602,
"log_odds_ratio": -0.47742602229118347,
"logits/chosen": -0.2706168293952942,
"logits/rejected": -0.5030975341796875,
"logps/chosen": -1.8207095861434937,
"logps/rejected": -2.2792084217071533,
"loss": 2.0652,
"nll_loss": 2.0174200534820557,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18207095563411713,
"rewards/margins": 0.045849889516830444,
"rewards/rejected": -0.22792083024978638,
"step": 117
},
{
"epoch": 0.18597320724980299,
"grad_norm": 0.47839877009391785,
"learning_rate": 3.68503937007874e-06,
"log_odds_chosen": 0.30926501750946045,
"log_odds_ratio": -0.5605666637420654,
"logits/chosen": -0.08142746239900589,
"logits/rejected": -0.2830328047275543,
"logps/chosen": -1.83827805519104,
"logps/rejected": -2.1060383319854736,
"loss": 2.0833,
"nll_loss": 2.0272810459136963,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18382780253887177,
"rewards/margins": 0.026776034384965897,
"rewards/rejected": -0.21060383319854736,
"step": 118
},
{
"epoch": 0.1875492513790386,
"grad_norm": 0.4611856937408447,
"learning_rate": 3.716535433070866e-06,
"log_odds_chosen": 0.30054840445518494,
"log_odds_ratio": -0.5679630041122437,
"logits/chosen": -0.13648174703121185,
"logits/rejected": -0.2771826982498169,
"logps/chosen": -1.8114385604858398,
"logps/rejected": -2.073773145675659,
"loss": 2.0738,
"nll_loss": 2.016970157623291,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1811438798904419,
"rewards/margins": 0.02623344026505947,
"rewards/rejected": -0.20737731456756592,
"step": 119
},
{
"epoch": 0.18912529550827423,
"grad_norm": 0.4521031677722931,
"learning_rate": 3.748031496062992e-06,
"log_odds_chosen": 0.6006217002868652,
"log_odds_ratio": -0.44487234950065613,
"logits/chosen": -0.18391111493110657,
"logits/rejected": -0.23368988931179047,
"logps/chosen": -1.713165283203125,
"logps/rejected": -2.2281906604766846,
"loss": 1.9444,
"nll_loss": 1.8999552726745605,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17131653428077698,
"rewards/margins": 0.05150254815816879,
"rewards/rejected": -0.22281907498836517,
"step": 120
},
{
"epoch": 0.19070133963750985,
"grad_norm": 0.4375765323638916,
"learning_rate": 3.779527559055118e-06,
"log_odds_chosen": 0.607134222984314,
"log_odds_ratio": -0.4383259415626526,
"logits/chosen": -0.2432423233985901,
"logits/rejected": -0.42568087577819824,
"logps/chosen": -1.8271077871322632,
"logps/rejected": -2.3578097820281982,
"loss": 2.0471,
"nll_loss": 2.0032753944396973,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18271078169345856,
"rewards/margins": 0.05307020992040634,
"rewards/rejected": -0.2357809841632843,
"step": 121
},
{
"epoch": 0.19227738376674547,
"grad_norm": 0.4764109253883362,
"learning_rate": 3.8110236220472436e-06,
"log_odds_chosen": 0.5355339050292969,
"log_odds_ratio": -0.47745591402053833,
"logits/chosen": -0.30387943983078003,
"logits/rejected": -0.23297454416751862,
"logps/chosen": -1.7462990283966064,
"logps/rejected": -2.2124016284942627,
"loss": 1.9695,
"nll_loss": 1.9217469692230225,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17462992668151855,
"rewards/margins": 0.04661024361848831,
"rewards/rejected": -0.22124016284942627,
"step": 122
},
{
"epoch": 0.1938534278959811,
"grad_norm": 0.4778152108192444,
"learning_rate": 3.8425196850393695e-06,
"log_odds_chosen": 0.39439237117767334,
"log_odds_ratio": -0.5168735384941101,
"logits/chosen": -0.06380043923854828,
"logits/rejected": -0.3037663400173187,
"logps/chosen": -1.8502298593521118,
"logps/rejected": -2.1919362545013428,
"loss": 2.0886,
"nll_loss": 2.036864757537842,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1850229799747467,
"rewards/margins": 0.03417064994573593,
"rewards/rejected": -0.21919363737106323,
"step": 123
},
{
"epoch": 0.1954294720252167,
"grad_norm": 0.42974886298179626,
"learning_rate": 3.874015748031496e-06,
"log_odds_chosen": 0.4634344279766083,
"log_odds_ratio": -0.5001177787780762,
"logits/chosen": -0.25694918632507324,
"logits/rejected": -0.22004857659339905,
"logps/chosen": -1.7307007312774658,
"logps/rejected": -2.1188673973083496,
"loss": 1.9529,
"nll_loss": 1.9028429985046387,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17307007312774658,
"rewards/margins": 0.03881664574146271,
"rewards/rejected": -0.2118867039680481,
"step": 124
},
{
"epoch": 0.19700551615445233,
"grad_norm": 0.4418664276599884,
"learning_rate": 3.905511811023622e-06,
"log_odds_chosen": 0.4357145130634308,
"log_odds_ratio": -0.505855143070221,
"logits/chosen": -0.15186086297035217,
"logits/rejected": -0.4162241220474243,
"logps/chosen": -1.7670007944107056,
"logps/rejected": -2.140122890472412,
"loss": 2.0107,
"nll_loss": 1.9601045846939087,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17670010030269623,
"rewards/margins": 0.03731219470500946,
"rewards/rejected": -0.2140122801065445,
"step": 125
},
{
"epoch": 0.19858156028368795,
"grad_norm": 0.47172582149505615,
"learning_rate": 3.937007874015748e-06,
"log_odds_chosen": 0.6300610899925232,
"log_odds_ratio": -0.4334939122200012,
"logits/chosen": -0.1414909064769745,
"logits/rejected": -0.23909084498882294,
"logps/chosen": -1.773337483406067,
"logps/rejected": -2.3242154121398926,
"loss": 1.998,
"nll_loss": 1.9546327590942383,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1773337423801422,
"rewards/margins": 0.05508778989315033,
"rewards/rejected": -0.23242153227329254,
"step": 126
},
{
"epoch": 0.20015760441292357,
"grad_norm": 0.4991217255592346,
"learning_rate": 3.9685039370078736e-06,
"log_odds_chosen": 0.4967314600944519,
"log_odds_ratio": -0.4841119050979614,
"logits/chosen": -0.11699728667736053,
"logits/rejected": -0.20840948820114136,
"logps/chosen": -1.8117221593856812,
"logps/rejected": -2.245494842529297,
"loss": 2.0711,
"nll_loss": 2.0227043628692627,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1811722218990326,
"rewards/margins": 0.04337725043296814,
"rewards/rejected": -0.22454948723316193,
"step": 127
},
{
"epoch": 0.2017336485421592,
"grad_norm": 0.45142993330955505,
"learning_rate": 4e-06,
"log_odds_chosen": 0.3186233639717102,
"log_odds_ratio": -0.5520759224891663,
"logits/chosen": -0.007777207065373659,
"logits/rejected": -0.0923052504658699,
"logps/chosen": -1.8868976831436157,
"logps/rejected": -2.1636693477630615,
"loss": 2.1041,
"nll_loss": 2.0488626956939697,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.18868978321552277,
"rewards/margins": 0.02767716720700264,
"rewards/rejected": -0.2163669317960739,
"step": 128
},
{
"epoch": 0.2033096926713948,
"grad_norm": 0.4361562430858612,
"learning_rate": 3.999992445477635e-06,
"log_odds_chosen": 0.45227304100990295,
"log_odds_ratio": -0.5005910396575928,
"logits/chosen": -0.04890897497534752,
"logits/rejected": -0.3455308675765991,
"logps/chosen": -1.8800231218338013,
"logps/rejected": -2.2768943309783936,
"loss": 2.0833,
"nll_loss": 2.0331974029541016,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1880023181438446,
"rewards/margins": 0.03968711942434311,
"rewards/rejected": -0.2276894450187683,
"step": 129
},
{
"epoch": 0.2048857368006304,
"grad_norm": 0.43126383423805237,
"learning_rate": 3.999969781967615e-06,
"log_odds_chosen": 0.27507030963897705,
"log_odds_ratio": -0.5684081315994263,
"logits/chosen": -0.03611285984516144,
"logits/rejected": -0.18150243163108826,
"logps/chosen": -1.980247139930725,
"logps/rejected": -2.222841501235962,
"loss": 2.1813,
"nll_loss": 2.1244537830352783,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.198024719953537,
"rewards/margins": 0.024259435012936592,
"rewards/rejected": -0.22228413820266724,
"step": 130
},
{
"epoch": 0.20646178092986603,
"grad_norm": 0.39808744192123413,
"learning_rate": 3.99993200964115e-06,
"log_odds_chosen": 0.37286853790283203,
"log_odds_ratio": -0.5256955623626709,
"logits/chosen": -0.13878293335437775,
"logits/rejected": -0.38629403710365295,
"logps/chosen": -1.8050813674926758,
"logps/rejected": -2.1250405311584473,
"loss": 2.0229,
"nll_loss": 1.9703779220581055,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18050813674926758,
"rewards/margins": 0.031995922327041626,
"rewards/rejected": -0.2125040590763092,
"step": 131
},
{
"epoch": 0.20803782505910165,
"grad_norm": 0.49582305550575256,
"learning_rate": 3.99987912878359e-06,
"log_odds_chosen": 0.1634901463985443,
"log_odds_ratio": -0.6200548410415649,
"logits/chosen": -0.09574344009160995,
"logits/rejected": -0.01687694527208805,
"logps/chosen": -1.7705621719360352,
"logps/rejected": -1.9035704135894775,
"loss": 2.0395,
"nll_loss": 1.977489709854126,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1770562380552292,
"rewards/margins": 0.013300813734531403,
"rewards/rejected": -0.19035704433918,
"step": 132
},
{
"epoch": 0.20961386918833727,
"grad_norm": 0.42744600772857666,
"learning_rate": 3.999811139794429e-06,
"log_odds_chosen": 0.3788083493709564,
"log_odds_ratio": -0.5247887372970581,
"logits/chosen": -0.06136152893304825,
"logits/rejected": -0.26565802097320557,
"logps/chosen": -1.8101221323013306,
"logps/rejected": -2.135667324066162,
"loss": 2.0189,
"nll_loss": 1.9664689302444458,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18101221323013306,
"rewards/margins": 0.032554514706134796,
"rewards/rejected": -0.21356670558452606,
"step": 133
},
{
"epoch": 0.2111899133175729,
"grad_norm": 0.4283032715320587,
"learning_rate": 3.999728043187288e-06,
"log_odds_chosen": 0.35587507486343384,
"log_odds_ratio": -0.5397139191627502,
"logits/chosen": -0.04053102061152458,
"logits/rejected": -0.2791404724121094,
"logps/chosen": -1.8541502952575684,
"logps/rejected": -2.164419412612915,
"loss": 2.0887,
"nll_loss": 2.0347094535827637,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18541501462459564,
"rewards/margins": 0.03102692775428295,
"rewards/rejected": -0.21644194424152374,
"step": 134
},
{
"epoch": 0.2127659574468085,
"grad_norm": 0.42146554589271545,
"learning_rate": 3.999629839589922e-06,
"log_odds_chosen": 0.22164756059646606,
"log_odds_ratio": -0.5917123556137085,
"logits/chosen": 0.012096976861357689,
"logits/rejected": -0.4592039883136749,
"logps/chosen": -1.9410529136657715,
"logps/rejected": -2.1342885494232178,
"loss": 2.1714,
"nll_loss": 2.1121792793273926,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19410529732704163,
"rewards/margins": 0.01932355761528015,
"rewards/rejected": -0.21342885494232178,
"step": 135
},
{
"epoch": 0.21434200157604413,
"grad_norm": 0.4338121712207794,
"learning_rate": 3.999516529744215e-06,
"log_odds_chosen": 0.3292469382286072,
"log_odds_ratio": -0.5459690690040588,
"logits/chosen": 0.07608616352081299,
"logits/rejected": -0.40737682580947876,
"logps/chosen": -1.8194847106933594,
"logps/rejected": -2.10274600982666,
"loss": 2.0403,
"nll_loss": 1.9856888055801392,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1819484829902649,
"rewards/margins": 0.02832612209022045,
"rewards/rejected": -0.2102746069431305,
"step": 136
},
{
"epoch": 0.21591804570527975,
"grad_norm": 0.3987630605697632,
"learning_rate": 3.999388114506166e-06,
"log_odds_chosen": 0.4010235667228699,
"log_odds_ratio": -0.518801212310791,
"logits/chosen": -0.046873223036527634,
"logits/rejected": -0.4238424599170685,
"logps/chosen": -1.7170600891113281,
"logps/rejected": -2.0600481033325195,
"loss": 1.931,
"nll_loss": 1.8791390657424927,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17170599102973938,
"rewards/margins": 0.034298814833164215,
"rewards/rejected": -0.206004798412323,
"step": 137
},
{
"epoch": 0.21749408983451538,
"grad_norm": 0.4006027579307556,
"learning_rate": 3.999244594845892e-06,
"log_odds_chosen": 0.25786924362182617,
"log_odds_ratio": -0.5790534019470215,
"logits/chosen": -0.023555610328912735,
"logits/rejected": -0.35695680975914,
"logps/chosen": -1.7636680603027344,
"logps/rejected": -1.985347032546997,
"loss": 1.9912,
"nll_loss": 1.9332479238510132,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17636680603027344,
"rewards/margins": 0.02216789685189724,
"rewards/rejected": -0.19853469729423523,
"step": 138
},
{
"epoch": 0.219070133963751,
"grad_norm": 0.4049820303916931,
"learning_rate": 3.999085971847616e-06,
"log_odds_chosen": 0.24562738835811615,
"log_odds_ratio": -0.58580482006073,
"logits/chosen": -0.012129198759794235,
"logits/rejected": -0.512136697769165,
"logps/chosen": -1.8218495845794678,
"logps/rejected": -2.0348801612854004,
"loss": 2.0389,
"nll_loss": 1.980314016342163,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18218494951725006,
"rewards/margins": 0.02130305953323841,
"rewards/rejected": -0.20348800718784332,
"step": 139
},
{
"epoch": 0.22064617809298662,
"grad_norm": 0.3915075361728668,
"learning_rate": 3.998912246709658e-06,
"log_odds_chosen": 0.3664124310016632,
"log_odds_ratio": -0.5369927287101746,
"logits/chosen": -0.08141148090362549,
"logits/rejected": -0.4445778429508209,
"logps/chosen": -1.7210732698440552,
"logps/rejected": -2.033111572265625,
"loss": 1.9395,
"nll_loss": 1.885791301727295,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17210730910301208,
"rewards/margins": 0.0312038566917181,
"rewards/rejected": -0.20331117510795593,
"step": 140
},
{
"epoch": 0.2222222222222222,
"grad_norm": 0.42792966961860657,
"learning_rate": 3.9987234207444295e-06,
"log_odds_chosen": 0.3271371126174927,
"log_odds_ratio": -0.5472476482391357,
"logits/chosen": -0.05070888251066208,
"logits/rejected": -0.30135494470596313,
"logps/chosen": -1.817905068397522,
"logps/rejected": -2.099316358566284,
"loss": 2.0393,
"nll_loss": 1.984588384628296,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1817905157804489,
"rewards/margins": 0.028141150251030922,
"rewards/rejected": -0.2099316567182541,
"step": 141
},
{
"epoch": 0.22379826635145783,
"grad_norm": 0.40370792150497437,
"learning_rate": 3.998519495378419e-06,
"log_odds_chosen": 0.34052586555480957,
"log_odds_ratio": -0.5426381826400757,
"logits/chosen": 0.0529387891292572,
"logits/rejected": -0.22841407358646393,
"logps/chosen": -1.896314024925232,
"logps/rejected": -2.194129705429077,
"loss": 2.0889,
"nll_loss": 2.0345916748046875,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.189631387591362,
"rewards/margins": 0.029781583696603775,
"rewards/rejected": -0.21941299736499786,
"step": 142
},
{
"epoch": 0.22537431048069345,
"grad_norm": 0.36527448892593384,
"learning_rate": 3.998300472152187e-06,
"log_odds_chosen": 0.3244752287864685,
"log_odds_ratio": -0.5529875755310059,
"logits/chosen": -0.07975448668003082,
"logits/rejected": -0.5476751923561096,
"logps/chosen": -1.6814563274383545,
"logps/rejected": -1.9556654691696167,
"loss": 1.9047,
"nll_loss": 1.8494292497634888,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16814564168453217,
"rewards/margins": 0.02742091566324234,
"rewards/rejected": -0.1955665647983551,
"step": 143
},
{
"epoch": 0.22695035460992907,
"grad_norm": 0.39946654438972473,
"learning_rate": 3.998066352720347e-06,
"log_odds_chosen": 0.4114699959754944,
"log_odds_ratio": -0.5125808715820312,
"logits/chosen": 0.10884220898151398,
"logits/rejected": -0.2769298553466797,
"logps/chosen": -1.807681679725647,
"logps/rejected": -2.163328170776367,
"loss": 2.0156,
"nll_loss": 1.9643856287002563,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1807681769132614,
"rewards/margins": 0.035564638674259186,
"rewards/rejected": -0.2163328230381012,
"step": 144
},
{
"epoch": 0.2285263987391647,
"grad_norm": 0.3883441388607025,
"learning_rate": 3.997817138851562e-06,
"log_odds_chosen": 0.3042789101600647,
"log_odds_ratio": -0.5641068816184998,
"logits/chosen": 0.047148481011390686,
"logits/rejected": -0.2664666771888733,
"logps/chosen": -1.894091248512268,
"logps/rejected": -2.1613922119140625,
"loss": 2.1037,
"nll_loss": 2.0473220348358154,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18940910696983337,
"rewards/margins": 0.0267301294952631,
"rewards/rejected": -0.21613925695419312,
"step": 145
},
{
"epoch": 0.23010244286840031,
"grad_norm": 0.3664163053035736,
"learning_rate": 3.997552832428522e-06,
"log_odds_chosen": 0.40218284726142883,
"log_odds_ratio": -0.5150843262672424,
"logits/chosen": -0.06997128576040268,
"logits/rejected": -0.30114221572875977,
"logps/chosen": -1.683500051498413,
"logps/rejected": -2.0218000411987305,
"loss": 1.9221,
"nll_loss": 1.8706339597702026,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1683500111103058,
"rewards/margins": 0.03382997214794159,
"rewards/rejected": -0.20217998325824738,
"step": 146
},
{
"epoch": 0.23167848699763594,
"grad_norm": 0.35738253593444824,
"learning_rate": 3.9972734354479366e-06,
"log_odds_chosen": 0.33341336250305176,
"log_odds_ratio": -0.5498776435852051,
"logits/chosen": 0.05102141201496124,
"logits/rejected": -0.588348925113678,
"logps/chosen": -1.6781635284423828,
"logps/rejected": -1.957244634628296,
"loss": 1.9026,
"nll_loss": 1.84757661819458,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1678163707256317,
"rewards/margins": 0.027908099815249443,
"rewards/rejected": -0.1957244724035263,
"step": 147
},
{
"epoch": 0.23325453112687156,
"grad_norm": 0.40279892086982727,
"learning_rate": 3.996978950020517e-06,
"log_odds_chosen": 0.21276284754276276,
"log_odds_ratio": -0.6048003435134888,
"logits/chosen": 0.1419999599456787,
"logits/rejected": -0.28115570545196533,
"logps/chosen": -1.8639137744903564,
"logps/rejected": -2.0437545776367188,
"loss": 2.0862,
"nll_loss": 2.0256824493408203,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.18639138340950012,
"rewards/margins": 0.01798408292233944,
"rewards/rejected": -0.2043754756450653,
"step": 148
},
{
"epoch": 0.23483057525610718,
"grad_norm": 0.35266202688217163,
"learning_rate": 3.996669378370959e-06,
"log_odds_chosen": 0.5316495299339294,
"log_odds_ratio": -0.46872708201408386,
"logits/chosen": 0.010169305838644505,
"logits/rejected": -0.3486666679382324,
"logps/chosen": -1.6488699913024902,
"logps/rejected": -2.1011159420013428,
"loss": 1.8663,
"nll_loss": 1.8194066286087036,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16488701105117798,
"rewards/margins": 0.045224592089653015,
"rewards/rejected": -0.2101115882396698,
"step": 149
},
{
"epoch": 0.2364066193853428,
"grad_norm": 0.39699894189834595,
"learning_rate": 3.996344722837929e-06,
"log_odds_chosen": 0.49894896149635315,
"log_odds_ratio": -0.4788591265678406,
"logits/chosen": -0.0008517892565578222,
"logits/rejected": -0.31570884585380554,
"logps/chosen": -1.8067049980163574,
"logps/rejected": -2.239896535873413,
"loss": 2.0051,
"nll_loss": 1.95723295211792,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18067049980163574,
"rewards/margins": 0.043319158256053925,
"rewards/rejected": -0.22398965060710907,
"step": 150
},
{
"epoch": 0.23798266351457842,
"grad_norm": 0.33054688572883606,
"learning_rate": 3.996004985874043e-06,
"log_odds_chosen": 0.2428944855928421,
"log_odds_ratio": -0.5818551778793335,
"logits/chosen": 0.11997652053833008,
"logits/rejected": -0.2498077005147934,
"logps/chosen": -1.698814868927002,
"logps/rejected": -1.9012809991836548,
"loss": 1.9266,
"nll_loss": 1.8684155941009521,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16988149285316467,
"rewards/margins": 0.020246637985110283,
"rewards/rejected": -0.19012810289859772,
"step": 151
},
{
"epoch": 0.23955870764381404,
"grad_norm": 0.3768727481365204,
"learning_rate": 3.995650170045855e-06,
"log_odds_chosen": 0.28013092279434204,
"log_odds_ratio": -0.5691770315170288,
"logits/chosen": 0.045284271240234375,
"logits/rejected": -0.3556942641735077,
"logps/chosen": -1.7669368982315063,
"logps/rejected": -2.0047860145568848,
"loss": 1.9804,
"nll_loss": 1.9234654903411865,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17669367790222168,
"rewards/margins": 0.023784920573234558,
"rewards/rejected": -0.20047861337661743,
"step": 152
},
{
"epoch": 0.24113475177304963,
"grad_norm": 0.3691398501396179,
"learning_rate": 3.995280278033825e-06,
"log_odds_chosen": 0.20656853914260864,
"log_odds_ratio": -0.6042188405990601,
"logits/chosen": 0.11217048764228821,
"logits/rejected": -0.24175406992435455,
"logps/chosen": -1.8632197380065918,
"logps/rejected": -2.039140462875366,
"loss": 2.0785,
"nll_loss": 2.0180578231811523,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.18632197380065918,
"rewards/margins": 0.017592042684555054,
"rewards/rejected": -0.20391403138637543,
"step": 153
},
{
"epoch": 0.24271079590228525,
"grad_norm": 0.35842734575271606,
"learning_rate": 3.994895312632314e-06,
"log_odds_chosen": 0.4981134533882141,
"log_odds_ratio": -0.4802461564540863,
"logits/chosen": 0.16929033398628235,
"logits/rejected": -0.7549564838409424,
"logps/chosen": -1.7397829294204712,
"logps/rejected": -2.1666393280029297,
"loss": 1.9514,
"nll_loss": 1.903334140777588,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1739783138036728,
"rewards/margins": 0.04268564283847809,
"rewards/rejected": -0.21666395664215088,
"step": 154
},
{
"epoch": 0.24428684003152087,
"grad_norm": 0.33668240904808044,
"learning_rate": 3.994495276749549e-06,
"log_odds_chosen": 0.47896263003349304,
"log_odds_ratio": -0.48850810527801514,
"logits/chosen": 0.07264027744531631,
"logits/rejected": -0.33893775939941406,
"logps/chosen": -1.7592413425445557,
"logps/rejected": -2.172550916671753,
"loss": 1.9568,
"nll_loss": 1.9079933166503906,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.175924152135849,
"rewards/margins": 0.04133095592260361,
"rewards/rejected": -0.217255100607872,
"step": 155
},
{
"epoch": 0.2458628841607565,
"grad_norm": 0.34656822681427,
"learning_rate": 3.994080173407612e-06,
"log_odds_chosen": 0.23366691172122955,
"log_odds_ratio": -0.5879085063934326,
"logits/chosen": 0.1407233029603958,
"logits/rejected": -0.3865904211997986,
"logps/chosen": -1.7929620742797852,
"logps/rejected": -1.9902442693710327,
"loss": 1.9994,
"nll_loss": 1.9406037330627441,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.17929621040821075,
"rewards/margins": 0.01972820609807968,
"rewards/rejected": -0.19902442395687103,
"step": 156
},
{
"epoch": 0.24743892828999212,
"grad_norm": 0.34390729665756226,
"learning_rate": 3.993650005742409e-06,
"log_odds_chosen": 0.3310392498970032,
"log_odds_ratio": -0.546916127204895,
"logits/chosen": 0.06716110557317734,
"logits/rejected": -0.23844899237155914,
"logps/chosen": -1.7643970251083374,
"logps/rejected": -2.046977996826172,
"loss": 1.9724,
"nll_loss": 1.9176855087280273,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17643971741199493,
"rewards/margins": 0.028258096426725388,
"rewards/rejected": -0.20469780266284943,
"step": 157
},
{
"epoch": 0.24901497241922774,
"grad_norm": 0.3150049149990082,
"learning_rate": 3.993204777003652e-06,
"log_odds_chosen": 0.337340384721756,
"log_odds_ratio": -0.5445095896720886,
"logits/chosen": 0.07445216178894043,
"logits/rejected": -0.441582053899765,
"logps/chosen": -1.6941965818405151,
"logps/rejected": -1.980376124382019,
"loss": 1.898,
"nll_loss": 1.843545913696289,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16941964626312256,
"rewards/margins": 0.02861795574426651,
"rewards/rejected": -0.19803762435913086,
"step": 158
},
{
"epoch": 0.25059101654846333,
"grad_norm": 0.3337211012840271,
"learning_rate": 3.992744490554832e-06,
"log_odds_chosen": 0.33180904388427734,
"log_odds_ratio": -0.5461182594299316,
"logits/chosen": 0.21238459646701813,
"logits/rejected": -0.269389808177948,
"logps/chosen": -1.7364294528961182,
"logps/rejected": -2.0181472301483154,
"loss": 1.9565,
"nll_loss": 1.9018971920013428,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17364296317100525,
"rewards/margins": 0.028171781450510025,
"rewards/rejected": -0.20181472599506378,
"step": 159
},
{
"epoch": 0.25216706067769895,
"grad_norm": 0.31171178817749023,
"learning_rate": 3.992269149873192e-06,
"log_odds_chosen": 0.32716354727745056,
"log_odds_ratio": -0.547857403755188,
"logits/chosen": 0.028445789590477943,
"logits/rejected": -0.35958513617515564,
"logps/chosen": -1.7089133262634277,
"logps/rejected": -1.9816714525222778,
"loss": 1.9143,
"nll_loss": 1.8594882488250732,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17089134454727173,
"rewards/margins": 0.02727578952908516,
"rewards/rejected": -0.19816714525222778,
"step": 160
},
{
"epoch": 0.25374310480693457,
"grad_norm": 0.30663299560546875,
"learning_rate": 3.991778758549705e-06,
"log_odds_chosen": 0.195327490568161,
"log_odds_ratio": -0.6073517799377441,
"logits/chosen": 0.2354024052619934,
"logits/rejected": -0.21214549243450165,
"logps/chosen": -1.8341491222381592,
"logps/rejected": -2.0042359828948975,
"loss": 2.023,
"nll_loss": 1.9623081684112549,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.18341490626335144,
"rewards/margins": 0.017008693888783455,
"rewards/rejected": -0.20042361319065094,
"step": 161
},
{
"epoch": 0.2553191489361702,
"grad_norm": 0.31510376930236816,
"learning_rate": 3.9912733202890415e-06,
"log_odds_chosen": 0.32201409339904785,
"log_odds_ratio": -0.548933744430542,
"logits/chosen": 0.10407942533493042,
"logits/rejected": -0.4005528688430786,
"logps/chosen": -1.7570445537567139,
"logps/rejected": -2.0296471118927,
"loss": 1.96,
"nll_loss": 1.905151128768921,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1757044792175293,
"rewards/margins": 0.027260230854153633,
"rewards/rejected": -0.20296470820903778,
"step": 162
},
{
"epoch": 0.2568951930654058,
"grad_norm": 0.3028022050857544,
"learning_rate": 3.990752838909548e-06,
"log_odds_chosen": 0.1706855297088623,
"log_odds_ratio": -0.6158415079116821,
"logits/chosen": 0.1903667449951172,
"logits/rejected": -0.2845366299152374,
"logps/chosen": -1.8539068698883057,
"logps/rejected": -1.9989856481552124,
"loss": 2.0475,
"nll_loss": 1.9859018325805664,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.18539069592952728,
"rewards/margins": 0.014507867395877838,
"rewards/rejected": -0.19989855587482452,
"step": 163
},
{
"epoch": 0.25847123719464143,
"grad_norm": 0.3319096267223358,
"learning_rate": 3.990217318343213e-06,
"log_odds_chosen": 0.5035750269889832,
"log_odds_ratio": -0.4748569130897522,
"logits/chosen": 0.15394604206085205,
"logits/rejected": -0.5184996128082275,
"logps/chosen": -1.921760082244873,
"logps/rejected": -2.365640163421631,
"loss": 2.1041,
"nll_loss": 2.0565714836120605,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19217601418495178,
"rewards/margins": 0.044387996196746826,
"rewards/rejected": -0.2365640103816986,
"step": 164
},
{
"epoch": 0.26004728132387706,
"grad_norm": 0.3104284405708313,
"learning_rate": 3.989666762635637e-06,
