FatDPO_ORPO_LoRA / trainer_state.json
jeiku's picture
Upload 8 files
0cf8653 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.0,
"eval_steps": 500,
"global_step": 530,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0037735849056603774,
"grad_norm": 1.4419530630111694,
"learning_rate": 4.9905660377358493e-05,
"log_odds_chosen": 0.4804525375366211,
"log_odds_ratio": -0.6413240432739258,
"logits/chosen": 0.3143516182899475,
"logits/rejected": -1.3078216314315796,
"logps/chosen": -1.7236464023590088,
"logps/rejected": -2.0679845809936523,
"loss": 2.2212,
"nll_loss": 2.1570305824279785,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.17236465215682983,
"rewards/margins": 0.03443381190299988,
"rewards/rejected": -0.2067984640598297,
"step": 1
},
{
"epoch": 0.007547169811320755,
"grad_norm": 1.0689440965652466,
"learning_rate": 4.9811320754716985e-05,
"log_odds_chosen": -0.09131823480129242,
"log_odds_ratio": -0.8326093554496765,
"logits/chosen": -0.3561238646507263,
"logits/rejected": -1.7676267623901367,
"logps/chosen": -1.9550193548202515,
"logps/rejected": -1.842592477798462,
"loss": 2.1604,
"nll_loss": 2.077115297317505,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.19550195336341858,
"rewards/margins": -0.011242689564824104,
"rewards/rejected": -0.18425926566123962,
"step": 2
},
{
"epoch": 0.011320754716981131,
"grad_norm": 1.0649460554122925,
"learning_rate": 4.9716981132075476e-05,
"log_odds_chosen": 0.6475443840026855,
"log_odds_ratio": -0.5298449993133545,
"logits/chosen": 0.7758256196975708,
"logits/rejected": -1.0398880243301392,
"logps/chosen": -1.5218658447265625,
"logps/rejected": -1.9791918992996216,
"loss": 1.9187,
"nll_loss": 1.8656777143478394,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1521865725517273,
"rewards/margins": 0.04573261737823486,
"rewards/rejected": -0.19791918992996216,
"step": 3
},
{
"epoch": 0.01509433962264151,
"grad_norm": 0.8160853981971741,
"learning_rate": 4.962264150943397e-05,
"log_odds_chosen": 0.4427639842033386,
"log_odds_ratio": -0.6302679777145386,
"logits/chosen": -1.3039603233337402,
"logits/rejected": -2.81551456451416,
"logps/chosen": -1.551461100578308,
"logps/rejected": -1.879504919052124,
"loss": 1.6384,
"nll_loss": 1.575362205505371,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.15514612197875977,
"rewards/margins": 0.032804377377033234,
"rewards/rejected": -0.1879504919052124,
"step": 4
},
{
"epoch": 0.018867924528301886,
"grad_norm": 0.8768725991249084,
"learning_rate": 4.952830188679246e-05,
"log_odds_chosen": 0.1845681071281433,
"log_odds_ratio": -0.7325757741928101,
"logits/chosen": -0.4272328019142151,
"logits/rejected": -1.7803056240081787,
"logps/chosen": -1.8399394750595093,
"logps/rejected": -1.96200692653656,
"loss": 2.0367,
"nll_loss": 1.9634612798690796,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.18399396538734436,
"rewards/margins": 0.012206735089421272,
"rewards/rejected": -0.19620069861412048,
"step": 5
},
{
"epoch": 0.022641509433962263,
"grad_norm": 3.759185314178467,
"learning_rate": 4.943396226415095e-05,
"log_odds_chosen": 0.4950372576713562,
"log_odds_ratio": -0.5588759183883667,
"logits/chosen": -0.019302427768707275,
"logits/rejected": -1.1816325187683105,
"logps/chosen": -1.6534287929534912,
"logps/rejected": -2.0343308448791504,
"loss": 1.9177,
"nll_loss": 1.8618510961532593,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.16534289717674255,
"rewards/margins": 0.038090191781520844,
"rewards/rejected": -0.2034330666065216,
"step": 6
},
{
"epoch": 0.026415094339622643,
"grad_norm": 0.9324076175689697,
"learning_rate": 4.933962264150943e-05,
"log_odds_chosen": 0.12158789485692978,
"log_odds_ratio": -0.7240866422653198,
"logits/chosen": -0.2951589822769165,
"logits/rejected": -2.193943500518799,
"logps/chosen": -1.9935435056686401,
"logps/rejected": -2.0926284790039062,
"loss": 1.9309,
"nll_loss": 1.8584506511688232,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.1993543654680252,
"rewards/margins": 0.009908500127494335,
"rewards/rejected": -0.20926286280155182,
"step": 7
},
{
"epoch": 0.03018867924528302,
"grad_norm": 0.7746831178665161,
"learning_rate": 4.9245283018867924e-05,
"log_odds_chosen": 1.0372132062911987,
"log_odds_ratio": -0.44096675515174866,
"logits/chosen": -0.7707222700119019,
"logits/rejected": -2.369938850402832,
"logps/chosen": -1.2799935340881348,
"logps/rejected": -2.0016367435455322,
"loss": 1.2616,
"nll_loss": 1.2174644470214844,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.12799936532974243,
"rewards/margins": 0.07216434180736542,
"rewards/rejected": -0.20016369223594666,
"step": 8
},
{
"epoch": 0.033962264150943396,
"grad_norm": 0.7088687419891357,
"learning_rate": 4.9150943396226415e-05,
"log_odds_chosen": 0.1920507550239563,
"log_odds_ratio": -0.729371964931488,
"logits/chosen": -0.6697689294815063,
"logits/rejected": -3.048678398132324,
"logps/chosen": -1.8056210279464722,
"logps/rejected": -1.9406684637069702,
"loss": 1.89,
"nll_loss": 1.8170198202133179,
"rewards/accuracies": 0.4375,
"rewards/chosen": -0.1805620938539505,
"rewards/margins": 0.01350475661456585,
"rewards/rejected": -0.1940668523311615,
"step": 9
},
{
"epoch": 0.03773584905660377,
"grad_norm": 0.7092457413673401,
"learning_rate": 4.9056603773584906e-05,
"log_odds_chosen": 0.8096474409103394,
"log_odds_ratio": -0.4400815963745117,
"logits/chosen": -0.08921952545642853,
"logits/rejected": -2.1929593086242676,
"logps/chosen": -1.4596810340881348,
"logps/rejected": -2.0628461837768555,
"loss": 1.6278,
"nll_loss": 1.583770751953125,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.14596810936927795,
"rewards/margins": 0.060316506773233414,
"rewards/rejected": -0.20628461241722107,
"step": 10
},
{
"epoch": 0.04150943396226415,
"grad_norm": 0.7152612209320068,
"learning_rate": 4.89622641509434e-05,
"log_odds_chosen": 0.5647962689399719,
"log_odds_ratio": -0.5348465442657471,
"logits/chosen": -0.022822290658950806,
"logits/rejected": -0.9105501174926758,
"logps/chosen": -1.738325595855713,
"logps/rejected": -2.1731629371643066,
"loss": 1.715,
"nll_loss": 1.6615500450134277,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.17383255064487457,
"rewards/margins": 0.043483734130859375,
"rewards/rejected": -0.21731628477573395,
"step": 11
},
{
"epoch": 0.045283018867924525,
"grad_norm": 0.49451878666877747,
"learning_rate": 4.886792452830189e-05,
"log_odds_chosen": 0.8111759424209595,
"log_odds_ratio": -0.5735799670219421,
"logits/chosen": -0.1950460523366928,
"logits/rejected": -1.8716578483581543,
"logps/chosen": -1.6253588199615479,
"logps/rejected": -2.223177671432495,
"loss": 1.6836,
"nll_loss": 1.6262810230255127,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.1625358760356903,
"rewards/margins": 0.05978189408779144,
"rewards/rejected": -0.22231778502464294,
"step": 12
},
{
"epoch": 0.04905660377358491,
"grad_norm": 0.49509674310684204,
"learning_rate": 4.877358490566038e-05,
"log_odds_chosen": 1.2461881637573242,
"log_odds_ratio": -0.3561839163303375,
"logits/chosen": -0.05543512478470802,
"logits/rejected": -1.6817309856414795,
"logps/chosen": -1.4552483558654785,
"logps/rejected": -2.393232583999634,
"loss": 1.6718,
"nll_loss": 1.6361980438232422,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.14552482962608337,
"rewards/margins": 0.09379842877388,
"rewards/rejected": -0.23932327330112457,
"step": 13
},
{
"epoch": 0.052830188679245285,
"grad_norm": 0.5647757649421692,
"learning_rate": 4.867924528301887e-05,
"log_odds_chosen": 0.9501932263374329,
"log_odds_ratio": -0.4295998215675354,
"logits/chosen": -0.2805134057998657,
"logits/rejected": -2.1965882778167725,
"logps/chosen": -1.7489657402038574,
"logps/rejected": -2.5288243293762207,
"loss": 1.7557,
"nll_loss": 1.7127894163131714,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17489658296108246,
"rewards/margins": 0.07798586785793304,
"rewards/rejected": -0.2528824508190155,
"step": 14
},
{
"epoch": 0.05660377358490566,
"grad_norm": 0.46898186206817627,
"learning_rate": 4.858490566037736e-05,
"log_odds_chosen": 1.263291597366333,
"log_odds_ratio": -0.4335542321205139,
"logits/chosen": -0.10565178096294403,
"logits/rejected": -1.839094638824463,
"logps/chosen": -1.3841191530227661,
"logps/rejected": -2.2567789554595947,
"loss": 1.4752,
"nll_loss": 1.431824803352356,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.13841190934181213,
"rewards/margins": 0.0872659832239151,
"rewards/rejected": -0.22567789256572723,
"step": 15
},
{
"epoch": 0.06037735849056604,
"grad_norm": 0.43158116936683655,
"learning_rate": 4.849056603773585e-05,
"log_odds_chosen": 0.6095455288887024,
"log_odds_ratio": -0.5792121887207031,
"logits/chosen": -0.17698495090007782,
"logits/rejected": -2.8387160301208496,
"logps/chosen": -2.014704942703247,
"logps/rejected": -2.574517250061035,
"loss": 1.9717,
"nll_loss": 1.9137517213821411,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.2014704942703247,
"rewards/margins": 0.05598121136426926,
"rewards/rejected": -0.25745171308517456,
"step": 16
},
{
"epoch": 0.06415094339622641,
"grad_norm": 0.4689409136772156,
"learning_rate": 4.8396226415094344e-05,
"log_odds_chosen": 1.2785823345184326,
"log_odds_ratio": -0.3914850652217865,
"logits/chosen": -0.17786982655525208,
"logits/rejected": -1.9287208318710327,
"logps/chosen": -1.5383832454681396,
"logps/rejected": -2.564413547515869,
"loss": 1.6308,
"nll_loss": 1.5916929244995117,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.15383832156658173,
"rewards/margins": 0.10260303318500519,
"rewards/rejected": -0.2564413547515869,
"step": 17
},
{
"epoch": 0.06792452830188679,
"grad_norm": 0.4632433354854584,
"learning_rate": 4.8301886792452835e-05,
"log_odds_chosen": 1.8498305082321167,
"log_odds_ratio": -0.2724185287952423,
"logits/chosen": 0.00803515687584877,
"logits/rejected": -1.558810830116272,
"logps/chosen": -1.5008854866027832,
"logps/rejected": -3.094841957092285,
"loss": 1.3575,
"nll_loss": 1.3302488327026367,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15008854866027832,
"rewards/margins": 0.15939566493034363,
"rewards/rejected": -0.30948421359062195,
"step": 18
},
{
"epoch": 0.07169811320754717,
"grad_norm": 0.35216155648231506,
"learning_rate": 4.8207547169811326e-05,
"log_odds_chosen": 2.0445966720581055,
"log_odds_ratio": -0.22769802808761597,
"logits/chosen": -1.055422306060791,
"logits/rejected": -2.8562209606170654,
"logps/chosen": -1.4856079816818237,
"logps/rejected": -3.2529525756835938,
"loss": 1.6494,
"nll_loss": 1.626587152481079,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1485608071088791,
"rewards/margins": 0.17673446238040924,
"rewards/rejected": -0.32529526948928833,
"step": 19
},
{
"epoch": 0.07547169811320754,
"grad_norm": 0.3028462827205658,
"learning_rate": 4.811320754716982e-05,
"log_odds_chosen": 1.1802749633789062,
"log_odds_ratio": -0.4493004083633423,
"logits/chosen": -1.202689290046692,
"logits/rejected": -3.1590261459350586,
"logps/chosen": -1.8469630479812622,
"logps/rejected": -2.9175806045532227,
"loss": 1.6889,
"nll_loss": 1.6439603567123413,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.18469631671905518,
"rewards/margins": 0.10706175863742828,
"rewards/rejected": -0.29175806045532227,
"step": 20
},
{
"epoch": 0.07924528301886792,
"grad_norm": 0.28935325145721436,
"learning_rate": 4.80188679245283e-05,
"log_odds_chosen": 2.2533977031707764,
"log_odds_ratio": -0.21968787908554077,
"logits/chosen": -0.12538594007492065,
"logits/rejected": -1.9605629444122314,
"logps/chosen": -1.555745244026184,
"logps/rejected": -3.573531150817871,
"loss": 1.479,
"nll_loss": 1.4570300579071045,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15557453036308289,
"rewards/margins": 0.2017785757780075,
"rewards/rejected": -0.3573530912399292,
"step": 21
},
{
"epoch": 0.0830188679245283,
"grad_norm": 0.39032241702079773,
"learning_rate": 4.792452830188679e-05,
"log_odds_chosen": 1.7034038305282593,
"log_odds_ratio": -0.41050586104393005,
"logits/chosen": -0.2803301215171814,
"logits/rejected": -2.3801870346069336,
"logps/chosen": -1.699625015258789,
"logps/rejected": -3.263505458831787,
"loss": 1.7973,
"nll_loss": 1.7562379837036133,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.16996252536773682,
"rewards/margins": 0.1563880443572998,
"rewards/rejected": -0.3263505697250366,
"step": 22
},
{
"epoch": 0.08679245283018867,
"grad_norm": 0.2868567705154419,
"learning_rate": 4.7830188679245284e-05,
"log_odds_chosen": 2.2137861251831055,
"log_odds_ratio": -0.3145076334476471,
"logits/chosen": -1.0021281242370605,
"logits/rejected": -2.4567623138427734,
"logps/chosen": -1.621106743812561,
"logps/rejected": -3.6742091178894043,
"loss": 1.5856,
"nll_loss": 1.5541696548461914,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.16211068630218506,
"rewards/margins": 0.20531021058559418,
"rewards/rejected": -0.36742085218429565,
"step": 23
},
{
"epoch": 0.09056603773584905,
"grad_norm": 0.3213948607444763,
"learning_rate": 4.7735849056603775e-05,
"log_odds_chosen": 3.161144256591797,
"log_odds_ratio": -0.2373344600200653,
"logits/chosen": -0.881264865398407,
"logits/rejected": -2.189188241958618,
"logps/chosen": -1.3406716585159302,
"logps/rejected": -3.9409539699554443,
"loss": 1.5735,
"nll_loss": 1.5498067140579224,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.13406717777252197,
"rewards/margins": 0.2600281834602356,
"rewards/rejected": -0.39409539103507996,
"step": 24
},
{
"epoch": 0.09433962264150944,
"grad_norm": 0.3009367287158966,
"learning_rate": 4.7641509433962266e-05,
"log_odds_chosen": 2.8621065616607666,
"log_odds_ratio": -0.2253154218196869,
"logits/chosen": -0.2625047266483307,
"logits/rejected": -2.8666999340057373,
"logps/chosen": -1.7110705375671387,
"logps/rejected": -4.400345325469971,
"loss": 1.7058,
"nll_loss": 1.683306097984314,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.17110705375671387,
"rewards/margins": 0.2689274847507477,
"rewards/rejected": -0.44003453850746155,
"step": 25
},
{
"epoch": 0.09811320754716982,
"grad_norm": 0.3267085552215576,
"learning_rate": 4.754716981132076e-05,
"log_odds_chosen": 2.815523386001587,
"log_odds_ratio": -0.2813441753387451,
"logits/chosen": -1.2153959274291992,
"logits/rejected": -3.1709775924682617,
"logps/chosen": -1.6071617603302002,
"logps/rejected": -4.192075729370117,
"loss": 1.642,
"nll_loss": 1.6138246059417725,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.16071617603302002,
"rewards/margins": 0.2584913671016693,
"rewards/rejected": -0.41920754313468933,
"step": 26
},
{
"epoch": 0.1018867924528302,
"grad_norm": 0.4692343771457672,
"learning_rate": 4.745283018867925e-05,
"log_odds_chosen": 3.2350544929504395,
"log_odds_ratio": -0.22966182231903076,
"logits/chosen": -0.6714242696762085,
"logits/rejected": -3.4772703647613525,
"logps/chosen": -1.573210597038269,
"logps/rejected": -4.57033634185791,
"loss": 1.5627,
"nll_loss": 1.5397582054138184,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.15732106566429138,
"rewards/margins": 0.29971253871917725,
"rewards/rejected": -0.45703360438346863,
"step": 27
},
{
"epoch": 0.10566037735849057,
"grad_norm": 0.31730151176452637,
"learning_rate": 4.735849056603774e-05,
"log_odds_chosen": 3.0557804107666016,
"log_odds_ratio": -0.19463254511356354,
"logits/chosen": -0.8169313669204712,
"logits/rejected": -1.874739646911621,
"logps/chosen": -1.7243003845214844,
"logps/rejected": -4.595163345336914,
"loss": 1.5311,
"nll_loss": 1.5116479396820068,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17243005335330963,
"rewards/margins": 0.2870863080024719,
"rewards/rejected": -0.45951637625694275,
"step": 28
},
{
"epoch": 0.10943396226415095,
"grad_norm": 0.25958406925201416,
"learning_rate": 4.726415094339623e-05,
"log_odds_chosen": 4.204806327819824,
"log_odds_ratio": -0.06744246184825897,
"logits/chosen": -0.7214003801345825,
"logits/rejected": -3.8571434020996094,
"logps/chosen": -1.5588371753692627,
"logps/rejected": -5.471315383911133,
"loss": 1.5991,
"nll_loss": 1.5923649072647095,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15588372945785522,
"rewards/margins": 0.39124780893325806,
"rewards/rejected": -0.5471315383911133,
"step": 29
},
{
"epoch": 0.11320754716981132,
"grad_norm": 0.2962592542171478,
"learning_rate": 4.716981132075472e-05,
"log_odds_chosen": 2.306314468383789,
"log_odds_ratio": -0.5868710279464722,
"logits/chosen": -0.9562402963638306,
"logits/rejected": -4.068589687347412,
"logps/chosen": -1.8381496667861938,
"logps/rejected": -4.1114959716796875,
"loss": 1.7932,
"nll_loss": 1.7344965934753418,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.18381497263908386,
"rewards/margins": 0.2273346334695816,
"rewards/rejected": -0.41114962100982666,
"step": 30
},
{
"epoch": 0.1169811320754717,
"grad_norm": 0.3253747224807739,
"learning_rate": 4.707547169811321e-05,
"log_odds_chosen": 3.867879867553711,
"log_odds_ratio": -0.2663135528564453,
"logits/chosen": -1.6323354244232178,
"logits/rejected": -2.8868775367736816,
"logps/chosen": -1.3807705640792847,
"logps/rejected": -4.98048210144043,
"loss": 1.3989,
"nll_loss": 1.3722314834594727,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.13807706534862518,
"rewards/margins": 0.35997116565704346,
"rewards/rejected": -0.49804821610450745,
"step": 31
},
{
"epoch": 0.12075471698113208,
"grad_norm": 0.25523456931114197,
"learning_rate": 4.6981132075471704e-05,
"log_odds_chosen": 4.133279800415039,
"log_odds_ratio": -0.20295441150665283,
"logits/chosen": -0.9551544785499573,
"logits/rejected": -4.06210470199585,
"logps/chosen": -1.683975338935852,
"logps/rejected": -5.5400261878967285,
"loss": 1.7012,
"nll_loss": 1.6808902025222778,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16839754581451416,
"rewards/margins": 0.3856050968170166,
"rewards/rejected": -0.5540026426315308,
"step": 32
},
{
"epoch": 0.12452830188679245,
"grad_norm": 0.2824070155620575,
"learning_rate": 4.6886792452830195e-05,
"log_odds_chosen": 4.0810699462890625,
"log_odds_ratio": -0.24137283861637115,
"logits/chosen": -0.1275748610496521,
"logits/rejected": -1.968299150466919,
"logps/chosen": -1.4146742820739746,
"logps/rejected": -5.19221830368042,
"loss": 1.4468,
"nll_loss": 1.422704815864563,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.14146743714809418,
"rewards/margins": 0.37775442004203796,
"rewards/rejected": -0.519221842288971,
"step": 33
},
{
"epoch": 0.12830188679245283,
"grad_norm": 0.3047614097595215,
"learning_rate": 4.679245283018868e-05,
"log_odds_chosen": 4.300058364868164,
"log_odds_ratio": -0.2853168845176697,
"logits/chosen": -2.589682102203369,
"logits/rejected": -3.8946940898895264,
"logps/chosen": -1.4936325550079346,
"logps/rejected": -5.401619911193848,
"loss": 1.6643,
"nll_loss": 1.6357437372207642,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.14936324954032898,
"rewards/margins": 0.3907987177371979,
"rewards/rejected": -0.5401619672775269,
"step": 34
},
{
"epoch": 0.1320754716981132,
"grad_norm": 0.298910528421402,
"learning_rate": 4.669811320754717e-05,
"log_odds_chosen": 6.439607620239258,
"log_odds_ratio": -0.03047255240380764,
"logits/chosen": -1.3439738750457764,
"logits/rejected": -3.9568533897399902,
"logps/chosen": -1.3088639974594116,
"logps/rejected": -7.318572044372559,
"loss": 1.6855,
"nll_loss": 1.6824290752410889,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13088640570640564,
"rewards/margins": 0.6009708642959595,
"rewards/rejected": -0.7318572402000427,
"step": 35
},
{
"epoch": 0.13584905660377358,
"grad_norm": 0.4024220407009125,
"learning_rate": 4.660377358490566e-05,
"log_odds_chosen": 5.031867980957031,
"log_odds_ratio": -0.17651405930519104,
"logits/chosen": -0.4531555771827698,
"logits/rejected": -3.1221275329589844,
"logps/chosen": -1.6552730798721313,
"logps/rejected": -6.324914932250977,
"loss": 1.7674,
"nll_loss": 1.749765157699585,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16552734375,
"rewards/margins": 0.46696415543556213,
"rewards/rejected": -0.6324914693832397,
"step": 36
},
{
"epoch": 0.13962264150943396,
"grad_norm": 0.2770419418811798,
"learning_rate": 4.650943396226415e-05,
"log_odds_chosen": 5.099140167236328,
"log_odds_ratio": -0.25729480385780334,
"logits/chosen": -1.506839632987976,
"logits/rejected": -3.454437732696533,
"logps/chosen": -1.6088337898254395,
"logps/rejected": -6.49648904800415,
"loss": 1.5488,
"nll_loss": 1.5230274200439453,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.160883367061615,
"rewards/margins": 0.48876553773880005,
"rewards/rejected": -0.649648904800415,
"step": 37
},
{
"epoch": 0.14339622641509434,
"grad_norm": 0.31073063611984253,
"learning_rate": 4.641509433962264e-05,
"log_odds_chosen": 5.445122241973877,
"log_odds_ratio": -0.18372483551502228,
"logits/chosen": -1.1133687496185303,
"logits/rejected": -3.4876227378845215,
"logps/chosen": -1.4217114448547363,
"logps/rejected": -6.529870986938477,
"loss": 1.4282,
"nll_loss": 1.4098114967346191,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.14217115938663483,
"rewards/margins": 0.5108159184455872,
"rewards/rejected": -0.6529870629310608,
"step": 38
},
{
"epoch": 0.1471698113207547,
"grad_norm": 0.29280173778533936,
"learning_rate": 4.6320754716981134e-05,
"log_odds_chosen": 6.195460319519043,
"log_odds_ratio": -0.035549942404031754,
"logits/chosen": -1.361371636390686,
"logits/rejected": -4.5884246826171875,
"logps/chosen": -1.7262309789657593,
"logps/rejected": -7.716752529144287,
"loss": 1.5956,
"nll_loss": 1.5920307636260986,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17262309789657593,
"rewards/margins": 0.5990520715713501,
"rewards/rejected": -0.7716752290725708,
"step": 39
},
{
"epoch": 0.1509433962264151,
"grad_norm": 0.28057682514190674,
"learning_rate": 4.6226415094339625e-05,
"log_odds_chosen": 6.020131587982178,
"log_odds_ratio": -0.03772380203008652,
"logits/chosen": -1.2496304512023926,
"logits/rejected": -4.043735504150391,
"logps/chosen": -1.566927433013916,
"logps/rejected": -7.254069805145264,
"loss": 1.4067,
"nll_loss": 1.402917742729187,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1566927433013916,
"rewards/margins": 0.5687142610549927,
"rewards/rejected": -0.7254070043563843,
"step": 40
},
{
"epoch": 0.15471698113207547,
"grad_norm": 0.2948552072048187,
"learning_rate": 4.6132075471698117e-05,
"log_odds_chosen": 4.575827598571777,
"log_odds_ratio": -0.2891075015068054,
"logits/chosen": -0.34096649289131165,
"logits/rejected": -2.514559507369995,
"logps/chosen": -1.75924813747406,
"logps/rejected": -6.152600288391113,
"loss": 1.6047,
"nll_loss": 1.575812816619873,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.17592480778694153,
"rewards/margins": 0.4393352270126343,
"rewards/rejected": -0.6152600049972534,
"step": 41
},
{
"epoch": 0.15849056603773584,
"grad_norm": 0.3365479111671448,
"learning_rate": 4.603773584905661e-05,
"log_odds_chosen": 6.562654972076416,
"log_odds_ratio": -0.16193996369838715,
"logits/chosen": -1.5872305631637573,
"logits/rejected": -3.203104019165039,
"logps/chosen": -1.6002566814422607,
"logps/rejected": -7.7676682472229,
"loss": 1.7498,
"nll_loss": 1.7336182594299316,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1600256711244583,
"rewards/margins": 0.6167411208152771,
"rewards/rejected": -0.7767667770385742,
"step": 42
},
{
"epoch": 0.16226415094339622,
"grad_norm": 0.28234297037124634,
"learning_rate": 4.59433962264151e-05,
"log_odds_chosen": 5.429837226867676,
"log_odds_ratio": -0.21709167957305908,
"logits/chosen": 0.13629187643527985,
"logits/rejected": -3.1108312606811523,
"logps/chosen": -1.7401251792907715,
"logps/rejected": -7.000548362731934,
"loss": 1.5603,
"nll_loss": 1.538613200187683,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17401251196861267,
"rewards/margins": 0.5260423421859741,
"rewards/rejected": -0.7000548243522644,
"step": 43
},
{
"epoch": 0.1660377358490566,
"grad_norm": 0.30918800830841064,
"learning_rate": 4.584905660377359e-05,
"log_odds_chosen": 5.783122539520264,
"log_odds_ratio": -0.11934540420770645,
"logits/chosen": -0.9094568490982056,
"logits/rejected": -3.6608901023864746,
"logps/chosen": -1.6641435623168945,
"logps/rejected": -7.227668285369873,
"loss": 1.5404,
"nll_loss": 1.5284289121627808,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16641436517238617,
"rewards/margins": 0.5563524961471558,
"rewards/rejected": -0.7227668762207031,
"step": 44
},
{
"epoch": 0.16981132075471697,
"grad_norm": 0.3235601782798767,
"learning_rate": 4.575471698113208e-05,
"log_odds_chosen": 6.502612113952637,
"log_odds_ratio": -0.10429678857326508,
"logits/chosen": -1.07891047000885,
"logits/rejected": -3.1915483474731445,
"logps/chosen": -1.5495665073394775,
"logps/rejected": -7.755338668823242,
"loss": 1.548,
"nll_loss": 1.5375723838806152,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1549566686153412,
"rewards/margins": 0.6205772161483765,
"rewards/rejected": -0.7755338549613953,
"step": 45
},
{
"epoch": 0.17358490566037735,
"grad_norm": 0.26763808727264404,
"learning_rate": 4.566037735849057e-05,
"log_odds_chosen": 6.199982166290283,
"log_odds_ratio": -0.07396085560321808,
"logits/chosen": -1.4093493223190308,
"logits/rejected": -5.298598289489746,
"logps/chosen": -1.7217578887939453,
"logps/rejected": -7.687525749206543,
"loss": 1.5789,
"nll_loss": 1.571506381034851,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.172175794839859,
"rewards/margins": 0.5965768098831177,
"rewards/rejected": -0.7687525749206543,
"step": 46
},
{
"epoch": 0.17735849056603772,
"grad_norm": 0.28478890657424927,
"learning_rate": 4.556603773584906e-05,
"log_odds_chosen": 5.49038028717041,
"log_odds_ratio": -0.17011021077632904,
"logits/chosen": -0.664116621017456,
"logits/rejected": -3.217937469482422,
"logps/chosen": -2.0000357627868652,
"logps/rejected": -7.373146057128906,
"loss": 1.7461,
"nll_loss": 1.7290459871292114,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.20000356435775757,
"rewards/margins": 0.5373110771179199,
"rewards/rejected": -0.7373145818710327,
"step": 47
},
{
"epoch": 0.1811320754716981,
"grad_norm": 0.29675430059432983,
"learning_rate": 4.547169811320755e-05,
"log_odds_chosen": 6.941376209259033,
"log_odds_ratio": -0.07056228816509247,
"logits/chosen": -0.9245410561561584,
"logits/rejected": -3.209742546081543,
"logps/chosen": -1.5620588064193726,
"logps/rejected": -8.255936622619629,
"loss": 1.6204,
"nll_loss": 1.6133677959442139,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1562058925628662,
"rewards/margins": 0.6693878173828125,
"rewards/rejected": -0.8255936503410339,
"step": 48
},
{
"epoch": 0.18490566037735848,
"grad_norm": 0.3024120330810547,
"learning_rate": 4.537735849056604e-05,
"log_odds_chosen": 7.8667497634887695,
"log_odds_ratio": -0.11265160888433456,
"logits/chosen": -1.589120864868164,
"logits/rejected": -4.3788957595825195,
"logps/chosen": -1.7432844638824463,
"logps/rejected": -9.341209411621094,
"loss": 1.7561,
"nll_loss": 1.7448220252990723,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.17432843148708344,
"rewards/margins": 0.7597925662994385,
"rewards/rejected": -0.9341210126876831,
"step": 49
},
{
"epoch": 0.18867924528301888,
"grad_norm": 0.27790966629981995,
"learning_rate": 4.528301886792453e-05,
"log_odds_chosen": 9.197233200073242,
"log_odds_ratio": -0.0743798017501831,
"logits/chosen": -2.399343729019165,
"logits/rejected": -5.902744770050049,
"logps/chosen": -1.7229493856430054,
"logps/rejected": -10.651434898376465,
"loss": 1.6465,
"nll_loss": 1.6390764713287354,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.17229494452476501,
"rewards/margins": 0.8928486108779907,
"rewards/rejected": -1.0651434659957886,
"step": 50
},
{
"epoch": 0.19245283018867926,
"grad_norm": 0.28730452060699463,
"learning_rate": 4.518867924528302e-05,
"log_odds_chosen": 7.1606950759887695,
"log_odds_ratio": -0.15011747181415558,
"logits/chosen": -0.7371699810028076,
"logits/rejected": -3.6732337474823,
"logps/chosen": -1.4196903705596924,
"logps/rejected": -8.275727272033691,
"loss": 1.5138,
"nll_loss": 1.4987900257110596,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.14196905493736267,
"rewards/margins": 0.685603678226471,
"rewards/rejected": -0.8275727033615112,
"step": 51
},
{
"epoch": 0.19622641509433963,
"grad_norm": 0.3072900176048279,
"learning_rate": 4.509433962264151e-05,
"log_odds_chosen": 9.726778030395508,
"log_odds_ratio": -0.0466623455286026,
"logits/chosen": -2.5045127868652344,
"logits/rejected": -5.068806171417236,
"logps/chosen": -1.6165649890899658,
"logps/rejected": -11.054706573486328,
"loss": 1.4102,
"nll_loss": 1.4055166244506836,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16165651381015778,
"rewards/margins": 0.9438142776489258,
"rewards/rejected": -1.1054707765579224,
"step": 52
},
{
"epoch": 0.2,
"grad_norm": 0.2943893074989319,
"learning_rate": 4.5e-05,
"log_odds_chosen": 10.128807067871094,
"log_odds_ratio": -0.0014914250932633877,
"logits/chosen": -1.9018776416778564,
"logits/rejected": -4.2667951583862305,
"logps/chosen": -1.5886954069137573,
"logps/rejected": -11.448442459106445,
"loss": 1.6031,
"nll_loss": 1.6029250621795654,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15886953473091125,
"rewards/margins": 0.9859746694564819,
"rewards/rejected": -1.1448442935943604,
"step": 53
},
{
"epoch": 0.2037735849056604,
"grad_norm": 0.2680610418319702,
"learning_rate": 4.4905660377358494e-05,
"log_odds_chosen": 8.853492736816406,
"log_odds_ratio": -0.04176551476120949,
"logits/chosen": -1.450791835784912,
"logits/rejected": -5.139822006225586,
"logps/chosen": -1.6527010202407837,
"logps/rejected": -10.218703269958496,
"loss": 1.6926,
"nll_loss": 1.6884214878082275,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16527009010314941,
"rewards/margins": 0.8566002249717712,
"rewards/rejected": -1.0218703746795654,
"step": 54
},
{
"epoch": 0.20754716981132076,
"grad_norm": 0.32189199328422546,
"learning_rate": 4.4811320754716985e-05,
"log_odds_chosen": 11.52000617980957,
"log_odds_ratio": -0.000579544750507921,
"logits/chosen": -1.3866709470748901,