"log_odds_chosen": 0.2950212061405182,
"log_odds_ratio": -0.5599942803382874,
"logits/chosen": 0.1402396708726883,
"logits/rejected": -0.27953556180000305,
"logps/chosen": -1.767073631286621,
"logps/rejected": -2.017169713973999,
"loss": 1.9826,
"nll_loss": 1.926632285118103,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17670737206935883,
"rewards/margins": 0.025009607896208763,
"rewards/rejected": -0.20171695947647095,
"step": 165
},
{
"epoch": 0.2616233254531127,
"grad_norm": 0.30276185274124146,
"learning_rate": 3.9891011759460056e-06,
"log_odds_chosen": 0.3634149432182312,
"log_odds_ratio": -0.5344185829162598,
"logits/chosen": 0.1359172910451889,
"logits/rejected": -0.3355481028556824,
"logps/chosen": -1.75217866897583,
"logps/rejected": -2.055960178375244,
"loss": 1.958,
"nll_loss": 1.9045321941375732,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1752178817987442,
"rewards/margins": 0.03037814423441887,
"rewards/rejected": -0.20559601485729218,
"step": 166
},
{
"epoch": 0.2631993695823483,
"grad_norm": 0.2769591212272644,
"learning_rate": 3.988520562547057e-06,
"log_odds_chosen": 0.3933558762073517,
"log_odds_ratio": -0.5260743498802185,
"logits/chosen": 0.15155696868896484,
"logits/rejected": -0.42779073119163513,
"logps/chosen": -1.7146203517913818,
"logps/rejected": -2.04492449760437,
"loss": 1.9032,
"nll_loss": 1.8506273031234741,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1714620441198349,
"rewards/margins": 0.033030424267053604,
"rewards/rejected": -0.2044924646615982,
"step": 167
},
{
"epoch": 0.2647754137115839,
"grad_norm": 0.29901036620140076,
"learning_rate": 3.987924926825047e-06,
"log_odds_chosen": 0.40397408604621887,
"log_odds_ratio": -0.5212303996086121,
"logits/chosen": 0.12876684963703156,
"logits/rejected": -0.32701608538627625,
"logps/chosen": -1.790124773979187,
"logps/rejected": -2.1377172470092773,
"loss": 2.0015,
"nll_loss": 1.9493852853775024,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1790124773979187,
"rewards/margins": 0.03475925698876381,
"rewards/rejected": -0.2137717306613922,
"step": 168
},
{
"epoch": 0.26635145784081954,
"grad_norm": 0.2879612147808075,
"learning_rate": 3.98731427327972e-06,
"log_odds_chosen": 0.38767939805984497,
"log_odds_ratio": -0.5189070105552673,
"logits/chosen": 0.09615111351013184,
"logits/rejected": -0.552105724811554,
"logps/chosen": -1.7201054096221924,
"logps/rejected": -2.047496795654297,
"loss": 1.9089,
"nll_loss": 1.8569788932800293,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17201054096221924,
"rewards/margins": 0.03273913264274597,
"rewards/rejected": -0.2047496885061264,
"step": 169
},
{
"epoch": 0.26792750197005516,
"grad_norm": 0.2887536585330963,
"learning_rate": 3.986688606524273e-06,
"log_odds_chosen": 0.32007887959480286,
"log_odds_ratio": -0.5568875670433044,
"logits/chosen": 0.20907297730445862,
"logits/rejected": -0.717089831829071,
"logps/chosen": -1.7479276657104492,
"logps/rejected": -2.0232884883880615,
"loss": 1.9428,
"nll_loss": 1.8871328830718994,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.17479278147220612,
"rewards/margins": 0.02753606252372265,
"rewards/rejected": -0.20232883095741272,
"step": 170
},
{
"epoch": 0.2695035460992908,
"grad_norm": 0.2829509675502777,
"learning_rate": 3.986047931285315e-06,
"log_odds_chosen": 0.493367999792099,
"log_odds_ratio": -0.4808065891265869,
"logits/chosen": 0.08703712373971939,
"logits/rejected": -0.6400361657142639,
"logps/chosen": -1.6581294536590576,
"logps/rejected": -2.072545051574707,
"loss": 1.848,
"nll_loss": 1.7999612092971802,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16581295430660248,
"rewards/margins": 0.041441574692726135,
"rewards/rejected": -0.20725451409816742,
"step": 171
},
{
"epoch": 0.2710795902285264,
"grad_norm": 0.27343836426734924,
"learning_rate": 3.985392252402847e-06,
"log_odds_chosen": 0.23880890011787415,
"log_odds_ratio": -0.5835365653038025,
"logits/chosen": 0.08178934454917908,
"logits/rejected": -0.431425541639328,
"logps/chosen": -1.655697226524353,
"logps/rejected": -1.8529317378997803,
"loss": 1.8557,
"nll_loss": 1.7973511219024658,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1655697375535965,
"rewards/margins": 0.0197234395891428,
"rewards/rejected": -0.18529316782951355,
"step": 172
},
{
"epoch": 0.272655634357762,
"grad_norm": 0.2862318754196167,
"learning_rate": 3.984721574830206e-06,
"log_odds_chosen": 0.27102503180503845,
"log_odds_ratio": -0.5708762407302856,
"logits/chosen": 0.14478133618831635,
"logits/rejected": -0.510442316532135,
"logps/chosen": -1.7600399255752563,
"logps/rejected": -1.9898476600646973,
"loss": 1.952,
"nll_loss": 1.894890308380127,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17600399255752563,
"rewards/margins": 0.02298077940940857,
"rewards/rejected": -0.1989847719669342,
"step": 173
},
{
"epoch": 0.27423167848699764,
"grad_norm": 0.26644009351730347,
"learning_rate": 3.984035903634041e-06,
"log_odds_chosen": 0.2790209650993347,
"log_odds_ratio": -0.5701457262039185,
"logits/chosen": 0.1971094161272049,
"logits/rejected": -0.5762545466423035,
"logps/chosen": -1.7363693714141846,
"logps/rejected": -1.9710581302642822,
"loss": 1.914,
"nll_loss": 1.8569939136505127,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.17363695800304413,
"rewards/margins": 0.023468857631087303,
"rewards/rejected": -0.19710581004619598,
"step": 174
},
{
"epoch": 0.27580772261623326,
"grad_norm": 0.2771873474121094,
"learning_rate": 3.983335243994273e-06,
"log_odds_chosen": 0.37392759323120117,
"log_odds_ratio": -0.5302526354789734,
"logits/chosen": 0.21129730343818665,
"logits/rejected": -0.5397756099700928,
"logps/chosen": -1.726618766784668,
"logps/rejected": -2.042837381362915,
"loss": 1.9182,
"nll_loss": 1.8651602268218994,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17266185581684113,
"rewards/margins": 0.03162187710404396,
"rewards/rejected": -0.20428375899791718,
"step": 175
},
{
"epoch": 0.2773837667454689,
"grad_norm": 0.2980581223964691,
"learning_rate": 3.982619601204049e-06,
"log_odds_chosen": 0.2845771908760071,
"log_odds_ratio": -0.565645158290863,
"logits/chosen": 0.1371021270751953,
"logits/rejected": -0.544967770576477,
"logps/chosen": -1.784250259399414,
"logps/rejected": -2.0266098976135254,
"loss": 1.9752,
"nll_loss": 1.9185881614685059,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17842502892017365,
"rewards/margins": 0.02423596940934658,
"rewards/rejected": -0.20266100764274597,
"step": 176
},
{
"epoch": 0.2789598108747045,
"grad_norm": 0.28183117508888245,
"learning_rate": 3.9818889806697085e-06,
"log_odds_chosen": 0.37239906191825867,
"log_odds_ratio": -0.5329021215438843,
"logits/chosen": 0.24540948867797852,
"logits/rejected": -0.537138044834137,
"logps/chosen": -1.824477195739746,
"logps/rejected": -2.1465721130371094,
"loss": 1.9904,
"nll_loss": 1.9371103048324585,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18244771659374237,
"rewards/margins": 0.032209500670433044,
"rewards/rejected": -0.21465720236301422,
"step": 177
},
{
"epoch": 0.2805358550039401,
"grad_norm": 0.2755196988582611,
"learning_rate": 3.98114338791074e-06,
"log_odds_chosen": 0.4614989757537842,
"log_odds_ratio": -0.5006829500198364,
"logits/chosen": 0.11655398458242416,
"logits/rejected": -0.4944975674152374,
"logps/chosen": -1.6953446865081787,
"logps/rejected": -2.088918685913086,
"loss": 1.89,
"nll_loss": 1.8398841619491577,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16953447461128235,
"rewards/margins": 0.03935740143060684,
"rewards/rejected": -0.2088918834924698,
"step": 178
},
{
"epoch": 0.28211189913317575,
"grad_norm": 0.262363463640213,
"learning_rate": 3.980382828559742e-06,
"log_odds_chosen": 0.37132197618484497,
"log_odds_ratio": -0.53047114610672,
"logits/chosen": 0.1178860068321228,
"logits/rejected": -0.5356312990188599,
"logps/chosen": -1.7367223501205444,
"logps/rejected": -2.051452159881592,
"loss": 1.9086,
"nll_loss": 1.8555084466934204,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17367224395275116,
"rewards/margins": 0.03147297352552414,
"rewards/rejected": -0.2051452100276947,
"step": 179
},
{
"epoch": 0.28368794326241137,
"grad_norm": 0.2845723032951355,
"learning_rate": 3.9796073083623774e-06,
"log_odds_chosen": 0.33220580220222473,
"log_odds_ratio": -0.5529859066009521,
"logits/chosen": 0.13935938477516174,
"logits/rejected": -0.5945897698402405,
"logps/chosen": -1.805295705795288,
"logps/rejected": -2.08857798576355,
"loss": 1.9753,
"nll_loss": 1.9200148582458496,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.18052956461906433,
"rewards/margins": 0.028328238055109978,
"rewards/rejected": -0.20885780453681946,
"step": 180
},
{
"epoch": 0.285263987391647,
"grad_norm": 0.2820357382297516,
"learning_rate": 3.978816833177329e-06,
"log_odds_chosen": 0.3633921444416046,
"log_odds_ratio": -0.5330725312232971,
"logits/chosen": 0.2247752994298935,
"logits/rejected": -0.42544880509376526,
"logps/chosen": -1.7320586442947388,
"logps/rejected": -2.0406675338745117,
"loss": 1.91,
"nll_loss": 1.8566887378692627,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17320585250854492,
"rewards/margins": 0.03086087293922901,
"rewards/rejected": -0.20406673848628998,
"step": 181
},
{
"epoch": 0.2868400315208826,
"grad_norm": 0.2795039713382721,
"learning_rate": 3.978011408976261e-06,
"log_odds_chosen": 0.2320139855146408,
"log_odds_ratio": -0.5891825556755066,
"logits/chosen": 0.19194500148296356,
"logits/rejected": -0.5741883516311646,
"logps/chosen": -1.750451683998108,
"logps/rejected": -1.9439057111740112,
"loss": 1.9425,
"nll_loss": 1.8835428953170776,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1750451624393463,
"rewards/margins": 0.019345413893461227,
"rewards/rejected": -0.19439058005809784,
"step": 182
},
{
"epoch": 0.28841607565011823,
"grad_norm": 0.2846316993236542,
"learning_rate": 3.9771910418437674e-06,
"log_odds_chosen": 0.3140004575252533,
"log_odds_ratio": -0.54939204454422,
"logits/chosen": 0.1681434065103531,
"logits/rejected": -0.3001997172832489,
"logps/chosen": -1.7808198928833008,
"logps/rejected": -2.049476385116577,
"loss": 1.956,
"nll_loss": 1.9010967016220093,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17808198928833008,
"rewards/margins": 0.026865659281611443,
"rewards/rejected": -0.20494765043258667,
"step": 183
},
{
"epoch": 0.2899921197793538,
"grad_norm": 0.2878468334674835,
"learning_rate": 3.976355737977332e-06,
"log_odds_chosen": 0.3773011863231659,
"log_odds_ratio": -0.5303803086280823,
"logits/chosen": 0.18860210478305817,
"logits/rejected": -0.5023236274719238,
"logps/chosen": -1.7276164293289185,
"logps/rejected": -2.046980381011963,
"loss": 1.9104,
"nll_loss": 1.8573479652404785,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17276166379451752,
"rewards/margins": 0.03193638473749161,
"rewards/rejected": -0.20469802618026733,
"step": 184
},
{
"epoch": 0.2915681639085894,
"grad_norm": 0.2533093988895416,
"learning_rate": 3.975505503687274e-06,
"log_odds_chosen": 0.3036259412765503,
"log_odds_ratio": -0.5556970238685608,
"logits/chosen": 0.23977245390415192,
"logits/rejected": -0.28354203701019287,
"logps/chosen": -1.7183681726455688,
"logps/rejected": -1.9746772050857544,
"loss": 1.8965,
"nll_loss": 1.8409373760223389,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17183682322502136,
"rewards/margins": 0.025630896911025047,
"rewards/rejected": -0.19746771454811096,
"step": 185
},
{
"epoch": 0.29314420803782504,
"grad_norm": 0.2554759383201599,
"learning_rate": 3.974640345396708e-06,
"log_odds_chosen": 0.458074152469635,
"log_odds_ratio": -0.4967803359031677,
"logits/chosen": 0.09675043821334839,
"logits/rejected": -0.8243634104728699,
"logps/chosen": -1.6910256147384644,
"logps/rejected": -2.077892780303955,
"loss": 1.8667,
"nll_loss": 1.8170145750045776,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16910257935523987,
"rewards/margins": 0.03868672996759415,
"rewards/rejected": -0.20778930187225342,
"step": 186
},
{
"epoch": 0.29472025216706066,
"grad_norm": 0.2935222387313843,
"learning_rate": 3.9737602696414925e-06,
"log_odds_chosen": 0.159349724650383,
"log_odds_ratio": -0.6268662214279175,
"logits/chosen": 0.21252146363258362,
"logits/rejected": -0.6300286650657654,
"logps/chosen": -1.78038489818573,
"logps/rejected": -1.914283037185669,
"loss": 1.9361,
"nll_loss": 1.8734400272369385,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.17803849279880524,
"rewards/margins": 0.013389825820922852,
"rewards/rejected": -0.1914283186197281,
"step": 187
},
{
"epoch": 0.2962962962962963,
"grad_norm": 0.2445816546678543,
"learning_rate": 3.972865283070179e-06,
"log_odds_chosen": 0.38453128933906555,
"log_odds_ratio": -0.5239704847335815,
"logits/chosen": 0.16407378017902374,
"logits/rejected": -0.7210597991943359,
"logps/chosen": -1.6779310703277588,
"logps/rejected": -2.003727436065674,
"loss": 1.8617,
"nll_loss": 1.8092904090881348,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16779311001300812,
"rewards/margins": 0.032579630613327026,
"rewards/rejected": -0.20037274062633514,
"step": 188
},
{
"epoch": 0.2978723404255319,
"grad_norm": 0.2602602243423462,
"learning_rate": 3.971955392443965e-06,
"log_odds_chosen": 0.3589805066585541,
"log_odds_ratio": -0.535193920135498,
"logits/chosen": 0.19359144568443298,
"logits/rejected": -0.5294786691665649,
"logps/chosen": -1.7104607820510864,
"logps/rejected": -2.017254114151001,
"loss": 1.8819,
"nll_loss": 1.8284274339675903,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17104607820510864,
"rewards/margins": 0.03067934513092041,
"rewards/rejected": -0.20172543823719025,
"step": 189
},
{
"epoch": 0.2994483845547675,
"grad_norm": 0.2570762634277344,
"learning_rate": 3.971030604636637e-06,
"log_odds_chosen": 0.20797011256217957,
"log_odds_ratio": -0.5995408892631531,
"logits/chosen": 0.24219730496406555,
"logits/rejected": -0.49577596783638,
"logps/chosen": -1.726780891418457,
"logps/rejected": -1.9002296924591064,
"loss": 1.8985,
"nll_loss": 1.8385004997253418,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17267809808254242,
"rewards/margins": 0.017344871535897255,
"rewards/rejected": -0.19002296030521393,
"step": 190
},
{
"epoch": 0.30102442868400314,
"grad_norm": 0.27055835723876953,
"learning_rate": 3.970090926634526e-06,
"log_odds_chosen": 0.29205840826034546,
"log_odds_ratio": -0.5609118938446045,
"logits/chosen": 0.3021741211414337,
"logits/rejected": -0.7257508635520935,
"logps/chosen": -1.7830076217651367,
"logps/rejected": -2.0307276248931885,
"loss": 1.9615,
"nll_loss": 1.9053698778152466,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17830076813697815,
"rewards/margins": 0.024771984666585922,
"rewards/rejected": -0.20307274162769318,
"step": 191
},
{
"epoch": 0.30260047281323876,
"grad_norm": 0.27686426043510437,
"learning_rate": 3.9691363655364526e-06,
"log_odds_chosen": 0.3535913825035095,
"log_odds_ratio": -0.5396946668624878,
"logits/chosen": 0.19939753413200378,
"logits/rejected": -0.527854859828949,
"logps/chosen": -1.7547531127929688,
"logps/rejected": -2.056049108505249,
"loss": 1.9315,
"nll_loss": 1.8775546550750732,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1754753142595291,
"rewards/margins": 0.030129600316286087,
"rewards/rejected": -0.2056049108505249,
"step": 192
},
{
"epoch": 0.3041765169424744,
"grad_norm": 0.24814869463443756,
"learning_rate": 3.968166928553666e-06,
"log_odds_chosen": 0.29018789529800415,
"log_odds_ratio": -0.5631506443023682,
"logits/chosen": 0.23991963267326355,
"logits/rejected": -0.6883436441421509,
"logps/chosen": -1.708446741104126,
"logps/rejected": -1.9508640766143799,
"loss": 1.9016,
"nll_loss": 1.8453017473220825,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1708446592092514,
"rewards/margins": 0.02424173429608345,
"rewards/rejected": -0.19508640468120575,
"step": 193
},
{
"epoch": 0.30575256107171,
"grad_norm": 0.2546519637107849,
"learning_rate": 3.967182623009804e-06,
"log_odds_chosen": 0.42340725660324097,
"log_odds_ratio": -0.5090901851654053,
"logits/chosen": 0.18889212608337402,
"logits/rejected": -0.5412212014198303,
"logps/chosen": -1.7333446741104126,
"logps/rejected": -2.0966084003448486,
"loss": 1.8946,
"nll_loss": 1.8436520099639893,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17333447933197021,
"rewards/margins": 0.03632635623216629,
"rewards/rejected": -0.2096608430147171,
"step": 194
},
{
"epoch": 0.3073286052009456,
"grad_norm": 0.27097785472869873,
"learning_rate": 3.966183456340821e-06,
"log_odds_chosen": 0.19589565694332123,
"log_odds_ratio": -0.6020736694335938,
"logits/chosen": 0.23710918426513672,
"logits/rejected": -0.5323060154914856,
"logps/chosen": -1.7581830024719238,
"logps/rejected": -1.9214140176773071,
"loss": 1.9344,
"nll_loss": 1.8741968870162964,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1758182942867279,
"rewards/margins": 0.01632309891283512,
"rewards/rejected": -0.19214141368865967,
"step": 195
},
{
"epoch": 0.30890464933018125,
"grad_norm": 0.2723829448223114,
"learning_rate": 3.965169436094947e-06,
"log_odds_chosen": 0.35698461532592773,
"log_odds_ratio": -0.5402282476425171,
"logits/chosen": 0.22953951358795166,
"logits/rejected": -0.668543815612793,
"logps/chosen": -1.6529057025909424,
"logps/rejected": -1.9540297985076904,
"loss": 1.8363,
"nll_loss": 1.7822985649108887,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16529057919979095,
"rewards/margins": 0.030112413689494133,
"rewards/rejected": -0.19540299475193024,
"step": 196
},
{
"epoch": 0.31048069345941687,
"grad_norm": 0.26106390357017517,
"learning_rate": 3.964140569932618e-06,
"log_odds_chosen": 0.29018843173980713,
"log_odds_ratio": -0.5803290009498596,
"logits/chosen": 0.1857983022928238,
"logits/rejected": -0.8773932456970215,
"logps/chosen": -1.7773064374923706,
"logps/rejected": -2.021932601928711,
"loss": 1.9428,
"nll_loss": 1.884739637374878,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.17773064970970154,
"rewards/margins": 0.02446262538433075,
"rewards/rejected": -0.2021932750940323,
"step": 197
},
{
"epoch": 0.3120567375886525,
"grad_norm": 0.25397989153862,
"learning_rate": 3.9630968656264285e-06,
"log_odds_chosen": 0.42598769068717957,
"log_odds_ratio": -0.509061872959137,
"logits/chosen": 0.21715356409549713,
"logits/rejected": -0.4817379415035248,
"logps/chosen": -1.7002034187316895,
"logps/rejected": -2.0615005493164062,
"loss": 1.8609,
"nll_loss": 1.810002326965332,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17002034187316895,
"rewards/margins": 0.036129724234342575,
"rewards/rejected": -0.20615006983280182,
"step": 198
},
{
"epoch": 0.3136327817178881,
"grad_norm": 0.26781895756721497,
"learning_rate": 3.962038331061065e-06,
"log_odds_chosen": 0.1461435854434967,
"log_odds_ratio": -0.6345757246017456,
"logits/chosen": 0.28389307856559753,
"logits/rejected": -0.6472858786582947,
"logps/chosen": -1.7552485466003418,
"logps/rejected": -1.883392333984375,
"loss": 1.9384,
"nll_loss": 1.8749713897705078,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.17552484571933746,
"rewards/margins": 0.012814389541745186,
"rewards/rejected": -0.1883392333984375,
"step": 199
},
{
"epoch": 0.31520882584712373,
"grad_norm": 0.26174989342689514,
"learning_rate": 3.96096497423325e-06,
"log_odds_chosen": 0.26020583510398865,
"log_odds_ratio": -0.5759395360946655,
"logits/chosen": 0.3217354714870453,
"logits/rejected": -0.2811959683895111,
"logps/chosen": -1.6606853008270264,
"logps/rejected": -1.8771640062332153,
"loss": 1.8637,
"nll_loss": 1.806121826171875,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16606852412223816,
"rewards/margins": 0.021647876128554344,
"rewards/rejected": -0.18771640956401825,
"step": 200
},
{
"epoch": 0.31678486997635935,
"grad_norm": 0.24145232141017914,
"learning_rate": 3.959876803251684e-06,
"log_odds_chosen": 0.3890625238418579,
"log_odds_ratio": -0.524219274520874,
"logits/chosen": 0.1805240660905838,
"logits/rejected": -0.5619311928749084,
"logps/chosen": -1.6790175437927246,
"logps/rejected": -2.00699782371521,
"loss": 1.8352,
"nll_loss": 1.782778263092041,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16790175437927246,
"rewards/margins": 0.03279803693294525,
"rewards/rejected": -0.2006998062133789,
"step": 201
},
{
"epoch": 0.31836091410559497,
"grad_norm": 0.25001904368400574,
"learning_rate": 3.958773826336977e-06,
"log_odds_chosen": 0.27317214012145996,
"log_odds_ratio": -0.5676698088645935,
"logits/chosen": 0.27888408303260803,
"logits/rejected": -0.415998250246048,
"logps/chosen": -1.7196669578552246,
"logps/rejected": -1.9488410949707031,
"loss": 1.8963,
"nll_loss": 1.8395037651062012,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17196668684482574,
"rewards/margins": 0.02291741594672203,
"rewards/rejected": -0.19488412141799927,
"step": 202
},
{
"epoch": 0.3199369582348306,
"grad_norm": 0.2554585933685303,
"learning_rate": 3.957656051821592e-06,
"log_odds_chosen": 0.23165369033813477,
"log_odds_ratio": -0.5892881751060486,
"logits/chosen": 0.3021984100341797,
"logits/rejected": -0.6948502063751221,
"logps/chosen": -1.7531200647354126,
"logps/rejected": -1.9506298303604126,
"loss": 1.9028,
"nll_loss": 1.8439098596572876,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17531201243400574,
"rewards/margins": 0.019750984385609627,
"rewards/rejected": -0.19506299495697021,
"step": 203
},
{
"epoch": 0.3215130023640662,
"grad_norm": 0.26098933815956116,
"learning_rate": 3.956523488149783e-06,
"log_odds_chosen": 0.3122096061706543,
"log_odds_ratio": -0.5535508394241333,
"logits/chosen": 0.3232036828994751,
"logits/rejected": -0.47777220606803894,
"logps/chosen": -1.8531510829925537,
"logps/rejected": -2.122044563293457,
"loss": 2.0121,
"nll_loss": 1.9567644596099854,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1853151172399521,
"rewards/margins": 0.02688935585319996,
"rewards/rejected": -0.2122044712305069,
"step": 204
},
{
"epoch": 0.32308904649330183,
"grad_norm": 0.2530037462711334,
"learning_rate": 3.9553761438775285e-06,
"log_odds_chosen": 0.5448867678642273,
"log_odds_ratio": -0.4623796045780182,
"logits/chosen": 0.1774434596300125,
"logits/rejected": -0.509428083896637,
"logps/chosen": -1.5850623846054077,
"logps/rejected": -2.0427145957946777,
"loss": 1.747,
"nll_loss": 1.7007546424865723,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15850622951984406,
"rewards/margins": 0.04576525092124939,
"rewards/rejected": -0.20427148044109344,
"step": 205
},
{
"epoch": 0.32466509062253746,
"grad_norm": 0.23750121891498566,
"learning_rate": 3.954214027672465e-06,
"log_odds_chosen": 0.48206406831741333,
"log_odds_ratio": -0.48811841011047363,
"logits/chosen": 0.18673132359981537,
"logits/rejected": -0.9018082022666931,
"logps/chosen": -1.67806077003479,
"logps/rejected": -2.0901315212249756,
"loss": 1.8315,
"nll_loss": 1.7826416492462158,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16780607402324677,
"rewards/margins": 0.041207075119018555,
"rewards/rejected": -0.2090131640434265,
"step": 206
},
{
"epoch": 0.3262411347517731,
"grad_norm": 0.25891193747520447,
"learning_rate": 3.953037148313825e-06,
"log_odds_chosen": 0.28911662101745605,
"log_odds_ratio": -0.5642296075820923,
"logits/chosen": 0.27622562646865845,
"logits/rejected": -0.581394612789154,
"logps/chosen": -1.6668041944503784,
"logps/rejected": -1.9111988544464111,
"loss": 1.8538,
"nll_loss": 1.7974015474319458,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16668042540550232,
"rewards/margins": 0.024439461529254913,
"rewards/rejected": -0.19111987948417664,
"step": 207
},
{
"epoch": 0.32781717888100864,
"grad_norm": 0.2777508497238159,
"learning_rate": 3.951845514692371e-06,
"log_odds_chosen": 0.39269790053367615,
"log_odds_ratio": -0.5250096321105957,
"logits/chosen": 0.2141711413860321,
"logits/rejected": -0.47686973214149475,
"logps/chosen": -1.729377269744873,
"logps/rejected": -2.064727783203125,
"loss": 1.9033,
"nll_loss": 1.8507862091064453,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17293773591518402,
"rewards/margins": 0.03353503346443176,
"rewards/rejected": -0.20647276937961578,
"step": 208
},
{
"epoch": 0.32939322301024426,
"grad_norm": 0.2944742441177368,
"learning_rate": 3.950639135810325e-06,
"log_odds_chosen": 0.3137563467025757,
"log_odds_ratio": -0.5539580583572388,
"logits/chosen": 0.3174862265586853,
"logits/rejected": -0.45376986265182495,
"logps/chosen": -1.850959062576294,
"logps/rejected": -2.1199288368225098,
"loss": 1.9963,
"nll_loss": 1.9408817291259766,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1850959062576294,
"rewards/margins": 0.02689695730805397,
"rewards/rejected": -0.21199287474155426,
"step": 209
},
{
"epoch": 0.3309692671394799,
"grad_norm": 0.3148304224014282,
"learning_rate": 3.9494180207813044e-06,
"log_odds_chosen": 0.23787932097911835,
"log_odds_ratio": -0.5915074348449707,
"logits/chosen": 0.30660662055015564,
"logits/rejected": -0.7329879403114319,
"logps/chosen": -1.717742919921875,
"logps/rejected": -1.916438102722168,
"loss": 1.8905,
"nll_loss": 1.8313184976577759,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17177429795265198,
"rewards/margins": 0.019869530573487282,
"rewards/rejected": -0.1916438341140747,
"step": 210
},
{
"epoch": 0.3325453112687155,
"grad_norm": 0.2588319778442383,
"learning_rate": 3.948182178830249e-06,
"log_odds_chosen": 0.29779791831970215,
"log_odds_ratio": -0.5746859908103943,
"logits/chosen": 0.2758365273475647,
"logits/rejected": -0.8558934926986694,
"logps/chosen": -1.7486122846603394,
"logps/rejected": -2.0061802864074707,
"loss": 1.904,
"nll_loss": 1.8465510606765747,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.17486125230789185,
"rewards/margins": 0.025756794959306717,
"rewards/rejected": -0.20061802864074707,
"step": 211
},
{
"epoch": 0.3341213553979511,
"grad_norm": 0.27347490191459656,
"learning_rate": 3.9469316192933545e-06,
"log_odds_chosen": 0.5377534627914429,
"log_odds_ratio": -0.4662073850631714,
"logits/chosen": 0.16830803453922272,
"logits/rejected": -0.490761935710907,
"logps/chosen": -1.6018506288528442,
"logps/rejected": -2.05351185798645,
"loss": 1.7708,
"nll_loss": 1.7242186069488525,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1601850688457489,
"rewards/margins": 0.04516611993312836,
"rewards/rejected": -0.20535118877887726,
"step": 212
},
{
"epoch": 0.33569739952718675,
"grad_norm": 0.2533399164676666,
"learning_rate": 3.945666351618001e-06,
"log_odds_chosen": 0.2882734537124634,