"logits/rejected": -3.310800790786743,
"logps/chosen": -1.6044374704360962,
"logps/rejected": -12.8070068359375,
"loss": 1.527,
"nll_loss": 1.5269696712493896,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1604437530040741,
"rewards/margins": 1.120257019996643,
"rewards/rejected": -1.2807008028030396,
"step": 55
},
{
"epoch": 0.21132075471698114,
"grad_norm": 0.2579837739467621,
"learning_rate": 4.4716981132075476e-05,
"log_odds_chosen": 8.566587448120117,
"log_odds_ratio": -0.025317970663309097,
"logits/chosen": -0.9709105491638184,
"logits/rejected": -5.532227993011475,
"logps/chosen": -1.6428523063659668,
"logps/rejected": -9.977346420288086,
"loss": 1.5057,
"nll_loss": 1.5031633377075195,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16428521275520325,
"rewards/margins": 0.8334494829177856,
"rewards/rejected": -0.9977346658706665,
"step": 56
},
{
"epoch": 0.21509433962264152,
"grad_norm": 0.32639437913894653,
"learning_rate": 4.462264150943397e-05,
"log_odds_chosen": 10.306001663208008,
"log_odds_ratio": -0.03766224905848503,
"logits/chosen": -1.1223177909851074,
"logits/rejected": -2.9572572708129883,
"logps/chosen": -1.6183445453643799,
"logps/rejected": -11.599272727966309,
"loss": 1.5893,
"nll_loss": 1.5855064392089844,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1618344485759735,
"rewards/margins": 0.998092770576477,
"rewards/rejected": -1.159927248954773,
"step": 57
},
{
"epoch": 0.2188679245283019,
"grad_norm": 0.25903022289276123,
"learning_rate": 4.452830188679246e-05,
"log_odds_chosen": 8.881806373596191,
"log_odds_ratio": -0.10347943007946014,
"logits/chosen": -0.9591866731643677,
"logits/rejected": -4.289064407348633,
"logps/chosen": -1.6897010803222656,
"logps/rejected": -10.346776962280273,
"loss": 1.6166,
"nll_loss": 1.606278419494629,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16897010803222656,
"rewards/margins": 0.8657075762748718,
"rewards/rejected": -1.0346777439117432,
"step": 58
},
{
"epoch": 0.22264150943396227,
"grad_norm": 0.2665920853614807,
"learning_rate": 4.443396226415095e-05,
"log_odds_chosen": 11.143839836120605,
"log_odds_ratio": -0.01865122653543949,
"logits/chosen": -2.2705538272857666,
"logits/rejected": -4.639202117919922,
"logps/chosen": -1.646267056465149,
"logps/rejected": -12.43774127960205,
"loss": 1.5823,
"nll_loss": 1.58046555519104,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16462671756744385,
"rewards/margins": 1.0791475772857666,
"rewards/rejected": -1.243774175643921,
"step": 59
},
{
"epoch": 0.22641509433962265,
"grad_norm": 0.2563314139842987,
"learning_rate": 4.433962264150944e-05,
"log_odds_chosen": 11.137109756469727,
"log_odds_ratio": -0.03539396822452545,
"logits/chosen": -0.5501875281333923,
"logits/rejected": -4.503236770629883,
"logps/chosen": -1.5203689336776733,
"logps/rejected": -12.256806373596191,
"loss": 1.5345,
"nll_loss": 1.5309462547302246,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1520369052886963,
"rewards/margins": 1.073643684387207,
"rewards/rejected": -1.2256807088851929,
"step": 60
},
{
"epoch": 0.23018867924528302,
"grad_norm": 0.3310481011867523,
"learning_rate": 4.4245283018867925e-05,
"log_odds_chosen": 10.03518295288086,
"log_odds_ratio": -0.09628309309482574,
"logits/chosen": -1.4740400314331055,
"logits/rejected": -4.722970962524414,
"logps/chosen": -1.992842435836792,
"logps/rejected": -11.883909225463867,
"loss": 1.7703,
"nll_loss": 1.7606245279312134,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.19928425550460815,
"rewards/margins": 0.9891066551208496,
"rewards/rejected": -1.1883909702301025,
"step": 61
},
{
"epoch": 0.2339622641509434,
"grad_norm": 0.3190458416938782,
"learning_rate": 4.4150943396226416e-05,
"log_odds_chosen": 9.210914611816406,
"log_odds_ratio": -0.040729597210884094,
"logits/chosen": -0.7013575434684753,
"logits/rejected": -3.3500924110412598,
"logps/chosen": -1.950926423072815,
"logps/rejected": -10.972383499145508,
"loss": 1.752,
"nll_loss": 1.7479382753372192,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19509264826774597,
"rewards/margins": 0.9021456837654114,
"rewards/rejected": -1.097238302230835,
"step": 62
},
{
"epoch": 0.23773584905660378,
"grad_norm": 0.3573385179042816,
"learning_rate": 4.405660377358491e-05,
"log_odds_chosen": 11.01914119720459,
"log_odds_ratio": -0.06296153366565704,
"logits/chosen": -1.7574726343154907,
"logits/rejected": -3.601536989212036,
"logps/chosen": -1.648729920387268,
"logps/rejected": -12.339969635009766,
"loss": 1.528,
"nll_loss": 1.5217459201812744,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16487298905849457,
"rewards/margins": 1.0691239833831787,
"rewards/rejected": -1.233996868133545,
"step": 63
},
{
"epoch": 0.24150943396226415,
"grad_norm": 0.2962454557418823,
"learning_rate": 4.39622641509434e-05,
"log_odds_chosen": 10.007362365722656,
"log_odds_ratio": -0.043272338807582855,
"logits/chosen": -1.2760871648788452,
"logits/rejected": -4.875994682312012,
"logps/chosen": -1.6131874322891235,
"logps/rejected": -11.287424087524414,
"loss": 1.5696,
"nll_loss": 1.5652244091033936,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16131874918937683,
"rewards/margins": 0.967423677444458,
"rewards/rejected": -1.1287423372268677,
"step": 64
},
{
"epoch": 0.24528301886792453,
"grad_norm": 0.3105560541152954,
"learning_rate": 4.386792452830189e-05,
"log_odds_chosen": 11.872138977050781,
"log_odds_ratio": -0.07399442046880722,
"logits/chosen": -0.6608960628509521,
"logits/rejected": -3.028031349182129,
"logps/chosen": -1.554811716079712,
"logps/rejected": -13.131336212158203,
"loss": 1.3613,
"nll_loss": 1.3538590669631958,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15548115968704224,
"rewards/margins": 1.1576523780822754,
"rewards/rejected": -1.313133716583252,
"step": 65
},
{
"epoch": 0.2490566037735849,
"grad_norm": 0.27170437574386597,
"learning_rate": 4.377358490566038e-05,
"log_odds_chosen": 9.97773265838623,
"log_odds_ratio": -0.05262986570596695,
"logits/chosen": -1.7461464405059814,
"logits/rejected": -3.9088220596313477,
"logps/chosen": -1.6022846698760986,
"logps/rejected": -11.324554443359375,
"loss": 1.5483,
"nll_loss": 1.542998194694519,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16022847592830658,
"rewards/margins": 0.9722269177436829,
"rewards/rejected": -1.1324553489685059,
"step": 66
},
{
"epoch": 0.2528301886792453,
"grad_norm": 0.34835413098335266,
"learning_rate": 4.367924528301887e-05,
"log_odds_chosen": 9.539581298828125,
"log_odds_ratio": -0.11755736172199249,
"logits/chosen": -1.313893437385559,
"logits/rejected": -3.734592914581299,
"logps/chosen": -1.4113194942474365,
"logps/rejected": -10.656464576721191,
"loss": 1.3774,
"nll_loss": 1.3656796216964722,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1411319375038147,
"rewards/margins": 0.9245145320892334,
"rewards/rejected": -1.0656465291976929,
"step": 67
},
{
"epoch": 0.25660377358490566,
"grad_norm": 0.2896377444267273,
"learning_rate": 4.358490566037736e-05,
"log_odds_chosen": 10.555123329162598,
"log_odds_ratio": -0.08397966623306274,
"logits/chosen": -0.9625846743583679,
"logits/rejected": -3.1470422744750977,
"logps/chosen": -1.5027086734771729,
"logps/rejected": -11.76999282836914,
"loss": 1.5365,
"nll_loss": 1.5281280279159546,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15027087926864624,
"rewards/margins": 1.0267283916473389,
"rewards/rejected": -1.1769993305206299,
"step": 68
},
{
"epoch": 0.26037735849056604,
"grad_norm": 0.2656661868095398,
"learning_rate": 4.3490566037735853e-05,
"log_odds_chosen": 11.847007751464844,
"log_odds_ratio": -0.0024143236223608255,
"logits/chosen": -2.249497890472412,
"logits/rejected": -4.6625542640686035,
"logps/chosen": -1.611348271369934,
"logps/rejected": -13.128585815429688,
"loss": 1.3686,
"nll_loss": 1.368332862854004,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16113483905792236,
"rewards/margins": 1.1517237424850464,
"rewards/rejected": -1.3128585815429688,
"step": 69
},
{
"epoch": 0.2641509433962264,
"grad_norm": 0.2848161458969116,
"learning_rate": 4.3396226415094345e-05,
"log_odds_chosen": 11.41952896118164,
"log_odds_ratio": -0.06205465644598007,
"logits/chosen": -1.0198203325271606,
"logits/rejected": -4.255819320678711,
"logps/chosen": -1.6085125207901,
"logps/rejected": -12.758248329162598,
"loss": 1.5198,
"nll_loss": 1.5136139392852783,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16085125505924225,
"rewards/margins": 1.114973545074463,
"rewards/rejected": -1.2758249044418335,
"step": 70
},
{
"epoch": 0.2679245283018868,
"grad_norm": 0.3033360540866852,
"learning_rate": 4.3301886792452836e-05,
"log_odds_chosen": 8.192865371704102,
"log_odds_ratio": -0.06372419744729996,
"logits/chosen": -2.1323366165161133,
"logits/rejected": -4.94224214553833,
"logps/chosen": -1.7140583992004395,
"logps/rejected": -9.679903030395508,
"loss": 1.6245,
"nll_loss": 1.6181257963180542,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1714058518409729,
"rewards/margins": 0.7965843677520752,
"rewards/rejected": -0.9679902791976929,
"step": 71
},
{
"epoch": 0.27169811320754716,
"grad_norm": 0.35890570282936096,
"learning_rate": 4.320754716981133e-05,
"log_odds_chosen": 9.654875755310059,
"log_odds_ratio": -0.13519693911075592,
"logits/chosen": -0.6182310581207275,
"logits/rejected": -2.691154718399048,
"logps/chosen": -1.8398606777191162,
"logps/rejected": -11.286558151245117,
"loss": 1.5057,
"nll_loss": 1.492200255393982,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.18398606777191162,
"rewards/margins": 0.9446697235107422,
"rewards/rejected": -1.1286557912826538,
"step": 72
},
{
"epoch": 0.27547169811320754,
"grad_norm": 1.4705555438995361,
"learning_rate": 4.311320754716982e-05,
"log_odds_chosen": 13.209151268005371,
"log_odds_ratio": -0.0005382616654969752,
"logits/chosen": -0.8367394804954529,
"logits/rejected": -4.00326681137085,
"logps/chosen": -1.536256194114685,
"logps/rejected": -14.438907623291016,
"loss": 1.4609,
"nll_loss": 1.4608838558197021,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15362560749053955,
"rewards/margins": 1.2902652025222778,
"rewards/rejected": -1.4438908100128174,
"step": 73
},
{
"epoch": 0.2792452830188679,
"grad_norm": 0.33669015765190125,
"learning_rate": 4.301886792452831e-05,
"log_odds_chosen": 11.09627628326416,
"log_odds_ratio": -0.09562374651432037,
"logits/chosen": -1.0604119300842285,
"logits/rejected": -2.2503695487976074,
"logps/chosen": -1.7556407451629639,
"logps/rejected": -12.555806159973145,
"loss": 1.3805,
"nll_loss": 1.3709644079208374,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.17556408047676086,
"rewards/margins": 1.0800164937973022,
"rewards/rejected": -1.2555806636810303,
"step": 74
},
{
"epoch": 0.2830188679245283,
"grad_norm": 0.26332709193229675,
"learning_rate": 4.292452830188679e-05,
"log_odds_chosen": 11.302566528320312,
"log_odds_ratio": -0.013619111850857735,
"logits/chosen": -1.1802372932434082,
"logits/rejected": -4.334959506988525,
"logps/chosen": -1.7836018800735474,
"logps/rejected": -12.880077362060547,
"loss": 1.6039,
"nll_loss": 1.6024914979934692,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17836017906665802,
"rewards/margins": 1.1096476316452026,
"rewards/rejected": -1.2880078554153442,
"step": 75
},
{
"epoch": 0.28679245283018867,
"grad_norm": 0.2604648470878601,
"learning_rate": 4.2830188679245284e-05,
"log_odds_chosen": 11.805458068847656,
"log_odds_ratio": -0.09639889001846313,
"logits/chosen": -1.4125428199768066,
"logits/rejected": -4.383103847503662,
"logps/chosen": -1.753788948059082,
"logps/rejected": -13.35352897644043,
"loss": 1.5627,
"nll_loss": 1.5530548095703125,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.17537888884544373,
"rewards/margins": 1.1599740982055664,
"rewards/rejected": -1.3353530168533325,
"step": 76
},
{
"epoch": 0.29056603773584905,
"grad_norm": 0.2621481418609619,
"learning_rate": 4.2735849056603775e-05,
"log_odds_chosen": 12.079008102416992,
"log_odds_ratio": -0.029307007789611816,
"logits/chosen": -1.0459429025650024,
"logits/rejected": -4.2921833992004395,
"logps/chosen": -1.7573282718658447,
"logps/rejected": -13.61059856414795,
"loss": 1.6019,
"nll_loss": 1.5990142822265625,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17573282122612,
"rewards/margins": 1.1853270530700684,
"rewards/rejected": -1.3610599040985107,
"step": 77
},
{
"epoch": 0.2943396226415094,
"grad_norm": 0.23861609399318695,
"learning_rate": 4.2641509433962266e-05,
"log_odds_chosen": 14.536327362060547,
"log_odds_ratio": -0.006748859770596027,
"logits/chosen": -1.6867283582687378,
"logits/rejected": -4.363768100738525,
"logps/chosen": -1.6174168586730957,
"logps/rejected": -15.880059242248535,
"loss": 1.5851,
"nll_loss": 1.5844483375549316,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16174167394638062,
"rewards/margins": 1.4262642860412598,
"rewards/rejected": -1.5880060195922852,
"step": 78
},
{
"epoch": 0.2981132075471698,
"grad_norm": 0.26430046558380127,
"learning_rate": 4.254716981132076e-05,
"log_odds_chosen": 10.643120765686035,
"log_odds_ratio": -0.15290699899196625,
"logits/chosen": -1.1397204399108887,
"logits/rejected": -4.667537689208984,
"logps/chosen": -1.5166621208190918,
"logps/rejected": -11.917654991149902,
"loss": 1.4211,
"nll_loss": 1.4057632684707642,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15166620910167694,
"rewards/margins": 1.0400993824005127,
"rewards/rejected": -1.191765546798706,
"step": 79
},
{
"epoch": 0.3018867924528302,
"grad_norm": 0.26361650228500366,
"learning_rate": 4.245283018867925e-05,
"log_odds_chosen": 14.004961967468262,
"log_odds_ratio": -0.0008121158462017775,
"logits/chosen": -1.747410774230957,
"logits/rejected": -5.366614818572998,
"logps/chosen": -1.7156703472137451,
"logps/rejected": -15.486541748046875,
"loss": 1.6296,
"nll_loss": 1.6295278072357178,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17156702280044556,
"rewards/margins": 1.377087116241455,
"rewards/rejected": -1.5486541986465454,
"step": 80
},
{
"epoch": 0.30566037735849055,
"grad_norm": 0.3431408405303955,
"learning_rate": 4.235849056603774e-05,
"log_odds_chosen": 9.586076736450195,
"log_odds_ratio": -0.07982049137353897,
"logits/chosen": -2.0183656215667725,
"logits/rejected": -4.584896564483643,
"logps/chosen": -1.6326284408569336,
"logps/rejected": -10.968765258789062,
"loss": 1.3615,
"nll_loss": 1.3535064458847046,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16326284408569336,
"rewards/margins": 0.9336137175559998,
"rewards/rejected": -1.096876621246338,
"step": 81
},
{
"epoch": 0.30943396226415093,
"grad_norm": 0.29136648774147034,
"learning_rate": 4.226415094339623e-05,
"log_odds_chosen": 12.734142303466797,
"log_odds_ratio": -0.004932960495352745,
"logits/chosen": -2.464571952819824,
"logits/rejected": -4.204010009765625,
"logps/chosen": -1.1688156127929688,
"logps/rejected": -13.363677978515625,
"loss": 1.0781,
"nll_loss": 1.0775874853134155,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.11688156425952911,
"rewards/margins": 1.2194862365722656,
"rewards/rejected": -1.3363678455352783,
"step": 82
},
{
"epoch": 0.3132075471698113,
"grad_norm": 0.3025481402873993,
"learning_rate": 4.216981132075472e-05,
"log_odds_chosen": 11.096177101135254,
"log_odds_ratio": -0.0837731808423996,
"logits/chosen": -0.643512487411499,
"logits/rejected": -2.385446071624756,
"logps/chosen": -1.6675033569335938,
"logps/rejected": -12.507808685302734,
"loss": 1.4565,
"nll_loss": 1.4480957984924316,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16675035655498505,
"rewards/margins": 1.0840305089950562,
"rewards/rejected": -1.2507808208465576,
"step": 83
},
{
"epoch": 0.3169811320754717,
"grad_norm": 0.2666255831718445,
"learning_rate": 4.207547169811321e-05,
"log_odds_chosen": 12.243534088134766,
"log_odds_ratio": -0.17083272337913513,
"logits/chosen": -1.1914499998092651,
"logits/rejected": -4.547676086425781,
"logps/chosen": -1.4934356212615967,
"logps/rejected": -13.507254600524902,
"loss": 1.5673,
"nll_loss": 1.5502480268478394,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1493435651063919,
"rewards/margins": 1.2013819217681885,
"rewards/rejected": -1.3507256507873535,
"step": 84
},
{
"epoch": 0.32075471698113206,
"grad_norm": 0.3045007884502411,
"learning_rate": 4.1981132075471704e-05,
"log_odds_chosen": 12.160042762756348,
"log_odds_ratio": -0.08180340379476547,
"logits/chosen": -2.2346179485321045,
"logits/rejected": -4.7149529457092285,
"logps/chosen": -1.7960240840911865,
"logps/rejected": -13.734954833984375,
"loss": 1.5391,
"nll_loss": 1.5308822393417358,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.17960241436958313,
"rewards/margins": 1.1938930749893188,
"rewards/rejected": -1.3734955787658691,
"step": 85
},
{
"epoch": 0.32452830188679244,
"grad_norm": 0.324097216129303,
"learning_rate": 4.1886792452830195e-05,
"log_odds_chosen": 13.820845603942871,
"log_odds_ratio": -0.0012615115847438574,
"logits/chosen": -1.7233232259750366,
"logits/rejected": -4.733954906463623,
"logps/chosen": -1.5087658166885376,
"logps/rejected": -15.039979934692383,
"loss": 1.4518,
"nll_loss": 1.451686143875122,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15087658166885376,
"rewards/margins": 1.353121280670166,
"rewards/rejected": -1.5039979219436646,
"step": 86
},
{
"epoch": 0.3283018867924528,
"grad_norm": 0.31188490986824036,
"learning_rate": 4.1792452830188686e-05,
"log_odds_chosen": 13.122318267822266,
"log_odds_ratio": -0.026753831654787064,
"logits/chosen": -1.6933764219284058,
"logits/rejected": -3.5955612659454346,
"logps/chosen": -1.3663079738616943,
"logps/rejected": -14.080284118652344,
"loss": 1.3399,
"nll_loss": 1.3372132778167725,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13663078844547272,
"rewards/margins": 1.271397590637207,
"rewards/rejected": -1.4080283641815186,
"step": 87
},
{
"epoch": 0.3320754716981132,
"grad_norm": 0.2510411739349365,
"learning_rate": 4.169811320754717e-05,
"log_odds_chosen": 13.573080062866211,
"log_odds_ratio": -0.07667991518974304,
"logits/chosen": -1.1544338464736938,
"logits/rejected": -5.001407146453857,
"logps/chosen": -1.331773042678833,
"logps/rejected": -14.553543090820312,
"loss": 1.4304,
"nll_loss": 1.4226869344711304,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.13317731022834778,
"rewards/margins": 1.3221771717071533,
"rewards/rejected": -1.4553543329238892,
"step": 88
},
{
"epoch": 0.33584905660377357,
"grad_norm": 0.31842777132987976,
"learning_rate": 4.160377358490566e-05,
"log_odds_chosen": 13.996376037597656,
"log_odds_ratio": -0.0015381659613922238,
"logits/chosen": -2.5720713138580322,
"logits/rejected": -4.779295921325684,
"logps/chosen": -1.5352387428283691,
"logps/rejected": -15.243785858154297,
"loss": 1.6801,
"nll_loss": 1.6799163818359375,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15352387726306915,
"rewards/margins": 1.3708547353744507,
"rewards/rejected": -1.524378776550293,
"step": 89
},
{
"epoch": 0.33962264150943394,
"grad_norm": 0.30600160360336304,
"learning_rate": 4.150943396226415e-05,
"log_odds_chosen": 14.847908973693848,
"log_odds_ratio": -0.0009415854001417756,
"logits/chosen": -0.8421066999435425,
"logits/rejected": -3.2322371006011963,
"logps/chosen": -1.3275384902954102,
"logps/rejected": -15.784387588500977,
"loss": 1.324,
"nll_loss": 1.323903203010559,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13275384902954102,
"rewards/margins": 1.4456850290298462,
"rewards/rejected": -1.5784387588500977,
"step": 90
},
{
"epoch": 0.3433962264150943,
"grad_norm": 0.30167537927627563,
"learning_rate": 4.1415094339622644e-05,
"log_odds_chosen": 12.118502616882324,
"log_odds_ratio": -0.08149873465299606,
"logits/chosen": -0.8174037337303162,
"logits/rejected": -3.398536443710327,
"logps/chosen": -1.4228649139404297,
"logps/rejected": -13.159191131591797,
"loss": 1.3606,
"nll_loss": 1.35248601436615,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1422865092754364,
"rewards/margins": 1.1736326217651367,
"rewards/rejected": -1.3159191608428955,
"step": 91
},
{
"epoch": 0.3471698113207547,
"grad_norm": 0.4155506491661072,
"learning_rate": 4.1320754716981135e-05,
"log_odds_chosen": 13.490291595458984,
"log_odds_ratio": -0.11288166791200638,
"logits/chosen": -0.7610112428665161,
"logits/rejected": -3.971395492553711,
"logps/chosen": -1.609310269355774,
"logps/rejected": -14.705669403076172,
"loss": 1.6664,
"nll_loss": 1.6551594734191895,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16093102097511292,
"rewards/margins": 1.3096359968185425,
"rewards/rejected": -1.470566987991333,
"step": 92
},
{
"epoch": 0.35094339622641507,
"grad_norm": 0.2882811427116394,
"learning_rate": 4.1226415094339626e-05,
"log_odds_chosen": 13.462080001831055,
"log_odds_ratio": -0.15872977674007416,
"logits/chosen": -1.3535653352737427,
"logits/rejected": -4.049402236938477,
"logps/chosen": -1.4857981204986572,
"logps/rejected": -14.58334732055664,
"loss": 1.4598,
"nll_loss": 1.4439499378204346,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.14857982099056244,
"rewards/margins": 1.3097549676895142,
"rewards/rejected": -1.4583346843719482,
"step": 93
},
{
"epoch": 0.35471698113207545,
"grad_norm": 0.3006541430950165,
"learning_rate": 4.113207547169812e-05,
"log_odds_chosen": 12.660806655883789,
"log_odds_ratio": -0.05375822260975838,
"logits/chosen": -2.8630645275115967,
"logits/rejected": -5.452940464019775,
"logps/chosen": -1.785542368888855,
"logps/rejected": -14.212843894958496,
"loss": 1.7117,
"nll_loss": 1.7063469886779785,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1785542368888855,
"rewards/margins": 1.2427302598953247,
"rewards/rejected": -1.4212844371795654,
"step": 94
},
{
"epoch": 0.3584905660377358,
"grad_norm": 0.2722157835960388,
"learning_rate": 4.103773584905661e-05,
"log_odds_chosen": 15.298785209655762,
"log_odds_ratio": -0.000986977363936603,
"logits/chosen": -1.7740228176116943,
"logits/rejected": -4.849878311157227,
"logps/chosen": -1.6763180494308472,
"logps/rejected": -16.691926956176758,
"loss": 1.6459,
"nll_loss": 1.645756721496582,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16763180494308472,
"rewards/margins": 1.5015610456466675,
"rewards/rejected": -1.6691927909851074,
"step": 95
},
{
"epoch": 0.3622641509433962,
"grad_norm": 0.5683091878890991,
"learning_rate": 4.09433962264151e-05,
"log_odds_chosen": 13.669389724731445,
"log_odds_ratio": -0.0002820601512212306,
"logits/chosen": -3.1163902282714844,
"logits/rejected": -6.106686115264893,
"logps/chosen": -1.557908058166504,
"logps/rejected": -14.852912902832031,
"loss": 1.4868,
"nll_loss": 1.4867894649505615,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1557908058166504,
"rewards/margins": 1.329500436782837,
"rewards/rejected": -1.4852913618087769,
"step": 96
},
{
"epoch": 0.3660377358490566,
"grad_norm": 0.2856326699256897,
"learning_rate": 4.084905660377359e-05,
"log_odds_chosen": 15.168952941894531,
"log_odds_ratio": -0.07142313569784164,
"logits/chosen": -1.4473392963409424,
"logits/rejected": -4.727592468261719,
"logps/chosen": -1.3984313011169434,
"logps/rejected": -16.263202667236328,
"loss": 1.581,
"nll_loss": 1.5738685131072998,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.13984312117099762,
"rewards/margins": 1.486477255821228,
"rewards/rejected": -1.6263203620910645,
"step": 97
},
{
"epoch": 0.36981132075471695,
"grad_norm": 0.29159900546073914,
"learning_rate": 4.075471698113208e-05,
"log_odds_chosen": 15.751093864440918,
"log_odds_ratio": -0.00026286751381121576,
"logits/chosen": -3.1162948608398438,
"logits/rejected": -5.55817985534668,
"logps/chosen": -1.6828699111938477,
"logps/rejected": -17.040075302124023,
"loss": 1.4492,
"nll_loss": 1.4491852521896362,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.168286994099617,
"rewards/margins": 1.5357205867767334,
"rewards/rejected": -1.704007625579834,
"step": 98
},
{
"epoch": 0.37358490566037733,
"grad_norm": 0.27507874369621277,
"learning_rate": 4.066037735849057e-05,
"log_odds_chosen": 13.987431526184082,
"log_odds_ratio": -0.07922342419624329,
"logits/chosen": -1.512285590171814,
"logits/rejected": -6.580392837524414,
"logps/chosen": -1.6363471746444702,
"logps/rejected": -15.355705261230469,
"loss": 1.4966,
"nll_loss": 1.4886995553970337,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16363471746444702,
"rewards/margins": 1.3719358444213867,
"rewards/rejected": -1.535570502281189,
"step": 99
},
{
"epoch": 0.37735849056603776,
"grad_norm": 0.25051409006118774,
"learning_rate": 4.0566037735849064e-05,
"log_odds_chosen": 15.439388275146484,
"log_odds_ratio": -5.675078136846423e-05,
"logits/chosen": -1.3515607118606567,
"logits/rejected": -5.860863208770752,
"logps/chosen": -1.4820207357406616,
"logps/rejected": -16.625112533569336,
"loss": 1.5335,
"nll_loss": 1.5334585905075073,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1482020765542984,
"rewards/margins": 1.5143091678619385,
"rewards/rejected": -1.6625111103057861,
"step": 100
},
{
"epoch": 0.38113207547169814,
"grad_norm": 0.3001623749732971,
"learning_rate": 4.047169811320755e-05,
"log_odds_chosen": 12.887317657470703,
"log_odds_ratio": -0.04469170421361923,
"logits/chosen": -2.702786684036255,
"logits/rejected": -4.913610458374023,
"logps/chosen": -1.787389874458313,
"logps/rejected": -14.465229988098145,
"loss": 1.5049,
"nll_loss": 1.5004353523254395,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17873898148536682,
"rewards/margins": 1.2677841186523438,
"rewards/rejected": -1.4465229511260986,
"step": 101
},
{
"epoch": 0.3849056603773585,
"grad_norm": 0.34954193234443665,
"learning_rate": 4.037735849056604e-05,
"log_odds_chosen": 13.664055824279785,
"log_odds_ratio": -0.1444215476512909,
"logits/chosen": -1.608425498008728,
"logits/rejected": -4.3469367027282715,
"logps/chosen": -1.8433992862701416,
"logps/rejected": -15.285165786743164,
"loss": 1.6168,
"nll_loss": 1.60233736038208,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.18433992564678192,
"rewards/margins": 1.3441766500473022,
"rewards/rejected": -1.5285166501998901,
"step": 102
},
{
"epoch": 0.3886792452830189,
"grad_norm": 0.3157506287097931,
"learning_rate": 4.028301886792453e-05,
"log_odds_chosen": 11.686209678649902,
"log_odds_ratio": -0.10109421610832214,
"logits/chosen": -0.873611330986023,
"logits/rejected": -3.7620019912719727,
"logps/chosen": -1.6527265310287476,
"logps/rejected": -13.108765602111816,
"loss": 1.5602,
"nll_loss": 1.5500413179397583,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16527265310287476,
"rewards/margins": 1.145603895187378,
"rewards/rejected": -1.3108766078948975,
"step": 103
},
{
"epoch": 0.39245283018867927,
"grad_norm": 0.2715020477771759,
"learning_rate": 4.018867924528302e-05,
"log_odds_chosen": 11.378678321838379,
"log_odds_ratio": -0.13346989452838898,
"logits/chosen": -1.1194851398468018,
"logits/rejected": -4.566447734832764,
"logps/chosen": -1.6644248962402344,
"logps/rejected": -12.777153015136719,
"loss": 1.5005,
"nll_loss": 1.487147331237793,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16644248366355896,
"rewards/margins": 1.1112728118896484,
"rewards/rejected": -1.2777153253555298,
"step": 104
},
{
"epoch": 0.39622641509433965,
"grad_norm": 0.29437255859375,
"learning_rate": 4.009433962264151e-05,
"log_odds_chosen": 15.541433334350586,
"log_odds_ratio": -5.782198059023358e-05,
"logits/chosen": -1.276192307472229,
"logits/rejected": -4.482449054718018,
"logps/chosen": -1.7318949699401855,
"logps/rejected": -17.009288787841797,
"loss": 1.5203,
"nll_loss": 1.5202912092208862,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17318949103355408,
"rewards/margins": 1.5277395248413086,
"rewards/rejected": -1.700929045677185,
"step": 105
},
{
"epoch": 0.4,
"grad_norm": 0.28076162934303284,
"learning_rate": 4e-05,
"log_odds_chosen": 15.423474311828613,
"log_odds_ratio": -0.003808586858212948,
"logits/chosen": -1.461142659187317,
"logits/rejected": -4.638182163238525,
"logps/chosen": -1.5100231170654297,
"logps/rejected": -16.47011947631836,
"loss": 1.497,
"nll_loss": 1.4965708255767822,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15100233256816864,
"rewards/margins": 1.4960097074508667,
"rewards/rejected": -1.6470119953155518,
"step": 106
},
{
"epoch": 0.4037735849056604,
"grad_norm": 0.28875720500946045,
"learning_rate": 3.9905660377358494e-05,
"log_odds_chosen": 12.895769119262695,
"log_odds_ratio": -0.060368865728378296,
"logits/chosen": -2.2072086334228516,
"logits/rejected": -5.86301851272583,
"logps/chosen": -1.8704140186309814,
"logps/rejected": -14.505237579345703,
"loss": 1.5992,
"nll_loss": 1.593145489692688,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.18704140186309814,
"rewards/margins": 1.2634824514389038,
"rewards/rejected": -1.4505239725112915,
"step": 107
},
{
"epoch": 0.4075471698113208,
"grad_norm": 0.2719891369342804,
"learning_rate": 3.9811320754716985e-05,
"log_odds_chosen": 15.212738037109375,
"log_odds_ratio": -0.06606274098157883,
"logits/chosen": -1.9831483364105225,
"logits/rejected": -5.709364414215088,
"logps/chosen": -1.6806718111038208,
"logps/rejected": -16.673917770385742,
"loss": 1.5861,
"nll_loss": 1.5794475078582764,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16806718707084656,
"rewards/margins": 1.4993247985839844,
"rewards/rejected": -1.6673917770385742,
"step": 108
},
{
"epoch": 0.41132075471698115,
"grad_norm": 0.30551427602767944,
"learning_rate": 3.9716981132075477e-05,
"log_odds_chosen": 13.554816246032715,
"log_odds_ratio": -0.17723311483860016,
"logits/chosen": -1.8992154598236084,
"logits/rejected": -4.842866897583008,
"logps/chosen": -1.408470869064331,
"logps/rejected": -14.5311861038208,
"loss": 1.4911,
"nll_loss": 1.4734236001968384,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1408470869064331,
"rewards/margins": 1.3122715950012207,
"rewards/rejected": -1.4531188011169434,
"step": 109
},
{
"epoch": 0.41509433962264153,
"grad_norm": 0.25901442766189575,
"learning_rate": 3.962264150943397e-05,
"log_odds_chosen": 14.055702209472656,