"log_odds_ratio": -0.562584638595581,
"logits/chosen": 0.2559697926044464,
"logits/rejected": -0.5387569069862366,
"logps/chosen": -1.6552236080169678,
"logps/rejected": -1.8957173824310303,
"loss": 1.823,
"nll_loss": 1.7667841911315918,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16552235186100006,
"rewards/margins": 0.02404937893152237,
"rewards/rejected": -0.18957173824310303,
"step": 213
},
{
"epoch": 0.33727344365642237,
"grad_norm": 0.25932711362838745,
"learning_rate": 3.9443863853626825e-06,
"log_odds_chosen": 0.3108881711959839,
"log_odds_ratio": -0.5589509010314941,
"logits/chosen": 0.2859116494655609,
"logits/rejected": -0.6345618367195129,
"logps/chosen": -1.6215680837631226,
"logps/rejected": -1.8729337453842163,
"loss": 1.8125,
"nll_loss": 1.7566089630126953,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1621568202972412,
"rewards/margins": 0.025136563926935196,
"rewards/rejected": -0.1872933804988861,
"step": 214
},
{
"epoch": 0.338849487785658,
"grad_norm": 0.250918984413147,
"learning_rate": 3.943091730196931e-06,
"log_odds_chosen": 0.2744632363319397,
"log_odds_ratio": -0.5730749368667603,
"logits/chosen": 0.3415977954864502,
"logits/rejected": -0.41705650091171265,
"logps/chosen": -1.6209944486618042,
"logps/rejected": -1.8511853218078613,
"loss": 1.7975,
"nll_loss": 1.740174651145935,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1620994359254837,
"rewards/margins": 0.023019105195999146,
"rewards/rejected": -0.18511852622032166,
"step": 215
},
{
"epoch": 0.3404255319148936,
"grad_norm": 0.27474015951156616,
"learning_rate": 3.941782395901249e-06,
"log_odds_chosen": 0.3491838574409485,
"log_odds_ratio": -0.5398426055908203,
"logits/chosen": 0.2213699072599411,
"logits/rejected": -0.549167811870575,
"logps/chosen": -1.69877028465271,
"logps/rejected": -1.9932255744934082,
"loss": 1.8755,
"nll_loss": 1.8215399980545044,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16987702250480652,
"rewards/margins": 0.02944553829729557,
"rewards/rejected": -0.19932258129119873,
"step": 216
},
{
"epoch": 0.34200157604412923,
"grad_norm": 0.25679028034210205,
"learning_rate": 3.940458392367032e-06,
"log_odds_chosen": 0.3104505240917206,
"log_odds_ratio": -0.5546755194664001,
"logits/chosen": 0.2595285475254059,
"logits/rejected": -0.8533272743225098,
"logps/chosen": -1.7510833740234375,
"logps/rejected": -2.0138750076293945,
"loss": 1.8934,
"nll_loss": 1.8379460573196411,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17510835826396942,
"rewards/margins": 0.026279138401150703,
"rewards/rejected": -0.20138749480247498,
"step": 217
},
{
"epoch": 0.34357762017336485,
"grad_norm": 0.2611715495586395,
"learning_rate": 3.939119729596493e-06,
"log_odds_chosen": 0.35739466547966003,
"log_odds_ratio": -0.5335213541984558,
"logits/chosen": 0.3199889659881592,
"logits/rejected": -0.6550436019897461,
"logps/chosen": -1.7533951997756958,
"logps/rejected": -2.057891368865967,
"loss": 1.9075,
"nll_loss": 1.8541964292526245,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17533953487873077,
"rewards/margins": 0.03044959530234337,
"rewards/rejected": -0.20578913390636444,
"step": 218
},
{
"epoch": 0.34515366430260047,
"grad_norm": 0.26392099261283875,
"learning_rate": 3.93776641770259e-06,
"log_odds_chosen": 0.2948194742202759,
"log_odds_ratio": -0.562857449054718,
"logits/chosen": 0.28813865780830383,
"logits/rejected": -0.7187209129333496,
"logps/chosen": -1.7230334281921387,
"logps/rejected": -1.973144292831421,
"loss": 1.8747,
"nll_loss": 1.8184067010879517,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17230333387851715,
"rewards/margins": 0.025011096149683,
"rewards/rejected": -0.19731444120407104,
"step": 219
},
{
"epoch": 0.3467297084318361,
"grad_norm": 0.26364782452583313,
"learning_rate": 3.93639846690895e-06,
"log_odds_chosen": 0.4254459738731384,
"log_odds_ratio": -0.5071126222610474,
"logits/chosen": 0.31134355068206787,
"logits/rejected": -0.9344062805175781,
"logps/chosen": -1.7416664361953735,
"logps/rejected": -2.10628080368042,
"loss": 1.8893,
"nll_loss": 1.8385860919952393,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17416664958000183,
"rewards/margins": 0.03646141290664673,
"rewards/rejected": -0.21062806248664856,
"step": 220
},
{
"epoch": 0.3483057525610717,
"grad_norm": 0.2517707943916321,
"learning_rate": 3.9350158875497855e-06,
"log_odds_chosen": 0.5023624300956726,
"log_odds_ratio": -0.4771158695220947,
"logits/chosen": 0.31009235978126526,
"logits/rejected": -0.771135687828064,
"logps/chosen": -1.6524683237075806,
"logps/rejected": -2.075108289718628,
"loss": 1.8136,
"nll_loss": 1.765908122062683,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.165246844291687,
"rewards/margins": 0.04226400703191757,
"rewards/rejected": -0.20751085877418518,
"step": 221
},
{
"epoch": 0.34988179669030733,
"grad_norm": 0.24314001202583313,
"learning_rate": 3.933618690069824e-06,
"log_odds_chosen": 0.4676588773727417,
"log_odds_ratio": -0.4905795454978943,
"logits/chosen": 0.17770220339298248,
"logits/rejected": -0.7084282636642456,
"logps/chosen": -1.6020326614379883,
"logps/rejected": -1.9884483814239502,
"loss": 1.7733,
"nll_loss": 1.7242058515548706,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16020327806472778,
"rewards/margins": 0.038641560822725296,
"rewards/rejected": -0.19884483516216278,
"step": 222
},
{
"epoch": 0.35145784081954295,
"grad_norm": 0.2550933361053467,
"learning_rate": 3.932206885024226e-06,
"log_odds_chosen": 0.23466067016124725,
"log_odds_ratio": -0.5852788090705872,
"logits/chosen": 0.22302168607711792,
"logits/rejected": -0.6709792017936707,
"logps/chosen": -1.7109097242355347,
"logps/rejected": -1.9076846837997437,
"loss": 1.8546,
"nll_loss": 1.7960247993469238,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1710909754037857,
"rewards/margins": 0.019677501171827316,
"rewards/rejected": -0.19076848030090332,
"step": 223
},
{
"epoch": 0.3530338849487786,
"grad_norm": 0.22619383037090302,
"learning_rate": 3.930780483078502e-06,
"log_odds_chosen": 0.3614213466644287,
"log_odds_ratio": -0.5334374904632568,
"logits/chosen": 0.1996021270751953,
"logits/rejected": -0.7123557329177856,
"logps/chosen": -1.6352108716964722,
"logps/rejected": -1.9351240396499634,
"loss": 1.807,
"nll_loss": 1.7537031173706055,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16352108120918274,
"rewards/margins": 0.02999131567776203,
"rewards/rejected": -0.19351240992546082,
"step": 224
},
{
"epoch": 0.3546099290780142,
"grad_norm": 0.2543198764324188,
"learning_rate": 3.92933949500844e-06,
"log_odds_chosen": 0.39015451073646545,
"log_odds_ratio": -0.5326002836227417,
"logits/chosen": 0.28930386900901794,
"logits/rejected": -0.7000013589859009,
"logps/chosen": -1.609468936920166,
"logps/rejected": -1.9277451038360596,
"loss": 1.7804,
"nll_loss": 1.727099895477295,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.16094687581062317,
"rewards/margins": 0.031827617436647415,
"rewards/rejected": -0.19277450442314148,
"step": 225
},
{
"epoch": 0.3561859732072498,
"grad_norm": 0.23875081539154053,
"learning_rate": 3.9278839317000155e-06,
"log_odds_chosen": 0.40506240725517273,
"log_odds_ratio": -0.5156422853469849,
"logits/chosen": 0.30614709854125977,
"logits/rejected": -0.6908950805664062,
"logps/chosen": -1.635833978652954,
"logps/rejected": -1.9742740392684937,
"loss": 1.7931,
"nll_loss": 1.7415223121643066,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1635833978652954,
"rewards/margins": 0.0338440015912056,
"rewards/rejected": -0.1974273920059204,
"step": 226
},
{
"epoch": 0.35776201733648544,
"grad_norm": 0.24680069088935852,
"learning_rate": 3.926413804149314e-06,
"log_odds_chosen": 0.1316530406475067,
"log_odds_ratio": -0.6378493309020996,
"logits/chosen": 0.28156548738479614,
"logits/rejected": -0.6600248217582703,
"logps/chosen": -1.7210869789123535,
"logps/rejected": -1.8304848670959473,
"loss": 1.8788,
"nll_loss": 1.815049171447754,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.1721086949110031,
"rewards/margins": 0.010939793661236763,
"rewards/rejected": -0.18304848670959473,
"step": 227
},
{
"epoch": 0.35933806146572106,
"grad_norm": 0.2515822649002075,
"learning_rate": 3.924929123462447e-06,
"log_odds_chosen": 0.2854158878326416,
"log_odds_ratio": -0.5666006207466125,
"logits/chosen": 0.33641237020492554,
"logits/rejected": -0.7290589809417725,
"logps/chosen": -1.7506561279296875,
"logps/rejected": -1.9922963380813599,
"loss": 1.8996,
"nll_loss": 1.84292733669281,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17506560683250427,
"rewards/margins": 0.024164030328392982,
"rewards/rejected": -0.1992296427488327,
"step": 228
},
{
"epoch": 0.3609141055949567,
"grad_norm": 0.26427891850471497,
"learning_rate": 3.923429900855468e-06,
"log_odds_chosen": 0.21544109284877777,
"log_odds_ratio": -0.5983462333679199,
"logits/chosen": 0.2909620702266693,
"logits/rejected": -0.438875675201416,
"logps/chosen": -1.8233458995819092,
"logps/rejected": -2.004505157470703,
"loss": 1.9767,
"nll_loss": 1.9168180227279663,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.18233460187911987,
"rewards/margins": 0.018115926533937454,
"rewards/rejected": -0.20045052468776703,
"step": 229
},
{
"epoch": 0.3624901497241923,
"grad_norm": 0.241230309009552,
"learning_rate": 3.921916147654287e-06,
"log_odds_chosen": 0.4597613215446472,
"log_odds_ratio": -0.49515679478645325,
"logits/chosen": 0.24242226779460907,
"logits/rejected": -0.7508520483970642,
"logps/chosen": -1.6597049236297607,
"logps/rejected": -2.044846773147583,
"loss": 1.7986,
"nll_loss": 1.7491000890731812,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16597048938274384,
"rewards/margins": 0.03851418197154999,
"rewards/rejected": -0.20448468625545502,
"step": 230
},
{
"epoch": 0.3640661938534279,
"grad_norm": 0.2590387761592865,
"learning_rate": 3.920387875294588e-06,
"log_odds_chosen": 0.30678579211235046,
"log_odds_ratio": -0.5545358657836914,
"logits/chosen": 0.2767443060874939,
"logits/rejected": -0.5900214910507202,
"logps/chosen": -1.6316543817520142,
"logps/rejected": -1.8850141763687134,
"loss": 1.7885,
"nll_loss": 1.7330236434936523,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16316545009613037,
"rewards/margins": 0.02533598057925701,
"rewards/rejected": -0.18850143253803253,
"step": 231
},
{
"epoch": 0.3656422379826635,
"grad_norm": 0.24193865060806274,
"learning_rate": 3.918845095321737e-06,
"log_odds_chosen": 0.29498574137687683,
"log_odds_ratio": -0.5670905113220215,
"logits/chosen": 0.23279045522212982,
"logits/rejected": -0.841462254524231,
"logps/chosen": -1.6641700267791748,
"logps/rejected": -1.9037166833877563,
"loss": 1.8125,
"nll_loss": 1.755805492401123,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16641701757907867,
"rewards/margins": 0.023954641073942184,
"rewards/rejected": -0.19037166237831116,
"step": 232
},
{
"epoch": 0.3672182821118991,
"grad_norm": 0.2594136595726013,
"learning_rate": 3.9172878193907004e-06,
"log_odds_chosen": 0.3280632793903351,
"log_odds_ratio": -0.551827609539032,
"logits/chosen": 0.18091654777526855,
"logits/rejected": -0.6931451559066772,
"logps/chosen": -1.7367044687271118,
"logps/rejected": -2.016188859939575,
"loss": 1.8981,
"nll_loss": 1.8429385423660278,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1736704558134079,
"rewards/margins": 0.02794845588505268,
"rewards/rejected": -0.20161890983581543,
"step": 233
},
{
"epoch": 0.36879432624113473,
"grad_norm": 0.2600264549255371,
"learning_rate": 3.915716059265955e-06,
"log_odds_chosen": 0.34488314390182495,
"log_odds_ratio": -0.543495774269104,
"logits/chosen": 0.25614839792251587,
"logits/rejected": -0.785013735294342,
"logps/chosen": -1.6906204223632812,
"logps/rejected": -1.9822051525115967,
"loss": 1.8406,
"nll_loss": 1.7862180471420288,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1690620481967926,
"rewards/margins": 0.02915847674012184,
"rewards/rejected": -0.19822052121162415,
"step": 234
},
{
"epoch": 0.37037037037037035,
"grad_norm": 0.22871002554893494,
"learning_rate": 3.9141298268213966e-06,
"log_odds_chosen": 0.4361989200115204,
"log_odds_ratio": -0.5028459429740906,
"logits/chosen": 0.2875756621360779,
"logits/rejected": -0.6794713735580444,
"logps/chosen": -1.6101529598236084,
"logps/rejected": -1.9754250049591064,
"loss": 1.7655,
"nll_loss": 1.7152522802352905,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16101528704166412,
"rewards/margins": 0.03652720898389816,
"rewards/rejected": -0.19754250347614288,
"step": 235
},
{
"epoch": 0.37194641449960597,
"grad_norm": 0.261535108089447,
"learning_rate": 3.912529134040255e-06,
"log_odds_chosen": 0.296495646238327,
"log_odds_ratio": -0.5591344833374023,
"logits/chosen": 0.27751684188842773,
"logits/rejected": -0.5663548707962036,
"logps/chosen": -1.6922770738601685,
"logps/rejected": -1.9417719841003418,
"loss": 1.8539,
"nll_loss": 1.7979625463485718,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1692277193069458,
"rewards/margins": 0.02494947426021099,
"rewards/rejected": -0.19417718052864075,
"step": 236
},
{
"epoch": 0.3735224586288416,
"grad_norm": 0.2609320878982544,
"learning_rate": 3.910913993014998e-06,
"log_odds_chosen": 0.4593818187713623,
"log_odds_ratio": -0.4922163188457489,
"logits/chosen": 0.3688610792160034,
"logits/rejected": -0.7247602939605713,
"logps/chosen": -1.687050223350525,
"logps/rejected": -2.0763425827026367,
"loss": 1.8363,
"nll_loss": 1.7870500087738037,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1687050312757492,
"rewards/margins": 0.038929231464862823,
"rewards/rejected": -0.20763425529003143,
"step": 237
},
{
"epoch": 0.3750985027580772,
"grad_norm": 0.25336939096450806,
"learning_rate": 3.909284415947246e-06,
"log_odds_chosen": 0.4314580261707306,
"log_odds_ratio": -0.5070245862007141,
"logits/chosen": 0.32482269406318665,
"logits/rejected": -0.8647500276565552,
"logps/chosen": -1.73568594455719,
"logps/rejected": -2.101022243499756,
"loss": 1.87,
"nll_loss": 1.819305419921875,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17356860637664795,
"rewards/margins": 0.036533623933792114,
"rewards/rejected": -0.21010223031044006,
"step": 238
},
{
"epoch": 0.37667454688731283,
"grad_norm": 0.24124158918857574,
"learning_rate": 3.907640415147674e-06,
"log_odds_chosen": 0.4658309817314148,
"log_odds_ratio": -0.49067068099975586,
"logits/chosen": 0.21756987273693085,
"logits/rejected": -0.9139111042022705,
"logps/chosen": -1.6312497854232788,
"logps/rejected": -2.020339250564575,
"loss": 1.788,
"nll_loss": 1.7389440536499023,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16312497854232788,
"rewards/margins": 0.038908950984478,
"rewards/rejected": -0.20203393697738647,
"step": 239
},
{
"epoch": 0.37825059101654845,
"grad_norm": 0.2780405282974243,
"learning_rate": 3.905982003035924e-06,
"log_odds_chosen": 0.3844011127948761,
"log_odds_ratio": -0.5241358280181885,
"logits/chosen": 0.23867914080619812,
"logits/rejected": -0.8461136817932129,
"logps/chosen": -1.7408959865570068,
"logps/rejected": -2.067878246307373,
"loss": 1.8775,
"nll_loss": 1.825044870376587,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17408961057662964,
"rewards/margins": 0.03269820660352707,
"rewards/rejected": -0.2067878246307373,
"step": 240
},
{
"epoch": 0.3798266351457841,
"grad_norm": 0.24185331165790558,
"learning_rate": 3.904309192140506e-06,
"log_odds_chosen": 0.278605192899704,
"log_odds_ratio": -0.571584165096283,
"logits/chosen": 0.24360942840576172,
"logits/rejected": -0.4919203519821167,
"logps/chosen": -1.6554006338119507,
"logps/rejected": -1.8880599737167358,
"loss": 1.8086,
"nll_loss": 1.7514057159423828,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16554008424282074,
"rewards/margins": 0.023265928030014038,
"rewards/rejected": -0.18880601227283478,
"step": 241
},
{
"epoch": 0.3814026792750197,
"grad_norm": 0.26548993587493896,
"learning_rate": 3.90262199509871e-06,
"log_odds_chosen": 0.2826857268810272,
"log_odds_ratio": -0.5648576021194458,
"logits/chosen": 0.22797901928424835,
"logits/rejected": -0.5838393568992615,
"logps/chosen": -1.6858417987823486,
"logps/rejected": -1.9206804037094116,
"loss": 1.8445,
"nll_loss": 1.7879817485809326,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1685841828584671,
"rewards/margins": 0.023483866825699806,
"rewards/rejected": -0.19206805527210236,
"step": 242
},
{
"epoch": 0.3829787234042553,
"grad_norm": 0.2655971944332123,
"learning_rate": 3.900920424656501e-06,
"log_odds_chosen": 0.4220297932624817,
"log_odds_ratio": -0.51133131980896,
"logits/chosen": 0.2471286654472351,
"logits/rejected": -0.7664991021156311,
"logps/chosen": -1.6024655103683472,
"logps/rejected": -1.953262209892273,
"loss": 1.7496,
"nll_loss": 1.6984376907348633,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16024655103683472,
"rewards/margins": 0.035079680383205414,
"rewards/rejected": -0.19532622396945953,
"step": 243
},
{
"epoch": 0.38455476753349094,
"grad_norm": 0.2719763517379761,
"learning_rate": 3.899204493668432e-06,
"log_odds_chosen": 0.34531235694885254,
"log_odds_ratio": -0.545647382736206,
"logits/chosen": 0.2916204631328583,
"logits/rejected": -0.6753207445144653,
"logps/chosen": -1.760868787765503,
"logps/rejected": -2.058903694152832,
"loss": 1.8996,
"nll_loss": 1.8450438976287842,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.17608687281608582,
"rewards/margins": 0.029803497716784477,
"rewards/rejected": -0.20589037239551544,
"step": 244
},
{
"epoch": 0.38613081166272656,
"grad_norm": 0.26759397983551025,
"learning_rate": 3.897474215097542e-06,
"log_odds_chosen": 0.34969383478164673,
"log_odds_ratio": -0.5378445386886597,
"logits/chosen": 0.30676940083503723,
"logits/rejected": -0.9074739217758179,
"logps/chosen": -1.7014522552490234,
"logps/rejected": -1.9977072477340698,
"loss": 1.8594,
"nll_loss": 1.805631160736084,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1701452136039734,
"rewards/margins": 0.029625503346323967,
"rewards/rejected": -0.1997707188129425,
"step": 245
},
{
"epoch": 0.3877068557919622,
"grad_norm": 0.24695803225040436,
"learning_rate": 3.8957296020152596e-06,
"log_odds_chosen": 0.4582657217979431,
"log_odds_ratio": -0.4948864281177521,
"logits/chosen": 0.3029223680496216,
"logits/rejected": -0.7880758047103882,
"logps/chosen": -1.6669402122497559,
"logps/rejected": -2.054736614227295,
"loss": 1.8203,
"nll_loss": 1.7708466053009033,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1666940301656723,
"rewards/margins": 0.03877962380647659,
"rewards/rejected": -0.2054736316204071,
"step": 246
},
{
"epoch": 0.3892828999211978,
"grad_norm": 0.2638940215110779,
"learning_rate": 3.893970667601303e-06,
"log_odds_chosen": 0.3049715757369995,
"log_odds_ratio": -0.5559093952178955,
"logits/chosen": 0.20208218693733215,
"logits/rejected": -1.0700570344924927,
"logps/chosen": -1.7513179779052734,
"logps/rejected": -2.0092613697052,
"loss": 1.8901,
"nll_loss": 1.83454430103302,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17513179779052734,
"rewards/margins": 0.02579433098435402,
"rewards/rejected": -0.20092612504959106,
"step": 247
},
{
"epoch": 0.3908589440504334,
"grad_norm": 0.260355681180954,
"learning_rate": 3.892197425143581e-06,
"log_odds_chosen": 0.4682731330394745,
"log_odds_ratio": -0.48940151929855347,
"logits/chosen": 0.255657821893692,
"logits/rejected": -0.671928882598877,
"logps/chosen": -1.6768563985824585,
"logps/rejected": -2.071852922439575,
"loss": 1.8127,
"nll_loss": 1.763710618019104,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16768565773963928,
"rewards/margins": 0.03949964791536331,
"rewards/rejected": -0.2071852833032608,
"step": 248
},
{
"epoch": 0.39243498817966904,
"grad_norm": 0.2769114673137665,
"learning_rate": 3.890409888038094e-06,
"log_odds_chosen": 0.35800671577453613,
"log_odds_ratio": -0.5353755354881287,
"logits/chosen": 0.27717655897140503,
"logits/rejected": -0.6495774984359741,
"logps/chosen": -1.718163013458252,
"logps/rejected": -2.023691415786743,
"loss": 1.8664,
"nll_loss": 1.812865138053894,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17181627452373505,
"rewards/margins": 0.030552847310900688,
"rewards/rejected": -0.20236913859844208,
"step": 249
},
{
"epoch": 0.39401103230890466,
"grad_norm": 0.2523289620876312,
"learning_rate": 3.888608069788831e-06,
"log_odds_chosen": 0.3482479751110077,
"log_odds_ratio": -0.5456973314285278,
"logits/chosen": 0.18573682010173798,
"logits/rejected": -0.5659343600273132,
"logps/chosen": -1.672256350517273,
"logps/rejected": -1.9685872793197632,
"loss": 1.8214,
"nll_loss": 1.7668198347091675,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16722562909126282,
"rewards/margins": 0.029633095487952232,
"rewards/rejected": -0.1968587189912796,
"step": 250
},
{
"epoch": 0.3955870764381403,
"grad_norm": 0.23957766592502594,
"learning_rate": 3.8867919840076685e-06,
"log_odds_chosen": 0.4184243083000183,
"log_odds_ratio": -0.5095526576042175,
"logits/chosen": 0.22751504182815552,
"logits/rejected": -0.8524928092956543,
"logps/chosen": -1.6037871837615967,
"logps/rejected": -1.9516103267669678,
"loss": 1.7432,
"nll_loss": 1.692209005355835,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16037872433662415,
"rewards/margins": 0.03478231281042099,
"rewards/rejected": -0.19516104459762573,
"step": 251
},
{
"epoch": 0.3971631205673759,
"grad_norm": 0.2576284110546112,
"learning_rate": 3.884961644414267e-06,
"log_odds_chosen": 0.34130245447158813,
"log_odds_ratio": -0.5387775897979736,
"logits/chosen": 0.24311238527297974,
"logits/rejected": -0.3957245647907257,
"logps/chosen": -1.7268153429031372,
"logps/rejected": -2.0152997970581055,
"loss": 1.848,
"nll_loss": 1.794105887413025,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1726815402507782,
"rewards/margins": 0.028848443180322647,
"rewards/rejected": -0.20152997970581055,
"step": 252
},
{
"epoch": 0.3987391646966115,
"grad_norm": 0.24223408102989197,
"learning_rate": 3.883117064835967e-06,
"log_odds_chosen": 0.3959362506866455,
"log_odds_ratio": -0.5255994200706482,
"logits/chosen": 0.1985481083393097,
"logits/rejected": -0.7414513230323792,
"logps/chosen": -1.6674610376358032,
"logps/rejected": -2.0037710666656494,
"loss": 1.8186,
"nll_loss": 1.7660025358200073,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1667460948228836,
"rewards/margins": 0.033631011843681335,
"rewards/rejected": -0.20037710666656494,
"step": 253
},
{
"epoch": 0.40031520882584715,
"grad_norm": 0.26332396268844604,
"learning_rate": 3.881258259207688e-06,
"log_odds_chosen": 0.3393678367137909,
"log_odds_ratio": -0.5432149171829224,
"logits/chosen": 0.30877983570098877,
"logits/rejected": -0.7882847785949707,
"logps/chosen": -1.6376413106918335,
"logps/rejected": -1.919863224029541,
"loss": 1.7661,
"nll_loss": 1.7118016481399536,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1637641191482544,
"rewards/margins": 0.028222184628248215,
"rewards/rejected": -0.1919863075017929,
"step": 254
},
{
"epoch": 0.40189125295508277,
"grad_norm": 0.25575923919677734,
"learning_rate": 3.8793852415718165e-06,
"log_odds_chosen": 0.4229394495487213,
"log_odds_ratio": -0.5054378509521484,
"logits/chosen": 0.21786652505397797,
"logits/rejected": -1.0462470054626465,
"logps/chosen": -1.561218023300171,
"logps/rejected": -1.9078154563903809,
"loss": 1.694,
"nll_loss": 1.6435015201568604,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15612182021141052,
"rewards/margins": 0.0346597358584404,
"rewards/rejected": -0.19078153371810913,
"step": 255
},
{
"epoch": 0.4034672970843184,
"grad_norm": 0.24122343957424164,
"learning_rate": 3.877498026078107e-06,
"log_odds_chosen": 0.2585289180278778,
"log_odds_ratio": -0.5783178210258484,
"logits/chosen": 0.22445496916770935,
"logits/rejected": -0.9274032711982727,
"logps/chosen": -1.7019916772842407,
"logps/rejected": -1.9216302633285522,
"loss": 1.8365,
"nll_loss": 1.7786757946014404,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1701991707086563,
"rewards/margins": 0.02196386270225048,
"rewards/rejected": -0.19216305017471313,
"step": 256
},
{
"epoch": 0.40504334121355395,
"grad_norm": 0.25378501415252686,
"learning_rate": 3.875596626983573e-06,
"log_odds_chosen": 0.48093894124031067,
"log_odds_ratio": -0.4933563768863678,
"logits/chosen": 0.24534080922603607,
"logits/rejected": -0.61388099193573,
"logps/chosen": -1.6251795291900635,
"logps/rejected": -2.02630352973938,
"loss": 1.7574,
"nll_loss": 1.708074688911438,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1625179499387741,
"rewards/margins": 0.040112413465976715,
"rewards/rejected": -0.20263037085533142,
"step": 257
},
{
"epoch": 0.4066193853427896,
"grad_norm": 0.26203182339668274,
"learning_rate": 3.873681058652374e-06,
"log_odds_chosen": 0.27626505494117737,
"log_odds_ratio": -0.5716174244880676,
"logits/chosen": 0.2242291271686554,
"logits/rejected": -0.705990731716156,
"logps/chosen": -1.742016077041626,
"logps/rejected": -1.976877212524414,
"loss": 1.8763,
"nll_loss": 1.819146990776062,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1742016226053238,
"rewards/margins": 0.023486101999878883,
"rewards/rejected": -0.19768773019313812,
"step": 258
},
{
"epoch": 0.4081954294720252,
"grad_norm": 0.26595333218574524,
"learning_rate": 3.871751335555715e-06,
"log_odds_chosen": 0.26442596316337585,
"log_odds_ratio": -0.5714837312698364,
"logits/chosen": 0.20778290927410126,
"logits/rejected": -0.8641613125801086,
"logps/chosen": -1.6800938844680786,
"logps/rejected": -1.9005059003829956,
"loss": 1.809,
"nll_loss": 1.7518802881240845,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16800937056541443,
"rewards/margins": 0.0220412015914917,
"rewards/rejected": -0.19005057215690613,
"step": 259
},
{
"epoch": 0.4097714736012608,
"grad_norm": 0.23981213569641113,
"learning_rate": 3.869807472271731e-06,
"log_odds_chosen": 0.3423271179199219,
"log_odds_ratio": -0.5396283864974976,
"logits/chosen": 0.23563726246356964,
"logits/rejected": -0.7463378310203552,
"logps/chosen": -1.6491752862930298,
"logps/rejected": -1.9362696409225464,
"loss": 1.8109,
"nll_loss": 1.7569705247879028,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16491752862930298,
"rewards/margins": 0.028709445148706436,
"rewards/rejected": -0.19362697005271912,
"step": 260
},
{
"epoch": 0.41134751773049644,
"grad_norm": 0.24014748632907867,
"learning_rate": 3.8678494834853826e-06,
"log_odds_chosen": 0.34554576873779297,
"log_odds_ratio": -0.5426000356674194,
"logits/chosen": 0.20350177586078644,