"log_odds_ratio": -0.07455631345510483,
"logits/chosen": -1.1471797227859497,
"logits/rejected": -4.136468887329102,
"logps/chosen": -1.398701786994934,
"logps/rejected": -15.004222869873047,
"loss": 1.4465,
"nll_loss": 1.4390912055969238,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.13987018167972565,
"rewards/margins": 1.3605520725250244,
"rewards/rejected": -1.5004222393035889,
"step": 110
},
{
"epoch": 0.4188679245283019,
"grad_norm": 0.3040740191936493,
"learning_rate": 3.952830188679246e-05,
"log_odds_chosen": 13.487771987915039,
"log_odds_ratio": -0.0007269117631949484,
"logits/chosen": -2.4711754322052,
"logits/rejected": -5.052443504333496,
"logps/chosen": -1.6260043382644653,
"logps/rejected": -14.823724746704102,
"loss": 1.432,
"nll_loss": 1.431890606880188,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16260044276714325,
"rewards/margins": 1.3197720050811768,
"rewards/rejected": -1.482372522354126,
"step": 111
},
{
"epoch": 0.4226415094339623,
"grad_norm": 0.25823676586151123,
"learning_rate": 3.943396226415095e-05,
"log_odds_chosen": 15.438188552856445,
"log_odds_ratio": -0.00011414527398301288,
"logits/chosen": -2.096723794937134,
"logits/rejected": -6.7537760734558105,
"logps/chosen": -1.8138432502746582,
"logps/rejected": -16.985055923461914,
"loss": 1.7199,
"nll_loss": 1.7199323177337646,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18138432502746582,
"rewards/margins": 1.517121434211731,
"rewards/rejected": -1.6985057592391968,
"step": 112
},
{
"epoch": 0.42641509433962266,
"grad_norm": 0.349231094121933,
"learning_rate": 3.933962264150944e-05,
"log_odds_chosen": 12.445716857910156,
"log_odds_ratio": -0.06478449702262878,
"logits/chosen": -2.650108575820923,
"logits/rejected": -5.185212135314941,
"logps/chosen": -1.6985199451446533,
"logps/rejected": -13.793228149414062,
"loss": 1.746,
"nll_loss": 1.739565372467041,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16985198855400085,
"rewards/margins": 1.2094708681106567,
"rewards/rejected": -1.3793230056762695,
"step": 113
},
{
"epoch": 0.43018867924528303,
"grad_norm": 0.2924027740955353,
"learning_rate": 3.924528301886793e-05,
"log_odds_chosen": 12.061296463012695,
"log_odds_ratio": -0.03779318928718567,
"logits/chosen": -0.749549150466919,
"logits/rejected": -3.7414801120758057,
"logps/chosen": -1.8810656070709229,
"logps/rejected": -13.728483200073242,
"loss": 1.7387,
"nll_loss": 1.7349181175231934,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18810656666755676,
"rewards/margins": 1.1847418546676636,
"rewards/rejected": -1.3728485107421875,
"step": 114
},
{
"epoch": 0.4339622641509434,
"grad_norm": 0.28793764114379883,
"learning_rate": 3.9150943396226416e-05,
"log_odds_chosen": 16.13731575012207,
"log_odds_ratio": -0.054478421807289124,
"logits/chosen": -1.7537418603897095,
"logits/rejected": -5.8226141929626465,
"logps/chosen": -1.4789087772369385,
"logps/rejected": -17.265403747558594,
"loss": 1.589,
"nll_loss": 1.5835916996002197,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.14789089560508728,
"rewards/margins": 1.5786495208740234,
"rewards/rejected": -1.7265403270721436,
"step": 115
},
{
"epoch": 0.4377358490566038,
"grad_norm": 0.31606027483940125,
"learning_rate": 3.905660377358491e-05,
"log_odds_chosen": 14.270502090454102,
"log_odds_ratio": -0.14657607674598694,
"logits/chosen": -1.5621134042739868,
"logits/rejected": -3.9824719429016113,
"logps/chosen": -1.5558480024337769,
"logps/rejected": -15.48137092590332,
"loss": 1.4384,
"nll_loss": 1.4237393140792847,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15558481216430664,
"rewards/margins": 1.392552375793457,
"rewards/rejected": -1.5481371879577637,
"step": 116
},
{
"epoch": 0.44150943396226416,
"grad_norm": 0.6901794672012329,
"learning_rate": 3.89622641509434e-05,
"log_odds_chosen": 12.610831260681152,
"log_odds_ratio": -0.11986835300922394,
"logits/chosen": -1.5961331129074097,
"logits/rejected": -4.9528584480285645,
"logps/chosen": -1.864069938659668,
"logps/rejected": -14.308099746704102,
"loss": 1.7146,
"nll_loss": 1.702580213546753,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.18640699982643127,
"rewards/margins": 1.2444028854370117,
"rewards/rejected": -1.4308099746704102,
"step": 117
},
{
"epoch": 0.44528301886792454,
"grad_norm": 0.28305572271347046,
"learning_rate": 3.886792452830189e-05,
"log_odds_chosen": 13.709443092346191,
"log_odds_ratio": -0.08205895870923996,
"logits/chosen": -1.5196449756622314,
"logits/rejected": -4.281819820404053,
"logps/chosen": -1.5432316064834595,
"logps/rejected": -14.945755958557129,
"loss": 1.5562,
"nll_loss": 1.5479600429534912,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15432317554950714,
"rewards/margins": 1.34025239944458,
"rewards/rejected": -1.4945755004882812,
"step": 118
},
{
"epoch": 0.4490566037735849,
"grad_norm": 0.30540400743484497,
"learning_rate": 3.877358490566038e-05,
"log_odds_chosen": 14.886051177978516,
"log_odds_ratio": -0.0001277975970879197,
"logits/chosen": -2.5971062183380127,
"logits/rejected": -6.4186625480651855,
"logps/chosen": -1.446777105331421,
"logps/rejected": -15.858993530273438,
"loss": 1.3599,
"nll_loss": 1.3598815202713013,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14467771351337433,
"rewards/margins": 1.4412215948104858,
"rewards/rejected": -1.5858993530273438,
"step": 119
},
{
"epoch": 0.4528301886792453,
"grad_norm": 0.279973566532135,
"learning_rate": 3.867924528301887e-05,
"log_odds_chosen": 12.182138442993164,
"log_odds_ratio": -0.007018540520220995,
"logits/chosen": -1.7155228853225708,
"logits/rejected": -5.174653053283691,
"logps/chosen": -1.7579047679901123,
"logps/rejected": -13.713508605957031,
"loss": 1.6865,
"nll_loss": 1.6858264207839966,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1757904589176178,
"rewards/margins": 1.195560336112976,
"rewards/rejected": -1.371350884437561,
"step": 120
},
{
"epoch": 0.45660377358490567,
"grad_norm": 0.2613874077796936,
"learning_rate": 3.858490566037736e-05,
"log_odds_chosen": 14.095218658447266,
"log_odds_ratio": -0.07463585585355759,
"logits/chosen": -1.3325459957122803,
"logits/rejected": -6.121455192565918,
"logps/chosen": -1.5543732643127441,
"logps/rejected": -15.366453170776367,
"loss": 1.5167,
"nll_loss": 1.509216070175171,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15543733537197113,
"rewards/margins": 1.3812079429626465,
"rewards/rejected": -1.5366454124450684,
"step": 121
},
{
"epoch": 0.46037735849056605,
"grad_norm": 0.30324679613113403,
"learning_rate": 3.8490566037735854e-05,
"log_odds_chosen": 16.875022888183594,
"log_odds_ratio": -0.00017375449533574283,
"logits/chosen": -1.8117984533309937,
"logits/rejected": -4.163735866546631,
"logps/chosen": -1.515238881111145,
"logps/rejected": -17.963499069213867,
"loss": 1.4792,
"nll_loss": 1.4791332483291626,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1515238881111145,
"rewards/margins": 1.6448261737823486,
"rewards/rejected": -1.7963500022888184,
"step": 122
},
{
"epoch": 0.4641509433962264,
"grad_norm": 0.3230510354042053,
"learning_rate": 3.8396226415094345e-05,
"log_odds_chosen": 17.355361938476562,
"log_odds_ratio": -8.270166631518805e-07,
"logits/chosen": -0.9121267795562744,
"logits/rejected": -4.532161712646484,
"logps/chosen": -1.351798415184021,
"logps/rejected": -18.30146026611328,
"loss": 1.4902,
"nll_loss": 1.4902395009994507,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13517984747886658,
"rewards/margins": 1.6949663162231445,
"rewards/rejected": -1.8301459550857544,
"step": 123
},
{
"epoch": 0.4679245283018868,
"grad_norm": 0.27051493525505066,
"learning_rate": 3.8301886792452836e-05,
"log_odds_chosen": 16.027782440185547,
"log_odds_ratio": -0.05309152603149414,
"logits/chosen": -2.257577896118164,
"logits/rejected": -6.006890296936035,
"logps/chosen": -1.417616844177246,
"logps/rejected": -17.086490631103516,
"loss": 1.439,
"nll_loss": 1.4336605072021484,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1417616754770279,
"rewards/margins": 1.5668874979019165,
"rewards/rejected": -1.7086491584777832,
"step": 124
},
{
"epoch": 0.4716981132075472,
"grad_norm": 0.28948161005973816,
"learning_rate": 3.820754716981133e-05,
"log_odds_chosen": 13.788773536682129,
"log_odds_ratio": -0.057656653225421906,
"logits/chosen": -1.0248003005981445,
"logits/rejected": -6.2845001220703125,
"logps/chosen": -1.651491641998291,
"logps/rejected": -15.156482696533203,
"loss": 1.6935,
"nll_loss": 1.6877108812332153,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16514916718006134,
"rewards/margins": 1.350499153137207,
"rewards/rejected": -1.515648365020752,
"step": 125
},
{
"epoch": 0.47547169811320755,
"grad_norm": 0.29785606265068054,
"learning_rate": 3.811320754716982e-05,
"log_odds_chosen": 15.78986930847168,
"log_odds_ratio": -8.410341251874343e-05,
"logits/chosen": -2.4442198276519775,
"logits/rejected": -6.754056453704834,
"logps/chosen": -1.6627094745635986,
"logps/rejected": -17.13760757446289,
"loss": 1.6297,
"nll_loss": 1.62972891330719,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16627094149589539,
"rewards/margins": 1.547489881515503,
"rewards/rejected": -1.7137608528137207,
"step": 126
},
{
"epoch": 0.47924528301886793,
"grad_norm": 0.3125203251838684,
"learning_rate": 3.801886792452831e-05,
"log_odds_chosen": 15.262259483337402,
"log_odds_ratio": -0.00025166559498757124,
"logits/chosen": -2.318103790283203,
"logits/rejected": -4.911985397338867,
"logps/chosen": -1.5119843482971191,
"logps/rejected": -16.430225372314453,
"loss": 1.5521,
"nll_loss": 1.5521178245544434,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15119843184947968,
"rewards/margins": 1.4918241500854492,
"rewards/rejected": -1.6430225372314453,
"step": 127
},
{
"epoch": 0.4830188679245283,
"grad_norm": 0.4130852222442627,
"learning_rate": 3.7924528301886794e-05,
"log_odds_chosen": 13.181659698486328,
"log_odds_ratio": -0.02264661341905594,
"logits/chosen": -1.5559453964233398,
"logits/rejected": -5.835235595703125,
"logps/chosen": -1.6575170755386353,
"logps/rejected": -14.44080638885498,
"loss": 1.531,
"nll_loss": 1.5287659168243408,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16575171053409576,
"rewards/margins": 1.2783288955688477,
"rewards/rejected": -1.4440807104110718,
"step": 128
},
{
"epoch": 0.4867924528301887,
"grad_norm": 0.27878594398498535,
"learning_rate": 3.7830188679245285e-05,
"log_odds_chosen": 15.444690704345703,
"log_odds_ratio": -0.024172237142920494,
"logits/chosen": -1.8147392272949219,
"logits/rejected": -5.994168758392334,
"logps/chosen": -1.4637315273284912,
"logps/rejected": -16.60024070739746,
"loss": 1.41,
"nll_loss": 1.4076224565505981,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14637315273284912,
"rewards/margins": 1.5136507749557495,
"rewards/rejected": -1.6600239276885986,
"step": 129
},
{
"epoch": 0.49056603773584906,
"grad_norm": 0.36368459463119507,
"learning_rate": 3.7735849056603776e-05,
"log_odds_chosen": 15.150455474853516,
"log_odds_ratio": -0.06176932156085968,
"logits/chosen": -2.638796806335449,
"logits/rejected": -6.133980751037598,
"logps/chosen": -1.5448698997497559,
"logps/rejected": -16.379032135009766,
"loss": 1.5112,
"nll_loss": 1.505061388015747,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1544869840145111,
"rewards/margins": 1.4834163188934326,
"rewards/rejected": -1.6379034519195557,
"step": 130
},
{
"epoch": 0.49433962264150944,
"grad_norm": 0.3242456316947937,
"learning_rate": 3.764150943396227e-05,
"log_odds_chosen": 13.985454559326172,
"log_odds_ratio": -4.5815289922757074e-05,
"logits/chosen": -1.494691014289856,
"logits/rejected": -5.898371696472168,
"logps/chosen": -1.9884130954742432,
"logps/rejected": -15.798318862915039,
"loss": 1.7428,
"nll_loss": 1.7428255081176758,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19884130358695984,
"rewards/margins": 1.3809905052185059,
"rewards/rejected": -1.579831838607788,
"step": 131
},
{
"epoch": 0.4981132075471698,
"grad_norm": 0.29855185747146606,
"learning_rate": 3.754716981132076e-05,
"log_odds_chosen": 12.895346641540527,
"log_odds_ratio": -0.0862637311220169,
"logits/chosen": -0.5306591391563416,
"logits/rejected": -3.3876893520355225,
"logps/chosen": -1.6508989334106445,
"logps/rejected": -14.312397956848145,
"loss": 1.5424,
"nll_loss": 1.533748745918274,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1650899052619934,
"rewards/margins": 1.2661499977111816,
"rewards/rejected": -1.4312398433685303,
"step": 132
},
{
"epoch": 0.5018867924528302,
"grad_norm": 0.3010408282279968,
"learning_rate": 3.745283018867924e-05,
"log_odds_chosen": 14.359496116638184,
"log_odds_ratio": -0.11943729966878891,
"logits/chosen": -1.3975955247879028,
"logits/rejected": -5.733314037322998,
"logps/chosen": -1.5766863822937012,
"logps/rejected": -15.659753799438477,
"loss": 1.5544,
"nll_loss": 1.5425056219100952,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15766863524913788,
"rewards/margins": 1.4083068370819092,
"rewards/rejected": -1.5659754276275635,
"step": 133
},
{
"epoch": 0.5056603773584906,
"grad_norm": 0.28366541862487793,
"learning_rate": 3.735849056603773e-05,
"log_odds_chosen": 16.42931365966797,
"log_odds_ratio": -1.6768943169154227e-05,
"logits/chosen": -1.0664993524551392,
"logits/rejected": -5.362708568572998,
"logps/chosen": -1.4823436737060547,
"logps/rejected": -17.60668182373047,
"loss": 1.4417,
"nll_loss": 1.441743016242981,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14823436737060547,
"rewards/margins": 1.612433671951294,
"rewards/rejected": -1.7606680393218994,
"step": 134
},
{
"epoch": 0.5094339622641509,
"grad_norm": 0.297313928604126,
"learning_rate": 3.7264150943396224e-05,
"log_odds_chosen": 15.513206481933594,
"log_odds_ratio": -0.03894759714603424,
"logits/chosen": -1.827707290649414,
"logits/rejected": -5.077935218811035,
"logps/chosen": -1.5281659364700317,
"logps/rejected": -16.74795150756836,
"loss": 1.4237,
"nll_loss": 1.4197759628295898,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15281659364700317,
"rewards/margins": 1.5219786167144775,
"rewards/rejected": -1.674795150756836,
"step": 135
},
{
"epoch": 0.5132075471698113,
"grad_norm": 0.2611730992794037,
"learning_rate": 3.7169811320754716e-05,
"log_odds_chosen": 14.942113876342773,
"log_odds_ratio": -0.001385436742566526,
"logits/chosen": -1.9019191265106201,
"logits/rejected": -5.761693000793457,
"logps/chosen": -1.4520561695098877,
"logps/rejected": -16.059383392333984,
"loss": 1.4831,
"nll_loss": 1.4829171895980835,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14520561695098877,
"rewards/margins": 1.4607326984405518,
"rewards/rejected": -1.60593843460083,
"step": 136
},
{
"epoch": 0.5169811320754717,
"grad_norm": 0.37624403834342957,
"learning_rate": 3.7075471698113207e-05,
"log_odds_chosen": 15.124824523925781,
"log_odds_ratio": -0.06929050385951996,
"logits/chosen": -0.5525396466255188,
"logits/rejected": -4.60980749130249,
"logps/chosen": -1.61146080493927,
"logps/rejected": -16.51461410522461,
"loss": 1.4183,
"nll_loss": 1.4113963842391968,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16114608943462372,
"rewards/margins": 1.4903154373168945,
"rewards/rejected": -1.6514614820480347,
"step": 137
},
{
"epoch": 0.5207547169811321,
"grad_norm": 0.3030368387699127,
"learning_rate": 3.69811320754717e-05,
"log_odds_chosen": 12.055831909179688,
"log_odds_ratio": -0.11039459705352783,
"logits/chosen": -0.8935803174972534,
"logits/rejected": -4.527928352355957,
"logps/chosen": -1.668050765991211,
"logps/rejected": -13.411079406738281,
"loss": 1.5587,
"nll_loss": 1.5476906299591064,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16680508852005005,
"rewards/margins": 1.1743026971817017,
"rewards/rejected": -1.3411078453063965,
"step": 138
},
{
"epoch": 0.5245283018867924,
"grad_norm": 0.2544783353805542,
"learning_rate": 3.688679245283019e-05,
"log_odds_chosen": 15.021339416503906,
"log_odds_ratio": -0.06613866984844208,
"logits/chosen": -2.3585753440856934,
"logits/rejected": -6.415126323699951,
"logps/chosen": -1.4854167699813843,
"logps/rejected": -16.204166412353516,
"loss": 1.3412,
"nll_loss": 1.3345497846603394,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1485416740179062,
"rewards/margins": 1.4718750715255737,
"rewards/rejected": -1.6204167604446411,
"step": 139
},
{
"epoch": 0.5283018867924528,
"grad_norm": 0.3188823461532593,
"learning_rate": 3.679245283018868e-05,
"log_odds_chosen": 15.247077941894531,
"log_odds_ratio": -0.0011911022011190653,
"logits/chosen": -2.5056328773498535,
"logits/rejected": -5.532197952270508,
"logps/chosen": -1.569747805595398,
"logps/rejected": -16.510929107666016,
"loss": 1.4605,
"nll_loss": 1.4603441953659058,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15697479248046875,
"rewards/margins": 1.494118094444275,
"rewards/rejected": -1.6510928869247437,
"step": 140
},
{
"epoch": 0.5320754716981132,
"grad_norm": 0.3175397515296936,
"learning_rate": 3.669811320754717e-05,
"log_odds_chosen": 13.892709732055664,
"log_odds_ratio": -0.0589115172624588,
"logits/chosen": -0.8055652379989624,
"logits/rejected": -3.4150259494781494,
"logps/chosen": -1.5232586860656738,
"logps/rejected": -14.959342956542969,
"loss": 1.5432,
"nll_loss": 1.5373512506484985,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15232588350772858,
"rewards/margins": 1.3436083793640137,
"rewards/rejected": -1.495934247970581,
"step": 141
},
{
"epoch": 0.5358490566037736,
"grad_norm": 0.27956297993659973,
"learning_rate": 3.660377358490566e-05,
"log_odds_chosen": 15.927668571472168,
"log_odds_ratio": -0.0016179109225049615,
"logits/chosen": -1.636272668838501,
"logits/rejected": -5.451028347015381,
"logps/chosen": -1.640097975730896,
"logps/rejected": -17.290578842163086,
"loss": 1.5572,
"nll_loss": 1.5570058822631836,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16400979459285736,
"rewards/margins": 1.565048098564148,
"rewards/rejected": -1.729057788848877,
"step": 142
},
{
"epoch": 0.539622641509434,
"grad_norm": 0.2930643558502197,
"learning_rate": 3.650943396226415e-05,
"log_odds_chosen": 16.596500396728516,
"log_odds_ratio": -9.716653585201129e-05,
"logits/chosen": -1.436676263809204,
"logits/rejected": -4.450935363769531,
"logps/chosen": -1.3673814535140991,
"logps/rejected": -17.530248641967773,
"loss": 1.3955,
"nll_loss": 1.3954448699951172,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1367381513118744,
"rewards/margins": 1.6162867546081543,
"rewards/rejected": -1.753024935722351,
"step": 143
},
{
"epoch": 0.5433962264150943,
"grad_norm": 0.4236195683479309,
"learning_rate": 3.641509433962264e-05,
"log_odds_chosen": 14.702760696411133,
"log_odds_ratio": -0.0004511699662543833,
"logits/chosen": -1.313840389251709,
"logits/rejected": -4.356607437133789,
"logps/chosen": -1.8211621046066284,
"logps/rejected": -16.25528335571289,
"loss": 1.547,
"nll_loss": 1.5469856262207031,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18211621046066284,
"rewards/margins": 1.443412184715271,
"rewards/rejected": -1.6255284547805786,
"step": 144
},
{
"epoch": 0.5471698113207547,
"grad_norm": 0.33278077840805054,
"learning_rate": 3.632075471698113e-05,
"log_odds_chosen": 15.362979888916016,
"log_odds_ratio": -0.0003507338115014136,
"logits/chosen": -1.3937156200408936,
"logits/rejected": -5.756740570068359,
"logps/chosen": -1.7482523918151855,
"logps/rejected": -16.875289916992188,
"loss": 1.608,
"nll_loss": 1.6079708337783813,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1748252511024475,
"rewards/margins": 1.5127036571502686,
"rewards/rejected": -1.6875288486480713,
"step": 145
},
{
"epoch": 0.5509433962264151,
"grad_norm": 0.32443860173225403,
"learning_rate": 3.622641509433962e-05,
"log_odds_chosen": 15.61093521118164,
"log_odds_ratio": -0.06764474511146545,
"logits/chosen": -1.9850184917449951,
"logits/rejected": -5.995416164398193,
"logps/chosen": -1.7201964855194092,
"logps/rejected": -17.044771194458008,
"loss": 1.6287,
"nll_loss": 1.6218953132629395,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.17201966047286987,
"rewards/margins": 1.5324573516845703,
"rewards/rejected": -1.704477071762085,
"step": 146
},
{
"epoch": 0.5547169811320755,
"grad_norm": 0.2766498029232025,
"learning_rate": 3.613207547169811e-05,
"log_odds_chosen": 14.549701690673828,
"log_odds_ratio": -0.000522086163982749,
"logits/chosen": -2.167712688446045,
"logits/rejected": -6.221152305603027,
"logps/chosen": -1.494560956954956,
"logps/rejected": -15.703763961791992,
"loss": 1.4152,
"nll_loss": 1.4151456356048584,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14945609867572784,
"rewards/margins": 1.4209203720092773,
"rewards/rejected": -1.5703763961791992,
"step": 147
},
{
"epoch": 0.5584905660377358,
"grad_norm": 0.4621659219264984,
"learning_rate": 3.60377358490566e-05,
"log_odds_chosen": 14.655370712280273,
"log_odds_ratio": -0.07225409895181656,
"logits/chosen": -0.9679865837097168,
"logits/rejected": -3.743302345275879,
"logps/chosen": -1.47898268699646,
"logps/rejected": -15.858048439025879,
"loss": 1.3764,
"nll_loss": 1.3691895008087158,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.14789827167987823,
"rewards/margins": 1.4379065036773682,
"rewards/rejected": -1.5858049392700195,
"step": 148
},
{
"epoch": 0.5622641509433962,
"grad_norm": 0.319667249917984,
"learning_rate": 3.594339622641509e-05,
"log_odds_chosen": 15.044527053833008,
"log_odds_ratio": -0.00024553845287300646,
"logits/chosen": -1.18473219871521,
"logits/rejected": -4.766110420227051,
"logps/chosen": -1.5453764200210571,
"logps/rejected": -16.291166305541992,
"loss": 1.5824,
"nll_loss": 1.5824246406555176,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1545376479625702,
"rewards/margins": 1.474579095840454,
"rewards/rejected": -1.6291167736053467,
"step": 149
},
{
"epoch": 0.5660377358490566,
"grad_norm": 0.3204805552959442,
"learning_rate": 3.5849056603773584e-05,
"log_odds_chosen": 17.485958099365234,
"log_odds_ratio": -0.10664539784193039,
"logits/chosen": -1.6424400806427002,
"logits/rejected": -3.842500686645508,
"logps/chosen": -1.4972755908966064,
"logps/rejected": -18.57717514038086,
"loss": 1.3499,
"nll_loss": 1.3392703533172607,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.14972755312919617,
"rewards/margins": 1.7079898118972778,
"rewards/rejected": -1.8577172756195068,
"step": 150
},
{
"epoch": 0.569811320754717,
"grad_norm": 0.26958513259887695,
"learning_rate": 3.5754716981132075e-05,
"log_odds_chosen": 15.958301544189453,
"log_odds_ratio": -0.0007034969748929143,
"logits/chosen": -1.0648162364959717,
"logits/rejected": -6.030667304992676,
"logps/chosen": -1.6111671924591064,
"logps/rejected": -17.287628173828125,
"loss": 1.6164,
"nll_loss": 1.6163535118103027,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16111671924591064,
"rewards/margins": 1.5676461458206177,
"rewards/rejected": -1.7287628650665283,
"step": 151
},
{
"epoch": 0.5735849056603773,
"grad_norm": 0.3305363059043884,
"learning_rate": 3.5660377358490566e-05,
"log_odds_chosen": 17.14991569519043,
"log_odds_ratio": -3.2623302104184404e-05,
"logits/chosen": -1.6892523765563965,
"logits/rejected": -4.232363224029541,
"logps/chosen": -1.5470982789993286,
"logps/rejected": -18.353713989257812,
"loss": 1.4759,
"nll_loss": 1.4759438037872314,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1547098308801651,
"rewards/margins": 1.6806614398956299,
"rewards/rejected": -1.8353712558746338,
"step": 152
},
{
"epoch": 0.5773584905660377,
"grad_norm": 0.3334077000617981,
"learning_rate": 3.556603773584906e-05,
"log_odds_chosen": 15.461456298828125,
"log_odds_ratio": -3.081683098571375e-05,
"logits/chosen": -1.5841903686523438,
"logits/rejected": -5.0937604904174805,
"logps/chosen": -1.774914264678955,
"logps/rejected": -16.993589401245117,
"loss": 1.5688,
"nll_loss": 1.5687741041183472,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1774914264678955,
"rewards/margins": 1.5218675136566162,
"rewards/rejected": -1.6993589401245117,
"step": 153
},
{
"epoch": 0.5811320754716981,
"grad_norm": 0.29705750942230225,
"learning_rate": 3.547169811320755e-05,
"log_odds_chosen": 14.837601661682129,
"log_odds_ratio": -4.0413448004983366e-05,
"logits/chosen": -2.566657781600952,
"logits/rejected": -6.176463603973389,
"logps/chosen": -1.7194881439208984,
"logps/rejected": -16.30776023864746,
"loss": 1.5701,
"nll_loss": 1.5700526237487793,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17194882035255432,
"rewards/margins": 1.4588273763656616,
"rewards/rejected": -1.6307761669158936,
"step": 154
},
{
"epoch": 0.5849056603773585,
"grad_norm": 0.4479868710041046,
"learning_rate": 3.537735849056604e-05,
"log_odds_chosen": 16.22223472595215,
"log_odds_ratio": -5.1474453357513994e-05,
"logits/chosen": -1.6525013446807861,
"logits/rejected": -5.60921573638916,
"logps/chosen": -1.4534053802490234,
"logps/rejected": -17.35995101928711,
"loss": 1.3998,
"nll_loss": 1.3998193740844727,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14534053206443787,
"rewards/margins": 1.5906546115875244,
"rewards/rejected": -1.7359951734542847,
"step": 155
},
{
"epoch": 0.5886792452830188,
"grad_norm": 0.3321553170681,
"learning_rate": 3.528301886792453e-05,
"log_odds_chosen": 14.45380973815918,
"log_odds_ratio": -0.00011496634397190064,
"logits/chosen": -1.5721373558044434,
"logits/rejected": -5.987687110900879,
"logps/chosen": -1.7010905742645264,
"logps/rejected": -15.914693832397461,
"loss": 1.4444,
"nll_loss": 1.4443511962890625,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17010906338691711,
"rewards/margins": 1.4213604927062988,
"rewards/rejected": -1.591469407081604,
"step": 156
},
{
"epoch": 0.5924528301886792,
"grad_norm": 0.32465389370918274,
"learning_rate": 3.518867924528302e-05,
"log_odds_chosen": 15.948369979858398,
"log_odds_ratio": -1.5907631677691825e-05,
"logits/chosen": -1.2226861715316772,
"logits/rejected": -4.106148719787598,
"logps/chosen": -1.8807780742645264,
"logps/rejected": -17.614301681518555,
"loss": 1.5042,
"nll_loss": 1.5042164325714111,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18807780742645264,
"rewards/margins": 1.5733524560928345,
"rewards/rejected": -1.761430263519287,
"step": 157
},
{
"epoch": 0.5962264150943396,
"grad_norm": 0.2754722237586975,
"learning_rate": 3.5094339622641506e-05,
"log_odds_chosen": 15.541213989257812,
"log_odds_ratio": -0.16899409890174866,
"logits/chosen": -2.1642675399780273,
"logits/rejected": -5.719496250152588,
"logps/chosen": -1.5195682048797607,
"logps/rejected": -16.678632736206055,
"loss": 1.6354,
"nll_loss": 1.6184601783752441,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15195682644844055,
"rewards/margins": 1.5159064531326294,
"rewards/rejected": -1.6678632497787476,
"step": 158
},
{
"epoch": 0.6,
"grad_norm": 0.28031638264656067,
"learning_rate": 3.5e-05,
"log_odds_chosen": 11.228550910949707,
"log_odds_ratio": -0.20393189787864685,
"logits/chosen": -0.9825246334075928,
"logits/rejected": -4.9232587814331055,
"logps/chosen": -1.7185983657836914,
"logps/rejected": -12.776618003845215,
"loss": 1.5923,
"nll_loss": 1.5718731880187988,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17185983061790466,
"rewards/margins": 1.1058019399642944,
"rewards/rejected": -1.2776618003845215,
"step": 159
},
{
"epoch": 0.6037735849056604,
"grad_norm": 0.2919527292251587,
"learning_rate": 3.490566037735849e-05,
"log_odds_chosen": 14.894678115844727,
"log_odds_ratio": -0.0009987247176468372,
"logits/chosen": -0.6535999178886414,
"logits/rejected": -4.017884254455566,
"logps/chosen": -1.5091729164123535,
"logps/rejected": -16.018918991088867,
"loss": 1.403,
"nll_loss": 1.40290367603302,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15091729164123535,
"rewards/margins": 1.450974702835083,
"rewards/rejected": -1.6018919944763184,
"step": 160
},
{
"epoch": 0.6075471698113207,
"grad_norm": 0.3071415424346924,
"learning_rate": 3.481132075471698e-05,
"log_odds_chosen": 16.471067428588867,
"log_odds_ratio": -2.074857911793515e-05,
"logits/chosen": -0.09290022403001785,
"logits/rejected": -3.7179436683654785,
"logps/chosen": -1.471862554550171,
"logps/rejected": -17.618274688720703,
"loss": 1.4793,
"nll_loss": 1.4793322086334229,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1471862643957138,
"rewards/margins": 1.6146413087844849,
"rewards/rejected": -1.7618277072906494,
"step": 161
},
{
"epoch": 0.6113207547169811,
"grad_norm": 0.2777910530567169,
"learning_rate": 3.471698113207547e-05,
"log_odds_chosen": 16.91250991821289,
"log_odds_ratio": -0.016157550737261772,
"logits/chosen": -1.235945463180542,
"logits/rejected": -5.316408157348633,
"logps/chosen": -1.6419563293457031,
"logps/rejected": -18.302946090698242,
"loss": 1.429,
"nll_loss": 1.4273896217346191,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16419564187526703,
"rewards/margins": 1.6660988330841064,
"rewards/rejected": -1.8302946090698242,
"step": 162
},
{
"epoch": 0.6150943396226415,
"grad_norm": 0.2712962329387665,
"learning_rate": 3.462264150943396e-05,
"log_odds_chosen": 16.41084098815918,
"log_odds_ratio": -2.9441296646837145e-05,
"logits/chosen": -1.025072693824768,
"logits/rejected": -5.848681449890137,
"logps/chosen": -1.5957037210464478,
"logps/rejected": -17.6140193939209,
"loss": 1.5493,
"nll_loss": 1.5493419170379639,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15957039594650269,
"rewards/margins": 1.6018316745758057,
"rewards/rejected": -1.7614020109176636,
"step": 163
},
{
"epoch": 0.6188679245283019,
"grad_norm": 0.30003705620765686,
"learning_rate": 3.452830188679245e-05,
"log_odds_chosen": 16.494930267333984,
"log_odds_ratio": -8.993155461212154e-06,
"logits/chosen": -1.4486334323883057,
"logits/rejected": -5.38150691986084,
"logps/chosen": -1.4585084915161133,
"logps/rejected": -17.579425811767578,
"loss": 1.5673,
"nll_loss": 1.5673246383666992,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14585085213184357,
"rewards/margins": 1.6120917797088623,
"rewards/rejected": -1.7579425573349,
"step": 164
},
{
"epoch": 0.6226415094339622,
"grad_norm": 0.2940906286239624,