"logits/rejected": -0.6877405643463135,
"logps/chosen": -1.6484317779541016,
"logps/rejected": -1.934998631477356,
"loss": 1.7826,
"nll_loss": 1.728297472000122,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16484320163726807,
"rewards/margins": 0.028656674548983574,
"rewards/rejected": -0.1934998780488968,
"step": 261
},
{
"epoch": 0.41292356185973206,
"grad_norm": 0.27845731377601624,
"learning_rate": 3.865877383988339e-06,
"log_odds_chosen": 0.41938862204551697,
"log_odds_ratio": -0.5120880603790283,
"logits/chosen": 0.20454849302768707,
"logits/rejected": -0.6759002208709717,
"logps/chosen": -1.6775566339492798,
"logps/rejected": -2.0273048877716064,
"loss": 1.8329,
"nll_loss": 1.7817234992980957,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16775566339492798,
"rewards/margins": 0.0349748432636261,
"rewards/rejected": -0.20273049175739288,
"step": 262
},
{
"epoch": 0.4144996059889677,
"grad_norm": 0.2552221417427063,
"learning_rate": 3.863891188678869e-06,
"log_odds_chosen": 0.16435928642749786,
"log_odds_ratio": -0.6247628927230835,
"logits/chosen": 0.17906969785690308,
"logits/rejected": -0.8283835649490356,
"logps/chosen": -1.5928330421447754,
"logps/rejected": -1.7304943799972534,
"loss": 1.7381,
"nll_loss": 1.6756339073181152,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1592833399772644,
"rewards/margins": 0.013766113668680191,
"rewards/rejected": -0.1730494201183319,
"step": 263
},
{
"epoch": 0.4160756501182033,
"grad_norm": 0.24987082183361053,
"learning_rate": 3.8618909125617305e-06,
"log_odds_chosen": 0.5285029411315918,
"log_odds_ratio": -0.4670139253139496,
"logits/chosen": 0.16765527427196503,
"logits/rejected": -1.05913507938385,
"logps/chosen": -1.5894949436187744,
"logps/rejected": -2.031071662902832,
"loss": 1.7407,
"nll_loss": 1.693982720375061,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15894947946071625,
"rewards/margins": 0.04415770620107651,
"rewards/rejected": -0.20310717821121216,
"step": 264
},
{
"epoch": 0.4176516942474389,
"grad_norm": 0.24382364749908447,
"learning_rate": 3.859876570748054e-06,
"log_odds_chosen": 0.3537558913230896,
"log_odds_ratio": -0.5339797735214233,
"logits/chosen": 0.2316458523273468,
"logits/rejected": -0.3801443874835968,
"logps/chosen": -1.6417627334594727,
"logps/rejected": -1.9363051652908325,
"loss": 1.792,
"nll_loss": 1.7385859489440918,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16417627036571503,
"rewards/margins": 0.02945425920188427,
"rewards/rejected": -0.19363053143024445,
"step": 265
},
{
"epoch": 0.41922773837667454,
"grad_norm": 0.23967225849628448,
"learning_rate": 3.857848178455231e-06,
"log_odds_chosen": 0.11959446966648102,
"log_odds_ratio": -0.6412428617477417,
"logits/chosen": 0.2559909522533417,
"logits/rejected": -1.0443730354309082,
"logps/chosen": -1.656022071838379,
"logps/rejected": -1.7599290609359741,
"loss": 1.8213,
"nll_loss": 1.7571834325790405,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.16560222208499908,
"rewards/margins": 0.010390684939920902,
"rewards/rejected": -0.1759929060935974,
"step": 266
},
{
"epoch": 0.42080378250591016,
"grad_norm": 0.27286192774772644,
"learning_rate": 3.855805751006794e-06,
"log_odds_chosen": 0.4806629419326782,
"log_odds_ratio": -0.49105098843574524,
"logits/chosen": 0.2224351465702057,
"logits/rejected": -1.0106374025344849,
"logps/chosen": -1.6792700290679932,
"logps/rejected": -2.0848042964935303,
"loss": 1.8049,
"nll_loss": 1.7558059692382812,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16792699694633484,
"rewards/margins": 0.04055342823266983,
"rewards/rejected": -0.20848044753074646,
"step": 267
},
{
"epoch": 0.4223798266351458,
"grad_norm": 0.22964340448379517,
"learning_rate": 3.853749303832308e-06,
"log_odds_chosen": 0.34756800532341003,
"log_odds_ratio": -0.5398483872413635,
"logits/chosen": 0.19423283636569977,
"logits/rejected": -0.5895228981971741,
"logps/chosen": -1.6056160926818848,
"logps/rejected": -1.892835021018982,
"loss": 1.7439,
"nll_loss": 1.6899384260177612,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16056160628795624,
"rewards/margins": 0.02872188575565815,
"rewards/rejected": -0.18928351998329163,
"step": 268
},
{
"epoch": 0.4239558707643814,
"grad_norm": 0.2715766429901123,
"learning_rate": 3.8516788524672495e-06,
"log_odds_chosen": 0.2077798843383789,
"log_odds_ratio": -0.6008598804473877,
"logits/chosen": 0.22610792517662048,
"logits/rejected": -0.9308719038963318,
"logps/chosen": -1.6854466199874878,
"logps/rejected": -1.8608953952789307,
"loss": 1.8208,
"nll_loss": 1.7607052326202393,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.16854466497898102,
"rewards/margins": 0.017544886097311974,
"rewards/rejected": -0.18608956038951874,
"step": 269
},
{
"epoch": 0.425531914893617,
"grad_norm": 0.23874664306640625,
"learning_rate": 3.849594412552889e-06,
"log_odds_chosen": 0.3525705933570862,
"log_odds_ratio": -0.5381215810775757,
"logits/chosen": 0.24739238619804382,
"logits/rejected": -0.6722042560577393,
"logps/chosen": -1.587384581565857,
"logps/rejected": -1.871896505355835,
"loss": 1.727,
"nll_loss": 1.673226237297058,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15873846411705017,
"rewards/margins": 0.028451191261410713,
"rewards/rejected": -0.18718963861465454,
"step": 270
},
{
"epoch": 0.42710795902285265,
"grad_norm": 0.2558384835720062,
"learning_rate": 3.847495999836175e-06,
"log_odds_chosen": 0.3543975353240967,
"log_odds_ratio": -0.5356498956680298,
"logits/chosen": 0.2624046504497528,
"logits/rejected": -0.9314246773719788,
"logps/chosen": -1.666176438331604,
"logps/rejected": -1.9635179042816162,
"loss": 1.7931,
"nll_loss": 1.7394909858703613,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16661766171455383,
"rewards/margins": 0.029734138399362564,
"rewards/rejected": -0.1963518112897873,
"step": 271
},
{
"epoch": 0.42868400315208827,
"grad_norm": 0.23578688502311707,
"learning_rate": 3.845383630169613e-06,
"log_odds_chosen": 0.3167238235473633,
"log_odds_ratio": -0.5535402894020081,
"logits/chosen": 0.11884280294179916,
"logits/rejected": -0.9415445923805237,
"logps/chosen": -1.5773344039916992,
"logps/rejected": -1.837627649307251,
"loss": 1.7049,
"nll_loss": 1.6495603322982788,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15773345530033112,
"rewards/margins": 0.026029333472251892,
"rewards/rejected": -0.18376277387142181,
"step": 272
},
{
"epoch": 0.4302600472813239,
"grad_norm": 0.2651219367980957,
"learning_rate": 3.843257319511147e-06,
"log_odds_chosen": 0.5530175566673279,
"log_odds_ratio": -0.46196067333221436,
"logits/chosen": 0.17332234978675842,
"logits/rejected": -1.021928310394287,
"logps/chosen": -1.5769469738006592,
"logps/rejected": -2.0343515872955322,
"loss": 1.7041,
"nll_loss": 1.6578803062438965,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15769469738006592,
"rewards/margins": 0.04574044048786163,
"rewards/rejected": -0.20343513786792755,
"step": 273
},
{
"epoch": 0.4318360914105595,
"grad_norm": 0.24971356987953186,
"learning_rate": 3.841117083924039e-06,
"log_odds_chosen": 0.32874542474746704,
"log_odds_ratio": -0.5475614666938782,
"logits/chosen": 0.16287937760353088,
"logits/rejected": -1.0936428308486938,
"logps/chosen": -1.6155405044555664,
"logps/rejected": -1.888677716255188,
"loss": 1.7473,
"nll_loss": 1.692505121231079,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16155406832695007,
"rewards/margins": 0.0273137167096138,
"rewards/rejected": -0.18886777758598328,
"step": 274
},
{
"epoch": 0.43341213553979513,
"grad_norm": 0.2504657506942749,
"learning_rate": 3.838962939576746e-06,
"log_odds_chosen": 0.4173721373081207,
"log_odds_ratio": -0.51390141248703,
"logits/chosen": 0.13147510588169098,
"logits/rejected": -0.8347434401512146,
"logps/chosen": -1.5679432153701782,
"logps/rejected": -1.9134660959243774,
"loss": 1.72,
"nll_loss": 1.668624758720398,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15679430961608887,
"rewards/margins": 0.03455227613449097,
"rewards/rejected": -0.19134658575057983,
"step": 275
},
{
"epoch": 0.43498817966903075,
"grad_norm": 0.2594338357448578,
"learning_rate": 3.8367949027427985e-06,
"log_odds_chosen": 0.3050197958946228,
"log_odds_ratio": -0.5566756129264832,
"logits/chosen": 0.19317705929279327,
"logits/rejected": -0.7772097587585449,
"logps/chosen": -1.683677077293396,
"logps/rejected": -1.9412704706192017,
"loss": 1.8395,
"nll_loss": 1.7838023900985718,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16836771368980408,
"rewards/margins": 0.02575933374464512,
"rewards/rejected": -0.19412705302238464,
"step": 276
},
{
"epoch": 0.43656422379826637,
"grad_norm": 0.26758840680122375,
"learning_rate": 3.834612989800681e-06,
"log_odds_chosen": 0.5289045572280884,
"log_odds_ratio": -0.46756529808044434,
"logits/chosen": 0.21106746792793274,
"logits/rejected": -1.300022006034851,
"logps/chosen": -1.6692185401916504,
"logps/rejected": -2.1159589290618896,
"loss": 1.7865,
"nll_loss": 1.7397515773773193,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16692185401916504,
"rewards/margins": 0.04467405378818512,
"rewards/rejected": -0.21159592270851135,
"step": 277
},
{
"epoch": 0.438140267927502,
"grad_norm": 0.37664347887039185,
"learning_rate": 3.832417217233703e-06,
"log_odds_chosen": 0.3042469024658203,
"log_odds_ratio": -0.555115282535553,
"logits/chosen": 0.17117249965667725,
"logits/rejected": -0.787204921245575,
"logps/chosen": -1.6579128503799438,
"logps/rejected": -1.9126213788986206,
"loss": 1.7924,
"nll_loss": 1.7368648052215576,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16579128801822662,
"rewards/margins": 0.02547084540128708,
"rewards/rejected": -0.1912621259689331,
"step": 278
},
{
"epoch": 0.4397163120567376,
"grad_norm": 0.2642952501773834,
"learning_rate": 3.8302076016298775e-06,
"log_odds_chosen": 0.4446471929550171,
"log_odds_ratio": -0.49887239933013916,
"logits/chosen": 0.11896737664937973,
"logits/rejected": -1.0080976486206055,
"logps/chosen": -1.5889983177185059,
"logps/rejected": -1.9574190378189087,
"loss": 1.7293,
"nll_loss": 1.6794246435165405,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15889984369277954,
"rewards/margins": 0.036842066794633865,
"rewards/rejected": -0.1957419216632843,
"step": 279
},
{
"epoch": 0.44129235618597323,
"grad_norm": 0.2610625922679901,
"learning_rate": 3.827984159681796e-06,
"log_odds_chosen": 0.34331244230270386,
"log_odds_ratio": -0.541645884513855,
"logits/chosen": 0.10684026777744293,
"logits/rejected": -0.8201386332511902,
"logps/chosen": -1.6439259052276611,
"logps/rejected": -1.9289183616638184,
"loss": 1.7716,
"nll_loss": 1.7174153327941895,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1643926203250885,
"rewards/margins": 0.028499236330389977,
"rewards/rejected": -0.19289185106754303,
"step": 280
},
{
"epoch": 0.4428684003152088,
"grad_norm": 0.24843472242355347,
"learning_rate": 3.825746908186498e-06,
"log_odds_chosen": 0.3994872272014618,
"log_odds_ratio": -0.5212578773498535,
"logits/chosen": 0.2376633733510971,
"logits/rejected": -1.0532779693603516,
"logps/chosen": -1.6814634799957275,
"logps/rejected": -2.018228769302368,
"loss": 1.7922,
"nll_loss": 1.740039348602295,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16814635694026947,
"rewards/margins": 0.03367652744054794,
"rewards/rejected": -0.201822891831398,
"step": 281
},
{
"epoch": 0.4444444444444444,
"grad_norm": 0.2748637795448303,
"learning_rate": 3.823495864045352e-06,
"log_odds_chosen": 0.4453040361404419,
"log_odds_ratio": -0.4974423050880432,
"logits/chosen": 0.20578338205814362,
"logits/rejected": -0.5144822001457214,
"logps/chosen": -1.6322267055511475,
"logps/rejected": -2.0018556118011475,
"loss": 1.7625,
"nll_loss": 1.7127240896224976,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16322267055511475,
"rewards/margins": 0.036962881684303284,
"rewards/rejected": -0.20018555223941803,
"step": 282
},
{
"epoch": 0.44602048857368004,
"grad_norm": 0.2273551970720291,
"learning_rate": 3.8212310442639205e-06,
"log_odds_chosen": 0.33169132471084595,
"log_odds_ratio": -0.5450347065925598,
"logits/chosen": 0.2068972885608673,
"logits/rejected": -0.695864200592041,
"logps/chosen": -1.5796043872833252,
"logps/rejected": -1.8531723022460938,
"loss": 1.7234,
"nll_loss": 1.6689225435256958,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1579604595899582,
"rewards/margins": 0.027356795966625214,
"rewards/rejected": -0.1853172332048416,
"step": 283
},
{
"epoch": 0.44759653270291566,
"grad_norm": 0.24065622687339783,
"learning_rate": 3.8189524659518355e-06,
"log_odds_chosen": 0.40192341804504395,
"log_odds_ratio": -0.5161522626876831,
"logits/chosen": 0.2697753310203552,
"logits/rejected": -0.8584244847297668,
"logps/chosen": -1.6505751609802246,
"logps/rejected": -1.9848883152008057,
"loss": 1.7755,
"nll_loss": 1.7239261865615845,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16505752503871918,
"rewards/margins": 0.03343129903078079,
"rewards/rejected": -0.19848881661891937,
"step": 284
},
{
"epoch": 0.4491725768321513,
"grad_norm": 0.244610995054245,
"learning_rate": 3.816660146322667e-06,
"log_odds_chosen": 0.36820292472839355,
"log_odds_ratio": -0.5295414328575134,
"logits/chosen": 0.1906885802745819,
"logits/rejected": -0.700128436088562,
"logps/chosen": -1.5470272302627563,
"logps/rejected": -1.847809076309204,
"loss": 1.6794,
"nll_loss": 1.626416802406311,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15470272302627563,
"rewards/margins": 0.03007819503545761,
"rewards/rejected": -0.18478091061115265,
"step": 285
},
{
"epoch": 0.4507486209613869,
"grad_norm": 0.23016570508480072,
"learning_rate": 3.814354102693797e-06,
"log_odds_chosen": 0.3836941719055176,
"log_odds_ratio": -0.5216916799545288,
"logits/chosen": 0.21484431624412537,
"logits/rejected": -0.9419076442718506,
"logps/chosen": -1.6132217645645142,
"logps/rejected": -1.9288195371627808,
"loss": 1.7432,
"nll_loss": 1.6910346746444702,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16132217645645142,
"rewards/margins": 0.0315597802400589,
"rewards/rejected": -0.19288195669651031,
"step": 286
},
{
"epoch": 0.4523246650906225,
"grad_norm": 0.23653464019298553,
"learning_rate": 3.8120343524862814e-06,
"log_odds_chosen": 0.3473433256149292,
"log_odds_ratio": -0.535927951335907,
"logits/chosen": 0.18108825385570526,
"logits/rejected": -0.833530843257904,
"logps/chosen": -1.6525256633758545,
"logps/rejected": -1.9398300647735596,
"loss": 1.7645,
"nll_loss": 1.710868239402771,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16525256633758545,
"rewards/margins": 0.028730444610118866,
"rewards/rejected": -0.1939830183982849,
"step": 287
},
{
"epoch": 0.45390070921985815,
"grad_norm": 0.23629866540431976,
"learning_rate": 3.809700913224726e-06,
"log_odds_chosen": 0.36637431383132935,
"log_odds_ratio": -0.5327513217926025,
"logits/chosen": 0.15463852882385254,
"logits/rejected": -1.287902593612671,
"logps/chosen": -1.5792927742004395,
"logps/rejected": -1.8798011541366577,
"loss": 1.7056,
"nll_loss": 1.6522890329360962,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15792928636074066,
"rewards/margins": 0.03005082532763481,
"rewards/rejected": -0.18798011541366577,
"step": 288
},
{
"epoch": 0.45547675334909377,
"grad_norm": 0.23628529906272888,
"learning_rate": 3.8073538025371494e-06,
"log_odds_chosen": 0.4771001935005188,
"log_odds_ratio": -0.48583680391311646,
"logits/chosen": 0.2631601393222809,
"logits/rejected": -1.0782270431518555,
"logps/chosen": -1.551235318183899,
"logps/rejected": -1.9436652660369873,
"loss": 1.6886,
"nll_loss": 1.6399903297424316,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1551235318183899,
"rewards/margins": 0.03924299776554108,
"rewards/rejected": -0.19436652958393097,
"step": 289
},
{
"epoch": 0.4570527974783294,
"grad_norm": 0.2573354244232178,
"learning_rate": 3.804993038154852e-06,
"log_odds_chosen": 0.3571556806564331,
"log_odds_ratio": -0.5380254983901978,
"logits/chosen": 0.18856900930404663,
"logits/rejected": -0.9549089670181274,
"logps/chosen": -1.5852500200271606,
"logps/rejected": -1.88179349899292,
"loss": 1.7309,
"nll_loss": 1.6771280765533447,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1585249900817871,
"rewards/margins": 0.029654357582330704,
"rewards/rejected": -0.1881793588399887,
"step": 290
},
{
"epoch": 0.458628841607565,
"grad_norm": 0.2339908331632614,
"learning_rate": 3.8026186379122816e-06,
"log_odds_chosen": 0.2612074613571167,
"log_odds_ratio": -0.5736123323440552,
"logits/chosen": 0.1575014591217041,
"logits/rejected": -0.9461207985877991,
"logps/chosen": -1.6632664203643799,
"logps/rejected": -1.8798828125,
"loss": 1.7845,
"nll_loss": 1.72710120677948,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16632665693759918,
"rewards/margins": 0.02166163921356201,
"rewards/rejected": -0.18798828125,
"step": 291
},
{
"epoch": 0.46020488573680063,
"grad_norm": 0.232927143573761,
"learning_rate": 3.8002306197468983e-06,
"log_odds_chosen": 0.16624774038791656,
"log_odds_ratio": -0.6159506440162659,
"logits/chosen": 0.15579693019390106,
"logits/rejected": -1.0309805870056152,
"logps/chosen": -1.6420388221740723,
"logps/rejected": -1.777790904045105,
"loss": 1.7816,
"nll_loss": 1.7199894189834595,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.16420388221740723,
"rewards/margins": 0.013575192540884018,
"rewards/rejected": -0.17777907848358154,
"step": 292
},
{
"epoch": 0.46178092986603625,
"grad_norm": 0.26111435890197754,
"learning_rate": 3.7978290016990367e-06,
"log_odds_chosen": 0.4844040274620056,
"log_odds_ratio": -0.4840275049209595,
"logits/chosen": 0.12877169251441956,
"logits/rejected": -1.0049679279327393,
"logps/chosen": -1.6433277130126953,
"logps/rejected": -2.0521345138549805,
"loss": 1.7643,
"nll_loss": 1.7158782482147217,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16433276236057281,
"rewards/margins": 0.040880680084228516,
"rewards/rejected": -0.20521345734596252,
"step": 293
},
{
"epoch": 0.46335697399527187,
"grad_norm": 0.2526698112487793,
"learning_rate": 3.795413801911776e-06,
"log_odds_chosen": 0.5316009521484375,
"log_odds_ratio": -0.46823519468307495,
"logits/chosen": 0.1174880787730217,
"logits/rejected": -1.0741727352142334,
"logps/chosen": -1.5681180953979492,
"logps/rejected": -2.0108184814453125,
"loss": 1.7117,
"nll_loss": 1.6648805141448975,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15681181848049164,
"rewards/margins": 0.04427003860473633,
"rewards/rejected": -0.20108187198638916,
"step": 294
},
{
"epoch": 0.4649330181245075,
"grad_norm": 0.26307472586631775,
"learning_rate": 3.7929850386307965e-06,
"log_odds_chosen": 0.32403603196144104,
"log_odds_ratio": -0.549573540687561,
"logits/chosen": 0.2209596186876297,
"logits/rejected": -0.8648887872695923,
"logps/chosen": -1.775384783744812,
"logps/rejected": -2.0520851612091064,
"loss": 1.8883,
"nll_loss": 1.8333498239517212,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17753848433494568,
"rewards/margins": 0.02767005003988743,
"rewards/rejected": -0.20520853996276855,
"step": 295
},
{
"epoch": 0.4665090622537431,
"grad_norm": 0.24105559289455414,
"learning_rate": 3.790542730204245e-06,
"log_odds_chosen": 0.391379177570343,
"log_odds_ratio": -0.5190234184265137,
"logits/chosen": 0.17468759417533875,
"logits/rejected": -1.050643801689148,
"logps/chosen": -1.705775260925293,
"logps/rejected": -2.037628412246704,
"loss": 1.8141,
"nll_loss": 1.7621614933013916,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1705775409936905,
"rewards/margins": 0.033185333013534546,
"rewards/rejected": -0.20376285910606384,
"step": 296
},
{
"epoch": 0.46808510638297873,
"grad_norm": 0.2376202493906021,
"learning_rate": 3.7880868950825935e-06,
"log_odds_chosen": 0.40684062242507935,
"log_odds_ratio": -0.510847806930542,
"logits/chosen": 0.13849994540214539,
"logits/rejected": -1.200305700302124,
"logps/chosen": -1.6610805988311768,
"logps/rejected": -2.001574754714966,
"loss": 1.7604,
"nll_loss": 1.7093130350112915,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16610805690288544,
"rewards/margins": 0.034049421548843384,
"rewards/rejected": -0.20015747845172882,
"step": 297
},
{
"epoch": 0.46966115051221435,
"grad_norm": 0.2616525888442993,
"learning_rate": 3.7856175518185058e-06,
"log_odds_chosen": 0.3999539613723755,
"log_odds_ratio": -0.5190368294715881,
"logits/chosen": 0.015357280150055885,
"logits/rejected": -1.1991338729858398,
"logps/chosen": -1.6416277885437012,
"logps/rejected": -1.9745447635650635,
"loss": 1.7682,
"nll_loss": 1.7162597179412842,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1641627848148346,
"rewards/margins": 0.033291686326265335,
"rewards/rejected": -0.19745448231697083,
"step": 298
},
{
"epoch": 0.47123719464145,
"grad_norm": 0.23752158880233765,
"learning_rate": 3.7831347190666883e-06,
"log_odds_chosen": 0.4781672954559326,
"log_odds_ratio": -0.488000750541687,
"logits/chosen": 0.08399657905101776,
"logits/rejected": -1.1212432384490967,
"logps/chosen": -1.6346526145935059,
"logps/rejected": -2.035963773727417,
"loss": 1.7627,
"nll_loss": 1.713881015777588,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16346527636051178,
"rewards/margins": 0.0401310995221138,
"rewards/rejected": -0.20359636843204498,
"step": 299
},
{
"epoch": 0.4728132387706856,
"grad_norm": 0.25213325023651123,
"learning_rate": 3.780638415583759e-06,
"log_odds_chosen": 0.2852388620376587,
"log_odds_ratio": -0.5705251097679138,
"logits/chosen": 0.21283775568008423,
"logits/rejected": -0.935249924659729,
"logps/chosen": -1.6187589168548584,
"logps/rejected": -1.8579552173614502,
"loss": 1.7571,
"nll_loss": 1.7000482082366943,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1618758887052536,
"rewards/margins": 0.023919638246297836,
"rewards/rejected": -0.18579553067684174,
"step": 300
},
{
"epoch": 0.4743892828999212,
"grad_norm": 0.24251310527324677,
"learning_rate": 3.7781286602280967e-06,
"log_odds_chosen": 0.17071868479251862,
"log_odds_ratio": -0.6216680407524109,
"logits/chosen": 0.14868459105491638,
"logits/rejected": -1.1415449380874634,
"logps/chosen": -1.5997627973556519,
"logps/rejected": -1.7335293292999268,
"loss": 1.7454,
"nll_loss": 1.6831833124160767,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1599762886762619,
"rewards/margins": 0.013376658782362938,
"rewards/rejected": -0.1733529418706894,
"step": 301
},
{
"epoch": 0.47596532702915684,
"grad_norm": 0.22967545688152313,
"learning_rate": 3.7756054719597044e-06,
"log_odds_chosen": 0.26898688077926636,
"log_odds_ratio": -0.5931567549705505,
"logits/chosen": 0.033872295171022415,
"logits/rejected": -1.1282514333724976,
"logps/chosen": -1.6345136165618896,
"logps/rejected": -1.8664180040359497,
"loss": 1.7516,
"nll_loss": 1.6922358274459839,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.16345134377479553,
"rewards/margins": 0.023190462961792946,
"rewards/rejected": -0.18664182722568512,
"step": 302
},
{
"epoch": 0.47754137115839246,
"grad_norm": 0.2190561145544052,
"learning_rate": 3.773068869840066e-06,
"log_odds_chosen": 0.19828103482723236,
"log_odds_ratio": -0.6074354648590088,
"logits/chosen": 0.2555277347564697,
"logits/rejected": -1.1666932106018066,
"logps/chosen": -1.6585826873779297,
"logps/rejected": -1.8199939727783203,
"loss": 1.7779,
"nll_loss": 1.7171512842178345,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.16585825383663177,
"rewards/margins": 0.016141142696142197,
"rewards/rejected": -0.18199938535690308,
"step": 303
},
{
"epoch": 0.4791174152876281,
"grad_norm": 0.21842867136001587,
"learning_rate": 3.770518873031997e-06,
"log_odds_chosen": 0.44869300723075867,
"log_odds_ratio": -0.4964669644832611,
"logits/chosen": 0.03254036605358124,
"logits/rejected": -0.8346494436264038,
"logps/chosen": -1.5328748226165771,
"logps/rejected": -1.8997446298599243,
"loss": 1.674,
"nll_loss": 1.6243867874145508,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15328750014305115,
"rewards/margins": 0.03668695688247681,
"rewards/rejected": -0.18997445702552795,
"step": 304
},
{
"epoch": 0.4806934594168637,
"grad_norm": 0.253165602684021,
"learning_rate": 3.7679555007995065e-06,
"log_odds_chosen": 0.41193148493766785,
"log_odds_ratio": -0.5260132551193237,
"logits/chosen": 0.14702853560447693,
"logits/rejected": -1.0447622537612915,
"logps/chosen": -1.6558337211608887,
"logps/rejected": -2.0012173652648926,
"loss": 1.7843,
"nll_loss": 1.731735110282898,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16558335721492767,
"rewards/margins": 0.0345383882522583,
"rewards/rejected": -0.20012176036834717,
"step": 305
},
{
"epoch": 0.48226950354609927,
"grad_norm": 0.24374498426914215,
"learning_rate": 3.7653787725076464e-06,
"log_odds_chosen": 0.2701460123062134,
"log_odds_ratio": -0.5720412731170654,
"logits/chosen": 0.1489667445421219,
"logits/rejected": -0.9628247618675232,
"logps/chosen": -1.6143256425857544,
"logps/rejected": -1.8385647535324097,
"loss": 1.7374,
"nll_loss": 1.680199146270752,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16143256425857544,
"rewards/margins": 0.022423917427659035,
"rewards/rejected": -0.18385647237300873,
"step": 306
},
{
"epoch": 0.4838455476753349,
"grad_norm": 0.2575761675834656,
"learning_rate": 3.7627887076223685e-06,
"log_odds_chosen": 0.3698280155658722,
"log_odds_ratio": -0.5293493270874023,
"logits/chosen": 0.17162802815437317,
"logits/rejected": -0.7795068621635437,
"logps/chosen": -1.6772853136062622,
"logps/rejected": -1.9871926307678223,
"loss": 1.783,
"nll_loss": 1.7300152778625488,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16772854328155518,
"rewards/margins": 0.03099072352051735,
"rewards/rejected": -0.19871927797794342,
"step": 307
},
{
"epoch": 0.4854215918045705,
"grad_norm": 0.2139917016029358,
"learning_rate": 3.7601853257103765e-06,
"log_odds_chosen": 0.22644855082035065,
"log_odds_ratio": -0.5928743481636047,
"logits/chosen": 0.06793497502803802,
"logits/rejected": -1.0903844833374023,
"logps/chosen": -1.582783579826355,
"logps/rejected": -1.7714552879333496,
"loss": 1.6918,
"nll_loss": 1.632529616355896,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15827836096286774,
"rewards/margins": 0.018867187201976776,
"rewards/rejected": -0.17714554071426392,
"step": 308
},
{
"epoch": 0.48699763593380613,
"grad_norm": 0.21651345491409302,
"learning_rate": 3.7575686464389767e-06,
"log_odds_chosen": 0.3462998867034912,
"log_odds_ratio": -0.5444170832633972,
"logits/chosen": 0.10276569426059723,