"learning_rate": 3.4433962264150943e-05,
"log_odds_chosen": 15.412787437438965,
"log_odds_ratio": -0.07001832127571106,
"logits/chosen": -1.34386146068573,
"logits/rejected": -5.122559070587158,
"logps/chosen": -1.8467469215393066,
"logps/rejected": -17.047271728515625,
"loss": 1.6754,
"nll_loss": 1.6683518886566162,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1846746802330017,
"rewards/margins": 1.5200525522232056,
"rewards/rejected": -1.7047271728515625,
"step": 165
},
{
"epoch": 0.6264150943396226,
"grad_norm": 0.3121996521949768,
"learning_rate": 3.4339622641509435e-05,
"log_odds_chosen": 16.29378318786621,
"log_odds_ratio": -0.05672796443104744,
"logits/chosen": -2.245677947998047,
"logits/rejected": -4.856801986694336,
"logps/chosen": -1.4243159294128418,
"logps/rejected": -17.36766815185547,
"loss": 1.5158,
"nll_loss": 1.5101523399353027,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1424316018819809,
"rewards/margins": 1.5943353176116943,
"rewards/rejected": -1.7367669343948364,
"step": 166
},
{
"epoch": 0.630188679245283,
"grad_norm": 0.28171584010124207,
"learning_rate": 3.4245283018867926e-05,
"log_odds_chosen": 15.490211486816406,
"log_odds_ratio": -0.024965543299913406,
"logits/chosen": -0.1043301597237587,
"logits/rejected": -4.409641742706299,
"logps/chosen": -1.771804690361023,
"logps/rejected": -17.02152442932129,
"loss": 1.6991,
"nll_loss": 1.6965808868408203,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1771804690361023,
"rewards/margins": 1.5249719619750977,
"rewards/rejected": -1.7021524906158447,
"step": 167
},
{
"epoch": 0.6339622641509434,
"grad_norm": 0.27846312522888184,
"learning_rate": 3.415094339622642e-05,
"log_odds_chosen": 16.138580322265625,
"log_odds_ratio": -0.00026864392566494644,
"logits/chosen": -1.5777232646942139,
"logits/rejected": -6.5474162101745605,
"logps/chosen": -1.6822800636291504,
"logps/rejected": -17.583065032958984,
"loss": 1.5533,
"nll_loss": 1.5532336235046387,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16822800040245056,
"rewards/margins": 1.5900784730911255,
"rewards/rejected": -1.7583065032958984,
"step": 168
},
{
"epoch": 0.6377358490566037,
"grad_norm": 0.3010658025741577,
"learning_rate": 3.405660377358491e-05,
"log_odds_chosen": 15.855169296264648,
"log_odds_ratio": -1.3776767445961013e-05,
"logits/chosen": -1.5479809045791626,
"logits/rejected": -5.985370635986328,
"logps/chosen": -1.7337384223937988,
"logps/rejected": -17.28085708618164,
"loss": 1.6455,
"nll_loss": 1.645505666732788,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17337384819984436,
"rewards/margins": 1.5547118186950684,
"rewards/rejected": -1.7280856370925903,
"step": 169
},
{
"epoch": 0.6415094339622641,
"grad_norm": 0.3382321000099182,
"learning_rate": 3.39622641509434e-05,
"log_odds_chosen": 15.886013984680176,
"log_odds_ratio": -8.158626769727562e-06,
"logits/chosen": -1.8764511346817017,
"logits/rejected": -5.525568962097168,
"logps/chosen": -1.3874365091323853,
"logps/rejected": -16.957935333251953,
"loss": 1.4466,
"nll_loss": 1.4466451406478882,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13874365389347076,
"rewards/margins": 1.5570499897003174,
"rewards/rejected": -1.695793628692627,
"step": 170
},
{
"epoch": 0.6452830188679245,
"grad_norm": 0.33056262135505676,
"learning_rate": 3.386792452830188e-05,
"log_odds_chosen": 15.484598159790039,
"log_odds_ratio": -8.27780422696378e-06,
"logits/chosen": -1.8570010662078857,
"logits/rejected": -4.6085357666015625,
"logps/chosen": -1.5708074569702148,
"logps/rejected": -16.722640991210938,
"loss": 1.5653,
"nll_loss": 1.5652695894241333,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.157080739736557,
"rewards/margins": 1.5151833295822144,
"rewards/rejected": -1.6722640991210938,
"step": 171
},
{
"epoch": 0.6490566037735849,
"grad_norm": 0.31156399846076965,
"learning_rate": 3.3773584905660374e-05,
"log_odds_chosen": 15.454925537109375,
"log_odds_ratio": -0.01993393711745739,
"logits/chosen": -1.1617259979248047,
"logits/rejected": -5.1523756980896,
"logps/chosen": -1.5937132835388184,
"logps/rejected": -16.762189865112305,
"loss": 1.3782,
"nll_loss": 1.3762309551239014,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15937133133411407,
"rewards/margins": 1.5168476104736328,
"rewards/rejected": -1.6762189865112305,
"step": 172
},
{
"epoch": 0.6528301886792452,
"grad_norm": 0.2925812005996704,
"learning_rate": 3.3679245283018865e-05,
"log_odds_chosen": 17.620845794677734,
"log_odds_ratio": -0.0001716611732263118,
"logits/chosen": -1.702080249786377,
"logits/rejected": -5.444385051727295,
"logps/chosen": -1.224086046218872,
"logps/rejected": -18.321523666381836,
"loss": 1.324,
"nll_loss": 1.3239673376083374,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.12240861356258392,
"rewards/margins": 1.709743857383728,
"rewards/rejected": -1.8321523666381836,
"step": 173
},
{
"epoch": 0.6566037735849056,
"grad_norm": 0.3308000862598419,
"learning_rate": 3.3584905660377356e-05,
"log_odds_chosen": 13.906808853149414,
"log_odds_ratio": -0.05518035590648651,
"logits/chosen": -1.4705300331115723,
"logits/rejected": -4.510253429412842,
"logps/chosen": -1.5690919160842896,
"logps/rejected": -15.069160461425781,
"loss": 1.3463,
"nll_loss": 1.3407379388809204,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15690919756889343,
"rewards/margins": 1.3500069379806519,
"rewards/rejected": -1.5069161653518677,
"step": 174
},
{
"epoch": 0.660377358490566,
"grad_norm": 0.2951977550983429,
"learning_rate": 3.349056603773585e-05,
"log_odds_chosen": 17.145343780517578,
"log_odds_ratio": -0.06170666217803955,
"logits/chosen": -2.072368860244751,
"logits/rejected": -5.481960773468018,
"logps/chosen": -1.3195624351501465,
"logps/rejected": -18.049631118774414,
"loss": 1.3551,
"nll_loss": 1.348900556564331,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.13195623457431793,
"rewards/margins": 1.6730068922042847,
"rewards/rejected": -1.8049631118774414,
"step": 175
},
{
"epoch": 0.6641509433962264,
"grad_norm": 0.37980836629867554,
"learning_rate": 3.339622641509434e-05,
"log_odds_chosen": 14.718152046203613,
"log_odds_ratio": -0.13777947425842285,
"logits/chosen": -1.4754526615142822,
"logits/rejected": -3.541693687438965,
"logps/chosen": -1.6201732158660889,
"logps/rejected": -15.98199462890625,
"loss": 1.4959,
"nll_loss": 1.4821652173995972,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1620173305273056,
"rewards/margins": 1.4361821413040161,
"rewards/rejected": -1.598199486732483,
"step": 176
},
{
"epoch": 0.6679245283018868,
"grad_norm": 0.2625764012336731,
"learning_rate": 3.330188679245283e-05,
"log_odds_chosen": 16.301048278808594,
"log_odds_ratio": -0.07389828562736511,
"logits/chosen": -1.6500033140182495,
"logits/rejected": -5.951912879943848,
"logps/chosen": -1.6418194770812988,
"logps/rejected": -17.625288009643555,
"loss": 1.6667,
"nll_loss": 1.659282922744751,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16418196260929108,
"rewards/margins": 1.5983469486236572,
"rewards/rejected": -1.762528896331787,
"step": 177
},
{
"epoch": 0.6716981132075471,
"grad_norm": 0.33599424362182617,
"learning_rate": 3.320754716981132e-05,
"log_odds_chosen": 18.26766014099121,
"log_odds_ratio": -2.645006361490232e-06,
"logits/chosen": -1.5161099433898926,
"logits/rejected": -4.741918087005615,
"logps/chosen": -1.222088098526001,
"logps/rejected": -18.985591888427734,
"loss": 1.4988,
"nll_loss": 1.4987843036651611,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.12220881879329681,
"rewards/margins": 1.7763502597808838,
"rewards/rejected": -1.8985592126846313,
"step": 178
},
{
"epoch": 0.6754716981132075,
"grad_norm": 0.3460422158241272,
"learning_rate": 3.311320754716981e-05,
"log_odds_chosen": 15.672019958496094,
"log_odds_ratio": -0.0003013765381183475,
"logits/chosen": -1.8128689527511597,
"logits/rejected": -5.0894670486450195,
"logps/chosen": -1.912743330001831,
"logps/rejected": -17.398412704467773,
"loss": 1.4925,
"nll_loss": 1.49249267578125,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19127434492111206,
"rewards/margins": 1.5485669374465942,
"rewards/rejected": -1.7398412227630615,
"step": 179
},
{
"epoch": 0.6792452830188679,
"grad_norm": 0.325990229845047,
"learning_rate": 3.30188679245283e-05,
"log_odds_chosen": 16.67650032043457,
"log_odds_ratio": -7.659475159016438e-06,
"logits/chosen": -1.7975369691848755,
"logits/rejected": -5.100127220153809,
"logps/chosen": -1.545456886291504,
"logps/rejected": -17.93629264831543,
"loss": 1.3418,
"nll_loss": 1.3418041467666626,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15454569458961487,
"rewards/margins": 1.639083743095398,
"rewards/rejected": -1.7936294078826904,
"step": 180
},
{
"epoch": 0.6830188679245283,
"grad_norm": 0.31531310081481934,
"learning_rate": 3.2924528301886794e-05,
"log_odds_chosen": 16.065807342529297,
"log_odds_ratio": -0.06276258826255798,
"logits/chosen": -1.224330186843872,
"logits/rejected": -4.1433610916137695,
"logps/chosen": -1.5791137218475342,
"logps/rejected": -17.382640838623047,
"loss": 1.2837,
"nll_loss": 1.277405858039856,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15791137516498566,
"rewards/margins": 1.580352783203125,
"rewards/rejected": -1.7382642030715942,
"step": 181
},
{
"epoch": 0.6867924528301886,
"grad_norm": 0.3160483241081238,
"learning_rate": 3.2830188679245285e-05,
"log_odds_chosen": 13.249960899353027,
"log_odds_ratio": -0.09639393538236618,
"logits/chosen": -1.4797954559326172,
"logits/rejected": -5.896228790283203,
"logps/chosen": -1.5692899227142334,
"logps/rejected": -14.452865600585938,
"loss": 1.4959,
"nll_loss": 1.4862685203552246,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15692900121212006,
"rewards/margins": 1.2883576154708862,
"rewards/rejected": -1.4452866315841675,
"step": 182
},
{
"epoch": 0.690566037735849,
"grad_norm": 0.3040304183959961,
"learning_rate": 3.2735849056603776e-05,
"log_odds_chosen": 18.430316925048828,
"log_odds_ratio": -0.00038310332456603646,
"logits/chosen": -1.5657317638397217,
"logits/rejected": -4.217155933380127,
"logps/chosen": -1.4011714458465576,
"logps/rejected": -19.451824188232422,
"loss": 1.5214,
"nll_loss": 1.5213794708251953,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14011713862419128,
"rewards/margins": 1.805065393447876,
"rewards/rejected": -1.9451824426651,
"step": 183
},
{
"epoch": 0.6943396226415094,
"grad_norm": 0.34743577241897583,
"learning_rate": 3.264150943396227e-05,
"log_odds_chosen": 17.73297882080078,
"log_odds_ratio": -2.7628839234239422e-05,
"logits/chosen": -0.9867445230484009,
"logits/rejected": -3.8493878841400146,
"logps/chosen": -1.577934741973877,
"logps/rejected": -19.020362854003906,
"loss": 1.4107,
"nll_loss": 1.4106800556182861,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15779347717761993,
"rewards/margins": 1.7442429065704346,
"rewards/rejected": -1.9020363092422485,
"step": 184
},
{
"epoch": 0.6981132075471698,
"grad_norm": 0.29433855414390564,
"learning_rate": 3.254716981132075e-05,
"log_odds_chosen": 18.695938110351562,
"log_odds_ratio": -2.980303406729945e-06,
"logits/chosen": -1.6250673532485962,
"logits/rejected": -5.061118125915527,
"logps/chosen": -1.1787736415863037,
"logps/rejected": -19.337848663330078,
"loss": 1.4456,
"nll_loss": 1.4456124305725098,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.11787736415863037,
"rewards/margins": 1.8159077167510986,
"rewards/rejected": -1.9337849617004395,
"step": 185
},
{
"epoch": 0.7018867924528301,
"grad_norm": 0.33233845233917236,
"learning_rate": 3.245283018867924e-05,
"log_odds_chosen": 18.042146682739258,
"log_odds_ratio": -3.9149457734311e-05,
"logits/chosen": -1.7575011253356934,
"logits/rejected": -5.358031272888184,
"logps/chosen": -1.8146780729293823,
"logps/rejected": -19.64087677001953,
"loss": 1.5776,
"nll_loss": 1.5775460004806519,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18146783113479614,
"rewards/margins": 1.7826199531555176,
"rewards/rejected": -1.964087724685669,
"step": 186
},
{
"epoch": 0.7056603773584905,
"grad_norm": 0.3320876657962799,
"learning_rate": 3.2358490566037734e-05,
"log_odds_chosen": 17.938766479492188,
"log_odds_ratio": -3.449665200605523e-06,
"logits/chosen": -1.9206900596618652,
"logits/rejected": -4.899381160736084,
"logps/chosen": -1.370758295059204,
"logps/rejected": -18.916780471801758,
"loss": 1.3438,
"nll_loss": 1.3438050746917725,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13707584142684937,
"rewards/margins": 1.754602313041687,
"rewards/rejected": -1.8916780948638916,
"step": 187
},
{
"epoch": 0.7094339622641509,
"grad_norm": 0.5529193878173828,
"learning_rate": 3.2264150943396225e-05,
"log_odds_chosen": 17.36458969116211,
"log_odds_ratio": -3.398192711756565e-05,
"logits/chosen": -2.5587615966796875,
"logits/rejected": -5.520333290100098,
"logps/chosen": -1.3868842124938965,
"logps/rejected": -18.35110855102539,
"loss": 1.3595,
"nll_loss": 1.3595402240753174,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13868843019008636,
"rewards/margins": 1.6964225769042969,
"rewards/rejected": -1.8351107835769653,
"step": 188
},
{
"epoch": 0.7132075471698113,
"grad_norm": 0.49584269523620605,
"learning_rate": 3.2169811320754716e-05,
"log_odds_chosen": 16.835905075073242,
"log_odds_ratio": -0.018066758289933205,
"logits/chosen": -1.943171739578247,
"logits/rejected": -5.371037006378174,
"logps/chosen": -1.5242998600006104,
"logps/rejected": -18.055362701416016,
"loss": 1.6063,
"nll_loss": 1.6044623851776123,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15242999792099,
"rewards/margins": 1.6531062126159668,
"rewards/rejected": -1.8055362701416016,
"step": 189
},
{
"epoch": 0.7169811320754716,
"grad_norm": 0.3044775128364563,
"learning_rate": 3.207547169811321e-05,
"log_odds_chosen": 17.216352462768555,
"log_odds_ratio": -1.419415457348805e-05,
"logits/chosen": -1.447345495223999,
"logits/rejected": -5.178557395935059,
"logps/chosen": -1.574653148651123,
"logps/rejected": -18.42108726501465,
"loss": 1.6679,
"nll_loss": 1.6678857803344727,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15746530890464783,
"rewards/margins": 1.6846433877944946,
"rewards/rejected": -1.8421087265014648,
"step": 190
},
{
"epoch": 0.720754716981132,
"grad_norm": 0.33981502056121826,
"learning_rate": 3.19811320754717e-05,
"log_odds_chosen": 16.49580192565918,
"log_odds_ratio": -4.55726585641969e-05,
"logits/chosen": -2.627636671066284,
"logits/rejected": -5.452616214752197,
"logps/chosen": -1.7808005809783936,
"logps/rejected": -18.012622833251953,
"loss": 1.6983,
"nll_loss": 1.6982595920562744,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17808005213737488,
"rewards/margins": 1.6231824159622192,
"rewards/rejected": -1.8012624979019165,
"step": 191
},
{
"epoch": 0.7245283018867924,
"grad_norm": 0.2643303871154785,
"learning_rate": 3.188679245283019e-05,
"log_odds_chosen": 18.62483024597168,
"log_odds_ratio": -2.2575586626771837e-06,
"logits/chosen": -0.962692141532898,
"logits/rejected": -5.363802433013916,
"logps/chosen": -1.6161587238311768,
"logps/rejected": -19.921184539794922,
"loss": 1.6401,
"nll_loss": 1.6400614976882935,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16161587834358215,
"rewards/margins": 1.8305026292800903,
"rewards/rejected": -1.99211847782135,
"step": 192
},
{
"epoch": 0.7283018867924528,
"grad_norm": 0.5224670171737671,
"learning_rate": 3.179245283018868e-05,
"log_odds_chosen": 17.83526039123535,
"log_odds_ratio": -2.4736191335250624e-06,
"logits/chosen": -2.249207019805908,
"logits/rejected": -5.893120765686035,
"logps/chosen": -1.5763590335845947,
"logps/rejected": -19.069358825683594,
"loss": 1.4385,
"nll_loss": 1.438537359237671,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.157635897397995,
"rewards/margins": 1.7493000030517578,
"rewards/rejected": -1.9069358110427856,
"step": 193
},
{
"epoch": 0.7320754716981132,
"grad_norm": 0.33392333984375,
"learning_rate": 3.169811320754717e-05,
"log_odds_chosen": 14.567859649658203,
"log_odds_ratio": -0.004244968760758638,
"logits/chosen": -1.951322317123413,
"logits/rejected": -6.3537211418151855,
"logps/chosen": -1.85011887550354,
"logps/rejected": -16.212106704711914,
"loss": 1.6572,
"nll_loss": 1.6567444801330566,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18501189351081848,
"rewards/margins": 1.4361987113952637,
"rewards/rejected": -1.6212105751037598,
"step": 194
},
{
"epoch": 0.7358490566037735,
"grad_norm": 0.33259832859039307,
"learning_rate": 3.160377358490566e-05,
"log_odds_chosen": 14.361299514770508,
"log_odds_ratio": -0.18310227990150452,
"logits/chosen": -1.6968700885772705,
"logits/rejected": -5.388698577880859,
"logps/chosen": -1.7790277004241943,
"logps/rejected": -15.95116901397705,
"loss": 1.571,
"nll_loss": 1.552640438079834,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17790275812149048,
"rewards/margins": 1.4172141551971436,
"rewards/rejected": -1.5951169729232788,
"step": 195
},
{
"epoch": 0.7396226415094339,
"grad_norm": 0.3773075342178345,
"learning_rate": 3.1509433962264154e-05,
"log_odds_chosen": 13.168961524963379,
"log_odds_ratio": -0.06529372930526733,
"logits/chosen": -1.13154935836792,
"logits/rejected": -4.5446624755859375,
"logps/chosen": -1.3777858018875122,
"logps/rejected": -14.206029891967773,
"loss": 1.3619,
"nll_loss": 1.3554112911224365,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.13777858018875122,
"rewards/margins": 1.2828243970870972,
"rewards/rejected": -1.4206030368804932,
"step": 196
},
{
"epoch": 0.7433962264150943,
"grad_norm": 0.3047720789909363,
"learning_rate": 3.1415094339622645e-05,
"log_odds_chosen": 18.703304290771484,
"log_odds_ratio": -3.1293072879634565e-06,
"logits/chosen": -1.948151707649231,
"logits/rejected": -5.263698101043701,
"logps/chosen": -1.3663289546966553,
"logps/rejected": -19.713842391967773,
"loss": 1.2498,
"nll_loss": 1.2497950792312622,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13663290441036224,
"rewards/margins": 1.8347513675689697,
"rewards/rejected": -1.9713842868804932,
"step": 197
},
{
"epoch": 0.7471698113207547,
"grad_norm": 0.3021809458732605,
"learning_rate": 3.132075471698113e-05,
"log_odds_chosen": 16.85780143737793,
"log_odds_ratio": -0.09297717362642288,
"logits/chosen": -2.2993781566619873,
"logits/rejected": -5.5674147605896,
"logps/chosen": -1.532175064086914,
"logps/rejected": -17.899972915649414,
"loss": 1.4076,
"nll_loss": 1.3982605934143066,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15321749448776245,
"rewards/margins": 1.63677978515625,
"rewards/rejected": -1.7899973392486572,
"step": 198
},
{
"epoch": 0.7509433962264151,
"grad_norm": 0.3545827269554138,
"learning_rate": 3.122641509433962e-05,
"log_odds_chosen": 16.710426330566406,
"log_odds_ratio": -1.721089915918128e-06,
"logits/chosen": -2.0543713569641113,
"logits/rejected": -6.516955375671387,
"logps/chosen": -1.7566440105438232,
"logps/rejected": -18.181135177612305,
"loss": 1.5498,
"nll_loss": 1.5497593879699707,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17566442489624023,
"rewards/margins": 1.642449140548706,
"rewards/rejected": -1.8181135654449463,
"step": 199
},
{
"epoch": 0.7547169811320755,
"grad_norm": 0.34778276085853577,
"learning_rate": 3.113207547169811e-05,
"log_odds_chosen": 18.424583435058594,
"log_odds_ratio": -7.215005462057889e-05,
"logits/chosen": -1.3918465375900269,
"logits/rejected": -3.4874095916748047,
"logps/chosen": -1.4977375268936157,
"logps/rejected": -19.452709197998047,
"loss": 1.5377,
"nll_loss": 1.5376887321472168,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1497737616300583,
"rewards/margins": 1.795497179031372,
"rewards/rejected": -1.9452710151672363,
"step": 200
},
{
"epoch": 0.7584905660377359,
"grad_norm": 0.2991779148578644,
"learning_rate": 3.10377358490566e-05,
"log_odds_chosen": 17.403640747070312,
"log_odds_ratio": -9.71866975305602e-05,
"logits/chosen": -1.1479456424713135,
"logits/rejected": -5.782166481018066,
"logps/chosen": -1.5020239353179932,
"logps/rejected": -18.525487899780273,
"loss": 1.4808,
"nll_loss": 1.4808180332183838,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15020239353179932,
"rewards/margins": 1.7023463249206543,
"rewards/rejected": -1.8525487184524536,
"step": 201
},
{
"epoch": 0.7622641509433963,
"grad_norm": 0.32272449135780334,
"learning_rate": 3.094339622641509e-05,
"log_odds_chosen": 18.042186737060547,
"log_odds_ratio": -1.0580668458715081e-05,
"logits/chosen": -1.1393158435821533,
"logits/rejected": -5.057635307312012,
"logps/chosen": -1.3160855770111084,
"logps/rejected": -18.88675308227539,
"loss": 1.3602,
"nll_loss": 1.3602066040039062,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13160854578018188,
"rewards/margins": 1.7570668458938599,
"rewards/rejected": -1.888675332069397,
"step": 202
},
{
"epoch": 0.7660377358490567,
"grad_norm": 0.32788515090942383,
"learning_rate": 3.0849056603773584e-05,
"log_odds_chosen": 16.405580520629883,
"log_odds_ratio": -0.06243692338466644,
"logits/chosen": -1.4627768993377686,
"logits/rejected": -5.623252868652344,
"logps/chosen": -1.8027026653289795,
"logps/rejected": -18.00653839111328,
"loss": 1.6046,
"nll_loss": 1.5983972549438477,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1802702695131302,
"rewards/margins": 1.6203837394714355,
"rewards/rejected": -1.8006539344787598,
"step": 203
},
{
"epoch": 0.769811320754717,
"grad_norm": 1.234761118888855,
"learning_rate": 3.0754716981132075e-05,
"log_odds_chosen": 18.839767456054688,
"log_odds_ratio": -4.664214884542162e-06,
"logits/chosen": -2.3654236793518066,
"logits/rejected": -5.212252140045166,
"logps/chosen": -1.4169280529022217,
"logps/rejected": -19.78860092163086,
"loss": 1.4248,
"nll_loss": 1.4247938394546509,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14169281721115112,
"rewards/margins": 1.8371672630310059,
"rewards/rejected": -1.9788599014282227,
"step": 204
},
{
"epoch": 0.7735849056603774,
"grad_norm": 0.31495201587677,
"learning_rate": 3.0660377358490567e-05,
"log_odds_chosen": 17.49294090270996,
"log_odds_ratio": -1.4975751128076809e-06,
"logits/chosen": -2.495401382446289,
"logits/rejected": -5.783306121826172,
"logps/chosen": -1.6663013696670532,
"logps/rejected": -18.882904052734375,
"loss": 1.684,
"nll_loss": 1.683968186378479,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16663014888763428,
"rewards/margins": 1.7216603755950928,
"rewards/rejected": -1.8882904052734375,
"step": 205
},
{
"epoch": 0.7773584905660378,
"grad_norm": 0.4705086350440979,
"learning_rate": 3.056603773584906e-05,
"log_odds_chosen": 18.32155990600586,
"log_odds_ratio": -0.05019821971654892,
"logits/chosen": -1.1568958759307861,
"logits/rejected": -4.691790580749512,
"logps/chosen": -1.4851261377334595,
"logps/rejected": -19.43079376220703,
"loss": 1.4827,
"nll_loss": 1.4777144193649292,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.14851261675357819,
"rewards/margins": 1.7945667505264282,
"rewards/rejected": -1.9430793523788452,
"step": 206
},
{
"epoch": 0.7811320754716982,
"grad_norm": 0.3913361728191376,
"learning_rate": 3.047169811320755e-05,
"log_odds_chosen": 16.319984436035156,
"log_odds_ratio": -0.05193231999874115,
"logits/chosen": -2.967703342437744,
"logits/rejected": -5.948724269866943,
"logps/chosen": -1.7976689338684082,
"logps/rejected": -17.946117401123047,
"loss": 1.7308,
"nll_loss": 1.7255607843399048,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.17976689338684082,
"rewards/margins": 1.6148450374603271,
"rewards/rejected": -1.7946120500564575,
"step": 207
},
{
"epoch": 0.7849056603773585,
"grad_norm": 0.309137761592865,
"learning_rate": 3.0377358490566036e-05,
"log_odds_chosen": 19.219402313232422,
"log_odds_ratio": -7.450602197422995e-07,
"logits/chosen": -0.8913941383361816,
"logits/rejected": -4.125641822814941,
"logps/chosen": -1.470247507095337,
"logps/rejected": -20.202777862548828,
"loss": 1.4043,
"nll_loss": 1.4043306112289429,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14702476561069489,
"rewards/margins": 1.8732528686523438,
"rewards/rejected": -2.020277500152588,
"step": 208
},
{
"epoch": 0.7886792452830189,
"grad_norm": 0.28629574179649353,
"learning_rate": 3.0283018867924528e-05,
"log_odds_chosen": 16.79374122619629,
"log_odds_ratio": -0.00662041874602437,
"logits/chosen": -0.6648188829421997,
"logits/rejected": -6.581351280212402,
"logps/chosen": -1.4513788223266602,
"logps/rejected": -17.918479919433594,
"loss": 1.4518,
"nll_loss": 1.4511276483535767,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14513790607452393,
"rewards/margins": 1.6467101573944092,
"rewards/rejected": -1.7918481826782227,
"step": 209
},
{
"epoch": 0.7924528301886793,
"grad_norm": 0.32280489802360535,
"learning_rate": 3.018867924528302e-05,
"log_odds_chosen": 17.51369857788086,
"log_odds_ratio": -3.799800651904661e-07,
"logits/chosen": -1.7065296173095703,
"logits/rejected": -5.712586402893066,
"logps/chosen": -1.4828323125839233,
"logps/rejected": -18.661277770996094,
"loss": 1.5052,
"nll_loss": 1.5052316188812256,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1482832282781601,
"rewards/margins": 1.7178447246551514,
"rewards/rejected": -1.8661279678344727,
"step": 210
},
{
"epoch": 0.7962264150943397,
"grad_norm": 1.3183481693267822,
"learning_rate": 3.009433962264151e-05,
"log_odds_chosen": 16.796680450439453,
"log_odds_ratio": -1.360561145702377e-05,
"logits/chosen": -1.5383379459381104,
"logits/rejected": -4.357212543487549,
"logps/chosen": -1.435273289680481,
"logps/rejected": -17.903779983520508,
"loss": 1.4249,
"nll_loss": 1.4248584508895874,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1435273289680481,
"rewards/margins": 1.646850824356079,
"rewards/rejected": -1.790378212928772,
"step": 211
},
{
"epoch": 0.8,
"grad_norm": 0.3218695819377899,
"learning_rate": 3e-05,
"log_odds_chosen": 17.916414260864258,
"log_odds_ratio": -1.38259565574117e-05,
"logits/chosen": -0.24590548872947693,
"logits/rejected": -4.629819869995117,
"logps/chosen": -1.4526543617248535,
"logps/rejected": -18.942493438720703,
"loss": 1.5065,
"nll_loss": 1.5065315961837769,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14526543021202087,
"rewards/margins": 1.7489840984344482,
"rewards/rejected": -1.894249439239502,
"step": 212
},
{
"epoch": 0.8037735849056604,
"grad_norm": 0.3358078896999359,
"learning_rate": 2.9905660377358492e-05,
"log_odds_chosen": 14.796281814575195,
"log_odds_ratio": -0.0924949049949646,
"logits/chosen": -1.8153976202011108,
"logits/rejected": -4.928350448608398,
"logps/chosen": -1.4929924011230469,
"logps/rejected": -15.970230102539062,
"loss": 1.4921,
"nll_loss": 1.4828470945358276,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1492992341518402,
"rewards/margins": 1.4477238655090332,
"rewards/rejected": -1.5970230102539062,
"step": 213
},
{
"epoch": 0.8075471698113208,
"grad_norm": 0.3094753921031952,
"learning_rate": 2.9811320754716983e-05,
"log_odds_chosen": 15.192641258239746,
"log_odds_ratio": -0.02841039001941681,
"logits/chosen": -2.0973596572875977,
"logits/rejected": -4.845073699951172,
"logps/chosen": -1.5211032629013062,
"logps/rejected": -16.33423614501953,
"loss": 1.5114,
"nll_loss": 1.5085326433181763,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15211032330989838,
"rewards/margins": 1.4813133478164673,
"rewards/rejected": -1.6334238052368164,
"step": 214
},
{
"epoch": 0.8113207547169812,
"grad_norm": 0.3006260395050049,
"learning_rate": 2.971698113207547e-05,
"log_odds_chosen": 16.960983276367188,
"log_odds_ratio": -0.05776922032237053,
"logits/chosen": -0.9070106744766235,
"logits/rejected": -4.620122909545898,
"logps/chosen": -1.5111815929412842,
"logps/rejected": -18.19387435913086,
"loss": 1.4915,
"nll_loss": 1.4857306480407715,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15111815929412842,
"rewards/margins": 1.6682692766189575,
"rewards/rejected": -1.819387435913086,
"step": 215
},
{
"epoch": 0.8150943396226416,
"grad_norm": 0.33227863907814026,
"learning_rate": 2.9622641509433962e-05,
"log_odds_chosen": 18.719661712646484,
"log_odds_ratio": -9.611276254872791e-07,
"logits/chosen": -1.1189182996749878,
"logits/rejected": -4.95672082901001,
"logps/chosen": -1.2981202602386475,
"logps/rejected": -19.64653778076172,
"loss": 1.3002,
"nll_loss": 1.3002173900604248,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.12981203198432922,
"rewards/margins": 1.8348416090011597,
"rewards/rejected": -1.9646536111831665,
"step": 216
},
{
"epoch": 0.8188679245283019,
"grad_norm": 0.2781373858451843,
"learning_rate": 2.9528301886792453e-05,
"log_odds_chosen": 17.168392181396484,
"log_odds_ratio": -0.0007286164909601212,
"logits/chosen": -2.3207314014434814,
"logits/rejected": -6.724469184875488,
"logps/chosen": -1.6115381717681885,
"logps/rejected": -18.52199935913086,
"loss": 1.576,
"nll_loss": 1.575928807258606,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16115380823612213,
"rewards/margins": 1.6910459995269775,
"rewards/rejected": -1.8521997928619385,
"step": 217
},
{
"epoch": 0.8226415094339623,
"grad_norm": 0.29558250308036804,
"learning_rate": 2.9433962264150944e-05,
"log_odds_chosen": 15.084373474121094,
"log_odds_ratio": -0.025247111916542053,
"logits/chosen": -0.26611119508743286,
"logits/rejected": -4.655291557312012,
"logps/chosen": -1.8194210529327393,
"logps/rejected": -16.723453521728516,
"loss": 1.6581,
"nll_loss": 1.6555492877960205,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1819421350955963,
"rewards/margins": 1.490403175354004,
"rewards/rejected": -1.6723453998565674,
"step": 218
},
{
"epoch": 0.8264150943396227,
"grad_norm": 0.2880527973175049,
"learning_rate": 2.9339622641509435e-05,
"log_odds_chosen": 17.260496139526367,
"log_odds_ratio": -4.0319591789739206e-05,
"logits/chosen": -0.7846205830574036,
"logits/rejected": -4.96677827835083,
"logps/chosen": -1.624293565750122,
"logps/rejected": -18.620319366455078,
"loss": 1.5828,
"nll_loss": 1.5828019380569458,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16242937743663788,
"rewards/margins": 1.6996028423309326,