"logits/rejected": -1.1056041717529297,
"logps/chosen": -1.5598326921463013,
"logps/rejected": -1.8447059392929077,
"loss": 1.6921,
"nll_loss": 1.6376224756240845,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15598325431346893,
"rewards/margins": 0.02848733589053154,
"rewards/rejected": -0.18447057902812958,
"step": 309
},
{
"epoch": 0.48857368006304175,
"grad_norm": 0.2089070826768875,
"learning_rate": 3.7549386895759315e-06,
"log_odds_chosen": 0.38229963183403015,
"log_odds_ratio": -0.5227848291397095,
"logits/chosen": 0.06746693700551987,
"logits/rejected": -1.028963327407837,
"logps/chosen": -1.5149016380310059,
"logps/rejected": -1.8246971368789673,
"loss": 1.6366,
"nll_loss": 1.5843473672866821,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15149016678333282,
"rewards/margins": 0.030979545786976814,
"rewards/rejected": -0.1824697107076645,
"step": 310
},
{
"epoch": 0.49014972419227737,
"grad_norm": 0.20956042408943176,
"learning_rate": 3.7522954749893086e-06,
"log_odds_chosen": 0.5883792042732239,
"log_odds_ratio": -0.45304739475250244,
"logits/chosen": -0.004524541087448597,
"logits/rejected": -1.15907621383667,
"logps/chosen": -1.5441385507583618,
"logps/rejected": -2.0315380096435547,
"loss": 1.6764,
"nll_loss": 1.631089448928833,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1544138640165329,
"rewards/margins": 0.048739951103925705,
"rewards/rejected": -0.2031538188457489,
"step": 311
},
{
"epoch": 0.491725768321513,
"grad_norm": 0.23161104321479797,
"learning_rate": 3.749639022647332e-06,
"log_odds_chosen": 0.35890865325927734,
"log_odds_ratio": -0.5340113639831543,
"logits/chosen": 0.06597714126110077,
"logits/rejected": -1.0388400554656982,
"logps/chosen": -1.6328809261322021,
"logps/rejected": -1.932662010192871,
"loss": 1.7561,
"nll_loss": 1.7027454376220703,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16328811645507812,
"rewards/margins": 0.02997809275984764,
"rewards/rejected": -0.19326619803905487,
"step": 312
},
{
"epoch": 0.4933018124507486,
"grad_norm": 0.20770005881786346,
"learning_rate": 3.7469693526182304e-06,
"log_odds_chosen": 0.513086199760437,
"log_odds_ratio": -0.485725075006485,
"logits/chosen": 0.0205635167658329,
"logits/rejected": -1.4314470291137695,
"logps/chosen": -1.581476092338562,
"logps/rejected": -2.0070858001708984,
"loss": 1.6904,
"nll_loss": 1.6417973041534424,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15814761817455292,
"rewards/margins": 0.04256095737218857,
"rewards/rejected": -0.2007085680961609,
"step": 313
},
{
"epoch": 0.49487785657998423,
"grad_norm": 0.21436423063278198,
"learning_rate": 3.744286485070085e-06,
"log_odds_chosen": 0.5086329579353333,
"log_odds_ratio": -0.47319746017456055,
"logits/chosen": 0.03913354501128197,
"logits/rejected": -1.1685231924057007,
"logps/chosen": -1.5282073020935059,
"logps/rejected": -1.9466618299484253,
"loss": 1.6421,
"nll_loss": 1.594788908958435,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15282073616981506,
"rewards/margins": 0.041845470666885376,
"rewards/rejected": -0.19466620683670044,
"step": 314
},
{
"epoch": 0.49645390070921985,
"grad_norm": 0.23891815543174744,
"learning_rate": 3.7415904402706795e-06,
"log_odds_chosen": 0.4192996621131897,
"log_odds_ratio": -0.514187216758728,
"logits/chosen": 0.06727111339569092,
"logits/rejected": -1.1843537092208862,
"logps/chosen": -1.7258471250534058,
"logps/rejected": -2.082641363143921,
"loss": 1.835,
"nll_loss": 1.7836283445358276,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17258471250534058,
"rewards/margins": 0.03567943722009659,
"rewards/rejected": -0.20826414227485657,
"step": 315
},
{
"epoch": 0.4980299448384555,
"grad_norm": 0.21085584163665771,
"learning_rate": 3.7388812385873435e-06,
"log_odds_chosen": 0.3644852042198181,
"log_odds_ratio": -0.5296192169189453,
"logits/chosen": 0.030755888670682907,
"logits/rejected": -1.189257264137268,
"logps/chosen": -1.5624669790267944,
"logps/rejected": -1.8610371351242065,
"loss": 1.6768,
"nll_loss": 1.6238601207733154,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15624670684337616,
"rewards/margins": 0.02985702082514763,
"rewards/rejected": -0.1861037313938141,
"step": 316
},
{
"epoch": 0.4996059889676911,
"grad_norm": 0.23275373876094818,
"learning_rate": 3.7361589004868033e-06,
"log_odds_chosen": 0.5013652443885803,
"log_odds_ratio": -0.479345440864563,
"logits/chosen": 0.14082355797290802,
"logits/rejected": -1.2382307052612305,
"logps/chosen": -1.6419178247451782,
"logps/rejected": -2.0661187171936035,
"loss": 1.7418,
"nll_loss": 1.6938456296920776,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16419179737567902,
"rewards/margins": 0.042420096695423126,
"rewards/rejected": -0.20661188662052155,
"step": 317
},
{
"epoch": 0.5011820330969267,
"grad_norm": 0.2070867121219635,
"learning_rate": 3.733423446535022e-06,
"log_odds_chosen": 0.5261310338973999,
"log_odds_ratio": -0.47375503182411194,
"logits/chosen": 0.12559077143669128,
"logits/rejected": -1.3008148670196533,
"logps/chosen": -1.634326457977295,
"logps/rejected": -2.0824835300445557,
"loss": 1.7359,
"nll_loss": 1.688564658164978,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16343267261981964,
"rewards/margins": 0.044815681874752045,
"rewards/rejected": -0.20824836194515228,
"step": 318
},
{
"epoch": 0.5027580772261623,
"grad_norm": 0.20222921669483185,
"learning_rate": 3.7306748973970476e-06,
"log_odds_chosen": 0.39474886655807495,
"log_odds_ratio": -0.5250571370124817,
"logits/chosen": 0.03667742758989334,
"logits/rejected": -1.4291538000106812,
"logps/chosen": -1.5048575401306152,
"logps/rejected": -1.823891282081604,
"loss": 1.6483,
"nll_loss": 1.5958224534988403,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15048575401306152,
"rewards/margins": 0.03190337494015694,
"rewards/rejected": -0.18238912522792816,
"step": 319
},
{
"epoch": 0.5043341213553979,
"grad_norm": 0.27620604634284973,
"learning_rate": 3.7279132738368564e-06,
"log_odds_chosen": 0.18474048376083374,
"log_odds_ratio": -0.6092777252197266,
"logits/chosen": 0.1035664826631546,
"logits/rejected": -1.2585889101028442,
"logps/chosen": -1.7019753456115723,
"logps/rejected": -1.8543686866760254,
"loss": 1.8124,
"nll_loss": 1.7514902353286743,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.17019754648208618,
"rewards/margins": 0.015239320695400238,
"rewards/rejected": -0.18543685972690582,
"step": 320
},
{
"epoch": 0.5059101654846335,
"grad_norm": 0.21710968017578125,
"learning_rate": 3.725138596717195e-06,
"log_odds_chosen": 0.44385021924972534,
"log_odds_ratio": -0.4991372227668762,
"logits/chosen": 0.023607883602380753,
"logits/rejected": -1.283747911453247,
"logps/chosen": -1.5559569597244263,
"logps/rejected": -1.9218320846557617,
"loss": 1.6659,
"nll_loss": 1.6159745454788208,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15559569001197815,
"rewards/margins": 0.03658752888441086,
"rewards/rejected": -0.1921832263469696,
"step": 321
},
{
"epoch": 0.5074862096138691,
"grad_norm": 0.20364168286323547,
"learning_rate": 3.7223508869994244e-06,
"log_odds_chosen": 0.4654800295829773,
"log_odds_ratio": -0.4937525987625122,
"logits/chosen": 0.04612987861037254,
"logits/rejected": -1.3605108261108398,
"logps/chosen": -1.6008849143981934,
"logps/rejected": -1.9879066944122314,
"loss": 1.7137,
"nll_loss": 1.6643142700195312,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16008850932121277,
"rewards/margins": 0.038702160120010376,
"rewards/rejected": -0.19879068434238434,
"step": 322
},
{
"epoch": 0.5090622537431048,
"grad_norm": 0.23145321011543274,
"learning_rate": 3.7195501657433594e-06,
"log_odds_chosen": 0.39249351620674133,
"log_odds_ratio": -0.5205338597297668,
"logits/chosen": -0.0046775080263614655,
"logits/rejected": -1.3155083656311035,
"logps/chosen": -1.6052358150482178,
"logps/rejected": -1.9290968179702759,
"loss": 1.7238,
"nll_loss": 1.6717445850372314,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16052357852458954,
"rewards/margins": 0.03238610923290253,
"rewards/rejected": -0.19290968775749207,
"step": 323
},
{
"epoch": 0.5106382978723404,
"grad_norm": 0.19913478195667267,
"learning_rate": 3.716736454107111e-06,
"log_odds_chosen": 0.5085100531578064,
"log_odds_ratio": -0.47588035464286804,
"logits/chosen": 0.002479949500411749,
"logits/rejected": -1.0518862009048462,
"logps/chosen": -1.5002402067184448,
"logps/rejected": -1.9133625030517578,
"loss": 1.6117,
"nll_loss": 1.5641216039657593,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15002401173114777,
"rewards/margins": 0.04131225496530533,
"rewards/rejected": -0.1913362741470337,
"step": 324
},
{
"epoch": 0.512214342001576,
"grad_norm": 0.21510443091392517,
"learning_rate": 3.7139097733469277e-06,
"log_odds_chosen": 0.5286773443222046,
"log_odds_ratio": -0.46826720237731934,
"logits/chosen": 0.02427489310503006,
"logits/rejected": -1.2997654676437378,
"logps/chosen": -1.54619562625885,
"logps/rejected": -1.982952356338501,
"loss": 1.6601,
"nll_loss": 1.6132692098617554,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15461957454681396,
"rewards/margins": 0.04367566481232643,
"rewards/rejected": -0.1982952207326889,
"step": 325
},
{
"epoch": 0.5137903861308116,
"grad_norm": 0.22013559937477112,
"learning_rate": 3.711070144817032e-06,
"log_odds_chosen": 0.4152149558067322,
"log_odds_ratio": -0.512770414352417,
"logits/chosen": 0.005936339497566223,
"logits/rejected": -1.122101068496704,
"logps/chosen": -1.633888840675354,
"logps/rejected": -1.9839468002319336,
"loss": 1.754,
"nll_loss": 1.7027238607406616,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16338886320590973,
"rewards/margins": 0.03500579297542572,
"rewards/rejected": -0.19839467108249664,
"step": 326
},
{
"epoch": 0.5153664302600472,
"grad_norm": 0.24476923048496246,
"learning_rate": 3.708217589969461e-06,
"log_odds_chosen": 0.5117456912994385,
"log_odds_ratio": -0.4778427481651306,
"logits/chosen": -0.02480306476354599,
"logits/rejected": -1.3842414617538452,
"logps/chosen": -1.6081162691116333,
"logps/rejected": -2.0358729362487793,
"loss": 1.7307,
"nll_loss": 1.682942509651184,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.160811647772789,
"rewards/margins": 0.042775679379701614,
"rewards/rejected": -0.20358730852603912,
"step": 327
},
{
"epoch": 0.5169424743892829,
"grad_norm": 0.2023211419582367,
"learning_rate": 3.705352130353904e-06,
"log_odds_chosen": 0.4642333984375,
"log_odds_ratio": -0.4990030825138092,
"logits/chosen": 0.04158155620098114,
"logits/rejected": -1.2979916334152222,
"logps/chosen": -1.5249441862106323,
"logps/rejected": -1.9054869413375854,
"loss": 1.6543,
"nll_loss": 1.6043576002120972,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.152494415640831,
"rewards/margins": 0.038054272532463074,
"rewards/rejected": -0.19054868817329407,
"step": 328
},
{
"epoch": 0.5185185185185185,
"grad_norm": 0.2157369703054428,
"learning_rate": 3.7024737876175404e-06,
"log_odds_chosen": 0.4267793893814087,
"log_odds_ratio": -0.5041587352752686,
"logits/chosen": -0.02120812237262726,
"logits/rejected": -1.1328731775283813,
"logps/chosen": -1.5772178173065186,
"logps/rejected": -1.9269858598709106,
"loss": 1.683,
"nll_loss": 1.6326076984405518,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15772177278995514,
"rewards/margins": 0.03497680649161339,
"rewards/rejected": -0.19269859790802002,
"step": 329
},
{
"epoch": 0.5200945626477541,
"grad_norm": 0.22613677382469177,
"learning_rate": 3.699582583504874e-06,
"log_odds_chosen": 0.2962914705276489,
"log_odds_ratio": -0.5632017254829407,
"logits/chosen": 0.033639900386333466,
"logits/rejected": -1.3186163902282715,
"logps/chosen": -1.7170500755310059,
"logps/rejected": -1.966399908065796,
"loss": 1.815,
"nll_loss": 1.7586567401885986,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17170500755310059,
"rewards/margins": 0.024934988468885422,
"rewards/rejected": -0.1966399997472763,
"step": 330
},
{
"epoch": 0.5216706067769897,
"grad_norm": 0.20476558804512024,
"learning_rate": 3.696678539857571e-06,
"log_odds_chosen": 0.40065276622772217,
"log_odds_ratio": -0.5229502320289612,
"logits/chosen": -0.00485864095389843,
"logits/rejected": -1.3043968677520752,
"logps/chosen": -1.6962597370147705,
"logps/rejected": -2.0372161865234375,
"loss": 1.8021,
"nll_loss": 1.7498358488082886,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16962596774101257,
"rewards/margins": 0.034095652401447296,
"rewards/rejected": -0.20372162759304047,
"step": 331
},
{
"epoch": 0.5232466509062254,
"grad_norm": 0.20712663233280182,
"learning_rate": 3.6937616786142956e-06,
"log_odds_chosen": 0.5233447551727295,
"log_odds_ratio": -0.47018271684646606,
"logits/chosen": 0.010211546905338764,
"logits/rejected": -1.0574318170547485,
"logps/chosen": -1.5134122371673584,
"logps/rejected": -1.9436025619506836,
"loss": 1.6422,
"nll_loss": 1.595203161239624,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15134122967720032,
"rewards/margins": 0.04301903396844864,
"rewards/rejected": -0.19436024129390717,
"step": 332
},
{
"epoch": 0.524822695035461,
"grad_norm": 0.18907472491264343,
"learning_rate": 3.6908320218105393e-06,
"log_odds_chosen": 0.3291381895542145,
"log_odds_ratio": -0.5515700578689575,
"logits/chosen": -0.01022535003721714,
"logits/rejected": -1.2733728885650635,
"logps/chosen": -1.5107743740081787,
"logps/rejected": -1.7811356782913208,
"loss": 1.6236,
"nll_loss": 1.5684043169021606,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15107741951942444,
"rewards/margins": 0.027036139741539955,
"rewards/rejected": -0.17811356484889984,
"step": 333
},
{
"epoch": 0.5263987391646966,
"grad_norm": 0.19910985231399536,
"learning_rate": 3.6878895915784607e-06,
"log_odds_chosen": 0.48929572105407715,
"log_odds_ratio": -0.483676016330719,
"logits/chosen": 0.03388974070549011,
"logits/rejected": -1.230672836303711,
"logps/chosen": -1.4776809215545654,
"logps/rejected": -1.8740699291229248,
"loss": 1.5862,
"nll_loss": 1.5378473997116089,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14776809513568878,
"rewards/margins": 0.03963891416788101,
"rewards/rejected": -0.187406986951828,
"step": 334
},
{
"epoch": 0.5279747832939322,
"grad_norm": 0.19112317264080048,
"learning_rate": 3.6849344101467147e-06,
"log_odds_chosen": 0.3993302285671234,
"log_odds_ratio": -0.5199868679046631,
"logits/chosen": 0.02075035311281681,
"logits/rejected": -1.4165470600128174,
"logps/chosen": -1.5604709386825562,
"logps/rejected": -1.890925407409668,
"loss": 1.6795,
"nll_loss": 1.6274938583374023,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15604707598686218,
"rewards/margins": 0.03304546698927879,
"rewards/rejected": -0.18909254670143127,
"step": 335
},
{
"epoch": 0.5295508274231678,
"grad_norm": 0.2065410017967224,
"learning_rate": 3.6819664998402857e-06,
"log_odds_chosen": 0.3870427906513214,
"log_odds_ratio": -0.521834671497345,
"logits/chosen": 0.007402241230010986,
"logits/rejected": -1.2406339645385742,
"logps/chosen": -1.597013235092163,
"logps/rejected": -1.9189107418060303,
"loss": 1.7091,
"nll_loss": 1.65691339969635,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15970134735107422,
"rewards/margins": 0.032189756631851196,
"rewards/rejected": -0.19189107418060303,
"step": 336
},
{
"epoch": 0.5311268715524035,
"grad_norm": 0.22015894949436188,
"learning_rate": 3.6789858830803186e-06,
"log_odds_chosen": 0.4236854314804077,
"log_odds_ratio": -0.5088356733322144,
"logits/chosen": 0.009572651237249374,
"logits/rejected": -1.2554011344909668,
"logps/chosen": -1.649095058441162,
"logps/rejected": -2.0025815963745117,
"loss": 1.7505,
"nll_loss": 1.6996192932128906,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16490954160690308,
"rewards/margins": 0.035348646342754364,
"rewards/rejected": -0.20025816559791565,
"step": 337
},
{
"epoch": 0.5327029156816391,
"grad_norm": 0.20396317541599274,
"learning_rate": 3.6759925823839486e-06,
"log_odds_chosen": 0.3307921886444092,
"log_odds_ratio": -0.5461194515228271,
"logits/chosen": -0.017404936254024506,
"logits/rejected": -1.1315640211105347,
"logps/chosen": -1.5301204919815063,
"logps/rejected": -1.7987494468688965,
"loss": 1.6634,
"nll_loss": 1.6087586879730225,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15301203727722168,
"rewards/margins": 0.02686290442943573,
"rewards/rejected": -0.1798749417066574,
"step": 338
},
{
"epoch": 0.5342789598108747,
"grad_norm": 0.19404453039169312,
"learning_rate": 3.672986620364134e-06,
"log_odds_chosen": 0.4503750205039978,
"log_odds_ratio": -0.49645406007766724,
"logits/chosen": 0.02146240696310997,
"logits/rejected": -1.14137601852417,
"logps/chosen": -1.6056911945343018,
"logps/rejected": -1.9799296855926514,
"loss": 1.7114,
"nll_loss": 1.6618030071258545,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16056913137435913,
"rewards/margins": 0.03742384910583496,
"rewards/rejected": -0.1979929804801941,
"step": 339
},
{
"epoch": 0.5358550039401103,
"grad_norm": 0.19405515491962433,
"learning_rate": 3.669968019729481e-06,
"log_odds_chosen": 0.5453786849975586,
"log_odds_ratio": -0.46696317195892334,
"logits/chosen": -0.03269782289862633,
"logits/rejected": -1.4044153690338135,
"logps/chosen": -1.4841482639312744,
"logps/rejected": -1.9323790073394775,
"loss": 1.5994,
"nll_loss": 1.5526580810546875,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14841482043266296,
"rewards/margins": 0.04482308030128479,
"rewards/rejected": -0.19323790073394775,
"step": 340
},
{
"epoch": 0.5374310480693459,
"grad_norm": 0.20472969114780426,
"learning_rate": 3.666936803284076e-06,
"log_odds_chosen": 0.47615405917167664,
"log_odds_ratio": -0.4855659306049347,
"logits/chosen": 0.01596236228942871,
"logits/rejected": -1.2273820638656616,
"logps/chosen": -1.6264029741287231,
"logps/rejected": -2.0267691612243652,
"loss": 1.7196,
"nll_loss": 1.6710734367370605,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1626402884721756,
"rewards/margins": 0.0400366336107254,
"rewards/rejected": -0.202676922082901,
"step": 341
},
{
"epoch": 0.5390070921985816,
"grad_norm": 0.21434363722801208,
"learning_rate": 3.663892993927312e-06,
"log_odds_chosen": 0.5617402791976929,
"log_odds_ratio": -0.4550952911376953,
"logits/chosen": -0.059917159378528595,
"logits/rejected": -1.359694004058838,
"logps/chosen": -1.575050950050354,
"logps/rejected": -2.043088674545288,
"loss": 1.6888,
"nll_loss": 1.643282175064087,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1575051099061966,
"rewards/margins": 0.046803757548332214,
"rewards/rejected": -0.20430885255336761,
"step": 342
},
{
"epoch": 0.5405831363278172,
"grad_norm": 0.20112687349319458,
"learning_rate": 3.6608366146537136e-06,
"log_odds_chosen": 0.6060886383056641,
"log_odds_ratio": -0.45338305830955505,
"logits/chosen": -0.08892233669757843,
"logits/rejected": -1.231791377067566,
"logps/chosen": -1.5024844408035278,
"logps/rejected": -2.009472608566284,
"loss": 1.6204,
"nll_loss": 1.575110912322998,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1502484530210495,
"rewards/margins": 0.05069882422685623,
"rewards/rejected": -0.20094728469848633,
"step": 343
},
{
"epoch": 0.5421591804570528,
"grad_norm": 0.20276731252670288,
"learning_rate": 3.6577676885527674e-06,
"log_odds_chosen": 0.47172704339027405,
"log_odds_ratio": -0.4901059567928314,
"logits/chosen": -0.064692422747612,
"logits/rejected": -1.493216633796692,
"logps/chosen": -1.5645720958709717,
"logps/rejected": -1.954676866531372,
"loss": 1.664,
"nll_loss": 1.6149814128875732,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15645720064640045,
"rewards/margins": 0.03901049122214317,
"rewards/rejected": -0.19546771049499512,
"step": 344
},
{
"epoch": 0.5437352245862884,
"grad_norm": 0.19876694679260254,
"learning_rate": 3.654686238808744e-06,
"log_odds_chosen": 0.4601633548736572,
"log_odds_ratio": -0.49714383482933044,
"logits/chosen": -0.09634008258581161,
"logits/rejected": -1.3992743492126465,
"logps/chosen": -1.657270908355713,
"logps/rejected": -2.0440926551818848,
"loss": 1.7581,
"nll_loss": 1.7084193229675293,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16572707891464233,
"rewards/margins": 0.03868217021226883,
"rewards/rejected": -0.20440925657749176,
"step": 345
},
{
"epoch": 0.545311268715524,
"grad_norm": 0.18044152855873108,
"learning_rate": 3.6515922887005245e-06,
"log_odds_chosen": 0.6581941843032837,
"log_odds_ratio": -0.42069223523139954,
"logits/chosen": -0.19447211921215057,
"logits/rejected": -1.4484854936599731,
"logps/chosen": -1.4948885440826416,
"logps/rejected": -2.034649133682251,
"loss": 1.6052,
"nll_loss": 1.563119888305664,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14948883652687073,
"rewards/margins": 0.053976062685251236,
"rewards/rejected": -0.20346491038799286,
"step": 346
},
{
"epoch": 0.5468873128447597,
"grad_norm": 0.1821555495262146,
"learning_rate": 3.6484858616014236e-06,
"log_odds_chosen": 0.29319724440574646,
"log_odds_ratio": -0.5593742728233337,
"logits/chosen": -0.02489875629544258,
"logits/rejected": -1.4301215410232544,
"logps/chosen": -1.6183066368103027,
"logps/rejected": -1.8595659732818604,
"loss": 1.7166,
"nll_loss": 1.6606316566467285,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16183066368103027,
"rewards/margins": 0.024125942960381508,
"rewards/rejected": -0.18595659732818604,
"step": 347
},
{
"epoch": 0.5484633569739953,
"grad_norm": 0.20152583718299866,
"learning_rate": 3.6453669809790154e-06,
"log_odds_chosen": 0.34444230794906616,
"log_odds_ratio": -0.5457963943481445,
"logits/chosen": 0.003659643232822418,
"logits/rejected": -1.1108278036117554,
"logps/chosen": -1.548649549484253,
"logps/rejected": -1.8318045139312744,
"loss": 1.6556,
"nll_loss": 1.6010490655899048,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15486496686935425,
"rewards/margins": 0.02831549569964409,
"rewards/rejected": -0.18318045139312744,
"step": 348
},
{
"epoch": 0.5500394011032309,
"grad_norm": 0.19325992465019226,
"learning_rate": 3.642235670394952e-06,
"log_odds_chosen": 0.43656274676322937,
"log_odds_ratio": -0.5032880902290344,
"logits/chosen": -0.02493971772491932,
"logits/rejected": -1.0908689498901367,
"logps/chosen": -1.5585966110229492,
"logps/rejected": -1.9134851694107056,
"loss": 1.6744,
"nll_loss": 1.6240770816802979,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15585967898368835,
"rewards/margins": 0.035488829016685486,
"rewards/rejected": -0.19134849309921265,
"step": 349
},
{
"epoch": 0.5516154452324665,
"grad_norm": 0.19222001731395721,
"learning_rate": 3.63909195350479e-06,
"log_odds_chosen": 0.41014277935028076,
"log_odds_ratio": -0.5183743238449097,
"logits/chosen": 0.04076732322573662,
"logits/rejected": -1.2510112524032593,
"logps/chosen": -1.6457445621490479,
"logps/rejected": -1.9860320091247559,
"loss": 1.7474,
"nll_loss": 1.6955746412277222,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16457447409629822,
"rewards/margins": 0.03402874246239662,
"rewards/rejected": -0.19860321283340454,
"step": 350
},
{
"epoch": 0.5531914893617021,
"grad_norm": 0.19920876622200012,
"learning_rate": 3.635935854057809e-06,
"log_odds_chosen": 0.3992159962654114,
"log_odds_ratio": -0.5221322774887085,
"logits/chosen": -0.14892584085464478,
"logits/rejected": -0.9987085461616516,
"logps/chosen": -1.5974595546722412,
"logps/rejected": -1.9293639659881592,
"loss": 1.7044,
"nll_loss": 1.6521823406219482,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1597459614276886,
"rewards/margins": 0.03319043666124344,
"rewards/rejected": -0.19293639063835144,
"step": 351
},
{
"epoch": 0.5547675334909378,
"grad_norm": 0.19184619188308716,
"learning_rate": 3.6327673958968327e-06,
"log_odds_chosen": 0.5156201720237732,
"log_odds_ratio": -0.4716494679450989,
"logits/chosen": -0.10240821540355682,
"logits/rejected": -1.2988749742507935,
"logps/chosen": -1.4805288314819336,
"logps/rejected": -1.8984272480010986,
"loss": 1.6073,
"nll_loss": 1.5601266622543335,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1480528861284256,
"rewards/margins": 0.04178984463214874,
"rewards/rejected": -0.18984271585941315,
"step": 352
},
{
"epoch": 0.5563435776201734,
"grad_norm": 0.17695419490337372,
"learning_rate": 3.6295866029580483e-06,
"log_odds_chosen": 0.49165239930152893,
"log_odds_ratio": -0.4805663824081421,
"logits/chosen": -0.09917198866605759,
"logits/rejected": -1.3583862781524658,
"logps/chosen": -1.5266773700714111,
"logps/rejected": -1.9281859397888184,
"loss": 1.623,
"nll_loss": 1.5749820470809937,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15266773104667664,
"rewards/margins": 0.04015086218714714,
"rewards/rejected": -0.19281861186027527,
"step": 353
},
{
"epoch": 0.557919621749409,
"grad_norm": 0.18373258411884308,
"learning_rate": 3.626393499270829e-06,
"log_odds_chosen": 0.45253658294677734,
"log_odds_ratio": -0.5001296997070312,
"logits/chosen": -0.09345138818025589,
"logits/rejected": -1.326229453086853,
"logps/chosen": -1.4635684490203857,
"logps/rejected": -1.8293596506118774,
"loss": 1.5919,
"nll_loss": 1.541857361793518,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14635683596134186,
"rewards/margins": 0.03657911717891693,
"rewards/rejected": -0.18293596804141998,
"step": 354
},
{
"epoch": 0.5594956658786446,
"grad_norm": 0.19679243862628937,
"learning_rate": 3.6231881089575466e-06,
"log_odds_chosen": 0.6074280142784119,
"log_odds_ratio": -0.43782979249954224,
"logits/chosen": -0.16901959478855133,
"logits/rejected": -1.4631741046905518,
"logps/chosen": -1.5110644102096558,
"logps/rejected": -2.0135817527770996,
"loss": 1.5938,
"nll_loss": 1.550012230873108,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15110644698143005,
"rewards/margins": 0.05025171861052513,
"rewards/rejected": -0.2013581544160843,
"step": 355
},
{
"epoch": 0.5610717100078803,
"grad_norm": 0.2090955525636673,
"learning_rate": 3.6199704562333945e-06,
"log_odds_chosen": 0.4590110778808594,
"log_odds_ratio": -0.4973413050174713,
"logits/chosen": -0.13931547105312347,
"logits/rejected": -1.3538612127304077,
"logps/chosen": -1.4821869134902954,
"logps/rejected": -1.8512073755264282,
"loss": 1.5913,
"nll_loss": 1.5415163040161133,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14821870625019073,
"rewards/margins": 0.03690203279256821,
"rewards/rejected": -0.18512074649333954,
"step": 356
},
{
"epoch": 0.5626477541371159,
"grad_norm": 0.20289309322834015,
"learning_rate": 3.6167405654062024e-06,
"log_odds_chosen": 0.4794601798057556,
"log_odds_ratio": -0.49567416310310364,
"logits/chosen": -0.07716728746891022,
"logits/rejected": -1.2874313592910767,
"logps/chosen": -1.541725516319275,
"logps/rejected": -1.9342597723007202,
"loss": 1.6395,
"nll_loss": 1.589914321899414,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15417256951332092,
"rewards/margins": 0.03925342112779617,
"rewards/rejected": -0.1934259682893753,
"step": 357
},
{
"epoch": 0.5642237982663515,
"grad_norm": 0.1778980940580368,
"learning_rate": 3.6134984608762515e-06,
"log_odds_chosen": 0.5081315636634827,
"log_odds_ratio": -0.4739688038825989,
"logits/chosen": -0.08499579131603241,
"logits/rejected": -1.3757938146591187,
"logps/chosen": -1.4640223979949951,
"logps/rejected": -1.8742594718933105,
"loss": 1.5833,
"nll_loss": 1.5359253883361816,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1464022397994995,
"rewards/margins": 0.041023701429367065,
"rewards/rejected": -0.18742592632770538,
"step": 358
},
{
"epoch": 0.5657998423955871,
"grad_norm": 0.2062489539384842,
"learning_rate": 3.6102441671360945e-06,
"log_odds_chosen": 0.4876347780227661,
"log_odds_ratio": -0.48857545852661133,
"logits/chosen": -0.12350551038980484,
"logits/rejected": -1.2674330472946167,
"logps/chosen": -1.6090378761291504,
"logps/rejected": -2.016418218612671,
"loss": 1.7003,
"nll_loss": 1.651473879814148,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16090378165245056,
"rewards/margins": 0.04073803871870041,
"rewards/rejected": -0.20164184272289276,
"step": 359
},
{
"epoch": 0.5673758865248227,
"grad_norm": 0.19282633066177368,
"learning_rate": 3.6069777087703654e-06,
"log_odds_chosen": 0.5007855296134949,
"log_odds_ratio": -0.48002350330352783,
"logits/chosen": -0.10314866900444031,
"logits/rejected": -1.2184945344924927,
"logps/chosen": -1.533939242362976,
"logps/rejected": -1.9453692436218262,
"loss": 1.6501,
"nll_loss": 1.602098822593689,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1533939242362976,
"rewards/margins": 0.041142985224723816,
"rewards/rejected": -0.19453692436218262,
"step": 360
},
{
"epoch": 0.5689519306540584,
"grad_norm": 0.1999576985836029,
"learning_rate": 3.6036991104555973e-06,
"log_odds_chosen": 0.4542830288410187,
"log_odds_ratio": -0.5044661164283752,
"logits/chosen": -0.08808690309524536,
"logits/rejected": -1.3358922004699707,
"logps/chosen": -1.6718225479125977,
"logps/rejected": -2.0570645332336426,
"loss": 1.7592,
"nll_loss": 1.7087852954864502,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16718226671218872,
"rewards/margins": 0.03852420300245285,
"rewards/rejected": -0.20570647716522217,
"step": 361
},
{
"epoch": 0.570527974783294,
"grad_norm": 0.19183827936649323,
"learning_rate": 3.600408396960034e-06,
"log_odds_chosen": 0.5055266618728638,
"log_odds_ratio": -0.4799540042877197,
"logits/chosen": -0.06735289096832275,
"logits/rejected": -0.9921278953552246,
"logps/chosen": -1.4769415855407715,
"logps/rejected": -1.8809959888458252,
"loss": 1.5713,
"nll_loss": 1.523301601409912,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1476941704750061,
"rewards/margins": 0.04040544107556343,
"rewards/rejected": -0.18809959292411804,
"step": 362
},
{
"epoch": 0.5721040189125296,
"grad_norm": 0.19298037886619568,
"learning_rate": 3.5971055931434447e-06,
"log_odds_chosen": 0.4434927701950073,
"log_odds_ratio": -0.5033714771270752,
"logits/chosen": -0.04916198179125786,
"logits/rejected": -1.233707308769226,
"logps/chosen": -1.5274627208709717,
"logps/rejected": -1.8934059143066406,
"loss": 1.6346,
"nll_loss": 1.5842169523239136,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1527462899684906,
"rewards/margins": 0.03659432381391525,
"rewards/rejected": -0.18934059143066406,
"step": 363
},
{
"epoch": 0.5736800630417652,
"grad_norm": 0.1897989809513092,
"learning_rate": 3.5937907239569343e-06,
"log_odds_chosen": 0.47198542952537537,
"log_odds_ratio": -0.4953606128692627,
"logits/chosen": -0.1299748420715332,
"logits/rejected": -1.3614236116409302,
"logps/chosen": -1.579548954963684,
"logps/rejected": -1.9681391716003418,
"loss": 1.6868,
"nll_loss": 1.6372454166412354,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15795490145683289,
"rewards/margins": 0.038859013468027115,
"rewards/rejected": -0.1968139261007309,
"step": 364
},
{
"epoch": 0.5752561071710008,
"grad_norm": 0.22458185255527496,
"learning_rate": 3.5904638144427572e-06,
"log_odds_chosen": 0.2747001647949219,
"log_odds_ratio": -0.5726035833358765,
"logits/chosen": -0.09000087529420853,
"logits/rejected": -1.1089969873428345,
"logps/chosen": -1.6369290351867676,
"logps/rejected": -1.8634740114212036,
"loss": 1.7333,
"nll_loss": 1.6760823726654053,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1636928915977478,
"rewards/margins": 0.02265450730919838,
"rewards/rejected": -0.18634741008281708,
"step": 365
},
{
"epoch": 0.5768321513002365,
"grad_norm": 0.2086506485939026,
"learning_rate": 3.5871248897341246e-06,
"log_odds_chosen": 0.5135898590087891,
"log_odds_ratio": -0.4752338230609894,
"logits/chosen": -0.12942443788051605,
"logits/rejected": -1.0795626640319824,
"logps/chosen": -1.4760322570800781,
"logps/rejected": -1.8992903232574463,
"loss": 1.598,
"nll_loss": 1.5504556894302368,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14760322868824005,
"rewards/margins": 0.042325813323259354,
"rewards/rejected": -0.1899290531873703,
"step": 366
},
{
"epoch": 0.578408195429472,
"grad_norm": 0.18183429539203644,
"learning_rate": 3.5837739750550182e-06,
"log_odds_chosen": 0.4922761619091034,
"log_odds_ratio": -0.4857975244522095,
"logits/chosen": -0.15072286128997803,
"logits/rejected": -1.6683162450790405,
"logps/chosen": -1.5550795793533325,
"logps/rejected": -1.9653193950653076,
"loss": 1.6554,
"nll_loss": 1.6068187952041626,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15550795197486877,
"rewards/margins": 0.04102398827672005,
"rewards/rejected": -0.19653193652629852,
"step": 367
},
{
"epoch": 0.5799842395587076,
"grad_norm": 0.1786677986383438,
"learning_rate": 3.5804110957199977e-06,
"log_odds_chosen": 0.5304347276687622,
"log_odds_ratio": -0.4684828817844391,
"logits/chosen": -0.031741030514240265,
"logits/rejected": -1.2201794385910034,
"logps/chosen": -1.533761739730835,
"logps/rejected": -1.97342848777771,
"loss": 1.6311,
"nll_loss": 1.584226369857788,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15337617695331573,
"rewards/margins": 0.04396669566631317,
"rewards/rejected": -0.1973428726196289,
"step": 368
},
{
"epoch": 0.5815602836879432,
"grad_norm": 0.19358626008033752,
"learning_rate": 3.577036277134011e-06,
"log_odds_chosen": 0.6033509373664856,
"log_odds_ratio": -0.44030019640922546,
"logits/chosen": -0.12189745157957077,
"logits/rejected": -1.4489309787750244,
"logps/chosen": -1.5189951658248901,
"logps/rejected": -2.0174660682678223,
"loss": 1.6305,
"nll_loss": 1.5864982604980469,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1518995314836502,
"rewards/margins": 0.04984709620475769,
"rewards/rejected": -0.2017466127872467,
"step": 369
},
{
"epoch": 0.5831363278171788,
"grad_norm": 0.20722126960754395,
"learning_rate": 3.5736495447922e-06,
"log_odds_chosen": 0.38122087717056274,
"log_odds_ratio": -0.5253958106040955,
"logits/chosen": -0.11854588240385056,
"logits/rejected": -1.445725440979004,
"logps/chosen": -1.643795371055603,
"logps/rejected": -1.960480809211731,
"loss": 1.7461,
"nll_loss": 1.693605899810791,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1643795371055603,
"rewards/margins": 0.031668562442064285,
"rewards/rejected": -0.19604811072349548,
"step": 370
},
{
"epoch": 0.5847123719464145,
"grad_norm": 0.1782001107931137,
"learning_rate": 3.5702509242797096e-06,
"log_odds_chosen": 0.7012959718704224,
"log_odds_ratio": -0.41842737793922424,
"logits/chosen": -0.13191047310829163,
"logits/rejected": -1.4372180700302124,
"logps/chosen": -1.4734841585159302,
"logps/rejected": -2.0532171726226807,
"loss": 1.5859,
"nll_loss": 1.5440880060195923,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14734841883182526,
"rewards/margins": 0.05797329545021057,
"rewards/rejected": -0.20532171428203583,
"step": 371
},
{
"epoch": 0.5862884160756501,
"grad_norm": 0.189020574092865,
"learning_rate": 3.566840441271495e-06,
"log_odds_chosen": 0.663593053817749,
"log_odds_ratio": -0.43108314275741577,
"logits/chosen": -0.09653455764055252,
"logits/rejected": -1.4999427795410156,
"logps/chosen": -1.4899102449417114,
"logps/rejected": -2.0343017578125,
"loss": 1.578,
"nll_loss": 1.5348646640777588,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14899101853370667,
"rewards/margins": 0.05443914607167244,
"rewards/rejected": -0.2034301608800888,
"step": 372
},
{
"epoch": 0.5878644602048857,
"grad_norm": 0.1943143755197525,
"learning_rate": 3.5634181215321265e-06,
"log_odds_chosen": 0.6287661790847778,
"log_odds_ratio": -0.43060097098350525,
"logits/chosen": -0.07774099707603455,
"logits/rejected": -1.3174422979354858,
"logps/chosen": -1.5436866283416748,
"logps/rejected": -2.067762613296509,
"loss": 1.6544,
"nll_loss": 1.6113276481628418,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15436868369579315,
"rewards/margins": 0.052407585084438324,
"rewards/rejected": -0.20677624642848969,
"step": 373
},
{
"epoch": 0.5894405043341213,
"grad_norm": 0.1871887594461441,
"learning_rate": 3.5599839909155947e-06,
"log_odds_chosen": 0.49737420678138733,
"log_odds_ratio": -0.4843059182167053,
"logits/chosen": -0.10240314900875092,
"logits/rejected": -1.391071081161499,
"logps/chosen": -1.5068795680999756,
"logps/rejected": -1.9151724576950073,
"loss": 1.6079,
"nll_loss": 1.5594788789749146,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15068796277046204,
"rewards/margins": 0.04082927852869034,
"rewards/rejected": -0.19151723384857178,
"step": 374
},
{
"epoch": 0.5910165484633569,
"grad_norm": 0.1926075965166092,
"learning_rate": 3.556538075365116e-06,
"log_odds_chosen": 0.5647552013397217,
"log_odds_ratio": -0.4604189991950989,
"logits/chosen": -0.05626612901687622,
"logits/rejected": -1.3970637321472168,
"logps/chosen": -1.5002104043960571,
"logps/rejected": -1.962066411972046,
"loss": 1.6008,
"nll_loss": 1.5547971725463867,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1500210165977478,
"rewards/margins": 0.046185605227947235,
"rewards/rejected": -0.19620664417743683,
"step": 375
},
{
"epoch": 0.5925925925925926,
"grad_norm": 0.20022441446781158,
"learning_rate": 3.5530804009129367e-06,
"log_odds_chosen": 0.5052685737609863,
"log_odds_ratio": -0.4745987355709076,
"logits/chosen": -0.11242527514696121,
"logits/rejected": -1.191476583480835,
"logps/chosen": -1.5387141704559326,
"logps/rejected": -1.9558299779891968,
"loss": 1.6266,
"nll_loss": 1.5791561603546143,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15387141704559326,
"rewards/margins": 0.041711580008268356,
"rewards/rejected": -0.19558300077915192,
"step": 376
},
{
"epoch": 0.5941686367218282,
"grad_norm": 0.1937384456396103,
"learning_rate": 3.5496109936801368e-06,
"log_odds_chosen": 0.49315646290779114,
"log_odds_ratio": -0.48415568470954895,
"logits/chosen": -0.17054055631160736,
"logits/rejected": -1.4667962789535522,
"logps/chosen": -1.5606714487075806,
"logps/rejected": -1.9743766784667969,
"loss": 1.6571,
"nll_loss": 1.6086863279342651,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1560671329498291,
"rewards/margins": 0.04137052595615387,
"rewards/rejected": -0.19743765890598297,
"step": 377
},
{
"epoch": 0.5957446808510638,
"grad_norm": 0.20795175433158875,
"learning_rate": 3.546129879876429e-06,
"log_odds_chosen": 0.3628261685371399,
"log_odds_ratio": -0.5408048629760742,
"logits/chosen": -0.05689922347664833,
"logits/rejected": -1.130873203277588,
"logps/chosen": -1.6209430694580078,
"logps/rejected": -1.9227240085601807,
"loss": 1.7149,
"nll_loss": 1.6608681678771973,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1620943248271942,
"rewards/margins": 0.03017808124423027,
"rewards/rejected": -0.19227240979671478,
"step": 378
},
{
"epoch": 0.5973207249802994,
"grad_norm": 0.19484388828277588,
"learning_rate": 3.5426370857999662e-06,
"log_odds_chosen": 0.3801731467247009,
"log_odds_ratio": -0.5251167416572571,
"logits/chosen": -0.10485132038593292,
"logits/rejected": -1.2549301385879517,
"logps/chosen": -1.6024380922317505,
"logps/rejected": -1.9180117845535278,
"loss": 1.7071,
"nll_loss": 1.6545681953430176,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16024382412433624,
"rewards/margins": 0.03155737742781639,
"rewards/rejected": -0.19180117547512054,
"step": 379
},
{
"epoch": 0.598896769109535,
"grad_norm": 0.19403113424777985,
"learning_rate": 3.53913263783714e-06,
"log_odds_chosen": 0.4171554744243622,
"log_odds_ratio": -0.5190439820289612,
"logits/chosen": -0.12759403884410858,
"logits/rejected": -1.392691731452942,
"logps/chosen": -1.555301308631897,
"logps/rejected": -1.904205083847046,
"loss": 1.6481,
"nll_loss": 1.5961991548538208,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15553012490272522,
"rewards/margins": 0.03489039093255997,
"rewards/rejected": -0.1904205083847046,
"step": 380
},
{
"epoch": 0.6004728132387707,
"grad_norm": 0.1852397322654724,
"learning_rate": 3.53561656246238e-06,
"log_odds_chosen": 0.6395785808563232,
"log_odds_ratio": -0.42720136046409607,
"logits/chosen": -0.2312593162059784,
"logits/rejected": -1.6027368307113647,
"logps/chosen": -1.5752849578857422,
"logps/rejected": -2.1103007793426514,
"loss": 1.6683,
"nll_loss": 1.6255991458892822,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15752847492694855,
"rewards/margins": 0.05350159481167793,
"rewards/rejected": -0.21103009581565857,
"step": 381
},
{
"epoch": 0.6020488573680063,
"grad_norm": 0.18580475449562073,
"learning_rate": 3.532088886237956e-06,
"log_odds_chosen": 0.539408802986145,
"log_odds_ratio": -0.4628356993198395,
"logits/chosen": -0.14560621976852417,
"logits/rejected": -1.546494483947754,
"logps/chosen": -1.6014196872711182,
"logps/rejected": -2.0526225566864014,
"loss": 1.6812,
"nll_loss": 1.634964942932129,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1601419895887375,
"rewards/margins": 0.045120254158973694,
"rewards/rejected": -0.20526225864887238,
"step": 382
},
{
"epoch": 0.6036249014972419,
"grad_norm": 0.17321372032165527,
"learning_rate": 3.528549635813778e-06,
"log_odds_chosen": 0.5498687028884888,
"log_odds_ratio": -0.46246442198753357,
"logits/chosen": -0.18676355481147766,
"logits/rejected": -1.3866394758224487,
"logps/chosen": -1.4379013776779175,
"logps/rejected": -1.884263277053833,
"loss": 1.5412,
"nll_loss": 1.494981288909912,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.143790140748024,
"rewards/margins": 0.04463617503643036,
"rewards/rejected": -0.18842631578445435,
"step": 383
},
{
"epoch": 0.6052009456264775,
"grad_norm": 0.20354455709457397,
"learning_rate": 3.524998837927192e-06,
"log_odds_chosen": 0.587373673915863,
"log_odds_ratio": -0.4441196024417877,
"logits/chosen": -0.14221185445785522,
"logits/rejected": -1.3197717666625977,
"logps/chosen": -1.5544791221618652,
"logps/rejected": -2.0435311794281006,
"loss": 1.6329,
"nll_loss": 1.5884504318237305,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15544790029525757,
"rewards/margins": 0.04890521243214607,
"rewards/rejected": -0.20435310900211334,
"step": 384
},
{
"epoch": 0.6067769897557131,
"grad_norm": 0.1994301825761795,
"learning_rate": 3.5214365194027797e-06,
"log_odds_chosen": 0.5964666604995728,
"log_odds_ratio": -0.44221487641334534,
"logits/chosen": -0.15480360388755798,
"logits/rejected": -1.4440878629684448,
"logps/chosen": -1.4780986309051514,
"logps/rejected": -1.9680850505828857,
"loss": 1.57,
"nll_loss": 1.5257560014724731,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14780986309051514,
"rewards/margins": 0.048998646438121796,
"rewards/rejected": -0.19680851697921753,
"step": 385
},
{
"epoch": 0.6083530338849488,
"grad_norm": 0.21158069372177124,
"learning_rate": 3.517862707152157e-06,
"log_odds_chosen": 0.45025360584259033,
"log_odds_ratio": -0.5009865164756775,
"logits/chosen": -0.06984852999448776,
"logits/rejected": -1.1637235879898071,
"logps/chosen": -1.6143461465835571,
"logps/rejected": -1.9897408485412598,
"loss": 1.695,
"nll_loss": 1.6448723077774048,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1614346206188202,
"rewards/margins": 0.03753947466611862,
"rewards/rejected": -0.1989741027355194,
"step": 386
},
{
"epoch": 0.6099290780141844,
"grad_norm": 0.21827860176563263,
"learning_rate": 3.5142774281737674e-06,
"log_odds_chosen": 0.6315152645111084,
"log_odds_ratio": -0.4288613796234131,
"logits/chosen": -0.15267856419086456,
"logits/rejected": -1.3205287456512451,
"logps/chosen": -1.6036657094955444,
"logps/rejected": -2.1343271732330322,
"loss": 1.6804,
"nll_loss": 1.6375137567520142,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16036657989025116,
"rewards/margins": 0.053066130727529526,
"rewards/rejected": -0.21343271434307098,
"step": 387
},
{
"epoch": 0.61150512214342,
"grad_norm": 0.17510953545570374,
"learning_rate": 3.5106807095526817e-06,
"log_odds_chosen": 0.6356069445610046,
"log_odds_ratio": -0.4359210133552551,
"logits/chosen": -0.1717139482498169,
"logits/rejected": -1.565706729888916,
"logps/chosen": -1.5656462907791138,
"logps/rejected": -2.093924045562744,
"loss": 1.6578,
"nll_loss": 1.6142207384109497,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1565646231174469,
"rewards/margins": 0.05282779037952423,
"rewards/rejected": -0.20939242839813232,
"step": 388
},
{
"epoch": 0.6130811662726556,
"grad_norm": 0.18727731704711914,
"learning_rate": 3.5070725784603905e-06,
"log_odds_chosen": 0.537490963935852,
"log_odds_ratio": -0.4669988751411438,
"logits/chosen": -0.24123258888721466,
"logits/rejected": -1.2403115034103394,
"logps/chosen": -1.439449429512024,
"logps/rejected": -1.8756340742111206,
"loss": 1.5259,
"nll_loss": 1.4791667461395264,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14394494891166687,
"rewards/margins": 0.04361846297979355,
"rewards/rejected": -0.18756340444087982,
"step": 389
},
{
"epoch": 0.6146572104018913,
"grad_norm": 0.22596481442451477,
"learning_rate": 3.503453062154602e-06,
"log_odds_chosen": 0.4628780484199524,
"log_odds_ratio": -0.49970224499702454,
"logits/chosen": -0.1650674045085907,
"logits/rejected": -1.2386726140975952,
"logps/chosen": -1.602417230606079,
"logps/rejected": -1.9950282573699951,
"loss": 1.6994,
"nll_loss": 1.6494615077972412,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1602417379617691,
"rewards/margins": 0.0392610989511013,
"rewards/rejected": -0.1995028257369995,
"step": 390
},
{
"epoch": 0.6162332545311269,
"grad_norm": 0.19225400686264038,
"learning_rate": 3.499822187979032e-06,
"log_odds_chosen": 0.45917797088623047,
"log_odds_ratio": -0.4991348385810852,
"logits/chosen": -0.09038020670413971,
"logits/rejected": -1.4409539699554443,
"logps/chosen": -1.5594383478164673,
"logps/rejected": -1.9403319358825684,
"loss": 1.6617,
"nll_loss": 1.6117753982543945,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1559438407421112,
"rewards/margins": 0.03808935359120369,
"rewards/rejected": -0.1940331906080246,
"step": 391
},
{
"epoch": 0.6178092986603625,
"grad_norm": 0.1856825351715088,
"learning_rate": 3.496179983363202e-06,
"log_odds_chosen": 0.41265982389450073,
"log_odds_ratio": -0.5119627118110657,
"logits/chosen": -0.11190656572580338,
"logits/rejected": -1.3574274778366089,
"logps/chosen": -1.5668977499008179,
"logps/rejected": -1.9084656238555908,
"loss": 1.6514,
"nll_loss": 1.6002510786056519,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15668979287147522,
"rewards/margins": 0.034156784415245056,
"rewards/rejected": -0.19084656238555908,
"step": 392
},
{
"epoch": 0.6193853427895981,
"grad_norm": 0.20805980265140533,
"learning_rate": 3.4925264758222268e-06,
"log_odds_chosen": 0.6294342279434204,
"log_odds_ratio": -0.43072307109832764,
"logits/chosen": -0.16953016817569733,
"logits/rejected": -1.2341816425323486,
"logps/chosen": -1.5001755952835083,
"logps/rejected": -2.0215494632720947,
"loss": 1.5961,
"nll_loss": 1.5530593395233154,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15001757442951202,
"rewards/margins": 0.05213739350438118,
"rewards/rejected": -0.2021549493074417,
"step": 393
},
{
"epoch": 0.6209613869188337,
"grad_norm": 0.24518869817256927,
"learning_rate": 3.488861692956611e-06,
"log_odds_chosen": 0.5471794009208679,
"log_odds_ratio": -0.4630282521247864,
"logits/chosen": -0.19086423516273499,
"logits/rejected": -1.339902639389038,
"logps/chosen": -1.5402213335037231,
"logps/rejected": -1.9952548742294312,
"loss": 1.6296,
"nll_loss": 1.5832523107528687,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15402214229106903,
"rewards/margins": 0.04550333693623543,
"rewards/rejected": -0.19952546060085297,
"step": 394
},
{
"epoch": 0.6225374310480694,
"grad_norm": 0.2510071098804474,
"learning_rate": 3.4851856624520394e-06,
"log_odds_chosen": 0.5228825807571411,
"log_odds_ratio": -0.46843814849853516,
"logits/chosen": -0.13727201521396637,
"logits/rejected": -1.3861788511276245,
"logps/chosen": -1.572906494140625,
"logps/rejected": -2.005622386932373,
"loss": 1.6575,
"nll_loss": 1.6106876134872437,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15729066729545593,
"rewards/margins": 0.04327157884836197,
"rewards/rejected": -0.2005622535943985,
"step": 395
},
{
"epoch": 0.624113475177305,
"grad_norm": 0.20526158809661865,
"learning_rate": 3.4814984120791664e-06,
"log_odds_chosen": 0.5153719782829285,
"log_odds_ratio": -0.4714086055755615,
"logits/chosen": -0.13660681247711182,
"logits/rejected": -1.2547574043273926,
"logps/chosen": -1.5410287380218506,
"logps/rejected": -1.964991807937622,
"loss": 1.636,
"nll_loss": 1.5889039039611816,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1541028618812561,
"rewards/margins": 0.04239630699157715,
"rewards/rejected": -0.19649919867515564,
"step": 396
},
{
"epoch": 0.6256895193065406,
"grad_norm": 0.18158891797065735,
"learning_rate": 3.477799969693407e-06,
"log_odds_chosen": 0.47999995946884155,
"log_odds_ratio": -0.490679532289505,
"logits/chosen": -0.09590557217597961,
"logits/rejected": -1.3949567079544067,
"logps/chosen": -1.541025996208191,
"logps/rejected": -1.9312851428985596,
"loss": 1.6301,
"nll_loss": 1.5809931755065918,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15410259366035461,
"rewards/margins": 0.039025940001010895,
"rewards/rejected": -0.1931285411119461,
"step": 397
},
{
"epoch": 0.6272655634357762,
"grad_norm": 0.2004203349351883,
"learning_rate": 3.474090363234728e-06,
"log_odds_chosen": 0.766968309879303,
"log_odds_ratio": -0.39646148681640625,
"logits/chosen": -0.2562759518623352,
"logits/rejected": -1.4337137937545776,
"logps/chosen": -1.4752384424209595,
"logps/rejected": -2.112752914428711,
"loss": 1.5586,
"nll_loss": 1.5189671516418457,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14752383530139923,
"rewards/margins": 0.0637514516711235,
"rewards/rejected": -0.21127529442310333,
"step": 398
},
{
"epoch": 0.6288416075650118,
"grad_norm": 0.19742122292518616,
"learning_rate": 3.4703696207274325e-06,
"log_odds_chosen": 0.5179776549339294,
"log_odds_ratio": -0.474680095911026,
"logits/chosen": -0.14589636027812958,
"logits/rejected": -1.2598799467086792,
"logps/chosen": -1.556343913078308,
"logps/rejected": -1.9830336570739746,
"loss": 1.6301,
"nll_loss": 1.5826400518417358,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15563438832759857,
"rewards/margins": 0.042668960988521576,
"rewards/rejected": -0.19830335676670074,
"step": 399
},
{
"epoch": 0.6304176516942475,
"grad_norm": 0.18192359805107117,
"learning_rate": 3.4666377702799545e-06,
"log_odds_chosen": 0.5299201011657715,
"log_odds_ratio": -0.4676084518432617,
"logits/chosen": -0.24987564980983734,
"logits/rejected": -1.5932269096374512,
"logps/chosen": -1.46696138381958,
"logps/rejected": -1.8967633247375488,
"loss": 1.568,
"nll_loss": 1.521193265914917,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14669615030288696,
"rewards/margins": 0.042980194091796875,
"rewards/rejected": -0.18967632949352264,
"step": 400
},
{
"epoch": 0.6319936958234831,
"grad_norm": 0.19527654349803925,
"learning_rate": 3.4628948400846417e-06,
"log_odds_chosen": 0.6314361691474915,
"log_odds_ratio": -0.4365447759628296,
"logits/chosen": -0.1739773005247116,
"logits/rejected": -1.4769560098648071,
"logps/chosen": -1.5479360818862915,
"logps/rejected": -2.0758628845214844,
"loss": 1.637,
"nll_loss": 1.5933518409729004,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1547936052083969,
"rewards/margins": 0.05279267579317093,
"rewards/rejected": -0.20758628845214844,
"step": 401
},
{
"epoch": 0.6335697399527187,
"grad_norm": 0.2035856693983078,
"learning_rate": 3.4591408584175426e-06,
"log_odds_chosen": 0.5278856158256531,
"log_odds_ratio": -0.4733438193798065,
"logits/chosen": -0.18014143407344818,
"logits/rejected": -1.3621004819869995,
"logps/chosen": -1.5849114656448364,
"logps/rejected": -2.027892589569092,
"loss": 1.6766,
"nll_loss": 1.6292688846588135,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1584911346435547,
"rewards/margins": 0.044298142194747925,
"rewards/rejected": -0.2027892768383026,
"step": 402
},
{
"epoch": 0.6351457840819543,
"grad_norm": 0.19495651125907898,
"learning_rate": 3.4553758536381974e-06,
"log_odds_chosen": 0.5365288853645325,
"log_odds_ratio": -0.4719133973121643,
"logits/chosen": -0.1622263491153717,
"logits/rejected": -1.3355566263198853,
"logps/chosen": -1.5399608612060547,
"logps/rejected": -1.984842300415039,
"loss": 1.6272,
"nll_loss": 1.580039381980896,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.153996080160141,
"rewards/margins": 0.04448813945055008,
"rewards/rejected": -0.19848422706127167,
"step": 403
},
{
"epoch": 0.6367218282111899,
"grad_norm": 0.20605534315109253,
"learning_rate": 3.451599854189418e-06,
"log_odds_chosen": 0.5477701425552368,
"log_odds_ratio": -0.46024447679519653,
"logits/chosen": -0.13585253059864044,
"logits/rejected": -1.0337761640548706,
"logps/chosen": -1.54197359085083,
"logps/rejected": -1.995017647743225,
"loss": 1.6418,
"nll_loss": 1.5958125591278076,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15419737994670868,
"rewards/margins": 0.045304395258426666,
"rewards/rejected": -0.19950176775455475,
"step": 404
},
{
"epoch": 0.6382978723404256,
"grad_norm": 0.19387783110141754,
"learning_rate": 3.4478128885970765e-06,
"log_odds_chosen": 0.6080644130706787,