"rewards/rejected": -1.8620320558547974,
"step": 219
},
{
"epoch": 0.8301886792452831,
"grad_norm": 0.24706144630908966,
"learning_rate": 2.9245283018867926e-05,
"log_odds_chosen": 18.422889709472656,
"log_odds_ratio": -8.091576091828756e-06,
"logits/chosen": -1.6815857887268066,
"logits/rejected": -6.923168659210205,
"logps/chosen": -1.4026367664337158,
"logps/rejected": -19.40614891052246,
"loss": 1.4331,
"nll_loss": 1.4331367015838623,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14026367664337158,
"rewards/margins": 1.8003512620925903,
"rewards/rejected": -1.940614938735962,
"step": 220
},
{
"epoch": 0.8339622641509434,
"grad_norm": 0.3142763078212738,
"learning_rate": 2.9150943396226417e-05,
"log_odds_chosen": 17.446882247924805,
"log_odds_ratio": -0.07998549938201904,
"logits/chosen": -1.7830405235290527,
"logits/rejected": -5.290676593780518,
"logps/chosen": -1.7186626195907593,
"logps/rejected": -18.9097957611084,
"loss": 1.672,
"nll_loss": 1.6639615297317505,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1718662679195404,
"rewards/margins": 1.7191133499145508,
"rewards/rejected": -1.890979528427124,
"step": 221
},
{
"epoch": 0.8377358490566038,
"grad_norm": 0.32208096981048584,
"learning_rate": 2.9056603773584905e-05,
"log_odds_chosen": 18.16156005859375,
"log_odds_ratio": -8.31518536870135e-06,
"logits/chosen": -2.01678729057312,
"logits/rejected": -5.499569892883301,
"logps/chosen": -1.758958101272583,
"logps/rejected": -19.644182205200195,
"loss": 1.5194,
"nll_loss": 1.5194087028503418,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1758958250284195,
"rewards/margins": 1.788522481918335,
"rewards/rejected": -1.9644180536270142,
"step": 222
},
{
"epoch": 0.8415094339622642,
"grad_norm": 0.3508431613445282,
"learning_rate": 2.8962264150943396e-05,
"log_odds_chosen": 16.593107223510742,
"log_odds_ratio": -4.483833254198544e-05,
"logits/chosen": -1.0808777809143066,
"logits/rejected": -5.002264022827148,
"logps/chosen": -1.6178133487701416,
"logps/rejected": -17.956409454345703,
"loss": 1.4168,
"nll_loss": 1.4167941808700562,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16178134083747864,
"rewards/margins": 1.633859395980835,
"rewards/rejected": -1.7956409454345703,
"step": 223
},
{
"epoch": 0.8452830188679246,
"grad_norm": 0.3514016270637512,
"learning_rate": 2.8867924528301887e-05,
"log_odds_chosen": 16.712297439575195,
"log_odds_ratio": -3.978675522375852e-06,
"logits/chosen": -1.4249446392059326,
"logits/rejected": -4.565216064453125,
"logps/chosen": -1.8810580968856812,
"logps/rejected": -18.380489349365234,
"loss": 1.4511,
"nll_loss": 1.4511330127716064,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18810580670833588,
"rewards/margins": 1.6499433517456055,
"rewards/rejected": -1.8380491733551025,
"step": 224
},
{
"epoch": 0.8490566037735849,
"grad_norm": 0.3235526978969574,
"learning_rate": 2.8773584905660378e-05,
"log_odds_chosen": 17.07125473022461,
"log_odds_ratio": -4.053184511576546e-06,
"logits/chosen": -0.9398770332336426,
"logits/rejected": -4.418087482452393,
"logps/chosen": -1.5705912113189697,
"logps/rejected": -18.33386993408203,
"loss": 1.5413,
"nll_loss": 1.5412647724151611,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15705913305282593,
"rewards/margins": 1.6763279438018799,
"rewards/rejected": -1.8333872556686401,
"step": 225
},
{
"epoch": 0.8528301886792453,
"grad_norm": 0.3555522561073303,
"learning_rate": 2.867924528301887e-05,
"log_odds_chosen": 16.7454833984375,
"log_odds_ratio": -0.09962432086467743,
"logits/chosen": -2.2143068313598633,
"logits/rejected": -5.618562698364258,
"logps/chosen": -1.8031433820724487,
"logps/rejected": -18.335561752319336,
"loss": 1.5886,
"nll_loss": 1.5786436796188354,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1803143322467804,
"rewards/margins": 1.6532416343688965,
"rewards/rejected": -1.833556056022644,
"step": 226
},
{
"epoch": 0.8566037735849057,
"grad_norm": 0.30511021614074707,
"learning_rate": 2.858490566037736e-05,
"log_odds_chosen": 16.435148239135742,
"log_odds_ratio": -0.05051492527127266,
"logits/chosen": -1.1615848541259766,
"logits/rejected": -4.774338722229004,
"logps/chosen": -1.4906309843063354,
"logps/rejected": -17.58717155456543,
"loss": 1.4203,
"nll_loss": 1.4152026176452637,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1490630954504013,
"rewards/margins": 1.6096539497375488,
"rewards/rejected": -1.7587170600891113,
"step": 227
},
{
"epoch": 0.8603773584905661,
"grad_norm": 0.2953304946422577,
"learning_rate": 2.8490566037735848e-05,
"log_odds_chosen": 17.240137100219727,
"log_odds_ratio": -0.00012110001989640296,
"logits/chosen": -2.184664726257324,
"logits/rejected": -6.46320915222168,
"logps/chosen": -1.8011096715927124,
"logps/rejected": -18.774145126342773,
"loss": 1.5879,
"nll_loss": 1.5879219770431519,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18011096119880676,
"rewards/margins": 1.6973035335540771,
"rewards/rejected": -1.8774144649505615,
"step": 228
},
{
"epoch": 0.8641509433962264,
"grad_norm": 0.3585098087787628,
"learning_rate": 2.839622641509434e-05,
"log_odds_chosen": 15.124954223632812,
"log_odds_ratio": -0.00011490716133266687,
"logits/chosen": -0.7336174249649048,
"logits/rejected": -3.8608899116516113,
"logps/chosen": -1.7269798517227173,
"logps/rejected": -16.582866668701172,
"loss": 1.6302,
"nll_loss": 1.6301769018173218,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1726979911327362,
"rewards/margins": 1.4855889081954956,
"rewards/rejected": -1.6582868099212646,
"step": 229
},
{
"epoch": 0.8679245283018868,
"grad_norm": 0.8896408677101135,
"learning_rate": 2.830188679245283e-05,
"log_odds_chosen": 18.62564468383789,
"log_odds_ratio": -1.3411050758804777e-07,
"logits/chosen": -0.7562139630317688,
"logits/rejected": -6.943488121032715,
"logps/chosen": -1.6029720306396484,
"logps/rejected": -19.97881507873535,
"loss": 1.5316,
"nll_loss": 1.531610131263733,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1602971851825714,
"rewards/margins": 1.837584376335144,
"rewards/rejected": -1.997881531715393,
"step": 230
},
{
"epoch": 0.8716981132075472,
"grad_norm": 0.3106488883495331,
"learning_rate": 2.820754716981132e-05,
"log_odds_chosen": 17.55572509765625,
"log_odds_ratio": -0.057248592376708984,
"logits/chosen": -1.0949640274047852,
"logits/rejected": -3.665689468383789,
"logps/chosen": -1.5709400177001953,
"logps/rejected": -18.767902374267578,
"loss": 1.3768,
"nll_loss": 1.3711098432540894,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15709398686885834,
"rewards/margins": 1.719696283340454,
"rewards/rejected": -1.8767902851104736,
"step": 231
},
{
"epoch": 0.8754716981132076,
"grad_norm": 0.329771488904953,
"learning_rate": 2.8113207547169812e-05,
"log_odds_chosen": 14.081006050109863,
"log_odds_ratio": -0.22504055500030518,
"logits/chosen": -1.2311842441558838,
"logits/rejected": -3.282421827316284,
"logps/chosen": -1.647598385810852,
"logps/rejected": -15.423025131225586,
"loss": 1.4569,
"nll_loss": 1.4343469142913818,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.16475984454154968,
"rewards/margins": 1.3775426149368286,
"rewards/rejected": -1.5423026084899902,
"step": 232
},
{
"epoch": 0.879245283018868,
"grad_norm": 0.3315247595310211,
"learning_rate": 2.8018867924528303e-05,
"log_odds_chosen": 16.6135196685791,
"log_odds_ratio": -0.015212212689220905,
"logits/chosen": -1.0095354318618774,
"logits/rejected": -5.442407608032227,
"logps/chosen": -1.6335803270339966,
"logps/rejected": -17.933931350708008,
"loss": 1.5397,
"nll_loss": 1.5382212400436401,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16335803270339966,
"rewards/margins": 1.6300350427627563,
"rewards/rejected": -1.7933931350708008,
"step": 233
},
{
"epoch": 0.8830188679245283,
"grad_norm": 0.2853158116340637,
"learning_rate": 2.7924528301886794e-05,
"log_odds_chosen": 17.776758193969727,
"log_odds_ratio": -0.00014782443759031594,
"logits/chosen": -1.9506645202636719,
"logits/rejected": -6.754343509674072,
"logps/chosen": -1.5541913509368896,
"logps/rejected": -19.017948150634766,
"loss": 1.5127,
"nll_loss": 1.5127075910568237,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15541914105415344,
"rewards/margins": 1.7463756799697876,
"rewards/rejected": -1.9017947912216187,
"step": 234
},
{
"epoch": 0.8867924528301887,
"grad_norm": 0.3323514759540558,
"learning_rate": 2.7830188679245282e-05,
"log_odds_chosen": 19.43785285949707,
"log_odds_ratio": -1.6764031443017302e-06,
"logits/chosen": -1.0080933570861816,
"logits/rejected": -4.819693565368652,
"logps/chosen": -1.4333504438400269,
"logps/rejected": -20.516963958740234,
"loss": 1.3111,
"nll_loss": 1.3111367225646973,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14333504438400269,
"rewards/margins": 1.9083614349365234,
"rewards/rejected": -2.051696538925171,
"step": 235
},
{
"epoch": 0.8905660377358491,
"grad_norm": 0.39661461114883423,
"learning_rate": 2.7735849056603773e-05,
"log_odds_chosen": 17.063642501831055,
"log_odds_ratio": -0.02693340554833412,
"logits/chosen": -2.2235286235809326,
"logits/rejected": -4.167008876800537,
"logps/chosen": -1.677812099456787,
"logps/rejected": -18.464336395263672,
"loss": 1.5569,
"nll_loss": 1.5542023181915283,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16778121888637543,
"rewards/margins": 1.6786524057388306,
"rewards/rejected": -1.8464335203170776,
"step": 236
},
{
"epoch": 0.8943396226415095,
"grad_norm": 0.2811707854270935,
"learning_rate": 2.7641509433962264e-05,
"log_odds_chosen": 17.162025451660156,
"log_odds_ratio": -0.0785740464925766,
"logits/chosen": -2.9656736850738525,
"logits/rejected": -5.683234214782715,
"logps/chosen": -1.4220517873764038,
"logps/rejected": -18.229448318481445,
"loss": 1.5842,
"nll_loss": 1.576366662979126,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.14220517873764038,
"rewards/margins": 1.6807397603988647,
"rewards/rejected": -1.8229451179504395,
"step": 237
},
{
"epoch": 0.8981132075471698,
"grad_norm": 0.29917973279953003,
"learning_rate": 2.7547169811320755e-05,
"log_odds_chosen": 16.159685134887695,
"log_odds_ratio": -0.022400004789233208,
"logits/chosen": -0.6307175755500793,
"logits/rejected": -5.9625935554504395,
"logps/chosen": -1.3831230401992798,
"logps/rejected": -17.1973876953125,
"loss": 1.3797,
"nll_loss": 1.377410888671875,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13831230998039246,
"rewards/margins": 1.5814265012741089,
"rewards/rejected": -1.7197388410568237,
"step": 238
},
{
"epoch": 0.9018867924528302,
"grad_norm": 0.4006063938140869,
"learning_rate": 2.7452830188679247e-05,
"log_odds_chosen": 17.939586639404297,
"log_odds_ratio": -1.753244941937737e-05,
"logits/chosen": -0.8297520875930786,
"logits/rejected": -5.683241844177246,
"logps/chosen": -1.3332468271255493,
"logps/rejected": -18.919038772583008,
"loss": 1.3098,
"nll_loss": 1.3098140954971313,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13332468271255493,
"rewards/margins": 1.7585792541503906,
"rewards/rejected": -1.8919038772583008,
"step": 239
},
{
"epoch": 0.9056603773584906,
"grad_norm": 0.30147290229797363,
"learning_rate": 2.7358490566037738e-05,
"log_odds_chosen": 15.91531753540039,
"log_odds_ratio": -0.03641377389431,
"logits/chosen": -0.7411012649536133,
"logits/rejected": -4.249468803405762,
"logps/chosen": -1.5858731269836426,
"logps/rejected": -17.18975830078125,
"loss": 1.5853,
"nll_loss": 1.5816415548324585,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15858730673789978,
"rewards/margins": 1.5603885650634766,
"rewards/rejected": -1.7189757823944092,
"step": 240
},
{
"epoch": 0.909433962264151,
"grad_norm": 0.31893160939216614,
"learning_rate": 2.726415094339623e-05,
"log_odds_chosen": 16.919687271118164,
"log_odds_ratio": -0.0328671857714653,
"logits/chosen": -1.2528479099273682,
"logits/rejected": -4.872306823730469,
"logps/chosen": -1.2921593189239502,
"logps/rejected": -17.771411895751953,
"loss": 1.4341,
"nll_loss": 1.43081533908844,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.12921592593193054,
"rewards/margins": 1.6479253768920898,
"rewards/rejected": -1.7771413326263428,
"step": 241
},
{
"epoch": 0.9132075471698113,
"grad_norm": 0.295743465423584,
"learning_rate": 2.7169811320754716e-05,
"log_odds_chosen": 17.80528450012207,
"log_odds_ratio": -0.00010832020052475855,
"logits/chosen": -0.645114541053772,
"logits/rejected": -5.190818786621094,
"logps/chosen": -1.5482758283615112,
"logps/rejected": -19.037681579589844,
"loss": 1.4926,
"nll_loss": 1.492592215538025,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15482759475708008,
"rewards/margins": 1.7489407062530518,
"rewards/rejected": -1.9037683010101318,
"step": 242
},
{
"epoch": 0.9169811320754717,
"grad_norm": 0.362249493598938,
"learning_rate": 2.7075471698113207e-05,
"log_odds_chosen": 17.395801544189453,
"log_odds_ratio": -0.007926247082650661,
"logits/chosen": -2.3562960624694824,
"logits/rejected": -6.801301956176758,
"logps/chosen": -1.1412396430969238,
"logps/rejected": -17.93915557861328,
"loss": 1.2173,
"nll_loss": 1.2165017127990723,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.11412396281957626,
"rewards/margins": 1.6797915697097778,
"rewards/rejected": -1.7939155101776123,
"step": 243
},
{
"epoch": 0.9207547169811321,
"grad_norm": 0.3344380259513855,
"learning_rate": 2.69811320754717e-05,
"log_odds_chosen": 18.15846824645996,
"log_odds_ratio": -1.184646635010722e-06,
"logits/chosen": -1.4028515815734863,
"logits/rejected": -5.666131019592285,
"logps/chosen": -1.6515846252441406,
"logps/rejected": -19.494230270385742,
"loss": 1.4627,
"nll_loss": 1.462727665901184,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1651584655046463,
"rewards/margins": 1.7842646837234497,
"rewards/rejected": -1.9494233131408691,
"step": 244
},
{
"epoch": 0.9245283018867925,
"grad_norm": 0.3528442084789276,
"learning_rate": 2.688679245283019e-05,
"log_odds_chosen": 16.469255447387695,
"log_odds_ratio": -0.16001646220684052,
"logits/chosen": -0.5904859304428101,
"logits/rejected": -4.443323135375977,
"logps/chosen": -1.5586724281311035,
"logps/rejected": -17.767322540283203,
"loss": 1.3949,
"nll_loss": 1.3788522481918335,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15586724877357483,
"rewards/margins": 1.620864987373352,
"rewards/rejected": -1.7767322063446045,
"step": 245
},
{
"epoch": 0.9283018867924528,
"grad_norm": 0.36119183897972107,
"learning_rate": 2.679245283018868e-05,
"log_odds_chosen": 16.416719436645508,
"log_odds_ratio": -0.20440763235092163,
"logits/chosen": -1.4841980934143066,
"logits/rejected": -4.816326141357422,
"logps/chosen": -1.387542724609375,
"logps/rejected": -17.60167121887207,
"loss": 1.4418,
"nll_loss": 1.4213593006134033,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.13875426352024078,
"rewards/margins": 1.6214128732681274,
"rewards/rejected": -1.7601672410964966,
"step": 246
},
{
"epoch": 0.9320754716981132,
"grad_norm": 0.3182571828365326,
"learning_rate": 2.6698113207547172e-05,
"log_odds_chosen": 16.2152099609375,
"log_odds_ratio": -0.03764305263757706,
"logits/chosen": -1.0904548168182373,
"logits/rejected": -3.9518043994903564,
"logps/chosen": -1.4248569011688232,
"logps/rejected": -17.271034240722656,
"loss": 1.4531,
"nll_loss": 1.4493829011917114,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14248570799827576,
"rewards/margins": 1.5846177339553833,
"rewards/rejected": -1.7271034717559814,
"step": 247
},
{
"epoch": 0.9358490566037736,
"grad_norm": 0.321162611246109,
"learning_rate": 2.6603773584905663e-05,
"log_odds_chosen": 18.773868560791016,
"log_odds_ratio": -1.393025740981102e-05,
"logits/chosen": -0.8928079009056091,
"logits/rejected": -5.256990432739258,
"logps/chosen": -1.6134278774261475,
"logps/rejected": -20.060321807861328,
"loss": 1.5017,
"nll_loss": 1.5017071962356567,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1613427847623825,
"rewards/margins": 1.8446893692016602,
"rewards/rejected": -2.0060322284698486,
"step": 248
},
{
"epoch": 0.939622641509434,
"grad_norm": 0.8356024026870728,
"learning_rate": 2.650943396226415e-05,
"log_odds_chosen": 16.179561614990234,
"log_odds_ratio": -0.04995302855968475,
"logits/chosen": -1.4648141860961914,
"logits/rejected": -4.60720682144165,
"logps/chosen": -1.707082748413086,
"logps/rejected": -17.653757095336914,
"loss": 1.5352,
"nll_loss": 1.5301692485809326,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.17070826888084412,
"rewards/margins": 1.5946673154830933,
"rewards/rejected": -1.7653756141662598,
"step": 249
},
{
"epoch": 0.9433962264150944,
"grad_norm": 0.3033410608768463,
"learning_rate": 2.641509433962264e-05,
"log_odds_chosen": 16.49986457824707,
"log_odds_ratio": -0.0050300052389502525,
"logits/chosen": -1.6823077201843262,
"logits/rejected": -5.105968475341797,
"logps/chosen": -1.6065022945404053,
"logps/rejected": -17.84182357788086,
"loss": 1.5878,
"nll_loss": 1.5873433351516724,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16065022349357605,
"rewards/margins": 1.6235322952270508,
"rewards/rejected": -1.7841825485229492,
"step": 250
},
{
"epoch": 0.9471698113207547,
"grad_norm": 0.35018157958984375,
"learning_rate": 2.6320754716981133e-05,
"log_odds_chosen": 19.928390502929688,
"log_odds_ratio": -8.717207720110309e-07,
"logits/chosen": -0.29920998215675354,
"logits/rejected": -4.818760395050049,
"logps/chosen": -1.2851896286010742,
"logps/rejected": -20.66608238220215,
"loss": 1.2603,
"nll_loss": 1.2603236436843872,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1285189688205719,
"rewards/margins": 1.93808913230896,
"rewards/rejected": -2.066608428955078,
"step": 251
},
{
"epoch": 0.9509433962264151,
"grad_norm": 0.3246742784976959,
"learning_rate": 2.6226415094339624e-05,
"log_odds_chosen": 17.97346305847168,
"log_odds_ratio": -1.0215319889539387e-05,
"logits/chosen": -1.2597813606262207,
"logits/rejected": -5.395329475402832,
"logps/chosen": -1.4511744976043701,
"logps/rejected": -19.067272186279297,
"loss": 1.492,
"nll_loss": 1.4920153617858887,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14511744678020477,
"rewards/margins": 1.761609673500061,
"rewards/rejected": -1.9067270755767822,
"step": 252
},
{
"epoch": 0.9547169811320755,
"grad_norm": 0.3272016942501068,
"learning_rate": 2.6132075471698115e-05,
"log_odds_chosen": 18.806455612182617,
"log_odds_ratio": -0.04364859312772751,
"logits/chosen": -1.467919945716858,
"logits/rejected": -4.545738220214844,
"logps/chosen": -1.4638400077819824,
"logps/rejected": -19.876646041870117,
"loss": 1.3501,
"nll_loss": 1.3457329273223877,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.14638400077819824,
"rewards/margins": 1.8412805795669556,
"rewards/rejected": -1.9876646995544434,
"step": 253
},
{
"epoch": 0.9584905660377359,
"grad_norm": 0.6240985989570618,
"learning_rate": 2.6037735849056606e-05,
"log_odds_chosen": 19.93435287475586,
"log_odds_ratio": -7.450581485102248e-09,
"logits/chosen": -0.8115564584732056,
"logits/rejected": -6.536435127258301,
"logps/chosen": -1.6901774406433105,
"logps/rejected": -21.379650115966797,
"loss": 1.674,
"nll_loss": 1.6740447282791138,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1690177470445633,
"rewards/margins": 1.968947410583496,
"rewards/rejected": -2.137965202331543,
"step": 254
},
{
"epoch": 0.9622641509433962,
"grad_norm": 0.3309653401374817,
"learning_rate": 2.5943396226415094e-05,
"log_odds_chosen": 19.6575927734375,
"log_odds_ratio": -0.01984320767223835,
"logits/chosen": -0.7965545654296875,
"logits/rejected": -3.3540658950805664,
"logps/chosen": -1.5116550922393799,
"logps/rejected": -20.71581268310547,
"loss": 1.5013,
"nll_loss": 1.4993512630462646,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15116551518440247,
"rewards/margins": 1.9204158782958984,
"rewards/rejected": -2.0715813636779785,
"step": 255
},
{
"epoch": 0.9660377358490566,
"grad_norm": 0.3074422776699066,
"learning_rate": 2.5849056603773585e-05,
"log_odds_chosen": 19.469036102294922,
"log_odds_ratio": -0.000835958169773221,
"logits/chosen": -0.7382791638374329,
"logits/rejected": -3.960020065307617,
"logps/chosen": -1.3473050594329834,
"logps/rejected": -20.31693458557129,
"loss": 1.4874,
"nll_loss": 1.4873261451721191,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1347305178642273,
"rewards/margins": 1.896963119506836,
"rewards/rejected": -2.031693696975708,
"step": 256
},
{
"epoch": 0.969811320754717,
"grad_norm": 0.2996930480003357,
"learning_rate": 2.5754716981132076e-05,
"log_odds_chosen": 19.313798904418945,
"log_odds_ratio": -3.367738599990844e-06,
"logits/chosen": -1.150359869003296,
"logits/rejected": -6.183455467224121,
"logps/chosen": -1.4993988275527954,
"logps/rejected": -20.48924446105957,
"loss": 1.4565,
"nll_loss": 1.4564992189407349,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1499398946762085,
"rewards/margins": 1.898984670639038,
"rewards/rejected": -2.048924446105957,
"step": 257
},
{
"epoch": 0.9735849056603774,
"grad_norm": 0.3131140470504761,
"learning_rate": 2.5660377358490567e-05,
"log_odds_chosen": 18.251684188842773,
"log_odds_ratio": -4.395888026920147e-06,
"logits/chosen": -0.7550607919692993,
"logits/rejected": -4.872794151306152,
"logps/chosen": -1.650226354598999,
"logps/rejected": -19.56356430053711,
"loss": 1.5635,
"nll_loss": 1.5635182857513428,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16502264142036438,
"rewards/margins": 1.791333794593811,
"rewards/rejected": -1.956356406211853,
"step": 258
},
{
"epoch": 0.9773584905660377,
"grad_norm": 0.31624630093574524,
"learning_rate": 2.5566037735849058e-05,
"log_odds_chosen": 17.00825309753418,
"log_odds_ratio": -0.0031702774576842785,
"logits/chosen": -1.8281440734863281,
"logits/rejected": -5.850688934326172,
"logps/chosen": -1.6439521312713623,
"logps/rejected": -18.4215087890625,
"loss": 1.4766,
"nll_loss": 1.4762651920318604,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16439521312713623,
"rewards/margins": 1.6777557134628296,
"rewards/rejected": -1.842151165008545,
"step": 259
},
{
"epoch": 0.9811320754716981,
"grad_norm": 0.34498074650764465,
"learning_rate": 2.547169811320755e-05,
"log_odds_chosen": 18.038841247558594,
"log_odds_ratio": -0.0025024032220244408,
"logits/chosen": -1.4132332801818848,
"logits/rejected": -5.574821949005127,
"logps/chosen": -1.6830906867980957,
"logps/rejected": -19.414941787719727,
"loss": 1.4516,
"nll_loss": 1.4513163566589355,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1683090627193451,
"rewards/margins": 1.773185133934021,
"rewards/rejected": -1.9414939880371094,
"step": 260
},
{
"epoch": 0.9849056603773585,
"grad_norm": 0.32891684770584106,
"learning_rate": 2.537735849056604e-05,
"log_odds_chosen": 19.396223068237305,
"log_odds_ratio": -9.61128534981981e-07,
"logits/chosen": -1.6232181787490845,
"logits/rejected": -6.2155914306640625,
"logps/chosen": -1.5656075477600098,
"logps/rejected": -20.644664764404297,
"loss": 1.3423,
"nll_loss": 1.3423418998718262,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1565607637166977,
"rewards/margins": 1.9079058170318604,
"rewards/rejected": -2.0644664764404297,
"step": 261
},
{
"epoch": 0.9886792452830189,
"grad_norm": 0.39039233326911926,
"learning_rate": 2.5283018867924528e-05,
"log_odds_chosen": 15.765559196472168,
"log_odds_ratio": -0.0038383540231734514,
"logits/chosen": -0.5578839778900146,
"logits/rejected": -5.357296466827393,
"logps/chosen": -1.6717973947525024,
"logps/rejected": -17.1871395111084,
"loss": 1.5226,
"nll_loss": 1.522261381149292,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16717973351478577,
"rewards/margins": 1.5515341758728027,
"rewards/rejected": -1.7187139987945557,
"step": 262
},
{
"epoch": 0.9924528301886792,
"grad_norm": 0.2830381989479065,
"learning_rate": 2.518867924528302e-05,
"log_odds_chosen": 21.367420196533203,
"log_odds_ratio": -7.450581485102248e-09,
"logits/chosen": -2.053356409072876,
"logits/rejected": -5.862036228179932,
"logps/chosen": -1.3459280729293823,
"logps/rejected": -22.330286026000977,
"loss": 1.2824,
"nll_loss": 1.2823607921600342,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13459281623363495,
"rewards/margins": 2.098435878753662,
"rewards/rejected": -2.2330286502838135,
"step": 263
},
{
"epoch": 0.9962264150943396,
"grad_norm": 0.34556370973587036,
"learning_rate": 2.509433962264151e-05,
"log_odds_chosen": 17.593475341796875,
"log_odds_ratio": -0.051692862063646317,
"logits/chosen": -1.9133423566818237,
"logits/rejected": -4.535181999206543,
"logps/chosen": -1.6120280027389526,
"logps/rejected": -18.84408950805664,
"loss": 1.5221,
"nll_loss": 1.5169554948806763,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1612028032541275,
"rewards/margins": 1.7232062816619873,
"rewards/rejected": -1.8844091892242432,
"step": 264
},
{
"epoch": 1.0,
"grad_norm": 0.3379838764667511,
"learning_rate": 2.5e-05,
"log_odds_chosen": 19.729963302612305,
"log_odds_ratio": -3.501869741739938e-06,
"logits/chosen": -1.0773916244506836,
"logits/rejected": -6.408069610595703,
"logps/chosen": -1.8509314060211182,
"logps/rejected": -21.368595123291016,
"loss": 1.8248,
"nll_loss": 1.824795126914978,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18509314954280853,
"rewards/margins": 1.9517664909362793,
"rewards/rejected": -2.136859655380249,
"step": 265
},
{
"epoch": 1.0037735849056604,
"grad_norm": 0.32272571325302124,
"learning_rate": 2.4905660377358492e-05,
"log_odds_chosen": 19.660472869873047,
"log_odds_ratio": -5.528574547497556e-06,
"logits/chosen": -2.0058681964874268,
"logits/rejected": -6.02784538269043,
"logps/chosen": -1.462933897972107,
"logps/rejected": -20.805423736572266,
"loss": 1.434,
"nll_loss": 1.43397057056427,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14629340171813965,
"rewards/margins": 1.9342491626739502,
"rewards/rejected": -2.080542802810669,
"step": 266
},
{
"epoch": 1.0075471698113208,
"grad_norm": 0.3159734308719635,
"learning_rate": 2.4811320754716983e-05,
"log_odds_chosen": 18.871841430664062,
"log_odds_ratio": -1.0505355021450669e-06,
"logits/chosen": -1.4568092823028564,
"logits/rejected": -5.773335933685303,
"logps/chosen": -1.4329451322555542,
"logps/rejected": -19.894126892089844,
"loss": 1.5099,
"nll_loss": 1.5098960399627686,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14329451322555542,
"rewards/margins": 1.8461179733276367,
"rewards/rejected": -1.989412546157837,
"step": 267
},
{
"epoch": 1.0113207547169811,
"grad_norm": 0.280902236700058,
"learning_rate": 2.4716981132075474e-05,
"log_odds_chosen": 20.24521255493164,
"log_odds_ratio": -2.2351767370309972e-07,
"logits/chosen": -1.0519423484802246,
"logits/rejected": -6.738117218017578,
"logps/chosen": -1.7242472171783447,
"logps/rejected": -21.655092239379883,
"loss": 1.678,
"nll_loss": 1.67795729637146,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17242471873760223,
"rewards/margins": 1.9930845499038696,
"rewards/rejected": -2.1655092239379883,
"step": 268
},
{
"epoch": 1.0150943396226415,
"grad_norm": 0.6097087264060974,
"learning_rate": 2.4622641509433962e-05,
"log_odds_chosen": 17.3674373626709,
"log_odds_ratio": -0.06029047444462776,
"logits/chosen": -2.0133872032165527,
"logits/rejected": -5.912678241729736,
"logps/chosen": -1.5788557529449463,
"logps/rejected": -18.655887603759766,
"loss": 1.5696,
"nll_loss": 1.5636141300201416,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1578855961561203,
"rewards/margins": 1.7077033519744873,
"rewards/rejected": -1.8655890226364136,
"step": 269
},
{
"epoch": 1.0188679245283019,
"grad_norm": 0.3251427710056305,
"learning_rate": 2.4528301886792453e-05,
"log_odds_chosen": 19.473735809326172,
"log_odds_ratio": -0.05202309042215347,
"logits/chosen": -2.5387845039367676,
"logits/rejected": -6.715412616729736,
"logps/chosen": -1.5151475667953491,
"logps/rejected": -20.52288055419922,
"loss": 1.5736,
"nll_loss": 1.5684118270874023,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15151476860046387,
"rewards/margins": 1.9007731676101685,
"rewards/rejected": -2.052288055419922,
"step": 270
},
{
"epoch": 1.0226415094339623,
"grad_norm": 0.28263339400291443,
"learning_rate": 2.4433962264150944e-05,
"log_odds_chosen": 21.164031982421875,
"log_odds_ratio": -8.56822566674964e-07,
"logits/chosen": -1.4600391387939453,
"logits/rejected": -5.919426441192627,
"logps/chosen": -1.4743690490722656,
"logps/rejected": -22.336339950561523,
"loss": 1.5115,
"nll_loss": 1.51149582862854,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14743690192699432,
"rewards/margins": 2.0861971378326416,
"rewards/rejected": -2.2336339950561523,
"step": 271
},
{
"epoch": 1.0264150943396226,
"grad_norm": 0.28004854917526245,
"learning_rate": 2.4339622641509435e-05,
"log_odds_chosen": 20.22897720336914,
"log_odds_ratio": -1.5720835335741867e-06,
"logits/chosen": -2.687819004058838,
"logits/rejected": -5.694319248199463,
"logps/chosen": -1.3161007165908813,
"logps/rejected": -21.028356552124023,
"loss": 1.1834,
"nll_loss": 1.1834136247634888,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13161008059978485,
"rewards/margins": 1.971225619316101,
"rewards/rejected": -2.1028356552124023,
"step": 272
},
{
"epoch": 1.030188679245283,
"grad_norm": 0.3043549358844757,
"learning_rate": 2.4245283018867926e-05,
"log_odds_chosen": 18.964893341064453,
"log_odds_ratio": -0.04836106672883034,
"logits/chosen": -1.09200918674469,
"logits/rejected": -5.859206199645996,
"logps/chosen": -1.600953221321106,
"logps/rejected": -20.301076889038086,
"loss": 1.5009,
"nll_loss": 1.4960479736328125,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16009533405303955,
"rewards/margins": 1.8700122833251953,
"rewards/rejected": -2.0301077365875244,
"step": 273
},
{
"epoch": 1.0339622641509434,
"grad_norm": 0.29216280579566956,
"learning_rate": 2.4150943396226418e-05,
"log_odds_chosen": 18.339717864990234,
"log_odds_ratio": -0.00336459930986166,
"logits/chosen": -2.02860689163208,
"logits/rejected": -6.949286460876465,
"logps/chosen": -1.6870245933532715,
"logps/rejected": -19.78402328491211,
"loss": 1.5731,