"log_odds_ratio": -0.4360560178756714,
"logits/chosen": -0.179177924990654,
"logits/rejected": -1.3908740282058716,
"logps/chosen": -1.604873776435852,
"logps/rejected": -2.1161084175109863,
"loss": 1.6909,
"nll_loss": 1.6472656726837158,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16048739850521088,
"rewards/margins": 0.051123470067977905,
"rewards/rejected": -0.21161086857318878,
"step": 405
},
{
"epoch": 0.6398739164696612,
"grad_norm": 0.190069779753685,
"learning_rate": 3.44401498546989e-06,
"log_odds_chosen": 0.42473104596138,
"log_odds_ratio": -0.5118191242218018,
"logits/chosen": -0.1188054233789444,
"logits/rejected": -1.49931001663208,
"logps/chosen": -1.5570849180221558,
"logps/rejected": -1.9073197841644287,
"loss": 1.6278,
"nll_loss": 1.5765697956085205,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15570849180221558,
"rewards/margins": 0.035023488104343414,
"rewards/rejected": -0.1907319873571396,
"step": 406
},
{
"epoch": 0.6414499605988968,
"grad_norm": 0.19107168912887573,
"learning_rate": 3.4402061734992005e-06,
"log_odds_chosen": 0.5350978374481201,
"log_odds_ratio": -0.46566373109817505,
"logits/chosen": -0.15249407291412354,
"logits/rejected": -1.402602195739746,
"logps/chosen": -1.5002124309539795,
"logps/rejected": -1.9390612840652466,
"loss": 1.5941,
"nll_loss": 1.5475372076034546,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1500212401151657,
"rewards/margins": 0.043884895741939545,
"rewards/rejected": -0.19390614330768585,
"step": 407
},
{
"epoch": 0.6430260047281324,
"grad_norm": 0.2040640264749527,
"learning_rate": 3.4363864814587656e-06,
"log_odds_chosen": 0.49293413758277893,
"log_odds_ratio": -0.48022550344467163,
"logits/chosen": -0.2637179493904114,
"logits/rejected": -0.9641510248184204,
"logps/chosen": -1.475205421447754,
"logps/rejected": -1.8777835369110107,
"loss": 1.5716,
"nll_loss": 1.5235683917999268,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1475205421447754,
"rewards/margins": 0.04025781527161598,
"rewards/rejected": -0.18777838349342346,
"step": 408
},
{
"epoch": 0.644602048857368,
"grad_norm": 0.20765496790409088,
"learning_rate": 3.4325559382045343e-06,
"log_odds_chosen": 0.4098273515701294,
"log_odds_ratio": -0.5131589770317078,
"logits/chosen": -0.23338492214679718,
"logits/rejected": -1.4860385656356812,
"logps/chosen": -1.4819140434265137,
"logps/rejected": -1.8101638555526733,
"loss": 1.5729,
"nll_loss": 1.5215588808059692,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1481914222240448,
"rewards/margins": 0.03282497450709343,
"rewards/rejected": -0.18101638555526733,
"step": 409
},
{
"epoch": 0.6461780929866037,
"grad_norm": 0.19961215555667877,
"learning_rate": 3.4287145726744295e-06,
"log_odds_chosen": 0.5392709374427795,
"log_odds_ratio": -0.4628680646419525,
"logits/chosen": -0.26481306552886963,
"logits/rejected": -1.3813103437423706,
"logps/chosen": -1.4664371013641357,
"logps/rejected": -1.904196858406067,
"loss": 1.5695,
"nll_loss": 1.523188829421997,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.146643728017807,
"rewards/margins": 0.04377596825361252,
"rewards/rejected": -0.19041968882083893,
"step": 410
},
{
"epoch": 0.6477541371158393,
"grad_norm": 0.20400294661521912,
"learning_rate": 3.4248624138881335e-06,
"log_odds_chosen": 0.42437130212783813,
"log_odds_ratio": -0.5254440307617188,
"logits/chosen": -0.1493159532546997,
"logits/rejected": -1.1513835191726685,
"logps/chosen": -1.6029716730117798,
"logps/rejected": -1.9682285785675049,
"loss": 1.6887,
"nll_loss": 1.6361618041992188,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16029717028141022,
"rewards/margins": 0.03652569651603699,
"rewards/rejected": -0.1968228816986084,
"step": 411
},
{
"epoch": 0.6493301812450749,
"grad_norm": 0.20339736342430115,
"learning_rate": 3.4209994909468672e-06,
"log_odds_chosen": 0.6561870574951172,
"log_odds_ratio": -0.42825421690940857,
"logits/chosen": -0.28305602073669434,
"logits/rejected": -1.04371976852417,
"logps/chosen": -1.5072612762451172,
"logps/rejected": -2.0518369674682617,
"loss": 1.5933,
"nll_loss": 1.5505071878433228,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15072615444660187,
"rewards/margins": 0.054457567632198334,
"rewards/rejected": -0.2051836997270584,
"step": 412
},
{
"epoch": 0.6509062253743105,
"grad_norm": 0.19428101181983948,
"learning_rate": 3.4171258330331667e-06,
"log_odds_chosen": 0.43498852849006653,
"log_odds_ratio": -0.5014755129814148,
"logits/chosen": -0.12146922200918198,
"logits/rejected": -1.1085965633392334,
"logps/chosen": -1.593203067779541,
"logps/rejected": -1.9516777992248535,
"loss": 1.6774,
"nll_loss": 1.627271056175232,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15932030975818634,
"rewards/margins": 0.03584747388958931,
"rewards/rejected": -0.19516779482364655,
"step": 413
},
{
"epoch": 0.6524822695035462,
"grad_norm": 0.19590893387794495,
"learning_rate": 3.4132414694106684e-06,
"log_odds_chosen": 0.6712747812271118,
"log_odds_ratio": -0.4207912087440491,
"logits/chosen": -0.16028505563735962,
"logits/rejected": -1.4717087745666504,
"logps/chosen": -1.498020052909851,
"logps/rejected": -2.0545201301574707,
"loss": 1.5811,
"nll_loss": 1.5389834642410278,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14980201423168182,
"rewards/margins": 0.055649999529123306,
"rewards/rejected": -0.20545199513435364,
"step": 414
},
{
"epoch": 0.6540583136327817,
"grad_norm": 0.20174475014209747,
"learning_rate": 3.409346429423884e-06,
"log_odds_chosen": 0.4537242650985718,
"log_odds_ratio": -0.4941532015800476,
"logits/chosen": -0.010996952652931213,
"logits/rejected": -1.3585155010223389,
"logps/chosen": -1.5672545433044434,
"logps/rejected": -1.9409823417663574,
"loss": 1.6455,
"nll_loss": 1.5961326360702515,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1567254513502121,
"rewards/margins": 0.03737279772758484,
"rewards/rejected": -0.19409826397895813,
"step": 415
},
{
"epoch": 0.6556343577620173,
"grad_norm": 0.19159899652004242,
"learning_rate": 3.40544074249798e-06,
"log_odds_chosen": 0.6167906522750854,
"log_odds_ratio": -0.4354286193847656,
"logits/chosen": -0.18299099802970886,
"logits/rejected": -1.46063232421875,
"logps/chosen": -1.5565929412841797,
"logps/rejected": -2.0718507766723633,
"loss": 1.6407,
"nll_loss": 1.5971307754516602,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15565930306911469,
"rewards/margins": 0.0515257902443409,
"rewards/rejected": -0.20718510448932648,
"step": 416
},
{
"epoch": 0.6572104018912529,
"grad_norm": 0.20851117372512817,
"learning_rate": 3.401524438138556e-06,
"log_odds_chosen": 0.45928677916526794,
"log_odds_ratio": -0.49514341354370117,
"logits/chosen": -0.13402409851551056,
"logits/rejected": -1.1145453453063965,
"logps/chosen": -1.5616354942321777,
"logps/rejected": -1.9419609308242798,
"loss": 1.6266,
"nll_loss": 1.5770922899246216,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1561635434627533,
"rewards/margins": 0.03803255409002304,
"rewards/rejected": -0.19419609010219574,
"step": 417
},
{
"epoch": 0.6587864460204885,
"grad_norm": 0.19792215526103973,
"learning_rate": 3.39759754593142e-06,
"log_odds_chosen": 0.4961496889591217,
"log_odds_ratio": -0.4824981689453125,
"logits/chosen": -0.1806434839963913,
"logits/rejected": -1.2893097400665283,
"logps/chosen": -1.6215399503707886,
"logps/rejected": -2.039522409439087,
"loss": 1.6951,
"nll_loss": 1.646848440170288,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16215398907661438,
"rewards/margins": 0.04179824888706207,
"rewards/rejected": -0.20395225286483765,
"step": 418
},
{
"epoch": 0.6603624901497241,
"grad_norm": 0.2001093477010727,
"learning_rate": 3.3936600955423683e-06,
"log_odds_chosen": 0.7080036997795105,
"log_odds_ratio": -0.4158882200717926,
"logits/chosen": -0.10217966884374619,
"logits/rejected": -1.1086851358413696,
"logps/chosen": -1.5155251026153564,
"logps/rejected": -2.1040117740631104,
"loss": 1.603,
"nll_loss": 1.5614415407180786,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15155251324176788,
"rewards/margins": 0.05884869024157524,
"rewards/rejected": -0.21040120720863342,
"step": 419
},
{
"epoch": 0.6619385342789598,
"grad_norm": 0.1921870857477188,
"learning_rate": 3.3897121167169573e-06,
"log_odds_chosen": 0.39313969016075134,
"log_odds_ratio": -0.5258656144142151,
"logits/chosen": -0.22022226452827454,
"logits/rejected": -1.3231024742126465,
"logps/chosen": -1.4893752336502075,
"logps/rejected": -1.808854341506958,
"loss": 1.5882,
"nll_loss": 1.535656452178955,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.14893752336502075,
"rewards/margins": 0.03194789960980415,
"rewards/rejected": -0.1808854341506958,
"step": 420
},
{
"epoch": 0.6635145784081954,
"grad_norm": 0.20021358132362366,
"learning_rate": 3.38575363928028e-06,
"log_odds_chosen": 0.6059412360191345,
"log_odds_ratio": -0.4419128894805908,
"logits/chosen": -0.14556629955768585,
"logits/rejected": -1.1496225595474243,
"logps/chosen": -1.4978344440460205,
"logps/rejected": -1.997003197669983,
"loss": 1.5713,
"nll_loss": 1.52711820602417,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1497834473848343,
"rewards/margins": 0.04991687089204788,
"rewards/rejected": -0.19970029592514038,
"step": 421
},
{
"epoch": 0.665090622537431,
"grad_norm": 0.19506679475307465,
"learning_rate": 3.3817846931367452e-06,
"log_odds_chosen": 0.415115624666214,
"log_odds_ratio": -0.5117157101631165,
"logits/chosen": -0.22510936856269836,
"logits/rejected": -1.2167201042175293,
"logps/chosen": -1.4807124137878418,
"logps/rejected": -1.8138527870178223,
"loss": 1.5701,
"nll_loss": 1.5189181566238403,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1480712592601776,
"rewards/margins": 0.033314019441604614,
"rewards/rejected": -0.18138529360294342,
"step": 422
},
{
"epoch": 0.6666666666666666,
"grad_norm": 0.19481457769870758,
"learning_rate": 3.377805308269844e-06,
"log_odds_chosen": 0.6872407793998718,
"log_odds_ratio": -0.42233163118362427,
"logits/chosen": -0.11342249810695648,
"logits/rejected": -1.5072932243347168,
"logps/chosen": -1.5372941493988037,
"logps/rejected": -2.1134226322174072,
"loss": 1.6132,
"nll_loss": 1.5709176063537598,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15372943878173828,
"rewards/margins": 0.05761285126209259,
"rewards/rejected": -0.21134227514266968,
"step": 423
},
{
"epoch": 0.6682427107959023,
"grad_norm": 0.18579819798469543,
"learning_rate": 3.3738155147419275e-06,
"log_odds_chosen": 0.6220219135284424,
"log_odds_ratio": -0.44024914503097534,
"logits/chosen": -0.235686257481575,
"logits/rejected": -1.2404701709747314,
"logps/chosen": -1.5340425968170166,
"logps/rejected": -2.0529518127441406,
"loss": 1.5954,
"nll_loss": 1.5513544082641602,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15340426564216614,
"rewards/margins": 0.05189090967178345,
"rewards/rejected": -0.2052951604127884,
"step": 424
},
{
"epoch": 0.6698187549251379,
"grad_norm": 0.19320693612098694,
"learning_rate": 3.3698153426939824e-06,
"log_odds_chosen": 0.7084161639213562,
"log_odds_ratio": -0.4107830226421356,
"logits/chosen": -0.2280699759721756,
"logits/rejected": -1.0708644390106201,
"logps/chosen": -1.463280439376831,
"logps/rejected": -2.0446677207946777,
"loss": 1.5378,
"nll_loss": 1.4967448711395264,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14632803201675415,
"rewards/margins": 0.05813872069120407,
"rewards/rejected": -0.2044667750597,
"step": 425
},
{
"epoch": 0.6713947990543735,
"grad_norm": 0.20736993849277496,
"learning_rate": 3.3658048223453954e-06,
"log_odds_chosen": 0.662844717502594,
"log_odds_ratio": -0.4233693480491638,
"logits/chosen": -0.1891903281211853,
"logits/rejected": -1.40634024143219,
"logps/chosen": -1.6040951013565063,
"logps/rejected": -2.1638996601104736,
"loss": 1.6514,
"nll_loss": 1.6090670824050903,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16040951013565063,
"rewards/margins": 0.05598045140504837,
"rewards/rejected": -0.2163899689912796,
"step": 426
},
{
"epoch": 0.6729708431836091,
"grad_norm": 0.1907954066991806,
"learning_rate": 3.3617839839937337e-06,
"log_odds_chosen": 0.6645872592926025,
"log_odds_ratio": -0.4254325330257416,
"logits/chosen": -0.11419974267482758,
"logits/rejected": -1.4858274459838867,
"logps/chosen": -1.5667697191238403,
"logps/rejected": -2.1278765201568604,
"loss": 1.6334,
"nll_loss": 1.5908530950546265,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15667694807052612,
"rewards/margins": 0.056110695004463196,
"rewards/rejected": -0.2127876579761505,
"step": 427
},
{
"epoch": 0.6745468873128447,
"grad_norm": 0.20678722858428955,
"learning_rate": 3.3577528580145107e-06,
"log_odds_chosen": 0.3481862545013428,
"log_odds_ratio": -0.5371396541595459,
"logits/chosen": -0.14534202218055725,
"logits/rejected": -1.29691481590271,
"logps/chosen": -1.585597276687622,
"logps/rejected": -1.8742812871932983,
"loss": 1.6676,
"nll_loss": 1.6138746738433838,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15855972468852997,
"rewards/margins": 0.028868405148386955,
"rewards/rejected": -0.18742814660072327,
"step": 428
},
{
"epoch": 0.6761229314420804,
"grad_norm": 0.19353458285331726,
"learning_rate": 3.353711474860956e-06,
"log_odds_chosen": 0.5981341600418091,
"log_odds_ratio": -0.45493775606155396,
"logits/chosen": -0.1791481226682663,
"logits/rejected": -1.231849193572998,
"logps/chosen": -1.5348036289215088,
"logps/rejected": -2.0394744873046875,
"loss": 1.597,
"nll_loss": 1.5514580011367798,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15348035097122192,
"rewards/margins": 0.05046708881855011,
"rewards/rejected": -0.20394745469093323,
"step": 429
},
{
"epoch": 0.677698975571316,
"grad_norm": 0.18463017046451569,
"learning_rate": 3.3496598650637916e-06,
"log_odds_chosen": 0.569009006023407,
"log_odds_ratio": -0.45516982674598694,
"logits/chosen": -0.20293018221855164,
"logits/rejected": -1.1590790748596191,
"logps/chosen": -1.435080647468567,
"logps/rejected": -1.8941256999969482,
"loss": 1.5186,
"nll_loss": 1.4731093645095825,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14350807666778564,
"rewards/margins": 0.0459044985473156,
"rewards/rejected": -0.18941256403923035,
"step": 430
},
{
"epoch": 0.6792750197005516,
"grad_norm": 0.18710088729858398,
"learning_rate": 3.3455980592309923e-06,
"log_odds_chosen": 0.619581401348114,
"log_odds_ratio": -0.4386385977268219,
"logits/chosen": -0.23650380969047546,
"logits/rejected": -1.2405811548233032,
"logps/chosen": -1.4119963645935059,
"logps/rejected": -1.9160082340240479,
"loss": 1.5042,
"nll_loss": 1.460310459136963,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14119963347911835,
"rewards/margins": 0.05040118098258972,
"rewards/rejected": -0.19160079956054688,
"step": 431
},
{
"epoch": 0.6808510638297872,
"grad_norm": 0.1992560178041458,
"learning_rate": 3.341526088047562e-06,
"log_odds_chosen": 0.5567487478256226,
"log_odds_ratio": -0.4619826078414917,
"logits/chosen": -0.17194384336471558,
"logits/rejected": -1.293556571006775,
"logps/chosen": -1.6019841432571411,
"logps/rejected": -2.070882558822632,
"loss": 1.662,
"nll_loss": 1.6158294677734375,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1601984202861786,
"rewards/margins": 0.046889837831258774,
"rewards/rejected": -0.20708826184272766,
"step": 432
},
{
"epoch": 0.6824271079590228,
"grad_norm": 0.226112961769104,
"learning_rate": 3.3374439822752972e-06,
"log_odds_chosen": 0.44421204924583435,
"log_odds_ratio": -0.49906378984451294,
"logits/chosen": -0.12680430710315704,
"logits/rejected": -1.1431465148925781,
"logps/chosen": -1.5594537258148193,
"logps/rejected": -1.9281821250915527,
"loss": 1.649,
"nll_loss": 1.5991100072860718,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15594536066055298,
"rewards/margins": 0.036872848868370056,
"rewards/rejected": -0.19281822443008423,
"step": 433
},
{
"epoch": 0.6840031520882585,
"grad_norm": 0.207797572016716,
"learning_rate": 3.333351772752559e-06,
"log_odds_chosen": 0.5869032144546509,
"log_odds_ratio": -0.4558618366718292,
"logits/chosen": -0.23912350833415985,
"logits/rejected": -1.287145972251892,
"logps/chosen": -1.6291857957839966,
"logps/rejected": -2.120633602142334,
"loss": 1.6932,
"nll_loss": 1.6475987434387207,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1629185825586319,
"rewards/margins": 0.049144770950078964,
"rewards/rejected": -0.21206337213516235,
"step": 434
},
{
"epoch": 0.6855791962174941,
"grad_norm": 0.20063838362693787,
"learning_rate": 3.3292494903940338e-06,
"log_odds_chosen": 0.7061201333999634,
"log_odds_ratio": -0.40912342071533203,
"logits/chosen": -0.1544758826494217,
"logits/rejected": -1.3016716241836548,
"logps/chosen": -1.5344663858413696,
"logps/rejected": -2.1211295127868652,
"loss": 1.6071,
"nll_loss": 1.5661424398422241,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15344664454460144,
"rewards/margins": 0.05866629630327225,
"rewards/rejected": -0.2121129333972931,
"step": 435
},
{
"epoch": 0.6871552403467297,
"grad_norm": 0.1838790476322174,
"learning_rate": 3.3251371661905063e-06,
"log_odds_chosen": 0.6065340638160706,
"log_odds_ratio": -0.44656994938850403,
"logits/chosen": -0.1865832805633545,
"logits/rejected": -1.140378713607788,
"logps/chosen": -1.3614064455032349,
"logps/rejected": -1.8547477722167969,
"loss": 1.4488,
"nll_loss": 1.4041305780410767,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1361406445503235,
"rewards/margins": 0.04933411255478859,
"rewards/rejected": -0.18547475337982178,
"step": 436
},
{
"epoch": 0.6887312844759653,
"grad_norm": 0.21491163969039917,
"learning_rate": 3.321014831208622e-06,
"log_odds_chosen": 0.5981911420822144,
"log_odds_ratio": -0.44269564747810364,
"logits/chosen": -0.19441677629947662,
"logits/rejected": -1.1200268268585205,
"logps/chosen": -1.4213942289352417,
"logps/rejected": -1.9019947052001953,
"loss": 1.5169,
"nll_loss": 1.4725940227508545,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14213941991329193,
"rewards/margins": 0.04806005209684372,
"rewards/rejected": -0.19019947946071625,
"step": 437
},
{
"epoch": 0.6903073286052009,
"grad_norm": 0.2163185477256775,
"learning_rate": 3.316882516590652e-06,
"log_odds_chosen": 0.6079857349395752,
"log_odds_ratio": -0.43692946434020996,
"logits/chosen": -0.18331696093082428,
"logits/rejected": -1.3628792762756348,
"logps/chosen": -1.4688149690628052,
"logps/rejected": -1.966677188873291,
"loss": 1.5508,
"nll_loss": 1.5071358680725098,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14688150584697723,
"rewards/margins": 0.04978622496128082,
"rewards/rejected": -0.19666773080825806,
"step": 438
},
{
"epoch": 0.6918833727344366,
"grad_norm": 0.21325580775737762,
"learning_rate": 3.31274025355426e-06,
"log_odds_chosen": 0.5819729566574097,
"log_odds_ratio": -0.4491526484489441,
"logits/chosen": -0.22322604060173035,
"logits/rejected": -1.189731478691101,
"logps/chosen": -1.5592323541641235,
"logps/rejected": -2.043937921524048,
"loss": 1.6116,
"nll_loss": 1.5666409730911255,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15592323243618011,
"rewards/margins": 0.04847054183483124,
"rewards/rejected": -0.20439377427101135,
"step": 439
},
{
"epoch": 0.6934594168636722,
"grad_norm": 0.20833250880241394,
"learning_rate": 3.308588073392265e-06,
"log_odds_chosen": 0.5521525144577026,
"log_odds_ratio": -0.46361684799194336,
"logits/chosen": -0.2142220437526703,
"logits/rejected": -1.1978386640548706,
"logps/chosen": -1.5905332565307617,
"logps/rejected": -2.054898738861084,
"loss": 1.6463,
"nll_loss": 1.599968433380127,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15905332565307617,
"rewards/margins": 0.046436551958322525,
"rewards/rejected": -0.2054898738861084,
"step": 440
},
{
"epoch": 0.6950354609929078,
"grad_norm": 0.19477160274982452,
"learning_rate": 3.3044260074724035e-06,
"log_odds_chosen": 0.6352304220199585,
"log_odds_ratio": -0.43725699186325073,
"logits/chosen": -0.20722348988056183,
"logits/rejected": -1.4436430931091309,
"logps/chosen": -1.5010461807250977,
"logps/rejected": -2.0235297679901123,
"loss": 1.5782,
"nll_loss": 1.5344798564910889,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15010464191436768,
"rewards/margins": 0.05224834755063057,
"rewards/rejected": -0.20235297083854675,
"step": 441
},
{
"epoch": 0.6966115051221434,
"grad_norm": 0.18534111976623535,
"learning_rate": 3.300254087237097e-06,
"log_odds_chosen": 0.5580976605415344,
"log_odds_ratio": -0.457084059715271,
"logits/chosen": -0.16514423489570618,
"logits/rejected": -1.3200604915618896,
"logps/chosen": -1.3929381370544434,
"logps/rejected": -1.8415474891662598,
"loss": 1.4993,
"nll_loss": 1.4535483121871948,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1392938196659088,
"rewards/margins": 0.044860921800136566,
"rewards/rejected": -0.18415474891662598,
"step": 442
},
{
"epoch": 0.698187549251379,
"grad_norm": 0.26748111844062805,
"learning_rate": 3.2960723442032105e-06,
"log_odds_chosen": 0.7100386619567871,
"log_odds_ratio": -0.4068644642829895,
"logits/chosen": -0.23274515569210052,
"logits/rejected": -1.637979507446289,
"logps/chosen": -1.590896725654602,
"logps/rejected": -2.1930923461914062,
"loss": 1.6465,
"nll_loss": 1.6058528423309326,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15908968448638916,
"rewards/margins": 0.06021953374147415,
"rewards/rejected": -0.21930919587612152,
"step": 443
},
{
"epoch": 0.6997635933806147,
"grad_norm": 0.1990683227777481,
"learning_rate": 3.291880809961814e-06,
"log_odds_chosen": 0.6279516220092773,
"log_odds_ratio": -0.4397643506526947,
"logits/chosen": -0.15238967537879944,
"logits/rejected": -1.2192305326461792,
"logps/chosen": -1.5531084537506104,
"logps/rejected": -2.074535369873047,
"loss": 1.6241,
"nll_loss": 1.5800902843475342,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15531083941459656,
"rewards/margins": 0.0521426796913147,
"rewards/rejected": -0.20745351910591125,
"step": 444
},
{
"epoch": 0.7013396375098503,
"grad_norm": 0.19820798933506012,
"learning_rate": 3.2876795161779473e-06,
"log_odds_chosen": 0.7250336408615112,
"log_odds_ratio": -0.3975331783294678,
"logits/chosen": -0.18375059962272644,
"logits/rejected": -1.543222188949585,
"logps/chosen": -1.5645023584365845,
"logps/rejected": -2.1766436100006104,
"loss": 1.6162,
"nll_loss": 1.5764946937561035,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15645024180412292,
"rewards/margins": 0.0612141489982605,
"rewards/rejected": -0.21766439080238342,
"step": 445
},
{
"epoch": 0.7029156816390859,
"grad_norm": 0.19689838588237762,
"learning_rate": 3.2834684945903776e-06,
"log_odds_chosen": 0.5597304105758667,
"log_odds_ratio": -0.45593225955963135,
"logits/chosen": -0.2338670790195465,
"logits/rejected": -1.2486504316329956,
"logps/chosen": -1.4915810823440552,
"logps/rejected": -1.9466543197631836,
"loss": 1.572,
"nll_loss": 1.5264508724212646,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14915812015533447,
"rewards/margins": 0.045507319271564484,
"rewards/rejected": -0.19466543197631836,
"step": 446
},
{
"epoch": 0.7044917257683215,
"grad_norm": 0.2056231051683426,
"learning_rate": 3.2792477770113624e-06,
"log_odds_chosen": 0.5060315728187561,
"log_odds_ratio": -0.47748908400535583,
"logits/chosen": -0.32436949014663696,
"logits/rejected": -1.376452922821045,
"logps/chosen": -1.5770741701126099,
"logps/rejected": -1.9976955652236938,
"loss": 1.6348,
"nll_loss": 1.5870327949523926,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15770742297172546,
"rewards/margins": 0.04206214100122452,
"rewards/rejected": -0.19976955652236938,
"step": 447
},
{
"epoch": 0.7060677698975572,
"grad_norm": 0.2036747932434082,
"learning_rate": 3.275017395326407e-06,
"log_odds_chosen": 0.4934311509132385,
"log_odds_ratio": -0.48520928621292114,
"logits/chosen": -0.13775332272052765,
"logits/rejected": -1.1801694631576538,
"logps/chosen": -1.4680533409118652,
"logps/rejected": -1.8637882471084595,
"loss": 1.5528,
"nll_loss": 1.5042613744735718,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.14680534601211548,
"rewards/margins": 0.039573490619659424,
"rewards/rejected": -0.1863788366317749,
"step": 448
},
{
"epoch": 0.7076438140267928,
"grad_norm": 0.1903351992368698,
"learning_rate": 3.2707773814940244e-06,
"log_odds_chosen": 0.6554431915283203,
"log_odds_ratio": -0.42446404695510864,
"logits/chosen": -0.15731389820575714,
"logits/rejected": -1.3429402112960815,
"logps/chosen": -1.4967751502990723,
"logps/rejected": -2.0305981636047363,
"loss": 1.5792,
"nll_loss": 1.5367555618286133,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14967751502990723,
"rewards/margins": 0.05338229984045029,
"rewards/rejected": -0.20305980741977692,
"step": 449
},
{
"epoch": 0.7092198581560284,
"grad_norm": 0.18632066249847412,
"learning_rate": 3.2665277675454935e-06,
"log_odds_chosen": 0.6589217185974121,
"log_odds_ratio": -0.43469709157943726,
"logits/chosen": -0.22264309227466583,
"logits/rejected": -1.58710515499115,
"logps/chosen": -1.5806881189346313,
"logps/rejected": -2.1421871185302734,
"loss": 1.6344,
"nll_loss": 1.590897798538208,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15806882083415985,
"rewards/margins": 0.05614989995956421,
"rewards/rejected": -0.21421872079372406,
"step": 450
},
{
"epoch": 0.710795902285264,
"grad_norm": 0.1898472160100937,
"learning_rate": 3.262268585584619e-06,
"log_odds_chosen": 0.6008990406990051,
"log_odds_ratio": -0.440044105052948,
"logits/chosen": -0.1463158279657364,
"logits/rejected": -1.3233999013900757,
"logps/chosen": -1.5258179903030396,
"logps/rejected": -2.0252106189727783,
"loss": 1.5941,
"nll_loss": 1.5501309633255005,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15258179605007172,
"rewards/margins": 0.04993927478790283,
"rewards/rejected": -0.20252105593681335,
"step": 451
},
{
"epoch": 0.7123719464144996,
"grad_norm": 0.20325587689876556,
"learning_rate": 3.2579998677874853e-06,
"log_odds_chosen": 0.691182017326355,
"log_odds_ratio": -0.41538918018341064,
"logits/chosen": -0.23037710785865784,
"logits/rejected": -1.2760132551193237,
"logps/chosen": -1.5629483461380005,
"logps/rejected": -2.1457886695861816,
"loss": 1.6173,
"nll_loss": 1.5757336616516113,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1562948226928711,
"rewards/margins": 0.05828403681516647,
"rewards/rejected": -0.21457885205745697,
"step": 452
},
{
"epoch": 0.7139479905437353,
"grad_norm": 0.18862774968147278,
"learning_rate": 3.2537216464022155e-06,
"log_odds_chosen": 0.6802084445953369,
"log_odds_ratio": -0.41921448707580566,
"logits/chosen": -0.20463165640830994,
"logits/rejected": -1.2661035060882568,
"logps/chosen": -1.5623295307159424,
"logps/rejected": -2.129488945007324,
"loss": 1.6273,
"nll_loss": 1.585338830947876,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15623293817043304,
"rewards/margins": 0.0567159429192543,
"rewards/rejected": -0.21294888854026794,
"step": 453
},
{
"epoch": 0.7155240346729709,
"grad_norm": 0.1871948093175888,
"learning_rate": 3.2494339537487314e-06,
"log_odds_chosen": 0.5408557653427124,
"log_odds_ratio": -0.46073442697525024,
"logits/chosen": -0.23007997870445251,
"logits/rejected": -1.5315394401550293,
"logps/chosen": -1.5641494989395142,
"logps/rejected": -2.0117013454437256,
"loss": 1.6397,
"nll_loss": 1.5936379432678223,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1564149558544159,
"rewards/margins": 0.04475518316030502,
"rewards/rejected": -0.2011701464653015,
"step": 454
},
{
"epoch": 0.7171000788022065,
"grad_norm": 0.18707512319087982,
"learning_rate": 3.2451368222185006e-06,
"log_odds_chosen": 0.454687237739563,
"log_odds_ratio": -0.4977053105831146,
"logits/chosen": -0.2190241664648056,
"logits/rejected": -1.1657060384750366,
"logps/chosen": -1.5094616413116455,
"logps/rejected": -1.8778409957885742,
"loss": 1.5921,
"nll_loss": 1.5422812700271606,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15094618499279022,
"rewards/margins": 0.03683791682124138,
"rewards/rejected": -0.1877840906381607,
"step": 455
},
{
"epoch": 0.7186761229314421,
"grad_norm": 0.17670530080795288,
"learning_rate": 3.2408302842743007e-06,
"log_odds_chosen": 0.6599798798561096,
"log_odds_ratio": -0.4245462715625763,
"logits/chosen": -0.1710800975561142,
"logits/rejected": -1.3067679405212402,
"logps/chosen": -1.4607981443405151,
"logps/rejected": -2.002277374267578,
"loss": 1.5484,
"nll_loss": 1.505940556526184,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14607983827590942,
"rewards/margins": 0.05414789542555809,
"rewards/rejected": -0.20022772252559662,
"step": 456
},
{
"epoch": 0.7202521670606777,
"grad_norm": 0.18531948328018188,
"learning_rate": 3.2365143724499684e-06,
"log_odds_chosen": 0.590911865234375,
"log_odds_ratio": -0.4425351023674011,
"logits/chosen": -0.1856268346309662,
"logits/rejected": -1.3595974445343018,
"logps/chosen": -1.5068074464797974,
"logps/rejected": -1.9935933351516724,
"loss": 1.5948,
"nll_loss": 1.5505702495574951,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1506807506084442,
"rewards/margins": 0.04867858439683914,
"rewards/rejected": -0.19935932755470276,
"step": 457
},
{
"epoch": 0.7218282111899134,
"grad_norm": 0.19168265163898468,
"learning_rate": 3.2321891193501564e-06,
"log_odds_chosen": 0.5883606672286987,
"log_odds_ratio": -0.4528968632221222,
"logits/chosen": -0.16482576727867126,
"logits/rejected": -0.8930314779281616,
"logps/chosen": -1.510907530784607,
"logps/rejected": -1.9958248138427734,
"loss": 1.5783,
"nll_loss": 1.532994031906128,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15109075605869293,
"rewards/margins": 0.04849172383546829,
"rewards/rejected": -0.19958247244358063,
"step": 458
},
{
"epoch": 0.723404255319149,
"grad_norm": 0.18463407456874847,
"learning_rate": 3.2278545576500858e-06,
"log_odds_chosen": 0.8210570812225342,
"log_odds_ratio": -0.37329068779945374,
"logits/chosen": -0.08877343684434891,
"logits/rejected": -1.078304648399353,
"logps/chosen": -1.3643009662628174,
"logps/rejected": -2.0353879928588867,
"loss": 1.4512,
"nll_loss": 1.4138658046722412,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13643008470535278,
"rewards/margins": 0.06710872799158096,
"rewards/rejected": -0.20353882014751434,
"step": 459
},
{
"epoch": 0.7249802994483846,
"grad_norm": 0.19575412571430206,
"learning_rate": 3.223510720095299e-06,
"log_odds_chosen": 0.785904049873352,
"log_odds_ratio": -0.38319867849349976,
"logits/chosen": -0.2679174542427063,
"logits/rejected": -1.3926105499267578,
"logps/chosen": -1.5412414073944092,
"logps/rejected": -2.202934980392456,
"loss": 1.5979,
"nll_loss": 1.5595486164093018,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15412414073944092,
"rewards/margins": 0.0661693587899208,
"rewards/rejected": -0.22029350697994232,
"step": 460
},
{
"epoch": 0.7265563435776202,
"grad_norm": 0.18656474351882935,
"learning_rate": 3.2191576395014158e-06,
"log_odds_chosen": 0.7210904359817505,
"log_odds_ratio": -0.3992398679256439,
"logits/chosen": -0.20938719809055328,
"logits/rejected": -1.283248782157898,
"logps/chosen": -1.4764115810394287,
"logps/rejected": -2.072150707244873,
"loss": 1.5461,
"nll_loss": 1.5061570405960083,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1476411670446396,
"rewards/margins": 0.059573911130428314,
"rewards/rejected": -0.2072150707244873,
"step": 461
},
{
"epoch": 0.7281323877068558,
"grad_norm": 0.18474119901657104,
"learning_rate": 3.2147953487538794e-06,
"log_odds_chosen": 0.6938648223876953,
"log_odds_ratio": -0.4105943739414215,
"logits/chosen": -0.14119039475917816,
"logits/rejected": -1.3704811334609985,
"logps/chosen": -1.4421080350875854,
"logps/rejected": -2.0102922916412354,
"loss": 1.514,
"nll_loss": 1.4729448556900024,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1442108154296875,
"rewards/margins": 0.05681842565536499,
"rewards/rejected": -0.2010292410850525,
"step": 462
},
{
"epoch": 0.7297084318360915,
"grad_norm": 0.19482626020908356,
"learning_rate": 3.2104238808077133e-06,
"log_odds_chosen": 0.5768538117408752,
"log_odds_ratio": -0.45441049337387085,
"logits/chosen": -0.169452965259552,
"logits/rejected": -1.0683661699295044,
"logps/chosen": -1.4814167022705078,
"logps/rejected": -1.9551403522491455,
"loss": 1.5563,
"nll_loss": 1.5108129978179932,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14814168214797974,
"rewards/margins": 0.04737236350774765,
"rewards/rejected": -0.1955140382051468,
"step": 463
},
{
"epoch": 0.731284475965327,
"grad_norm": 0.18169742822647095,
"learning_rate": 3.2060432686872704e-06,
"log_odds_chosen": 0.8345743417739868,
"log_odds_ratio": -0.3699982464313507,
"logits/chosen": -0.2313491553068161,
"logits/rejected": -1.2104275226593018,
"logps/chosen": -1.387764811515808,
"logps/rejected": -2.067291736602783,
"loss": 1.4733,
"nll_loss": 1.436316728591919,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.138776496052742,
"rewards/margins": 0.06795267760753632,
"rewards/rejected": -0.20672915875911713,
"step": 464
},
{
"epoch": 0.7328605200945626,
"grad_norm": 0.18539521098136902,
"learning_rate": 3.201653545485982e-06,
"log_odds_chosen": 0.6590836048126221,
"log_odds_ratio": -0.42436298727989197,
"logits/chosen": -0.14550940692424774,
"logits/rejected": -1.2948503494262695,
"logps/chosen": -1.5323055982589722,
"logps/rejected": -2.083099365234375,
"loss": 1.5889,
"nll_loss": 1.5465004444122314,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15323056280612946,
"rewards/margins": 0.0550793781876564,
"rewards/rejected": -0.20830994844436646,
"step": 465
},
{
"epoch": 0.7344365642237982,
"grad_norm": 0.2062307447195053,
"learning_rate": 3.197254744366111e-06,
"log_odds_chosen": 0.67624831199646,
"log_odds_ratio": -0.41811689734458923,
"logits/chosen": -0.1269284039735794,
"logits/rejected": -1.2436720132827759,
"logps/chosen": -1.442671775817871,
"logps/rejected": -1.994502305984497,
"loss": 1.519,
"nll_loss": 1.4771640300750732,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14426717162132263,
"rewards/margins": 0.0551830530166626,
"rewards/rejected": -0.19945020973682404,
"step": 466
},
{
"epoch": 0.7360126083530338,
"grad_norm": 0.19510377943515778,
"learning_rate": 3.192846898558498e-06,
"log_odds_chosen": 0.5479187369346619,
"log_odds_ratio": -0.4584289491176605,
"logits/chosen": -0.17342276871204376,
"logits/rejected": -1.2843399047851562,
"logps/chosen": -1.5753339529037476,
"logps/rejected": -2.0328989028930664,
"loss": 1.6463,
"nll_loss": 1.600473403930664,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15753339231014252,
"rewards/margins": 0.045756496489048004,
"rewards/rejected": -0.20328989624977112,
"step": 467
},
{
"epoch": 0.7375886524822695,
"grad_norm": 0.19615499675273895,
"learning_rate": 3.188430041362313e-06,
"log_odds_chosen": 0.5022854804992676,
"log_odds_ratio": -0.4808230698108673,
"logits/chosen": -0.1259315460920334,
"logits/rejected": -1.3108328580856323,
"logps/chosen": -1.5695524215698242,
"logps/rejected": -1.9884154796600342,
"loss": 1.6285,
"nll_loss": 1.580439567565918,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15695525705814362,
"rewards/margins": 0.04188628867268562,
"rewards/rejected": -0.19884154200553894,
"step": 468
},
{
"epoch": 0.7391646966115051,
"grad_norm": 0.20162400603294373,
"learning_rate": 3.184004206144803e-06,
"log_odds_chosen": 0.7329556941986084,
"log_odds_ratio": -0.4027223289012909,
"logits/chosen": -0.23247480392456055,
"logits/rejected": -1.2680878639221191,
"logps/chosen": -1.4653428792953491,
"logps/rejected": -2.0662801265716553,
"loss": 1.5357,
"nll_loss": 1.4954301118850708,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1465342789888382,
"rewards/margins": 0.06009373068809509,
"rewards/rejected": -0.20662802457809448,
"step": 469
},
{
"epoch": 0.7407407407407407,
"grad_norm": 0.18583819270133972,
"learning_rate": 3.1795694263410386e-06,
"log_odds_chosen": 0.7982565760612488,
"log_odds_ratio": -0.38432639837265015,
"logits/chosen": -0.19295667111873627,
"logits/rejected": -1.293751835823059,
"logps/chosen": -1.4557034969329834,
"logps/rejected": -2.122990608215332,
"loss": 1.5197,
"nll_loss": 1.48124098777771,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14557035267353058,
"rewards/margins": 0.06672872602939606,
"rewards/rejected": -0.21229907870292664,
"step": 470
},
{
"epoch": 0.7423167848699763,
"grad_norm": 0.1854601353406906,
"learning_rate": 3.1751257354536634e-06,
"log_odds_chosen": 0.552662193775177,
"log_odds_ratio": -0.4665309488773346,
"logits/chosen": -0.1528901308774948,
"logits/rejected": -1.371885061264038,
"logps/chosen": -1.4518961906433105,
"logps/rejected": -1.9063466787338257,
"loss": 1.5332,
"nll_loss": 1.4865120649337769,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.14518961310386658,
"rewards/margins": 0.04544505476951599,
"rewards/rejected": -0.19063468277454376,
"step": 471
},
{
"epoch": 0.7438928289992119,
"grad_norm": 0.1899978667497635,
"learning_rate": 3.1706731670526394e-06,
"log_odds_chosen": 0.6217374205589294,
"log_odds_ratio": -0.43196773529052734,
"logits/chosen": -0.2525237500667572,
"logits/rejected": -1.3043723106384277,
"logps/chosen": -1.5042320489883423,
"logps/rejected": -2.0150396823883057,
"loss": 1.5598,
"nll_loss": 1.5165841579437256,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15042319893836975,
"rewards/margins": 0.05108076333999634,
"rewards/rejected": -0.2015039622783661,
"step": 472
},
{
"epoch": 0.7454688731284476,
"grad_norm": 0.1951638162136078,
"learning_rate": 3.166211754774994e-06,
"log_odds_chosen": 0.6629724502563477,
"log_odds_ratio": -0.422730028629303,
"logits/chosen": -0.20018966495990753,
"logits/rejected": -1.4212433099746704,
"logps/chosen": -1.5557457208633423,
"logps/rejected": -2.1103403568267822,
"loss": 1.6437,
"nll_loss": 1.6013872623443604,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15557457506656647,
"rewards/margins": 0.055459462106227875,
"rewards/rejected": -0.21103402972221375,
"step": 473
},
{
"epoch": 0.7470449172576832,
"grad_norm": 0.19168664515018463,
"learning_rate": 3.1617415323245665e-06,
"log_odds_chosen": 0.6726161241531372,
"log_odds_ratio": -0.42290619015693665,
"logits/chosen": -0.2400115728378296,
"logits/rejected": -1.3783491849899292,
"logps/chosen": -1.4658328294754028,
"logps/rejected": -2.0105373859405518,
"loss": 1.5266,
"nll_loss": 1.4843122959136963,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14658328890800476,
"rewards/margins": 0.054470453411340714,
"rewards/rejected": -0.20105375349521637,
"step": 474
},
{
"epoch": 0.7486209613869188,
"grad_norm": 0.18189279735088348,
"learning_rate": 3.157262533471752e-06,
"log_odds_chosen": 0.7140947580337524,
"log_odds_ratio": -0.40235432982444763,
"logits/chosen": -0.15209892392158508,
"logits/rejected": -1.3825089931488037,
"logps/chosen": -1.5169684886932373,
"logps/rejected": -2.1144258975982666,
"loss": 1.5855,
"nll_loss": 1.54523503780365,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15169686079025269,
"rewards/margins": 0.05974572151899338,
"rewards/rejected": -0.21144257485866547,
"step": 475
},
{
"epoch": 0.7501970055161544,
"grad_norm": 0.19609789550304413,
"learning_rate": 3.1527747920532468e-06,
"log_odds_chosen": 0.565768837928772,
"log_odds_ratio": -0.4514068365097046,
"logits/chosen": -0.16712833940982819,
"logits/rejected": -1.1320550441741943,
"logps/chosen": -1.4766864776611328,
"logps/rejected": -1.9391474723815918,
"loss": 1.5527,
"nll_loss": 1.5075533390045166,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14766864478588104,
"rewards/margins": 0.04624609276652336,
"rewards/rejected": -0.1939147412776947,
"step": 476
},
{
"epoch": 0.75177304964539,
"grad_norm": 0.19328206777572632,
"learning_rate": 3.148278341971795e-06,
"log_odds_chosen": 0.65244460105896,
"log_odds_ratio": -0.4249870181083679,
"logits/chosen": -0.22436018288135529,
"logits/rejected": -1.223185420036316,
"logps/chosen": -1.5077931880950928,
"logps/rejected": -2.0486738681793213,
"loss": 1.5739,
"nll_loss": 1.5314006805419922,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15077932178974152,
"rewards/margins": 0.054088056087493896,
"rewards/rejected": -0.2048673778772354,
"step": 477
},
{
"epoch": 0.7533490937746257,
"grad_norm": 0.20173610746860504,
"learning_rate": 3.143773217195929e-06,
"log_odds_chosen": 0.73219895362854,
"log_odds_ratio": -0.4004945158958435,
"logits/chosen": -0.2401634305715561,
"logits/rejected": -1.3643122911453247,
"logps/chosen": -1.5717616081237793,
"logps/rejected": -2.18953800201416,
"loss": 1.6288,
"nll_loss": 1.5887385606765747,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1571761518716812,
"rewards/margins": 0.06177765130996704,
"rewards/rejected": -0.21895381808280945,
"step": 478
},
{
"epoch": 0.7549251379038613,
"grad_norm": 0.19656141102313995,
"learning_rate": 3.139259451759714e-06,
"log_odds_chosen": 0.5744882225990295,
"log_odds_ratio": -0.4510755240917206,
"logits/chosen": -0.24257460236549377,
"logits/rejected": -1.3083908557891846,
"logps/chosen": -1.5712709426879883,
"logps/rejected": -2.050100564956665,
"loss": 1.644,
"nll_loss": 1.5989316701889038,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15712709724903107,
"rewards/margins": 0.047882966697216034,
"rewards/rejected": -0.2050100713968277,
"step": 479
},
{
"epoch": 0.7565011820330969,
"grad_norm": 0.18602579832077026,
"learning_rate": 3.134737079762493e-06,
"log_odds_chosen": 0.6003292202949524,
"log_odds_ratio": -0.4424300193786621,
"logits/chosen": -0.1941242814064026,
"logits/rejected": -1.311800479888916,
"logps/chosen": -1.501556634902954,
"logps/rejected": -1.9944710731506348,
"loss": 1.5673,
"nll_loss": 1.5230939388275146,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15015564858913422,
"rewards/margins": 0.04929143935441971,
"rewards/rejected": -0.19944709539413452,
"step": 480
},
{
"epoch": 0.7580772261623325,
"grad_norm": 0.20337559282779694,
"learning_rate": 3.130206135368626e-06,
"log_odds_chosen": 0.6041734218597412,
"log_odds_ratio": -0.4474312365055084,
"logits/chosen": -0.24175474047660828,
"logits/rejected": -0.9721249341964722,
"logps/chosen": -1.425898790359497,
"logps/rejected": -1.9149250984191895,
"loss": 1.5167,
"nll_loss": 1.471928596496582,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14258988201618195,
"rewards/margins": 0.04890260472893715,
"rewards/rejected": -0.19149249792099,
"step": 481
},
{
"epoch": 0.7596532702915682,
"grad_norm": 0.19193829596042633,
"learning_rate": 3.1256666528072327e-06,
"log_odds_chosen": 0.7867165803909302,
"log_odds_ratio": -0.38077130913734436,
"logits/chosen": -0.23528993129730225,
"logits/rejected": -1.075020670890808,
"logps/chosen": -1.472687840461731,
"logps/rejected": -2.124319553375244,
"loss": 1.5375,
"nll_loss": 1.499396800994873,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14726878702640533,
"rewards/margins": 0.06516318768262863,
"rewards/rejected": -0.21243198215961456,
"step": 482
},
{
"epoch": 0.7612293144208038,
"grad_norm": 0.18599654734134674,
"learning_rate": 3.121118666371937e-06,
"log_odds_chosen": 0.595000147819519,
"log_odds_ratio": -0.4452923834323883,
"logits/chosen": -0.15184305608272552,
"logits/rejected": -1.538995623588562,
"logps/chosen": -1.5383219718933105,
"logps/rejected": -2.033515453338623,
"loss": 1.5951,
"nll_loss": 1.550559639930725,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15383221209049225,
"rewards/margins": 0.049519333988428116,
"rewards/rejected": -0.20335155725479126,
"step": 483
},
{
"epoch": 0.7628053585500394,
"grad_norm": 0.21291442215442657,
"learning_rate": 3.1165622104206034e-06,
"log_odds_chosen": 0.770659863948822,
"log_odds_ratio": -0.3848019242286682,
"logits/chosen": -0.3678995370864868,
"logits/rejected": -1.2183688879013062,
"logps/chosen": -1.4637348651885986,
"logps/rejected": -2.1029365062713623,
"loss": 1.5321,
"nll_loss": 1.4936531782150269,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14637349545955658,
"rewards/margins": 0.06392017006874084,
"rewards/rejected": -0.21029365062713623,
"step": 484
},
{
"epoch": 0.764381402679275,
"grad_norm": 0.18177950382232666,
"learning_rate": 3.1119973193750816e-06,
"log_odds_chosen": 0.6704604029655457,
"log_odds_ratio": -0.41766875982284546,
"logits/chosen": -0.2663368880748749,
"logits/rejected": -1.2851604223251343,
"logps/chosen": -1.475534439086914,
"logps/rejected": -2.027827739715576,
"loss": 1.536,
"nll_loss": 1.4942355155944824,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1475534588098526,
"rewards/margins": 0.05522932484745979,
"rewards/rejected": -0.2027827799320221,
"step": 485
},
{
"epoch": 0.7659574468085106,
"grad_norm": 0.21737127006053925,
"learning_rate": 3.1074240277209408e-06,
"log_odds_chosen": 0.6194983124732971,
"log_odds_ratio": -0.4379619061946869,
"logits/chosen": -0.24437181651592255,
"logits/rejected": -1.2440481185913086,
"logps/chosen": -1.494814157485962,
"logps/rejected": -2.00154709815979,
"loss": 1.5423,
"nll_loss": 1.4985466003417969,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1494814157485962,
"rewards/margins": 0.05067329481244087,
"rewards/rejected": -0.20015469193458557,
"step": 486
},
{
"epoch": 0.7675334909377463,
"grad_norm": 0.20863457024097443,
"learning_rate": 3.102842370007217e-06,
"log_odds_chosen": 0.6833222508430481,
"log_odds_ratio": -0.41733595728874207,
"logits/chosen": -0.16676893830299377,
"logits/rejected": -1.15752375125885,
"logps/chosen": -1.5041608810424805,
"logps/rejected": -2.0647218227386475,
"loss": 1.5703,
"nll_loss": 1.5285258293151855,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1504160761833191,
"rewards/margins": 0.05605611205101013,
"rewards/rejected": -0.20647220313549042,
"step": 487
},
{
"epoch": 0.7691095350669819,
"grad_norm": 0.1944776177406311,
"learning_rate": 3.0982523808461454e-06,
"log_odds_chosen": 0.5529102683067322,
"log_odds_ratio": -0.4605258107185364,
"logits/chosen": -0.15431943535804749,
"logits/rejected": -1.2589623928070068,
"logps/chosen": -1.5484925508499146,
"logps/rejected": -2.0021567344665527,
"loss": 1.6075,
"nll_loss": 1.561496376991272,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15484926104545593,
"rewards/margins": 0.04536642134189606,
"rewards/rejected": -0.2002156674861908,
"step": 488
},
{
"epoch": 0.7706855791962175,
"grad_norm": 0.18545880913734436,
"learning_rate": 3.0936540949129006e-06,
"log_odds_chosen": 0.6402296423912048,
"log_odds_ratio": -0.4339551031589508,
"logits/chosen": -0.2207900583744049,
"logits/rejected": -1.1871285438537598,
"logps/chosen": -1.5241451263427734,
"logps/rejected": -2.057253122329712,
"loss": 1.5905,
"nll_loss": 1.5471105575561523,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15241453051567078,
"rewards/margins": 0.053310781717300415,
"rewards/rejected": -0.20572529733181,
"step": 489
},
{
"epoch": 0.7722616233254531,
"grad_norm": 0.178094744682312,
"learning_rate": 3.0890475469453378e-06,
"log_odds_chosen": 0.7427234053611755,
"log_odds_ratio": -0.40040361881256104,
"logits/chosen": -0.2081877589225769,
"logits/rejected": -1.3924274444580078,
"logps/chosen": -1.4031827449798584,
"logps/rejected": -2.006999969482422,
"loss": 1.4721,
"nll_loss": 1.4321045875549316,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14031827449798584,
"rewards/margins": 0.06038173660635948,
"rewards/rejected": -0.20069998502731323,
"step": 490
},
{
"epoch": 0.7738376674546887,
"grad_norm": 0.18201249837875366,
"learning_rate": 3.0844327717437263e-06,
"log_odds_chosen": 0.6974368691444397,
"log_odds_ratio": -0.41516321897506714,
"logits/chosen": -0.21113747358322144,
"logits/rejected": -1.2565526962280273,
"logps/chosen": -1.4272940158843994,
"logps/rejected": -2.0010313987731934,
"loss": 1.4995,
"nll_loss": 1.4580097198486328,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14272941648960114,
"rewards/margins": 0.057373758405447006,
"rewards/rejected": -0.20010316371917725,
"step": 491
},
{
"epoch": 0.7754137115839244,
"grad_norm": 0.1853957176208496,
"learning_rate": 3.0798098041704892e-06,
"log_odds_chosen": 0.5092182159423828,
"log_odds_ratio": -0.47541776299476624,
"logits/chosen": -0.15838466584682465,
"logits/rejected": -1.3052377700805664,
"logps/chosen": -1.361024260520935,
"logps/rejected": -1.7661519050598145,
"loss": 1.4535,
"nll_loss": 1.405916690826416,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13610242307186127,
"rewards/margins": 0.04051277041435242,
"rewards/rejected": -0.1766151785850525,
"step": 492
},
{
"epoch": 0.77698975571316,
"grad_norm": 0.1893599033355713,
"learning_rate": 3.0751786791499368e-06,
"log_odds_chosen": 0.6547857522964478,
"log_odds_ratio": -0.42588385939598083,
"logits/chosen": -0.17713405191898346,
"logits/rejected": -1.2708368301391602,
"logps/chosen": -1.5448169708251953,
"logps/rejected": -2.0935890674591064,
"loss": 1.6139,
"nll_loss": 1.5712815523147583,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1544817090034485,
"rewards/margins": 0.05487719178199768,
"rewards/rejected": -0.20935890078544617,
"step": 493
},
{
"epoch": 0.7785657998423956,
"grad_norm": 0.18722397089004517,
"learning_rate": 3.070539431668008e-06,
"log_odds_chosen": 0.6233894228935242,
"log_odds_ratio": -0.43421441316604614,
"logits/chosen": -0.1989862620830536,
"logits/rejected": -1.2033601999282837,
"logps/chosen": -1.5463478565216064,
"logps/rejected": -2.0686848163604736,
"loss": 1.6068,
"nll_loss": 1.5633586645126343,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15463480353355408,
"rewards/margins": 0.05223367363214493,
"rewards/rejected": -0.2068684697151184,
"step": 494
},
{
"epoch": 0.7801418439716312,
"grad_norm": 0.19007954001426697,
"learning_rate": 3.0658920967720018e-06,
"log_odds_chosen": 0.7926431894302368,
"log_odds_ratio": -0.38211071491241455,
"logits/chosen": -0.3403100073337555,
"logits/rejected": -1.2223634719848633,
"logps/chosen": -1.4851934909820557,
"logps/rejected": -2.1482720375061035,
"loss": 1.5559,
"nll_loss": 1.5176681280136108,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.148519366979599,
"rewards/margins": 0.06630785018205643,
"rewards/rejected": -0.21482720971107483,
"step": 495
},
{
"epoch": 0.7817178881008668,
"grad_norm": 0.1839301884174347,
"learning_rate": 3.0612367095703116e-06,
"log_odds_chosen": 0.7492038607597351,
"log_odds_ratio": -0.39107295870780945,
"logits/chosen": -0.18261493742465973,
"logits/rejected": -1.4831030368804932,
"logps/chosen": -1.5249770879745483,
"logps/rejected": -2.153897762298584,
"loss": 1.5805,
"nll_loss": 1.5413737297058105,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15249772369861603,
"rewards/margins": 0.06289205700159073,
"rewards/rejected": -0.21538978815078735,
"step": 496
},
{
"epoch": 0.7832939322301025,
"grad_norm": 0.19066324830055237,
"learning_rate": 3.056573305232167e-06,
"log_odds_chosen": 0.7923998832702637,
"log_odds_ratio": -0.380237877368927,
"logits/chosen": -0.20253872871398926,
"logits/rejected": -1.41769278049469,
"logps/chosen": -1.4783601760864258,
"logps/rejected": -2.1322264671325684,
"loss": 1.5446,
"nll_loss": 1.5065717697143555,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14783601462841034,
"rewards/margins": 0.06538661569356918,
"rewards/rejected": -0.21322263777256012,
"step": 497
},
{
"epoch": 0.7848699763593381,
"grad_norm": 0.19244952499866486,
"learning_rate": 3.051901918987359e-06,
"log_odds_chosen": 0.7435587048530579,
"log_odds_ratio": -0.39330264925956726,
"logits/chosen": -0.33268722891807556,
"logits/rejected": -1.3394025564193726,
"logps/chosen": -1.4223885536193848,
"logps/rejected": -2.0310678482055664,
"loss": 1.4913,
"nll_loss": 1.4519734382629395,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14223885536193848,
"rewards/margins": 0.06086793541908264,
"rewards/rejected": -0.20310677587985992,
"step": 498
},
{
"epoch": 0.7864460204885737,
"grad_norm": 0.22321587800979614,
"learning_rate": 3.047222586125979e-06,
"log_odds_chosen": 0.7735339999198914,
"log_odds_ratio": -0.3840080797672272,
"logits/chosen": -0.1548488885164261,
"logits/rejected": -0.9053974151611328,
"logps/chosen": -1.431384801864624,
"logps/rejected": -2.068115234375,
"loss": 1.51,
"nll_loss": 1.4715591669082642,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14313849806785583,
"rewards/margins": 0.06367303431034088,
"rewards/rejected": -0.20681151747703552,
"step": 499
},
{
"epoch": 0.7880220646178093,
"grad_norm": 0.18041643500328064,
"learning_rate": 3.042535341998152e-06,
"log_odds_chosen": 0.5752748847007751,
"log_odds_ratio": -0.4517236649990082,
"logits/chosen": -0.07348179817199707,
"logits/rejected": -1.2616822719573975,
"logps/chosen": -1.578834056854248,
"logps/rejected": -2.061535120010376,
"loss": 1.6247,
"nll_loss": 1.579504370689392,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1578833907842636,
"rewards/margins": 0.04827011376619339,
"rewards/rejected": -0.2061535120010376,
"step": 500
}
],
"logging_steps": 1,
"max_steps": 1270,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"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
}