"nll_loss": 1.5727555751800537,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16870248317718506,
"rewards/margins": 1.8096998929977417,
"rewards/rejected": -1.9784023761749268,
"step": 274
},
{
"epoch": 1.0377358490566038,
"grad_norm": 0.2941288352012634,
"learning_rate": 2.405660377358491e-05,
"log_odds_chosen": 22.276226043701172,
"log_odds_ratio": 0.0,
"logits/chosen": -2.0723040103912354,
"logits/rejected": -5.220292568206787,
"logps/chosen": -1.4046821594238281,
"logps/rejected": -23.268173217773438,
"loss": 1.3712,
"nll_loss": 1.371185541152954,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14046822488307953,
"rewards/margins": 2.1863491535186768,
"rewards/rejected": -2.326817512512207,
"step": 275
},
{
"epoch": 1.0415094339622641,
"grad_norm": 0.29630234837532043,
"learning_rate": 2.3962264150943396e-05,
"log_odds_chosen": 22.004188537597656,
"log_odds_ratio": 0.0,
"logits/chosen": -1.7940192222595215,
"logits/rejected": -7.220149040222168,
"logps/chosen": -1.6048816442489624,
"logps/rejected": -23.359508514404297,
"loss": 1.4889,
"nll_loss": 1.4889363050460815,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16048815846443176,
"rewards/margins": 2.1754627227783203,
"rewards/rejected": -2.3359508514404297,
"step": 276
},
{
"epoch": 1.0452830188679245,
"grad_norm": 0.3317105174064636,
"learning_rate": 2.3867924528301887e-05,
"log_odds_chosen": 17.85032081604004,
"log_odds_ratio": -2.0116699488426093e-06,
"logits/chosen": -0.9674760699272156,
"logits/rejected": -5.794704914093018,
"logps/chosen": -1.475320816040039,
"logps/rejected": -18.945966720581055,
"loss": 1.3938,
"nll_loss": 1.3938060998916626,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14753207564353943,
"rewards/margins": 1.7470645904541016,
"rewards/rejected": -1.8945965766906738,
"step": 277
},
{
"epoch": 1.049056603773585,
"grad_norm": 0.30001261830329895,
"learning_rate": 2.377358490566038e-05,
"log_odds_chosen": 20.2429256439209,
"log_odds_ratio": -4.4703490686970326e-08,
"logits/chosen": -0.8269245624542236,
"logits/rejected": -5.202162265777588,
"logps/chosen": -1.4923174381256104,
"logps/rejected": -21.42803192138672,
"loss": 1.3961,
"nll_loss": 1.3961377143859863,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14923176169395447,
"rewards/margins": 1.9935715198516846,
"rewards/rejected": -2.142803192138672,
"step": 278
},
{
"epoch": 1.0528301886792453,
"grad_norm": 0.35864245891571045,
"learning_rate": 2.367924528301887e-05,
"log_odds_chosen": 18.86318588256836,
"log_odds_ratio": -0.06927872449159622,
"logits/chosen": -1.760351538658142,
"logits/rejected": -5.471229553222656,
"logps/chosen": -1.6287931203842163,
"logps/rejected": -20.220352172851562,
"loss": 1.4988,
"nll_loss": 1.4918767213821411,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1628793179988861,
"rewards/margins": 1.8591558933258057,
"rewards/rejected": -2.0220353603363037,
"step": 279
},
{
"epoch": 1.0566037735849056,
"grad_norm": 0.3526778519153595,
"learning_rate": 2.358490566037736e-05,
"log_odds_chosen": 21.42916488647461,
"log_odds_ratio": -1.3187654985813424e-06,
"logits/chosen": -1.282188057899475,
"logits/rejected": -4.82998514175415,
"logps/chosen": -1.2494144439697266,
"logps/rejected": -22.1539306640625,
"loss": 1.2298,
"nll_loss": 1.2297983169555664,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.12494143843650818,
"rewards/margins": 2.090451240539551,
"rewards/rejected": -2.21539306640625,
"step": 280
},
{
"epoch": 1.060377358490566,
"grad_norm": 0.28869158029556274,
"learning_rate": 2.3490566037735852e-05,
"log_odds_chosen": 20.201156616210938,
"log_odds_ratio": -1.1175878711355836e-07,
"logits/chosen": -1.7291052341461182,
"logits/rejected": -7.483429431915283,
"logps/chosen": -1.5107462406158447,
"logps/rejected": -21.40848159790039,
"loss": 1.4772,
"nll_loss": 1.4772439002990723,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15107461810112,
"rewards/margins": 1.9897735118865967,
"rewards/rejected": -2.140848159790039,
"step": 281
},
{
"epoch": 1.0641509433962264,
"grad_norm": 0.31443092226982117,
"learning_rate": 2.339622641509434e-05,
"log_odds_chosen": 19.611129760742188,
"log_odds_ratio": -3.022684541065246e-05,
"logits/chosen": -1.9202224016189575,
"logits/rejected": -6.25022029876709,
"logps/chosen": -1.5923198461532593,
"logps/rejected": -20.92377471923828,
"loss": 1.5393,
"nll_loss": 1.539318561553955,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15923196077346802,
"rewards/margins": 1.933145523071289,
"rewards/rejected": -2.0923774242401123,
"step": 282
},
{
"epoch": 1.0679245283018868,
"grad_norm": 0.33393925428390503,
"learning_rate": 2.330188679245283e-05,
"log_odds_chosen": 19.191036224365234,
"log_odds_ratio": -0.051419854164123535,
"logits/chosen": -1.4811030626296997,
"logits/rejected": -5.777919769287109,
"logps/chosen": -1.5946496725082397,
"logps/rejected": -20.5208797454834,
"loss": 1.6181,
"nll_loss": 1.6130021810531616,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1594649851322174,
"rewards/margins": 1.8926230669021606,
"rewards/rejected": -2.0520882606506348,
"step": 283
},
{
"epoch": 1.0716981132075472,
"grad_norm": 0.2892889082431793,
"learning_rate": 2.320754716981132e-05,
"log_odds_chosen": 20.433334350585938,
"log_odds_ratio": -2.980237638894323e-07,
"logits/chosen": -0.5262770652770996,
"logits/rejected": -5.781856536865234,
"logps/chosen": -1.421867847442627,
"logps/rejected": -21.404203414916992,
"loss": 1.409,
"nll_loss": 1.408989667892456,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14218679070472717,
"rewards/margins": 1.998233675956726,
"rewards/rejected": -2.140420436859131,
"step": 284
},
{
"epoch": 1.0754716981132075,
"grad_norm": 0.3078497350215912,
"learning_rate": 2.3113207547169813e-05,
"log_odds_chosen": 18.807418823242188,
"log_odds_ratio": -0.0475982129573822,
"logits/chosen": -1.0839369297027588,
"logits/rejected": -3.862969398498535,
"logps/chosen": -1.431301474571228,
"logps/rejected": -19.818119049072266,
"loss": 1.5687,
"nll_loss": 1.563894510269165,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.14313015341758728,
"rewards/margins": 1.838681936264038,
"rewards/rejected": -1.9818120002746582,
"step": 285
},
{
"epoch": 1.079245283018868,
"grad_norm": 0.9563864469528198,
"learning_rate": 2.3018867924528304e-05,
"log_odds_chosen": 19.715076446533203,
"log_odds_ratio": -7.935160283523146e-06,
"logits/chosen": -2.1578338146209717,
"logits/rejected": -4.625355243682861,
"logps/chosen": -1.4262417554855347,
"logps/rejected": -20.82624053955078,
"loss": 1.4746,
"nll_loss": 1.4745495319366455,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14262418448925018,
"rewards/margins": 1.9399999380111694,
"rewards/rejected": -2.0826241970062256,
"step": 286
},
{
"epoch": 1.0830188679245283,
"grad_norm": 0.29822468757629395,
"learning_rate": 2.2924528301886795e-05,
"log_odds_chosen": 20.520158767700195,
"log_odds_ratio": -5.96046660916727e-08,
"logits/chosen": -1.959904670715332,
"logits/rejected": -5.203014850616455,
"logps/chosen": -1.3631935119628906,
"logps/rejected": -21.410646438598633,
"loss": 1.4288,
"nll_loss": 1.4287554025650024,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1363193392753601,
"rewards/margins": 2.0047454833984375,
"rewards/rejected": -2.1410648822784424,
"step": 287
},
{
"epoch": 1.0867924528301887,
"grad_norm": 0.31314578652381897,
"learning_rate": 2.2830188679245286e-05,
"log_odds_chosen": 20.801912307739258,
"log_odds_ratio": 0.0,
"logits/chosen": -2.6818971633911133,
"logits/rejected": -7.3150763511657715,
"logps/chosen": -1.596503496170044,
"logps/rejected": -22.121469497680664,
"loss": 1.4044,
"nll_loss": 1.4043606519699097,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15965035557746887,
"rewards/margins": 2.0524966716766357,
"rewards/rejected": -2.2121472358703613,
"step": 288
},
{
"epoch": 1.090566037735849,
"grad_norm": 0.3501436412334442,
"learning_rate": 2.2735849056603774e-05,
"log_odds_chosen": 19.989656448364258,
"log_odds_ratio": -1.385821519761521e-06,
"logits/chosen": -2.4570391178131104,
"logits/rejected": -6.362525463104248,
"logps/chosen": -1.6581473350524902,
"logps/rejected": -21.345970153808594,
"loss": 1.5318,
"nll_loss": 1.53177011013031,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16581472754478455,
"rewards/margins": 1.9687824249267578,
"rewards/rejected": -2.134597063064575,
"step": 289
},
{
"epoch": 1.0943396226415094,
"grad_norm": 0.33854931592941284,
"learning_rate": 2.2641509433962265e-05,
"log_odds_chosen": 19.100032806396484,
"log_odds_ratio": -5.811464234284358e-07,
"logits/chosen": -0.7418297529220581,
"logits/rejected": -4.735363960266113,
"logps/chosen": -1.7482143640518188,
"logps/rejected": -20.59360694885254,
"loss": 1.581,
"nll_loss": 1.5810353755950928,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17482145130634308,
"rewards/margins": 1.884539246559143,
"rewards/rejected": -2.0593605041503906,
"step": 290
},
{
"epoch": 1.0981132075471698,
"grad_norm": 0.3294488489627838,
"learning_rate": 2.2547169811320756e-05,
"log_odds_chosen": 17.665861129760742,
"log_odds_ratio": -0.08739493787288666,
"logits/chosen": -0.7531633377075195,
"logits/rejected": -5.106237888336182,
"logps/chosen": -1.710965871810913,
"logps/rejected": -19.166933059692383,
"loss": 1.5588,
"nll_loss": 1.5500967502593994,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1710965931415558,
"rewards/margins": 1.7455967664718628,
"rewards/rejected": -1.9166933298110962,
"step": 291
},
{
"epoch": 1.1018867924528302,
"grad_norm": 0.3396783769130707,
"learning_rate": 2.2452830188679247e-05,
"log_odds_chosen": 18.452682495117188,
"log_odds_ratio": -0.07296016067266464,
"logits/chosen": -0.9494550228118896,
"logits/rejected": -4.494377613067627,
"logps/chosen": -1.316884994506836,
"logps/rejected": -19.361968994140625,
"loss": 1.1822,
"nll_loss": 1.1749296188354492,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.13168850541114807,
"rewards/margins": 1.8045084476470947,
"rewards/rejected": -1.9361971616744995,
"step": 292
},
{
"epoch": 1.1056603773584905,
"grad_norm": 0.3204915523529053,
"learning_rate": 2.2358490566037738e-05,
"log_odds_chosen": 21.57263946533203,
"log_odds_ratio": -3.62108698936936e-06,
"logits/chosen": -0.9738727807998657,
"logits/rejected": -5.303981304168701,
"logps/chosen": -1.5682296752929688,
"logps/rejected": -22.808374404907227,
"loss": 1.462,
"nll_loss": 1.462032437324524,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15682296454906464,
"rewards/margins": 2.1240146160125732,
"rewards/rejected": -2.280837297439575,
"step": 293
},
{
"epoch": 1.109433962264151,
"grad_norm": 0.3384822607040405,
"learning_rate": 2.226415094339623e-05,
"log_odds_chosen": 18.951013565063477,
"log_odds_ratio": -5.208150469115935e-06,
"logits/chosen": -1.3006513118743896,
"logits/rejected": -4.634444236755371,
"logps/chosen": -1.5552959442138672,
"logps/rejected": -20.169574737548828,
"loss": 1.5539,
"nll_loss": 1.5539031028747559,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15552960336208344,
"rewards/margins": 1.8614277839660645,
"rewards/rejected": -2.0169572830200195,
"step": 294
},
{
"epoch": 1.1132075471698113,
"grad_norm": 1.1054335832595825,
"learning_rate": 2.216981132075472e-05,
"log_odds_chosen": 18.913715362548828,
"log_odds_ratio": -4.693871460403898e-07,
"logits/chosen": -0.9914897680282593,
"logits/rejected": -5.061755180358887,
"logps/chosen": -1.4729012250900269,
"logps/rejected": -20.089599609375,
"loss": 1.3732,
"nll_loss": 1.373215913772583,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14729014039039612,
"rewards/margins": 1.8616697788238525,
"rewards/rejected": -2.008960247039795,
"step": 295
},
{
"epoch": 1.1169811320754717,
"grad_norm": 0.41514357924461365,
"learning_rate": 2.2075471698113208e-05,
"log_odds_chosen": 20.548986434936523,
"log_odds_ratio": -5.2154071283894154e-08,
"logits/chosen": -1.4249317646026611,
"logits/rejected": -5.687395095825195,
"logps/chosen": -1.5955243110656738,
"logps/rejected": -21.793594360351562,
"loss": 1.4593,
"nll_loss": 1.4593462944030762,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1595524251461029,
"rewards/margins": 2.0198073387145996,
"rewards/rejected": -2.1793594360351562,
"step": 296
},
{
"epoch": 1.120754716981132,
"grad_norm": 0.33038532733917236,
"learning_rate": 2.19811320754717e-05,
"log_odds_chosen": 19.50081443786621,
"log_odds_ratio": -1.6255449736490846e-05,
"logits/chosen": -2.909379482269287,
"logits/rejected": -7.22389030456543,
"logps/chosen": -1.4944963455200195,
"logps/rejected": -20.643218994140625,
"loss": 1.3455,
"nll_loss": 1.345502257347107,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1494496464729309,
"rewards/margins": 1.914872169494629,
"rewards/rejected": -2.064321756362915,
"step": 297
},
{
"epoch": 1.1245283018867924,
"grad_norm": 0.29738759994506836,
"learning_rate": 2.188679245283019e-05,
"log_odds_chosen": 20.67082405090332,
"log_odds_ratio": -0.0003345193399582058,
"logits/chosen": -1.593409776687622,
"logits/rejected": -5.576838493347168,
"logps/chosen": -1.746903657913208,
"logps/rejected": -22.0710506439209,
"loss": 1.3993,
"nll_loss": 1.3992958068847656,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1746903657913208,
"rewards/margins": 2.032414674758911,
"rewards/rejected": -2.2071051597595215,
"step": 298
},
{
"epoch": 1.1283018867924528,
"grad_norm": 0.4151991605758667,
"learning_rate": 2.179245283018868e-05,
"log_odds_chosen": 17.427101135253906,
"log_odds_ratio": -0.11041603237390518,
"logits/chosen": -0.4509122669696808,
"logits/rejected": -4.386507034301758,
"logps/chosen": -1.5307447910308838,
"logps/rejected": -18.620765686035156,
"loss": 1.4554,
"nll_loss": 1.4443206787109375,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1530744880437851,
"rewards/margins": 1.7090023756027222,
"rewards/rejected": -1.862076759338379,
"step": 299
},
{
"epoch": 1.1320754716981132,
"grad_norm": 0.327581524848938,
"learning_rate": 2.1698113207547172e-05,
"log_odds_chosen": 17.98784065246582,
"log_odds_ratio": -4.783316398970783e-06,
"logits/chosen": -1.5300605297088623,
"logits/rejected": -6.33521842956543,
"logps/chosen": -1.5090255737304688,
"logps/rejected": -19.206493377685547,
"loss": 1.401,
"nll_loss": 1.4010220766067505,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15090256929397583,
"rewards/margins": 1.7697467803955078,
"rewards/rejected": -1.9206492900848389,
"step": 300
},
{
"epoch": 1.1358490566037736,
"grad_norm": 0.33648499846458435,
"learning_rate": 2.1603773584905663e-05,
"log_odds_chosen": 20.12273597717285,
"log_odds_ratio": -5.960480393696344e-07,
"logits/chosen": -0.9786512851715088,
"logits/rejected": -5.974287986755371,
"logps/chosen": -1.504534363746643,
"logps/rejected": -21.347797393798828,
"loss": 1.4384,
"nll_loss": 1.4384446144104004,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15045343339443207,
"rewards/margins": 1.9843263626098633,
"rewards/rejected": -2.134779930114746,
"step": 301
},
{
"epoch": 1.139622641509434,
"grad_norm": 0.4433332681655884,
"learning_rate": 2.1509433962264154e-05,
"log_odds_chosen": 18.096084594726562,
"log_odds_ratio": -0.07683061808347702,
"logits/chosen": -1.3693748712539673,
"logits/rejected": -5.253786087036133,
"logps/chosen": -1.7420886754989624,
"logps/rejected": -19.588783264160156,
"loss": 1.4441,
"nll_loss": 1.4364526271820068,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1742088496685028,
"rewards/margins": 1.7846695184707642,
"rewards/rejected": -1.9588782787322998,
"step": 302
},
{
"epoch": 1.1433962264150943,
"grad_norm": 0.2902994155883789,
"learning_rate": 2.1415094339622642e-05,
"log_odds_chosen": 22.471710205078125,
"log_odds_ratio": -1.1175880842984043e-07,
"logits/chosen": -1.3224496841430664,
"logits/rejected": -6.421805381774902,
"logps/chosen": -1.4149394035339355,
"logps/rejected": -23.561216354370117,
"loss": 1.3612,
"nll_loss": 1.36124587059021,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14149394631385803,
"rewards/margins": 2.214627742767334,
"rewards/rejected": -2.356121778488159,
"step": 303
},
{
"epoch": 1.1471698113207547,
"grad_norm": 0.35126540064811707,
"learning_rate": 2.1320754716981133e-05,
"log_odds_chosen": 20.562442779541016,
"log_odds_ratio": -9.716237400425598e-06,
"logits/chosen": -0.9620351791381836,
"logits/rejected": -6.237518787384033,
"logps/chosen": -1.6788274049758911,
"logps/rejected": -21.986478805541992,
"loss": 1.4306,
"nll_loss": 1.4306453466415405,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1678827404975891,
"rewards/margins": 2.0307650566101074,
"rewards/rejected": -2.1986477375030518,
"step": 304
},
{
"epoch": 1.150943396226415,
"grad_norm": 0.3252897560596466,
"learning_rate": 2.1226415094339624e-05,
"log_odds_chosen": 20.130619049072266,
"log_odds_ratio": -0.00018702806846704334,
"logits/chosen": -1.5204761028289795,
"logits/rejected": -4.985358238220215,
"logps/chosen": -1.533634901046753,
"logps/rejected": -21.230838775634766,
"loss": 1.6996,
"nll_loss": 1.6996041536331177,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15336349606513977,
"rewards/margins": 1.9697201251983643,
"rewards/rejected": -2.1230835914611816,
"step": 305
},
{
"epoch": 1.1547169811320754,
"grad_norm": 0.4213809072971344,
"learning_rate": 2.1132075471698115e-05,
"log_odds_chosen": 22.56460952758789,
"log_odds_ratio": -1.7881419012155675e-07,
"logits/chosen": -1.123805046081543,
"logits/rejected": -4.740581035614014,
"logps/chosen": -1.4182591438293457,
"logps/rejected": -23.586502075195312,
"loss": 1.3603,
"nll_loss": 1.360290288925171,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14182589948177338,
"rewards/margins": 2.2168242931365967,
"rewards/rejected": -2.3586502075195312,
"step": 306
},
{
"epoch": 1.1584905660377358,
"grad_norm": 0.3281170725822449,
"learning_rate": 2.1037735849056606e-05,
"log_odds_chosen": 19.17583656311035,
"log_odds_ratio": -2.123439571732888e-06,
"logits/chosen": -0.20950892567634583,
"logits/rejected": -4.772949695587158,
"logps/chosen": -1.5309257507324219,
"logps/rejected": -20.371936798095703,
"loss": 1.4766,
"nll_loss": 1.476578950881958,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15309256315231323,
"rewards/margins": 1.884101390838623,
"rewards/rejected": -2.037193775177002,
"step": 307
},
{
"epoch": 1.1622641509433962,
"grad_norm": 0.3546697497367859,
"learning_rate": 2.0943396226415098e-05,
"log_odds_chosen": 23.225772857666016,
"log_odds_ratio": -7.450581485102248e-09,
"logits/chosen": -0.4746635854244232,
"logits/rejected": -2.8694400787353516,
"logps/chosen": -1.2704362869262695,
"logps/rejected": -24.101713180541992,
"loss": 1.2967,
"nll_loss": 1.2966803312301636,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.12704363465309143,
"rewards/margins": 2.283127546310425,
"rewards/rejected": -2.4101712703704834,
"step": 308
},
{
"epoch": 1.1660377358490566,
"grad_norm": 1.0174925327301025,
"learning_rate": 2.0849056603773585e-05,
"log_odds_chosen": 19.03729248046875,
"log_odds_ratio": -0.10631541162729263,
"logits/chosen": -2.604809522628784,
"logits/rejected": -5.2874040603637695,
"logps/chosen": -2.420698642730713,
"logps/rejected": -21.103818893432617,
"loss": 1.6129,
"nll_loss": 1.6022355556488037,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.24206990003585815,
"rewards/margins": 1.8683120012283325,
"rewards/rejected": -2.110382080078125,
"step": 309
},
{
"epoch": 1.169811320754717,
"grad_norm": 0.29557734727859497,
"learning_rate": 2.0754716981132076e-05,
"log_odds_chosen": 16.802288055419922,
"log_odds_ratio": -0.02574881538748741,
"logits/chosen": -1.028839111328125,
"logits/rejected": -4.576976776123047,
"logps/chosen": -1.665901780128479,
"logps/rejected": -18.113601684570312,
"loss": 1.5644,
"nll_loss": 1.561858057975769,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16659018397331238,
"rewards/margins": 1.6447699069976807,
"rewards/rejected": -1.8113601207733154,
"step": 310
},
{
"epoch": 1.1735849056603773,
"grad_norm": 0.3357307016849518,
"learning_rate": 2.0660377358490567e-05,
"log_odds_chosen": 18.195541381835938,
"log_odds_ratio": -0.05465611815452576,
"logits/chosen": -2.216845989227295,
"logits/rejected": -5.583156108856201,
"logps/chosen": -1.4862521886825562,
"logps/rejected": -19.355636596679688,
"loss": 1.4894,
"nll_loss": 1.4839730262756348,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1486252397298813,
"rewards/margins": 1.7869385480880737,
"rewards/rejected": -1.9355638027191162,
"step": 311
},
{
"epoch": 1.1773584905660377,
"grad_norm": 0.43319424986839294,
"learning_rate": 2.056603773584906e-05,
"log_odds_chosen": 20.049089431762695,
"log_odds_ratio": -8.195664804588887e-07,
"logits/chosen": -1.8693492412567139,
"logits/rejected": -6.614497184753418,
"logps/chosen": -1.4651036262512207,
"logps/rejected": -21.203433990478516,
"loss": 1.3849,
"nll_loss": 1.384904384613037,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14651036262512207,
"rewards/margins": 1.9738330841064453,
"rewards/rejected": -2.1203434467315674,
"step": 312
},
{
"epoch": 1.181132075471698,
"grad_norm": 0.3610400855541229,
"learning_rate": 2.047169811320755e-05,
"log_odds_chosen": 18.725547790527344,
"log_odds_ratio": -0.0023514274507761,
"logits/chosen": -0.6479529738426208,
"logits/rejected": -3.6973776817321777,
"logps/chosen": -1.5272626876831055,
"logps/rejected": -19.948326110839844,
"loss": 1.4189,
"nll_loss": 1.4186826944351196,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15272627770900726,
"rewards/margins": 1.8421063423156738,
"rewards/rejected": -1.9948326349258423,
"step": 313
},
{
"epoch": 1.1849056603773584,
"grad_norm": 0.3513402044773102,
"learning_rate": 2.037735849056604e-05,
"log_odds_chosen": 17.298198699951172,
"log_odds_ratio": -0.002320481464266777,
"logits/chosen": -1.4968432188034058,
"logits/rejected": -5.039670944213867,
"logps/chosen": -1.4913229942321777,
"logps/rejected": -18.48564910888672,
"loss": 1.4084,
"nll_loss": 1.4081535339355469,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14913231134414673,
"rewards/margins": 1.6994324922561646,
"rewards/rejected": -1.848564863204956,
"step": 314
},
{
"epoch": 1.1886792452830188,
"grad_norm": 0.3477972447872162,
"learning_rate": 2.0283018867924532e-05,
"log_odds_chosen": 18.793106079101562,
"log_odds_ratio": -8.6095547885634e-05,
"logits/chosen": -2.6079301834106445,
"logits/rejected": -5.591516494750977,
"logps/chosen": -1.5191848278045654,
"logps/rejected": -19.985429763793945,
"loss": 1.5223,
"nll_loss": 1.5223308801651,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1519184708595276,
"rewards/margins": 1.8466243743896484,
"rewards/rejected": -1.9985430240631104,
"step": 315
},
{
"epoch": 1.1924528301886792,
"grad_norm": 0.35549482703208923,
"learning_rate": 2.018867924528302e-05,
"log_odds_chosen": 18.749046325683594,
"log_odds_ratio": -5.386953034758335e-06,
"logits/chosen": -0.9891374707221985,
"logits/rejected": -5.465826988220215,
"logps/chosen": -1.3818237781524658,
"logps/rejected": -19.735633850097656,
"loss": 1.4975,
"nll_loss": 1.4974541664123535,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1381823718547821,
"rewards/margins": 1.835381031036377,
"rewards/rejected": -1.973563313484192,
"step": 316
},
{
"epoch": 1.1962264150943396,
"grad_norm": 0.34496262669563293,
"learning_rate": 2.009433962264151e-05,
"log_odds_chosen": 15.491250038146973,
"log_odds_ratio": -0.20011255145072937,
"logits/chosen": -1.0274806022644043,
"logits/rejected": -5.168841361999512,
"logps/chosen": -1.7780976295471191,
"logps/rejected": -17.050220489501953,
"loss": 1.6287,
"nll_loss": 1.6086933612823486,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17780977487564087,
"rewards/margins": 1.527212142944336,
"rewards/rejected": -1.7050219774246216,
"step": 317
},
{
"epoch": 1.2,
"grad_norm": 0.32769209146499634,
"learning_rate": 2e-05,
"log_odds_chosen": 16.972900390625,
"log_odds_ratio": -0.00024161383043974638,
"logits/chosen": -0.6759935617446899,
"logits/rejected": -5.338271141052246,
"logps/chosen": -1.6053509712219238,
"logps/rejected": -18.34259796142578,
"loss": 1.4908,
"nll_loss": 1.4907457828521729,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16053511202335358,
"rewards/margins": 1.6737247705459595,
"rewards/rejected": -1.8342599868774414,
"step": 318
},
{
"epoch": 1.2037735849056603,
"grad_norm": 0.33330532908439636,
"learning_rate": 1.9905660377358493e-05,
"log_odds_chosen": 18.765106201171875,
"log_odds_ratio": -0.007183433044701815,
"logits/chosen": -3.5232436656951904,
"logits/rejected": -7.270813465118408,
"logps/chosen": -1.494720697402954,
"logps/rejected": -19.954343795776367,
"loss": 1.4191,
"nll_loss": 1.4183340072631836,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14947207272052765,
"rewards/margins": 1.8459622859954834,
"rewards/rejected": -1.9954345226287842,
"step": 319
},
{
"epoch": 1.2075471698113207,
"grad_norm": 0.35828787088394165,
"learning_rate": 1.9811320754716984e-05,
"log_odds_chosen": 15.424698829650879,
"log_odds_ratio": -0.006778358481824398,
"logits/chosen": -2.202425956726074,
"logits/rejected": -7.088006973266602,
"logps/chosen": -1.7607218027114868,
"logps/rejected": -16.957534790039062,
"loss": 1.6174,
"nll_loss": 1.616753339767456,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17607218027114868,
"rewards/margins": 1.5196813344955444,
"rewards/rejected": -1.6957534551620483,
"step": 320
},
{
"epoch": 1.211320754716981,
"grad_norm": 0.34036463499069214,
"learning_rate": 1.9716981132075475e-05,
"log_odds_chosen": 18.791336059570312,
"log_odds_ratio": -3.852009740512585e-06,
"logits/chosen": -1.9099091291427612,
"logits/rejected": -7.520392417907715,
"logps/chosen": -1.561924934387207,
"logps/rejected": -19.94056510925293,
"loss": 1.471,
"nll_loss": 1.4709736108779907,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15619248151779175,
"rewards/margins": 1.8378640413284302,
"rewards/rejected": -1.9940567016601562,
"step": 321
},
{
"epoch": 1.2150943396226415,
"grad_norm": 0.3238532841205597,
"learning_rate": 1.9622641509433966e-05,
"log_odds_chosen": 17.8562068939209,
"log_odds_ratio": -8.6430118244607e-06,
"logits/chosen": -0.658406138420105,
"logits/rejected": -4.410658836364746,
"logps/chosen": -1.2828052043914795,
"logps/rejected": -18.753049850463867,
"loss": 1.3437,
"nll_loss": 1.3436634540557861,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.12828052043914795,
"rewards/margins": 1.7470245361328125,
"rewards/rejected": -1.875304937362671,
"step": 322
},
{
"epoch": 1.2188679245283018,
"grad_norm": 0.3303399682044983,
"learning_rate": 1.9528301886792454e-05,
"log_odds_chosen": 14.796485900878906,
"log_odds_ratio": -0.055772680789232254,
"logits/chosen": -1.0683139562606812,
"logits/rejected": -7.066629409790039,
"logps/chosen": -1.8164430856704712,
"logps/rejected": -16.421266555786133,
"loss": 1.613,
"nll_loss": 1.6073920726776123,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.18164430558681488,
"rewards/margins": 1.4604823589324951,
"rewards/rejected": -1.6421265602111816,
"step": 323
},
{
"epoch": 1.2226415094339622,
"grad_norm": 0.35494038462638855,
"learning_rate": 1.9433962264150945e-05,
"log_odds_chosen": 17.69695472717285,
"log_odds_ratio": -0.005540680605918169,
"logits/chosen": -1.0188480615615845,
"logits/rejected": -4.988126277923584,
"logps/chosen": -1.4110499620437622,
"logps/rejected": -18.787839889526367,
"loss": 1.3428,
"nll_loss": 1.3422446250915527,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14110499620437622,
"rewards/margins": 1.7376790046691895,
"rewards/rejected": -1.878783941268921,
"step": 324
},
{
"epoch": 1.2264150943396226,
"grad_norm": 0.35788360238075256,
"learning_rate": 1.9339622641509436e-05,
"log_odds_chosen": 14.982892990112305,
"log_odds_ratio": -0.049817949533462524,
"logits/chosen": -1.3258823156356812,
"logits/rejected": -4.249111652374268,
"logps/chosen": -1.2381442785263062,
"logps/rejected": -15.715821266174316,
"loss": 1.2877,
"nll_loss": 1.2827649116516113,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.12381443381309509,
"rewards/margins": 1.447767734527588,
"rewards/rejected": -1.5715820789337158,
"step": 325
},
{
"epoch": 1.230188679245283,
"grad_norm": 0.39687126874923706,
"learning_rate": 1.9245283018867927e-05,
"log_odds_chosen": 16.164493560791016,
"log_odds_ratio": -0.054912034422159195,
"logits/chosen": -1.1859130859375,
"logits/rejected": -4.106606960296631,
"logps/chosen": -1.3010371923446655,
"logps/rejected": -16.917049407958984,
"loss": 1.4605,
"nll_loss": 1.4549907445907593,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1301037073135376,
"rewards/margins": 1.561601161956787,
"rewards/rejected": -1.6917049884796143,
"step": 326
},
{
"epoch": 1.2339622641509433,
"grad_norm": 0.3798399865627289,
"learning_rate": 1.9150943396226418e-05,
"log_odds_chosen": 15.7042875289917,
"log_odds_ratio": -0.00472813518717885,
"logits/chosen": 0.0541728138923645,
"logits/rejected": -4.224884033203125,
"logps/chosen": -1.7171664237976074,
"logps/rejected": -17.207719802856445,
"loss": 1.5367,
"nll_loss": 1.5362149477005005,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1717166304588318,
"rewards/margins": 1.5490553379058838,
"rewards/rejected": -1.7207720279693604,
"step": 327
},
{
"epoch": 1.2377358490566037,
"grad_norm": 0.3624114990234375,
"learning_rate": 1.905660377358491e-05,
"log_odds_chosen": 16.288440704345703,
"log_odds_ratio": -0.010537970811128616,
"logits/chosen": -0.5711084008216858,
"logits/rejected": -4.729043006896973,
"logps/chosen": -1.5029557943344116,
"logps/rejected": -17.498348236083984,
"loss": 1.3411,
"nll_loss": 1.340043544769287,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15029558539390564,
"rewards/margins": 1.5995393991470337,
"rewards/rejected": -1.7498348951339722,
"step": 328
},
{
"epoch": 1.241509433962264,
"grad_norm": 0.3725604712963104,
"learning_rate": 1.8962264150943397e-05,
"log_odds_chosen": 15.354927062988281,
"log_odds_ratio": -0.02149188332259655,
"logits/chosen": -1.4398235082626343,
"logits/rejected": -4.620980739593506,
"logps/chosen": -1.4068002700805664,
"logps/rejected": -16.40656280517578,
"loss": 1.2986,
"nll_loss": 1.2964096069335938,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14068001508712769,
"rewards/margins": 1.4999765157699585,
"rewards/rejected": -1.640656590461731,
"step": 329
},
{
"epoch": 1.2452830188679245,
"grad_norm": 0.370463490486145,
"learning_rate": 1.8867924528301888e-05,
"log_odds_chosen": 14.876998901367188,
"log_odds_ratio": -0.05089250206947327,
"logits/chosen": -1.1731348037719727,
"logits/rejected": -4.278902530670166,
"logps/chosen": -1.3039909601211548,
"logps/rejected": -15.608766555786133,
"loss": 1.4343,
"nll_loss": 1.4292408227920532,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13039910793304443,
"rewards/margins": 1.4304776191711426,
"rewards/rejected": -1.5608766078948975,
"step": 330
},
{
"epoch": 1.2490566037735849,
"grad_norm": 0.3387150466442108,
"learning_rate": 1.877358490566038e-05,
"log_odds_chosen": 15.262411117553711,
"log_odds_ratio": -6.615633174078539e-05,
"logits/chosen": -2.305337905883789,
"logits/rejected": -5.992312431335449,
"logps/chosen": -1.5741474628448486,
"logps/rejected": -16.528202056884766,
"loss": 1.4408,
"nll_loss": 1.440836787223816,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1574147343635559,
"rewards/margins": 1.4954053163528442,
"rewards/rejected": -1.652820110321045,
"step": 331
},
{
"epoch": 1.2528301886792452,
"grad_norm": 0.37555763125419617,
"learning_rate": 1.8679245283018867e-05,
"log_odds_chosen": 18.441608428955078,
"log_odds_ratio": -1.5244633686961606e-05,
"logits/chosen": -2.3576619625091553,
"logits/rejected": -4.850790500640869,
"logps/chosen": -1.4474366903305054,
"logps/rejected": -19.480161666870117,
"loss": 1.397,
"nll_loss": 1.3970084190368652,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1447436660528183,
"rewards/margins": 1.8032724857330322,
"rewards/rejected": -1.9480161666870117,
"step": 332
},
{
"epoch": 1.2566037735849056,
"grad_norm": 0.37432780861854553,
"learning_rate": 1.8584905660377358e-05,
"log_odds_chosen": 16.72200584411621,
"log_odds_ratio": -2.3494829292758368e-05,
"logits/chosen": -0.13491854071617126,
"logits/rejected": -3.951087474822998,
"logps/chosen": -1.4685932397842407,
"logps/rejected": -17.866477966308594,
"loss": 1.5172,
"nll_loss": 1.5171722173690796,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1468593329191208,
"rewards/margins": 1.6397886276245117,
"rewards/rejected": -1.7866477966308594,
"step": 333
},
{
"epoch": 1.260377358490566,
"grad_norm": 0.3993586301803589,
"learning_rate": 1.849056603773585e-05,
"log_odds_chosen": 15.771561622619629,
"log_odds_ratio": -3.5672157537192106e-05,
"logits/chosen": -1.266805648803711,
"logits/rejected": -4.964008808135986,
"logps/chosen": -1.478631854057312,
"logps/rejected": -16.92867088317871,
"loss": 1.5646,
"nll_loss": 1.5645837783813477,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14786317944526672,
"rewards/margins": 1.545003890991211,
"rewards/rejected": -1.6928670406341553,
"step": 334
},
{
"epoch": 1.2641509433962264,
"grad_norm": 0.34949737787246704,
"learning_rate": 1.839622641509434e-05,
"log_odds_chosen": 14.615917205810547,
"log_odds_ratio": -0.18080249428749084,
"logits/chosen": -2.7243032455444336,
"logits/rejected": -6.512376308441162,
"logps/chosen": -1.9120750427246094,
"logps/rejected": -16.35055160522461,
"loss": 1.6374,
"nll_loss": 1.6193275451660156,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.19120751321315765,
"rewards/margins": 1.44384765625,
"rewards/rejected": -1.6350551843643188,
"step": 335
},
{
"epoch": 1.2679245283018867,
"grad_norm": 0.3900119960308075,
"learning_rate": 1.830188679245283e-05,
"log_odds_chosen": 15.152379035949707,
"log_odds_ratio": -0.07246612012386322,
"logits/chosen": -2.0148720741271973,
"logits/rejected": -5.148480415344238,
"logps/chosen": -1.7054778337478638,
"logps/rejected": -16.5804443359375,
"loss": 1.4763,
"nll_loss": 1.4690625667572021,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.17054779827594757,
"rewards/margins": 1.4874964952468872,
"rewards/rejected": -1.6580443382263184,
"step": 336
},
{
"epoch": 1.271698113207547,
"grad_norm": 0.3045012354850769,
"learning_rate": 1.820754716981132e-05,
"log_odds_chosen": 14.322192192077637,
"log_odds_ratio": -0.0033934221137315035,
"logits/chosen": -1.8886581659317017,
"logits/rejected": -7.217465877532959,
"logps/chosen": -1.5194714069366455,
"logps/rejected": -15.524861335754395,
"loss": 1.6423,
"nll_loss": 1.6419386863708496,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15194714069366455,
"rewards/margins": 1.4005389213562012,
"rewards/rejected": -1.5524861812591553,
"step": 337
},
{
"epoch": 1.2754716981132075,
"grad_norm": 0.27202707529067993,
"learning_rate": 1.811320754716981e-05,
"log_odds_chosen": 17.09416389465332,
"log_odds_ratio": -0.01481825951486826,
"logits/chosen": -2.577331066131592,
"logits/rejected": -6.886340141296387,
"logps/chosen": -1.345436930656433,
"logps/rejected": -17.985553741455078,
"loss": 1.3331,
"nll_loss": 1.3316359519958496,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13454370200634003,
"rewards/margins": 1.6640119552612305,
"rewards/rejected": -1.7985554933547974,
"step": 338
},
{
"epoch": 1.2792452830188679,
"grad_norm": 0.35674479603767395,
"learning_rate": 1.80188679245283e-05,
"log_odds_chosen": 17.09935760498047,
"log_odds_ratio": -0.04254509136080742,
"logits/chosen": -1.4332383871078491,
"logits/rejected": -6.181375503540039,
"logps/chosen": -1.9294726848602295,
"logps/rejected": -18.842205047607422,
"loss": 1.6018,
"nll_loss": 1.597574234008789,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.19294726848602295,
"rewards/margins": 1.6912733316421509,
"rewards/rejected": -1.8842206001281738,
"step": 339
},
{
"epoch": 1.2830188679245282,
"grad_norm": 0.33350715041160583,
"learning_rate": 1.7924528301886792e-05,
"log_odds_chosen": 17.224140167236328,
"log_odds_ratio": -2.8834060685767327e-06,
"logits/chosen": -1.3092684745788574,
"logits/rejected": -7.616753578186035,
"logps/chosen": -1.7400907278060913,
"logps/rejected": -18.713537216186523,
"loss": 1.5776,
"nll_loss": 1.5776221752166748,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1740090698003769,
"rewards/margins": 1.6973445415496826,
"rewards/rejected": -1.8713536262512207,
"step": 340
},
{
"epoch": 1.2867924528301886,
"grad_norm": 0.31082576513290405,
"learning_rate": 1.7830188679245283e-05,
"log_odds_chosen": 16.050865173339844,
"log_odds_ratio": -0.00012872874503955245,
"logits/chosen": -1.133547067642212,
"logits/rejected": -5.591988563537598,
"logps/chosen": -1.5776842832565308,
"logps/rejected": -17.315185546875,
"loss": 1.3479,
"nll_loss": 1.347907543182373,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15776842832565308,
"rewards/margins": 1.5737502574920654,
"rewards/rejected": -1.7315186262130737,
"step": 341
},
{
"epoch": 1.290566037735849,
"grad_norm": 0.37589412927627563,
"learning_rate": 1.7735849056603774e-05,
"log_odds_chosen": 15.594026565551758,
"log_odds_ratio": -0.08674325048923492,
"logits/chosen": -1.7124016284942627,
"logits/rejected": -4.819745063781738,
"logps/chosen": -1.672520637512207,
"logps/rejected": -17.02138328552246,
"loss": 1.4996,
"nll_loss": 1.4909459352493286,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1672520637512207,
"rewards/margins": 1.534886360168457,
"rewards/rejected": -1.7021384239196777,
"step": 342
},
{
"epoch": 1.2943396226415094,
"grad_norm": 0.3507995307445526,
"learning_rate": 1.7641509433962265e-05,
"log_odds_chosen": 16.04661750793457,
"log_odds_ratio": -0.007493563462048769,
"logits/chosen": -1.3305596113204956,
"logits/rejected": -5.7259111404418945,
"logps/chosen": -1.474043846130371,
"logps/rejected": -17.168848037719727,
"loss": 1.4017,
"nll_loss": 1.4009082317352295,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14740438759326935,
"rewards/margins": 1.569480538368225,
"rewards/rejected": -1.7168848514556885,
"step": 343
},
{
"epoch": 1.2981132075471697,
"grad_norm": 0.37658241391181946,
"learning_rate": 1.7547169811320753e-05,
"log_odds_chosen": 17.842830657958984,
"log_odds_ratio": -1.3475509149429854e-05,
"logits/chosen": -2.2428817749023438,
"logits/rejected": -5.751044273376465,
"logps/chosen": -1.5199084281921387,
"logps/rejected": -19.044391632080078,
"loss": 1.4837,
"nll_loss": 1.4836554527282715,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1519908457994461,
"rewards/margins": 1.752448320388794,
"rewards/rejected": -1.9044389724731445,
"step": 344
},
{
"epoch": 1.3018867924528301,
"grad_norm": 0.35167789459228516,
"learning_rate": 1.7452830188679244e-05,
"log_odds_chosen": 15.047775268554688,
"log_odds_ratio": -0.008871389552950859,
"logits/chosen": -1.8355504274368286,
"logits/rejected": -6.161797046661377,
"logps/chosen": -1.6795439720153809,
"logps/rejected": -16.46615982055664,
"loss": 1.488,
"nll_loss": 1.4871084690093994,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16795440018177032,
"rewards/margins": 1.4786615371704102,
"rewards/rejected": -1.646615982055664,
"step": 345
},
{
"epoch": 1.3056603773584905,
"grad_norm": 0.35509005188941956,
"learning_rate": 1.7358490566037735e-05,
"log_odds_chosen": 13.43126106262207,
"log_odds_ratio": -0.05844378471374512,
"logits/chosen": -0.840483546257019,
"logits/rejected": -5.20952844619751,
"logps/chosen": -1.282634973526001,
"logps/rejected": -14.154840469360352,
"loss": 1.3724,
"nll_loss": 1.3665363788604736,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.12826348841190338,
"rewards/margins": 1.2872204780578613,
"rewards/rejected": -1.4154839515686035,
"step": 346
},
{
"epoch": 1.3094339622641509,
"grad_norm": 0.3561452031135559,
"learning_rate": 1.7264150943396226e-05,
"log_odds_chosen": 18.20964813232422,
"log_odds_ratio": -4.6193684966056026e-07,
"logits/chosen": -2.4437973499298096,
"logits/rejected": -6.393451690673828,
"logps/chosen": -1.564558982849121,
"logps/rejected": -19.447908401489258,
"loss": 1.6323,
"nll_loss": 1.6323482990264893,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1564558893442154,
"rewards/margins": 1.788334846496582,
"rewards/rejected": -1.9447907209396362,
"step": 347
},
{
"epoch": 1.3132075471698113,
"grad_norm": 0.2806137204170227,
"learning_rate": 1.7169811320754717e-05,
"log_odds_chosen": 18.3279972076416,
"log_odds_ratio": -1.1324946171953343e-06,
"logits/chosen": -1.2502247095108032,
"logits/rejected": -6.719517230987549,
"logps/chosen": -1.4687795639038086,
"logps/rejected": -19.442962646484375,
"loss": 1.2923,
"nll_loss": 1.2922688722610474,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1468779593706131,
"rewards/margins": 1.7974183559417725,
"rewards/rejected": -1.9442962408065796,
"step": 348
},
{
"epoch": 1.3169811320754716,
"grad_norm": 0.3450721502304077,
"learning_rate": 1.707547169811321e-05,
"log_odds_chosen": 14.217653274536133,
"log_odds_ratio": -0.10679548978805542,
"logits/chosen": -1.2525098323822021,
"logits/rejected": -4.723179340362549,
"logps/chosen": -1.651379942893982,
"logps/rejected": -15.504240036010742,
"loss": 1.5352,
"nll_loss": 1.5244849920272827,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16513800621032715,
"rewards/margins": 1.3852860927581787,
"rewards/rejected": -1.5504240989685059,
"step": 349
},
{
"epoch": 1.320754716981132,
"grad_norm": 0.33943700790405273,
"learning_rate": 1.69811320754717e-05,
"log_odds_chosen": 16.3641300201416,
"log_odds_ratio": -0.061000462621450424,
"logits/chosen": -1.4788792133331299,
"logits/rejected": -5.779600143432617,
"logps/chosen": -1.4309313297271729,
"logps/rejected": -17.400362014770508,
"loss": 1.4527,
"nll_loss": 1.4466229677200317,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.14309315383434296,
"rewards/margins": 1.5969431400299072,
"rewards/rejected": -1.7400362491607666,
"step": 350
},
{
"epoch": 1.3245283018867924,
"grad_norm": 0.32729095220565796,
"learning_rate": 1.6886792452830187e-05,
"log_odds_chosen": 15.319132804870605,
"log_odds_ratio": -0.056552521884441376,
"logits/chosen": -0.8255969285964966,
"logits/rejected": -5.052517890930176,
"logps/chosen": -1.3465955257415771,
"logps/rejected": -16.29346466064453,
"loss": 1.4911,
"nll_loss": 1.4854259490966797,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1346595585346222,
"rewards/margins": 1.4946870803833008,
"rewards/rejected": -1.6293466091156006,
"step": 351
},
{
"epoch": 1.3283018867924528,
"grad_norm": 0.3329434394836426,
"learning_rate": 1.6792452830188678e-05,
"log_odds_chosen": 16.610496520996094,
"log_odds_ratio": -3.859699427266605e-05,
"logits/chosen": -2.8449089527130127,
"logits/rejected": -7.928000450134277,
"logps/chosen": -1.596476435661316,
"logps/rejected": -17.940351486206055,
"loss": 1.4448,
"nll_loss": 1.4447648525238037,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15964765846729279,
"rewards/margins": 1.634387493133545,
"rewards/rejected": -1.7940351963043213,
"step": 352
},
{
"epoch": 1.3320754716981131,
"grad_norm": 0.3026202321052551,
"learning_rate": 1.669811320754717e-05,
"log_odds_chosen": 16.932235717773438,
"log_odds_ratio": -0.03685789927840233,
"logits/chosen": -1.5450692176818848,
"logits/rejected": -4.212255001068115,
"logps/chosen": -1.2493813037872314,
"logps/rejected": -17.693649291992188,
"loss": 1.4363,
"nll_loss": 1.4326279163360596,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.12493812292814255,
"rewards/margins": 1.644426941871643,
"rewards/rejected": -1.7693649530410767,
"step": 353
},
{
"epoch": 1.3358490566037735,
"grad_norm": 0.2926734685897827,
"learning_rate": 1.660377358490566e-05,
"log_odds_chosen": 17.678836822509766,
"log_odds_ratio": -8.680287464812864e-06,
"logits/chosen": -2.915510654449463,
"logits/rejected": -7.7731733322143555,
"logps/chosen": -1.7902599573135376,
"logps/rejected": -19.27121353149414,
"loss": 1.5991,
"nll_loss": 1.5991185903549194,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17902600765228271,
"rewards/margins": 1.7480952739715576,
"rewards/rejected": -1.9271214008331299,
"step": 354
},
{
"epoch": 1.3396226415094339,
"grad_norm": 0.340572327375412,
"learning_rate": 1.650943396226415e-05,
"log_odds_chosen": 17.665685653686523,
"log_odds_ratio": -0.02697843872010708,
"logits/chosen": -1.601323127746582,
"logits/rejected": -5.377262592315674,
"logps/chosen": -1.6611402034759521,
"logps/rejected": -19.03213119506836,
"loss": 1.4539,
"nll_loss": 1.4511959552764893,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16611401736736298,
"rewards/margins": 1.7370991706848145,
"rewards/rejected": -1.9032131433486938,
"step": 355
},
{
"epoch": 1.3433962264150943,
"grad_norm": 1.6760241985321045,
"learning_rate": 1.6415094339622643e-05,
"log_odds_chosen": 18.934751510620117,
"log_odds_ratio": -1.7136345320523105e-07,
"logits/chosen": -2.097600221633911,
"logits/rejected": -6.444354057312012,
"logps/chosen": -1.4015324115753174,
"logps/rejected": -19.90370750427246,
"loss": 1.4334,
"nll_loss": 1.4333903789520264,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14015324413776398,
"rewards/margins": 1.850217580795288,
"rewards/rejected": -1.990370750427246,
"step": 356
},
{
"epoch": 1.3471698113207546,
"grad_norm": 0.3295755386352539,
"learning_rate": 1.6320754716981134e-05,
"log_odds_chosen": 19.07089614868164,
"log_odds_ratio": -5.215408194203519e-08,
"logits/chosen": -0.8209649324417114,
"logits/rejected": -4.625163555145264,
"logps/chosen": -1.615844964981079,
"logps/rejected": -20.342763900756836,
"loss": 1.7516,
"nll_loss": 1.7515586614608765,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1615844964981079,
"rewards/margins": 1.8726916313171387,
"rewards/rejected": -2.034276247024536,
"step": 357
},
{
"epoch": 1.350943396226415,
"grad_norm": 0.3472559154033661,
"learning_rate": 1.622641509433962e-05,
"log_odds_chosen": 16.93330955505371,
"log_odds_ratio": -0.0761246532201767,
"logits/chosen": -1.1531541347503662,
"logits/rejected": -4.757190704345703,
"logps/chosen": -1.6295312643051147,
"logps/rejected": -18.269256591796875,
"loss": 1.452,
"nll_loss": 1.4443397521972656,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16295313835144043,
"rewards/margins": 1.6639723777770996,
"rewards/rejected": -1.82692551612854,
"step": 358
},
{
"epoch": 1.3547169811320754,
"grad_norm": 0.3795447051525116,
"learning_rate": 1.6132075471698112e-05,
"log_odds_chosen": 18.87713623046875,
"log_odds_ratio": -2.4379320166190155e-05,
"logits/chosen": -3.349454402923584,
"logits/rejected": -7.033996105194092,
"logps/chosen": -1.5019011497497559,
"logps/rejected": -20.02358055114746,
"loss": 1.4321,
"nll_loss": 1.4321367740631104,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15019011497497559,
"rewards/margins": 1.8521679639816284,
"rewards/rejected": -2.0023581981658936,
"step": 359
},
{
"epoch": 1.3584905660377358,
"grad_norm": 0.40565335750579834,
"learning_rate": 1.6037735849056604e-05,
"log_odds_chosen": 15.8294095993042,
"log_odds_ratio": -0.08895085752010345,
"logits/chosen": -1.62678861618042,
"logits/rejected": -4.0340166091918945,
"logps/chosen": -1.596952199935913,
"logps/rejected": -17.054447174072266,
"loss": 1.4759,
"nll_loss": 1.4669734239578247,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15969522297382355,
"rewards/margins": 1.5457494258880615,
"rewards/rejected": -1.705444574356079,
"step": 360
},
{
"epoch": 1.3622641509433961,
"grad_norm": 0.363148033618927,
"learning_rate": 1.5943396226415095e-05,
"log_odds_chosen": 18.350400924682617,
"log_odds_ratio": -2.0042255073349224e-06,
"logits/chosen": -0.9768545031547546,
"logits/rejected": -5.307037353515625,
"logps/chosen": -1.459285020828247,
"logps/rejected": -19.410093307495117,
"loss": 1.3574,
"nll_loss": 1.3574293851852417,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1459285020828247,
"rewards/margins": 1.7950809001922607,
"rewards/rejected": -1.941009283065796,
"step": 361
},
{
"epoch": 1.3660377358490565,
"grad_norm": 0.3766542673110962,
"learning_rate": 1.5849056603773586e-05,
"log_odds_chosen": 16.924793243408203,
"log_odds_ratio": -6.675824806734454e-06,
"logits/chosen": -2.368605136871338,
"logits/rejected": -7.107510566711426,
"logps/chosen": -1.487027883529663,
"logps/rejected": -18.08290672302246,
"loss": 1.4416,
"nll_loss": 1.4416375160217285,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14870277047157288,
"rewards/margins": 1.6595878601074219,
"rewards/rejected": -1.8082906007766724,
"step": 362
},
{
"epoch": 1.369811320754717,
"grad_norm": 0.36386638879776,
"learning_rate": 1.5754716981132077e-05,
"log_odds_chosen": 18.357254028320312,
"log_odds_ratio": -3.6732189983013086e-06,
"logits/chosen": -1.288124680519104,
"logits/rejected": -5.227755546569824,
"logps/chosen": -1.0662007331848145,
"logps/rejected": -18.76520538330078,
"loss": 1.0875,
"nll_loss": 1.0875431299209595,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.10662007331848145,
"rewards/margins": 1.7699006795883179,
"rewards/rejected": -1.8765207529067993,
"step": 363
},
{
"epoch": 1.3735849056603773,
"grad_norm": 0.3877186179161072,
"learning_rate": 1.5660377358490564e-05,
"log_odds_chosen": 17.02109718322754,
"log_odds_ratio": -9.321229299530387e-06,
"logits/chosen": -1.9352566003799438,
"logits/rejected": -5.730966567993164,
"logps/chosen": -1.414484977722168,
"logps/rejected": -18.040632247924805,
"loss": 1.3428,
"nll_loss": 1.3428069353103638,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1414484977722168,
"rewards/margins": 1.6626148223876953,
"rewards/rejected": -1.804063320159912,
"step": 364
},
{
"epoch": 1.3773584905660377,
"grad_norm": 1.2752561569213867,
"learning_rate": 1.5566037735849056e-05,
"log_odds_chosen": 16.804956436157227,
"log_odds_ratio": -0.0012535701971501112,
"logits/chosen": -1.5871500968933105,
"logits/rejected": -5.9648027420043945,
"logps/chosen": -1.3970187902450562,
"logps/rejected": -17.712099075317383,
"loss": 1.5305,
"nll_loss": 1.5304073095321655,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13970187306404114,
"rewards/margins": 1.6315079927444458,
"rewards/rejected": -1.7712098360061646,
"step": 365
},
{
"epoch": 1.3811320754716983,
"grad_norm": 0.33747872710227966,
"learning_rate": 1.5471698113207547e-05,
"log_odds_chosen": 18.296247482299805,
"log_odds_ratio": -6.034980515323696e-07,
"logits/chosen": -0.8064440488815308,
"logits/rejected": -5.531889915466309,
"logps/chosen": -1.6278749704360962,
"logps/rejected": -19.650760650634766,
"loss": 1.5717,
"nll_loss": 1.5716900825500488,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16278749704360962,
"rewards/margins": 1.8022886514663696,
"rewards/rejected": -1.965076208114624,
"step": 366
},
{
"epoch": 1.3849056603773584,
"grad_norm": 0.4021959900856018,
"learning_rate": 1.5377358490566038e-05,
"log_odds_chosen": 17.49240493774414,
"log_odds_ratio": -0.0005192816024646163,
"logits/chosen": -1.245084285736084,
"logits/rejected": -5.5846710205078125,
"logps/chosen": -1.4274266958236694,
"logps/rejected": -18.540454864501953,
"loss": 1.3852,
"nll_loss": 1.3851439952850342,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14274267852306366,
"rewards/margins": 1.7113028764724731,
"rewards/rejected": -1.8540456295013428,
"step": 367
},
{
"epoch": 1.388679245283019,
"grad_norm": 0.4979332685470581,
"learning_rate": 1.528301886792453e-05,
"log_odds_chosen": 16.063011169433594,
"log_odds_ratio": -0.06267835944890976,
"logits/chosen": -1.0869340896606445,
"logits/rejected": -4.391514778137207,
"logps/chosen": -1.4439592361450195,
"logps/rejected": -17.10696029663086,
"loss": 1.4131,
"nll_loss": 1.406846046447754,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.14439593255519867,
"rewards/margins": 1.5663002729415894,
"rewards/rejected": -1.7106962203979492,
"step": 368
},
{
"epoch": 1.3924528301886792,
"grad_norm": 1.9730045795440674,
"learning_rate": 1.5188679245283018e-05,
"log_odds_chosen": 15.107033729553223,
"log_odds_ratio": -0.06132856011390686,
"logits/chosen": -1.2891849279403687,
"logits/rejected": -4.657969951629639,
"logps/chosen": -1.6434388160705566,
"logps/rejected": -16.488523483276367,
"loss": 1.4781,
"nll_loss": 1.4719549417495728,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16434389352798462,
"rewards/margins": 1.4845083951950073,
"rewards/rejected": -1.6488523483276367,
"step": 369
},
{
"epoch": 1.3962264150943398,
"grad_norm": 0.3606973886489868,
"learning_rate": 1.509433962264151e-05,
"log_odds_chosen": 18.26697540283203,
"log_odds_ratio": -2.7567176630327594e-07,
"logits/chosen": -0.31442150473594666,
"logits/rejected": -4.0157999992370605,
"logps/chosen": -1.3492426872253418,
"logps/rejected": -19.13711929321289,
"loss": 1.4501,
"nll_loss": 1.4500954151153564,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1349242627620697,
"rewards/margins": 1.778787612915039,
"rewards/rejected": -1.9137119054794312,
"step": 370
},
{
"epoch": 1.4,
"grad_norm": 0.7839713096618652,
"learning_rate": 1.5e-05,
"log_odds_chosen": 18.16492462158203,
"log_odds_ratio": -0.06902754306793213,
"logits/chosen": -2.4219627380371094,
"logits/rejected": -5.2284698486328125,
"logps/chosen": -1.5058531761169434,
"logps/rejected": -19.259458541870117,
"loss": 1.5453,
"nll_loss": 1.5383750200271606,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1505853235721588,
"rewards/margins": 1.7753607034683228,
"rewards/rejected": -1.9259459972381592,
"step": 371
},
{
"epoch": 1.4037735849056605,
"grad_norm": 0.3838679790496826,
"learning_rate": 1.4905660377358491e-05,
"log_odds_chosen": 16.6832218170166,
"log_odds_ratio": -0.06244910508394241,
"logits/chosen": -1.3226251602172852,
"logits/rejected": -6.231893539428711,
"logps/chosen": -1.6935502290725708,
"logps/rejected": -18.128742218017578,
"loss": 1.6526,
"nll_loss": 1.6463897228240967,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16935503482818604,
"rewards/margins": 1.6435191631317139,
"rewards/rejected": -1.8128741979599,
"step": 372
},
{
"epoch": 1.4075471698113207,
"grad_norm": 0.38915058970451355,
"learning_rate": 1.4811320754716981e-05,
"log_odds_chosen": 16.009296417236328,
"log_odds_ratio": -1.8557118892204016e-05,
"logits/chosen": -1.1935677528381348,
"logits/rejected": -5.60360050201416,
"logps/chosen": -1.3822619915008545,
"logps/rejected": -17.003089904785156,
"loss": 1.3057,
"nll_loss": 1.305704951286316,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1382262110710144,
"rewards/margins": 1.5620828866958618,
"rewards/rejected": -1.7003090381622314,
"step": 373
},
{
"epoch": 1.4113207547169813,
"grad_norm": 0.31976625323295593,
"learning_rate": 1.4716981132075472e-05,
"log_odds_chosen": 18.23406982421875,
"log_odds_ratio": -0.0007746726041659713,
"logits/chosen": -2.0351643562316895,
"logits/rejected": -5.288931846618652,
"logps/chosen": -1.3592875003814697,
"logps/rejected": -19.12387466430664,
"loss": 1.4888,
"nll_loss": 1.4887058734893799,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13592875003814697,
"rewards/margins": 1.776458978652954,
"rewards/rejected": -1.9123876094818115,
"step": 374
},
{
"epoch": 1.4150943396226414,
"grad_norm": 0.3839375376701355,
"learning_rate": 1.4622641509433963e-05,
"log_odds_chosen": 16.633087158203125,
"log_odds_ratio": -0.18126872181892395,
"logits/chosen": -1.4143149852752686,
"logits/rejected": -4.575836658477783,
"logps/chosen": -1.3354686498641968,
"logps/rejected": -17.673622131347656,
"loss": 1.4257,
"nll_loss": 1.4075796604156494,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1335468739271164,
"rewards/margins": 1.6338152885437012,
"rewards/rejected": -1.7673622369766235,
"step": 375
},
{
"epoch": 1.418867924528302,
"grad_norm": 0.3899994194507599,
"learning_rate": 1.4528301886792452e-05,
"log_odds_chosen": 18.611719131469727,
"log_odds_ratio": -5.587950795415964e-07,
"logits/chosen": -1.2232400178909302,
"logits/rejected": -5.466405868530273,
"logps/chosen": -1.512157917022705,
"logps/rejected": -19.77882194519043,
"loss": 1.5621,
"nll_loss": 1.5620990991592407,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1512157917022705,
"rewards/margins": 1.8266665935516357,
"rewards/rejected": -1.9778823852539062,
"step": 376
},
{
"epoch": 1.4226415094339622,
"grad_norm": 0.31554582715034485,
"learning_rate": 1.4433962264150944e-05,
"log_odds_chosen": 17.60826301574707,
"log_odds_ratio": -3.2186994758376386e-06,
"logits/chosen": -1.3368791341781616,
"logits/rejected": -6.721653938293457,
"logps/chosen": -1.4264721870422363,
"logps/rejected": -18.68645477294922,
"loss": 1.37,
"nll_loss": 1.3699793815612793,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14264723658561707,
"rewards/margins": 1.7259981632232666,
"rewards/rejected": -1.868645429611206,
"step": 377
},
{
"epoch": 1.4264150943396228,
"grad_norm": 0.3350052535533905,
"learning_rate": 1.4339622641509435e-05,
"log_odds_chosen": 15.84263801574707,
"log_odds_ratio": -0.05588989332318306,
"logits/chosen": -2.5817272663116455,
"logits/rejected": -7.034815311431885,
"logps/chosen": -1.5793471336364746,
"logps/rejected": -17.11503791809082,
"loss": 1.4305,
"nll_loss": 1.4249083995819092,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15793471038341522,
"rewards/margins": 1.5535691976547241,
"rewards/rejected": -1.7115037441253662,
"step": 378
},
{
"epoch": 1.430188679245283,
"grad_norm": 0.3305111825466156,
"learning_rate": 1.4245283018867924e-05,
"log_odds_chosen": 17.508865356445312,
"log_odds_ratio": -4.04572256229585e-06,
"logits/chosen": -1.8427374362945557,
"logits/rejected": -7.116032600402832,
"logps/chosen": -1.6888805627822876,
"logps/rejected": -18.881450653076172,
"loss": 1.4913,
"nll_loss": 1.491302490234375,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16888806223869324,
"rewards/margins": 1.719257116317749,
"rewards/rejected": -1.888145089149475,
"step": 379
},
{
"epoch": 1.4339622641509435,
"grad_norm": 0.5024957656860352,
"learning_rate": 1.4150943396226415e-05,
"log_odds_chosen": 16.263195037841797,
"log_odds_ratio": -0.0004213673819322139,
"logits/chosen": -1.309700608253479,
"logits/rejected": -5.588892459869385,
"logps/chosen": -1.4294378757476807,
"logps/rejected": -17.363752365112305,
"loss": 1.3392,
"nll_loss": 1.339144229888916,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14294379949569702,
"rewards/margins": 1.5934314727783203,
"rewards/rejected": -1.736375331878662,
"step": 380
},
{
"epoch": 1.4377358490566037,
"grad_norm": 0.3632482588291168,
"learning_rate": 1.4056603773584906e-05,
"log_odds_chosen": 17.019535064697266,
"log_odds_ratio": -5.90098215980106e-06,
"logits/chosen": -1.0069024562835693,
"logits/rejected": -5.128624439239502,
"logps/chosen": -1.4965219497680664,
"logps/rejected": -18.188318252563477,
"loss": 1.4404,
"nll_loss": 1.440355896949768,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14965218305587769,
"rewards/margins": 1.6691796779632568,
"rewards/rejected": -1.8188319206237793,
"step": 381
},
{
"epoch": 1.4415094339622643,
"grad_norm": 0.3996204435825348,
"learning_rate": 1.3962264150943397e-05,
"log_odds_chosen": 16.09657096862793,
"log_odds_ratio": -0.050253380089998245,
"logits/chosen": -1.884864330291748,
"logits/rejected": -5.274215221405029,
"logps/chosen": -1.6420609951019287,
"logps/rejected": -17.452035903930664,
"loss": 1.556,
"nll_loss": 1.5509570837020874,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1642061024904251,
"rewards/margins": 1.5809974670410156,
"rewards/rejected": -1.7452034950256348,
"step": 382
},
{
"epoch": 1.4452830188679244,
"grad_norm": 0.38239994645118713,
"learning_rate": 1.3867924528301887e-05,
"log_odds_chosen": 17.446556091308594,
"log_odds_ratio": -0.021241456270217896,
"logits/chosen": -1.7642720937728882,
"logits/rejected": -4.676743507385254,
"logps/chosen": -1.4043127298355103,
"logps/rejected": -18.422576904296875,
"loss": 1.4152,
"nll_loss": 1.4130536317825317,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14043128490447998,
"rewards/margins": 1.7018264532089233,
"rewards/rejected": -1.8422577381134033,
"step": 383
},
{
"epoch": 1.449056603773585,
"grad_norm": 0.32839640974998474,
"learning_rate": 1.3773584905660378e-05,
"log_odds_chosen": 16.299285888671875,
"log_odds_ratio": -4.187269496469526e-06,
"logits/chosen": -2.137349843978882,
"logits/rejected": -8.301512718200684,
"logps/chosen": -1.6436606645584106,
"logps/rejected": -17.667858123779297,
"loss": 1.5086,
"nll_loss": 1.5085557699203491,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16436606645584106,
"rewards/margins": 1.6024198532104492,
"rewards/rejected": -1.7667858600616455,
"step": 384
},
{
"epoch": 1.4528301886792452,
"grad_norm": 0.3542121648788452,
"learning_rate": 1.3679245283018869e-05,
"log_odds_chosen": 15.682723999023438,
"log_odds_ratio": -2.753584340098314e-05,
"logits/chosen": -1.6206883192062378,
"logits/rejected": -5.171188831329346,
"logps/chosen": -1.3008809089660645,
"logps/rejected": -16.44762420654297,
"loss": 1.2725,
"nll_loss": 1.2724624872207642,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13008809089660645,
"rewards/margins": 1.514674425125122,
"rewards/rejected": -1.644762396812439,
"step": 385
},
{
"epoch": 1.4566037735849058,
"grad_norm": 0.3344227373600006,
"learning_rate": 1.3584905660377358e-05,
"log_odds_chosen": 16.07890510559082,
"log_odds_ratio": -0.00016838790907058865,
"logits/chosen": -1.1785922050476074,
"logits/rejected": -5.77630090713501,
"logps/chosen": -1.60475492477417,
"logps/rejected": -17.42005157470703,
"loss": 1.4546,
"nll_loss": 1.4545769691467285,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.160475492477417,
"rewards/margins": 1.5815297365188599,
"rewards/rejected": -1.7420052289962769,
"step": 386
},
{
"epoch": 1.460377358490566,
"grad_norm": 0.3509989380836487,
"learning_rate": 1.349056603773585e-05,
"log_odds_chosen": 17.025386810302734,
"log_odds_ratio": -4.601534237735905e-05,
"logits/chosen": -1.3022335767745972,
"logits/rejected": -5.336873531341553,
"logps/chosen": -1.3712674379348755,
"logps/rejected": -17.987762451171875,
"loss": 1.2749,
"nll_loss": 1.2749412059783936,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13712672889232635,
"rewards/margins": 1.6616493463516235,
"rewards/rejected": -1.7987761497497559,
"step": 387
},
{
"epoch": 1.4641509433962265,
"grad_norm": 0.35176974534988403,
"learning_rate": 1.339622641509434e-05,
"log_odds_chosen": 18.190950393676758,
"log_odds_ratio": -1.4410341464099474e-05,
"logits/chosen": -2.399728298187256,
"logits/rejected": -6.07908296585083,
"logps/chosen": -1.1239242553710938,
"logps/rejected": -18.555187225341797,
"loss": 1.3573,
"nll_loss": 1.3572728633880615,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.11239242553710938,
"rewards/margins": 1.7431262731552124,
"rewards/rejected": -1.8555186986923218,
"step": 388
},
{
"epoch": 1.4679245283018867,
"grad_norm": 0.35888710618019104,
"learning_rate": 1.3301886792452831e-05,
"log_odds_chosen": 16.425899505615234,
"log_odds_ratio": -0.062047071754932404,
"logits/chosen": -2.126614570617676,
"logits/rejected": -6.522840976715088,
"logps/chosen": -1.5465141534805298,
"logps/rejected": -17.640609741210938,
"loss": 1.4328,
"nll_loss": 1.4265483617782593,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15465140342712402,
"rewards/margins": 1.6094096899032593,
"rewards/rejected": -1.7640612125396729,
"step": 389
},
{
"epoch": 1.4716981132075473,
"grad_norm": 0.4034729301929474,
"learning_rate": 1.320754716981132e-05,
"log_odds_chosen": 15.568371772766113,
"log_odds_ratio": -3.7845486076548696e-05,
"logits/chosen": -0.5623266696929932,
"logits/rejected": -4.619795799255371,
"logps/chosen": -1.4479458332061768,
"logps/rejected": -16.667530059814453,
"loss": 1.3254,
"nll_loss": 1.3253822326660156,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14479456841945648,
"rewards/margins": 1.5219584703445435,
"rewards/rejected": -1.6667530536651611,
"step": 390
},
{
"epoch": 1.4754716981132074,
"grad_norm": 0.3520585596561432,
"learning_rate": 1.3113207547169812e-05,
"log_odds_chosen": 14.612020492553711,
"log_odds_ratio": -0.0023431943263858557,
"logits/chosen": -2.2221946716308594,
"logits/rejected": -6.539764881134033,
"logps/chosen": -1.4834494590759277,
"logps/rejected": -15.739931106567383,
"loss": 1.4064,
"nll_loss": 1.4061977863311768,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.148344948887825,
"rewards/margins": 1.4256480932235718,
"rewards/rejected": -1.5739930868148804,
"step": 391
},
{
"epoch": 1.479245283018868,
"grad_norm": 0.36251071095466614,
"learning_rate": 1.3018867924528303e-05,
"log_odds_chosen": 16.438549041748047,
"log_odds_ratio": -0.09116589277982712,
"logits/chosen": -2.7520642280578613,
"logits/rejected": -7.930850028991699,
"logps/chosen": -1.535467267036438,
"logps/rejected": -17.66228485107422,
"loss": 1.4605,
"nll_loss": 1.4513870477676392,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15354672074317932,
"rewards/margins": 1.6126817464828491,
"rewards/rejected": -1.766228437423706,
"step": 392
},
{
"epoch": 1.4830188679245282,
"grad_norm": 0.34293168783187866,
"learning_rate": 1.2924528301886792e-05,
"log_odds_chosen": 16.164066314697266,
"log_odds_ratio": -0.0003289695887360722,
"logits/chosen": -1.1001029014587402,
"logits/rejected": -6.432461738586426,
"logps/chosen": -1.488477349281311,
"logps/rejected": -17.351869583129883,
"loss": 1.3264,
"nll_loss": 1.3263657093048096,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14884772896766663,
"rewards/margins": 1.5863392353057861,
"rewards/rejected": -1.73518705368042,
"step": 393
},
{
"epoch": 1.4867924528301888,
"grad_norm": 0.3245314359664917,
"learning_rate": 1.2830188679245283e-05,
"log_odds_chosen": 17.266613006591797,
"log_odds_ratio": -0.00029518589144572616,
"logits/chosen": -1.3167364597320557,
"logits/rejected": -5.948409080505371,
"logps/chosen": -1.1578724384307861,
"logps/rejected": -17.86254119873047,
"loss": 1.2429,
"nll_loss": 1.2429192066192627,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.11578723788261414,
"rewards/margins": 1.6704668998718262,
"rewards/rejected": -1.7862541675567627,
"step": 394
},
{
"epoch": 1.490566037735849,
"grad_norm": 0.3447877764701843,
"learning_rate": 1.2735849056603775e-05,
"log_odds_chosen": 15.028793334960938,
"log_odds_ratio": -0.1531548947095871,
"logits/chosen": -2.7501590251922607,
"logits/rejected": -6.11824369430542,
"logps/chosen": -1.720342993736267,
"logps/rejected": -16.537107467651367,
"loss": 1.6015,
"nll_loss": 1.5861958265304565,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17203430831432343,
"rewards/margins": 1.481676459312439,
"rewards/rejected": -1.6537107229232788,
"step": 395
},
{
"epoch": 1.4943396226415095,
"grad_norm": 0.3362826704978943,
"learning_rate": 1.2641509433962264e-05,
"log_odds_chosen": 17.786401748657227,
"log_odds_ratio": -3.323038754388108e-06,
"logits/chosen": -1.742743730545044,
"logits/rejected": -5.260643482208252,
"logps/chosen": -1.4009883403778076,
"logps/rejected": -18.64501190185547,
"loss": 1.4321,
"nll_loss": 1.4321409463882446,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14009883999824524,
"rewards/margins": 1.7244024276733398,
"rewards/rejected": -1.8645012378692627,
"step": 396
},
{
"epoch": 1.4981132075471697,
"grad_norm": 0.3690617084503174,
"learning_rate": 1.2547169811320755e-05,
"log_odds_chosen": 18.10134506225586,
"log_odds_ratio": -0.050154730677604675,
"logits/chosen": -1.2822532653808594,
"logits/rejected": -4.944076061248779,
"logps/chosen": -1.3214470148086548,
"logps/rejected": -19.015644073486328,
"loss": 1.219,
"nll_loss": 1.2139716148376465,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.13214470446109772,
"rewards/margins": 1.7694196701049805,
"rewards/rejected": -1.9015644788742065,
"step": 397
},
{
"epoch": 1.5018867924528303,
"grad_norm": 0.3674098551273346,
"learning_rate": 1.2452830188679246e-05,
"log_odds_chosen": 15.48218059539795,
"log_odds_ratio": -0.06490848958492279,
"logits/chosen": -1.954714298248291,
"logits/rejected": -6.481346130371094,
"logps/chosen": -1.3896689414978027,
"logps/rejected": -16.480030059814453,
"loss": 1.4439,
"nll_loss": 1.4373944997787476,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1389668881893158,
"rewards/margins": 1.5090359449386597,
"rewards/rejected": -1.6480028629302979,
"step": 398
},
{
"epoch": 1.5056603773584905,
"grad_norm": 0.36629459261894226,
"learning_rate": 1.2358490566037737e-05,
"log_odds_chosen": 16.291677474975586,
"log_odds_ratio": -0.08093234151601791,
"logits/chosen": -0.5040303468704224,
"logits/rejected": -4.219124794006348,
"logps/chosen": -1.6667414903640747,
"logps/rejected": -17.61298370361328,
"loss": 1.6038,
"nll_loss": 1.5956670045852661,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1666741520166397,
"rewards/margins": 1.594624400138855,
"rewards/rejected": -1.761298418045044,
"step": 399
},
{
"epoch": 1.509433962264151,
"grad_norm": 0.4143602252006531,
"learning_rate": 1.2264150943396227e-05,
"log_odds_chosen": 16.119110107421875,
"log_odds_ratio": -0.0009257096680812538,
"logits/chosen": -0.8463224172592163,
"logits/rejected": -4.586696624755859,
"logps/chosen": -1.3971197605133057,
"logps/rejected": -17.024677276611328,
"loss": 1.4237,
"nll_loss": 1.423575758934021,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13971199095249176,
"rewards/margins": 1.5627559423446655,
"rewards/rejected": -1.702467918395996,
"step": 400
},
{
"epoch": 1.5132075471698112,
"grad_norm": 0.3812906742095947,
"learning_rate": 1.2169811320754718e-05,
"log_odds_chosen": 14.85096549987793,
"log_odds_ratio": -0.05318630486726761,
"logits/chosen": -0.3447520136833191,
"logits/rejected": -4.831925868988037,
"logps/chosen": -1.568985939025879,
"logps/rejected": -16.13321304321289,
"loss": 1.4211,
"nll_loss": 1.4157638549804688,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1568985879421234,
"rewards/margins": 1.4564226865768433,
"rewards/rejected": -1.613321304321289,
"step": 401
},
{
"epoch": 1.5169811320754718,
"grad_norm": 0.3796207010746002,
"learning_rate": 1.2075471698113209e-05,
"log_odds_chosen": 14.776948928833008,
"log_odds_ratio": -0.13772796094417572,
"logits/chosen": -2.6085329055786133,
"logits/rejected": -5.985441207885742,
"logps/chosen": -1.564119815826416,
"logps/rejected": -16.096309661865234,
"loss": 1.3687,
"nll_loss": 1.3549602031707764,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15641197562217712,
"rewards/margins": 1.4532190561294556,
"rewards/rejected": -1.6096309423446655,
"step": 402
},
{
"epoch": 1.520754716981132,
"grad_norm": 0.6695427894592285,
"learning_rate": 1.1981132075471698e-05,
"log_odds_chosen": 15.497125625610352,
"log_odds_ratio": -7.294305760296993e-06,
"logits/chosen": -0.9591398239135742,
"logits/rejected": -5.608551979064941,
"logps/chosen": -1.536940097808838,
"logps/rejected": -16.734256744384766,
"loss": 1.3521,
"nll_loss": 1.3521382808685303,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1536940038204193,
"rewards/margins": 1.5197317600250244,
"rewards/rejected": -1.6734257936477661,
"step": 403
},
{
"epoch": 1.5245283018867926,
"grad_norm": 0.5477868318557739,
"learning_rate": 1.188679245283019e-05,
"log_odds_chosen": 16.6223087310791,
"log_odds_ratio": -0.0006986000225879252,
"logits/chosen": -1.419918179512024,
"logits/rejected": -4.885513782501221,
"logps/chosen": -1.3331007957458496,
"logps/rejected": -17.54778480529785,
"loss": 1.3093,
"nll_loss": 1.3091806173324585,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13331007957458496,
"rewards/margins": 1.621468424797058,
"rewards/rejected": -1.7547786235809326,
"step": 404
},
{
"epoch": 1.5283018867924527,
"grad_norm": 0.333292156457901,
"learning_rate": 1.179245283018868e-05,
"log_odds_chosen": 18.347660064697266,
"log_odds_ratio": -1.3806706192553975e-05,
"logits/chosen": -1.4701156616210938,
"logits/rejected": -6.684134483337402,
"logps/chosen": -1.4482415914535522,
"logps/rejected": -19.408504486083984,
"loss": 1.3793,
"nll_loss": 1.3793381452560425,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14482416212558746,
"rewards/margins": 1.7960264682769775,
"rewards/rejected": -1.9408507347106934,
"step": 405
},
{
"epoch": 1.5320754716981133,
"grad_norm": 0.3505614399909973,
"learning_rate": 1.169811320754717e-05,
"log_odds_chosen": 18.066612243652344,
"log_odds_ratio": -6.847452823421918e-06,
"logits/chosen": -2.2876837253570557,
"logits/rejected": -6.302707195281982,
"logps/chosen": -1.4648969173431396,
"logps/rejected": -19.111392974853516,
"loss": 1.6152,
"nll_loss": 1.6151999235153198,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1464896947145462,
"rewards/margins": 1.7646493911743164,
"rewards/rejected": -1.9111392498016357,
"step": 406
},
{
"epoch": 1.5358490566037735,
"grad_norm": 0.356839120388031,
"learning_rate": 1.160377358490566e-05,
"log_odds_chosen": 15.59557819366455,
"log_odds_ratio": -0.0001072017039405182,
"logits/chosen": -0.30015936493873596,
"logits/rejected": -6.0860114097595215,
"logps/chosen": -1.6691813468933105,
"logps/rejected": -17.017396926879883,
"loss": 1.4345,
"nll_loss": 1.434523344039917,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16691814363002777,
"rewards/margins": 1.534821629524231,
"rewards/rejected": -1.70173978805542,
"step": 407
},
{
"epoch": 1.539622641509434,
"grad_norm": 0.36700817942619324,
"learning_rate": 1.1509433962264152e-05,
"log_odds_chosen": 17.510787963867188,
"log_odds_ratio": -1.4975793192206766e-06,
"logits/chosen": -1.014905571937561,
"logits/rejected": -5.991207122802734,
"logps/chosen": -1.4787532091140747,
"logps/rejected": -18.62135887145996,
"loss": 1.4582,
"nll_loss": 1.4582459926605225,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1478753238916397,
"rewards/margins": 1.7142606973648071,
"rewards/rejected": -1.8621360063552856,
"step": 408
},
{
"epoch": 1.5433962264150942,
"grad_norm": 0.355266273021698,
"learning_rate": 1.1415094339622643e-05,
"log_odds_chosen": 16.35395050048828,
"log_odds_ratio": -7.43008786230348e-05,
"logits/chosen": -1.7440268993377686,
"logits/rejected": -6.251463890075684,
"logps/chosen": -1.4440710544586182,
"logps/rejected": -17.370180130004883,
"loss": 1.6528,
"nll_loss": 1.6527677774429321,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14440712332725525,
"rewards/margins": 1.5926108360290527,
"rewards/rejected": -1.73701810836792,
"step": 409
},
{
"epoch": 1.5471698113207548,
"grad_norm": 0.3182680904865265,
"learning_rate": 1.1320754716981132e-05,
"log_odds_chosen": 15.64585018157959,
"log_odds_ratio": -0.013405690900981426,
"logits/chosen": -1.3935030698776245,
"logits/rejected": -5.763398170471191,
"logps/chosen": -1.366743564605713,
"logps/rejected": -16.644275665283203,
"loss": 1.4451,
"nll_loss": 1.443767786026001,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13667435944080353,
"rewards/margins": 1.5277531147003174,
"rewards/rejected": -1.6644275188446045,
"step": 410
},
{
"epoch": 1.550943396226415,
"grad_norm": 0.3500687777996063,
"learning_rate": 1.1226415094339623e-05,
"log_odds_chosen": 16.229408264160156,
"log_odds_ratio": -5.9309946664143354e-05,
"logits/chosen": -2.4652421474456787,
"logits/rejected": -6.496399879455566,
"logps/chosen": -1.3401199579238892,
"logps/rejected": -17.225467681884766,
"loss": 1.5363,
"nll_loss": 1.5363225936889648,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13401198387145996,
"rewards/margins": 1.5885348320007324,
"rewards/rejected": -1.7225468158721924,
"step": 411
},
{
"epoch": 1.5547169811320756,
"grad_norm": 0.427716463804245,
"learning_rate": 1.1132075471698115e-05,
"log_odds_chosen": 16.30535888671875,
"log_odds_ratio": -0.02768136002123356,
"logits/chosen": -1.3295129537582397,
"logits/rejected": -5.655584335327148,
"logps/chosen": -1.5316812992095947,
"logps/rejected": -17.553970336914062,
"loss": 1.3546,
"nll_loss": 1.3517988920211792,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15316811203956604,
"rewards/margins": 1.602229118347168,
"rewards/rejected": -1.7553972005844116,
"step": 412
},
{
"epoch": 1.5584905660377357,
"grad_norm": 0.3918018639087677,
"learning_rate": 1.1037735849056604e-05,
"log_odds_chosen": 17.692913055419922,
"log_odds_ratio": -9.760309467310435e-07,
"logits/chosen": -0.8427505493164062,
"logits/rejected": -4.809689521789551,
"logps/chosen": -1.7556877136230469,
"logps/rejected": -19.217164993286133,
"loss": 1.3681,
"nll_loss": 1.368148684501648,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17556877434253693,
"rewards/margins": 1.746147632598877,
"rewards/rejected": -1.9217164516448975,
"step": 413
},
{
"epoch": 1.5622641509433963,
"grad_norm": 0.3670799434185028,
"learning_rate": 1.0943396226415095e-05,
"log_odds_chosen": 17.20905303955078,
"log_odds_ratio": -0.020199574530124664,
"logits/chosen": -1.5989339351654053,
"logits/rejected": -5.965342044830322,
"logps/chosen": -1.5377042293548584,
"logps/rejected": -18.371013641357422,
"loss": 1.5964,
"nll_loss": 1.5944006443023682,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15377041697502136,
"rewards/margins": 1.68333101272583,
"rewards/rejected": -1.8371014595031738,
"step": 414
},
{
"epoch": 1.5660377358490565,
"grad_norm": 0.35576295852661133,
"learning_rate": 1.0849056603773586e-05,
"log_odds_chosen": 15.496426582336426,
"log_odds_ratio": -0.049891576170921326,
"logits/chosen": -0.8893330097198486,
"logits/rejected": -5.53910493850708,
"logps/chosen": -1.553781270980835,
"logps/rejected": -16.68655776977539,
"loss": 1.5559,
"nll_loss": 1.5509579181671143,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15537814795970917,
"rewards/margins": 1.5132777690887451,
"rewards/rejected": -1.6686559915542603,
"step": 415
},
{
"epoch": 1.569811320754717,
"grad_norm": 0.32513803243637085,
"learning_rate": 1.0754716981132077e-05,
"log_odds_chosen": 17.47341537475586,
"log_odds_ratio": -0.0002301803178852424,
"logits/chosen": -1.110384225845337,
"logits/rejected": -6.549045562744141,
"logps/chosen": -1.5312076807022095,
"logps/rejected": -18.674665451049805,
"loss": 1.5338,
"nll_loss": 1.5337331295013428,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15312077105045319,
"rewards/margins": 1.7143458127975464,
"rewards/rejected": -1.867466688156128,
"step": 416
},
{
"epoch": 1.5735849056603772,
"grad_norm": 0.3317815959453583,
"learning_rate": 1.0660377358490567e-05,
"log_odds_chosen": 17.113929748535156,
"log_odds_ratio": -0.00808227900415659,
"logits/chosen": -1.527684211730957,
"logits/rejected": -6.576870918273926,
"logps/chosen": -1.5478432178497314,
"logps/rejected": -18.27701187133789,
"loss": 1.5447,
"nll_loss": 1.5438923835754395,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15478432178497314,
"rewards/margins": 1.6729168891906738,
"rewards/rejected": -1.8277010917663574,
"step": 417
},
{
"epoch": 1.5773584905660378,
"grad_norm": 0.3883110582828522,
"learning_rate": 1.0566037735849058e-05,
"log_odds_chosen": 16.38488006591797,
"log_odds_ratio": -0.01739910989999771,
"logits/chosen": -2.0036072731018066,
"logits/rejected": -6.386971473693848,
"logps/chosen": -1.778263807296753,
"logps/rejected": -17.913105010986328,
"loss": 1.5031,
"nll_loss": 1.5013105869293213,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1778264045715332,
"rewards/margins": 1.6134843826293945,
"rewards/rejected": -1.7913107872009277,
"step": 418
},
{
"epoch": 1.581132075471698,
"grad_norm": 0.3066483736038208,
"learning_rate": 1.0471698113207549e-05,
"log_odds_chosen": 19.433700561523438,
"log_odds_ratio": -1.0430817098949774e-07,
"logits/chosen": -1.0177987813949585,
"logits/rejected": -6.801183700561523,
"logps/chosen": -1.4599207639694214,
"logps/rejected": -20.431859970092773,
"loss": 1.3882,
"nll_loss": 1.3881503343582153,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14599208533763885,
"rewards/margins": 1.8971937894821167,
"rewards/rejected": -2.0431859493255615,
"step": 419
},
{
"epoch": 1.5849056603773586,
"grad_norm": 0.40567484498023987,
"learning_rate": 1.0377358490566038e-05,
"log_odds_chosen": 13.711181640625,
"log_odds_ratio": -0.05775582417845726,
"logits/chosen": -0.9496514797210693,
"logits/rejected": -4.353882789611816,
"logps/chosen": -1.6069025993347168,
"logps/rejected": -15.056055068969727,
"loss": 1.3888,
"nll_loss": 1.3829797506332397,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16069024801254272,
"rewards/margins": 1.3449151515960693,
"rewards/rejected": -1.5056054592132568,
"step": 420
},
{
"epoch": 1.5886792452830187,
"grad_norm": 0.34252166748046875,
"learning_rate": 1.028301886792453e-05,
"log_odds_chosen": 19.151784896850586,
"log_odds_ratio": -1.4499668395728804e-05,
"logits/chosen": -1.0097556114196777,
"logits/rejected": -5.88823127746582,
"logps/chosen": -1.4318840503692627,
"logps/rejected": -20.140625,
"loss": 1.4247,
"nll_loss": 1.4247350692749023,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14318840205669403,
"rewards/margins": 1.870874285697937,
"rewards/rejected": -2.0140626430511475,
"step": 421
},
{
"epoch": 1.5924528301886793,
"grad_norm": 0.4356919527053833,
"learning_rate": 1.018867924528302e-05,
"log_odds_chosen": 18.37000274658203,
"log_odds_ratio": -4.2022656998597085e-06,
"logits/chosen": -2.5176637172698975,
"logits/rejected": -5.485119819641113,
"logps/chosen": -1.699413776397705,
"logps/rejected": -19.841552734375,
"loss": 1.4642,
"nll_loss": 1.4642114639282227,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16994138062000275,
"rewards/margins": 1.8142141103744507,
"rewards/rejected": -1.9841554164886475,
"step": 422
},
{
"epoch": 1.5962264150943395,
"grad_norm": 0.3340418338775635,
"learning_rate": 1.009433962264151e-05,
"log_odds_chosen": 17.76876449584961,
"log_odds_ratio": -1.0319773537048604e-05,
"logits/chosen": -1.0354773998260498,
"logits/rejected": -6.589682579040527,
"logps/chosen": -1.570968508720398,
"logps/rejected": -19.032516479492188,
"loss": 1.5785,
"nll_loss": 1.578470230102539,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15709686279296875,
"rewards/margins": 1.74615478515625,
"rewards/rejected": -1.9032516479492188,
"step": 423
},
{
"epoch": 1.6,
"grad_norm": 0.3202284574508667,
"learning_rate": 1e-05,
"log_odds_chosen": 17.717674255371094,
"log_odds_ratio": -2.1941097656963393e-05,
"logits/chosen": -1.6594294309616089,
"logits/rejected": -5.430027484893799,
"logps/chosen": -1.4836103916168213,
"logps/rejected": -18.895004272460938,
"loss": 1.6185,
"nll_loss": 1.6185247898101807,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14836102724075317,
"rewards/margins": 1.7411394119262695,
"rewards/rejected": -1.889500379562378,
"step": 424
},
{
"epoch": 1.6037735849056602,
"grad_norm": 0.3735036849975586,
"learning_rate": 9.905660377358492e-06,
"log_odds_chosen": 17.9097843170166,
"log_odds_ratio": -2.078727447951678e-06,
"logits/chosen": -1.3803160190582275,
"logits/rejected": -5.703221797943115,
"logps/chosen": -1.2858457565307617,
"logps/rejected": -18.800107955932617,
"loss": 1.1804,
"nll_loss": 1.1804119348526,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1285845786333084,
"rewards/margins": 1.751426339149475,
"rewards/rejected": -1.880010962486267,
"step": 425
},
{
"epoch": 1.6075471698113208,
"grad_norm": 0.5746561288833618,
"learning_rate": 9.811320754716983e-06,
"log_odds_chosen": 17.22057342529297,
"log_odds_ratio": -3.956294222007273e-06,
"logits/chosen": -1.1769520044326782,
"logits/rejected": -5.837000846862793,
"logps/chosen": -1.6268696784973145,
"logps/rejected": -18.550504684448242,
"loss": 1.653,
"nll_loss": 1.6529629230499268,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16268697381019592,
"rewards/margins": 1.6923635005950928,
"rewards/rejected": -1.8550504446029663,
"step": 426
},
{
"epoch": 1.611320754716981,
"grad_norm": 0.363587349653244,
"learning_rate": 9.716981132075472e-06,
"log_odds_chosen": 14.084954261779785,
"log_odds_ratio": -0.07172351330518723,
"logits/chosen": -1.2698285579681396,
"logits/rejected": -4.79845666885376,
"logps/chosen": -1.500381350517273,
"logps/rejected": -15.104291915893555,
"loss": 1.4545,
"nll_loss": 1.4473347663879395,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15003815293312073,
"rewards/margins": 1.3603911399841309,
"rewards/rejected": -1.5104291439056396,
"step": 427
},
{
"epoch": 1.6150943396226416,
"grad_norm": 0.3944726884365082,
"learning_rate": 9.622641509433963e-06,
"log_odds_chosen": 16.034841537475586,
"log_odds_ratio": -2.1783518604934216e-05,
"logits/chosen": -0.33273932337760925,
"logits/rejected": -4.981014251708984,
"logps/chosen": -1.6068644523620605,
"logps/rejected": -17.390186309814453,
"loss": 1.5603,
"nll_loss": 1.5603405237197876,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1606864631175995,
"rewards/margins": 1.5783324241638184,
"rewards/rejected": -1.7390189170837402,
"step": 428
},
{
"epoch": 1.6188679245283017,
"grad_norm": 0.3169495463371277,
"learning_rate": 9.528301886792455e-06,
"log_odds_chosen": 17.651002883911133,
"log_odds_ratio": -0.08082529902458191,
"logits/chosen": -1.4152789115905762,
"logits/rejected": -5.411661148071289,
"logps/chosen": -1.3260101079940796,
"logps/rejected": -18.322998046875,
"loss": 1.4476,
"nll_loss": 1.439566731452942,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.13260100781917572,
"rewards/margins": 1.6996989250183105,
"rewards/rejected": -1.8322997093200684,
"step": 429
},
{
"epoch": 1.6226415094339623,
"grad_norm": 0.3842892348766327,
"learning_rate": 9.433962264150944e-06,
"log_odds_chosen": 18.826631546020508,
"log_odds_ratio": -5.513446694749291e-07,
"logits/chosen": -1.7618896961212158,
"logits/rejected": -6.005032062530518,
"logps/chosen": -1.2967678308486938,
"logps/rejected": -19.6664981842041,
"loss": 1.4705,
"nll_loss": 1.4704593420028687,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.12967678904533386,
"rewards/margins": 1.8369731903076172,
"rewards/rejected": -1.9666498899459839,
"step": 430
},
{
"epoch": 1.6264150943396225,
"grad_norm": 0.38966941833496094,
"learning_rate": 9.339622641509433e-06,
"log_odds_chosen": 13.728066444396973,
"log_odds_ratio": -0.15279138088226318,
"logits/chosen": -1.5197663307189941,
"logits/rejected": -5.018875598907471,
"logps/chosen": -1.4149266481399536,
"logps/rejected": -14.751510620117188,
"loss": 1.2838,
"nll_loss": 1.2685606479644775,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.14149266481399536,
"rewards/margins": 1.3336584568023682,
"rewards/rejected": -1.4751511812210083,
"step": 431
},
{
"epoch": 1.630188679245283,
"grad_norm": 0.3509092330932617,
"learning_rate": 9.245283018867924e-06,
"log_odds_chosen": 16.643827438354492,
"log_odds_ratio": -0.06135103851556778,
"logits/chosen": -2.0793323516845703,
"logits/rejected": -6.901341915130615,
"logps/chosen": -1.5519651174545288,
"logps/rejected": -17.924978256225586,
"loss": 1.3948,
"nll_loss": 1.3886274099349976,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15519651770591736,
"rewards/margins": 1.6373014450073242,
"rewards/rejected": -1.7924977540969849,
"step": 432
},
{
"epoch": 1.6339622641509433,
"grad_norm": 0.374100923538208,
"learning_rate": 9.150943396226416e-06,
"log_odds_chosen": 16.738018035888672,
"log_odds_ratio": -9.723278708406724e-06,
"logits/chosen": -0.8103672862052917,
"logits/rejected": -5.695781230926514,
"logps/chosen": -1.61466646194458,
"logps/rejected": -18.09328269958496,
"loss": 1.5884,
"nll_loss": 1.5884230136871338,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16146664321422577,
"rewards/margins": 1.6478617191314697,
"rewards/rejected": -1.809328317642212,
"step": 433
},
{
"epoch": 1.6377358490566039,
"grad_norm": 0.348984956741333,
"learning_rate": 9.056603773584905e-06,
"log_odds_chosen": 15.502889633178711,
"log_odds_ratio": -0.0006235900218598545,
"logits/chosen": -0.9150213003158569,
"logits/rejected": -5.924428939819336,
"logps/chosen": -1.5006028413772583,
"logps/rejected": -16.678489685058594,
"loss": 1.3731,
"nll_loss": 1.3730108737945557,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15006029605865479,
"rewards/margins": 1.5177886486053467,
"rewards/rejected": -1.6678489446640015,
"step": 434
},
{
"epoch": 1.641509433962264,
"grad_norm": 0.3935360312461853,
"learning_rate": 8.962264150943396e-06,
"log_odds_chosen": 15.421186447143555,
"log_odds_ratio": -4.3803676817333326e-05,
"logits/chosen": -2.2583491802215576,
"logits/rejected": -6.483213424682617,
"logps/chosen": -1.748963475227356,
"logps/rejected": -16.925840377807617,
"loss": 1.528,
"nll_loss": 1.52796471118927,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17489634454250336,
"rewards/margins": 1.5176876783370972,
"rewards/rejected": -1.6925840377807617,
"step": 435
},
{
"epoch": 1.6452830188679246,
"grad_norm": 0.3647741973400116,
"learning_rate": 8.867924528301887e-06,
"log_odds_chosen": 18.231021881103516,
"log_odds_ratio": -1.4775437193748076e-05,
"logits/chosen": -1.7286620140075684,
"logits/rejected": -3.7592124938964844,
"logps/chosen": -1.3811477422714233,
"logps/rejected": -19.279529571533203,
"loss": 1.5674,
"nll_loss": 1.5673856735229492,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1381147801876068,
"rewards/margins": 1.7898383140563965,
"rewards/rejected": -1.9279532432556152,
"step": 436
},
{
"epoch": 1.6490566037735848,
"grad_norm": 0.3980993330478668,
"learning_rate": 8.773584905660376e-06,
"log_odds_chosen": 18.004159927368164,
"log_odds_ratio": -9.220154606737196e-05,
"logits/chosen": -0.7007821202278137,
"logits/rejected": -4.726859092712402,
"logps/chosen": -1.5681556463241577,
"logps/rejected": -19.279773712158203,
"loss": 1.4785,
"nll_loss": 1.478508472442627,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1568155735731125,
"rewards/margins": 1.7711620330810547,
"rewards/rejected": -1.9279775619506836,
"step": 437
},
{
"epoch": 1.6528301886792454,
"grad_norm": 0.38771992921829224,
"learning_rate": 8.679245283018868e-06,
"log_odds_chosen": 15.956310272216797,
"log_odds_ratio": -0.08997757732868195,
"logits/chosen": -2.8503222465515137,
"logits/rejected": -6.6901421546936035,
"logps/chosen": -1.7365132570266724,
"logps/rejected": -17.479639053344727,
"loss": 1.4619,
"nll_loss": 1.4529235363006592,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.173651322722435,
"rewards/margins": 1.574312448501587,
"rewards/rejected": -1.747963786125183,
"step": 438
},
{
"epoch": 1.6566037735849055,
"grad_norm": 0.40687695145606995,
"learning_rate": 8.584905660377359e-06,
"log_odds_chosen": 17.718862533569336,
"log_odds_ratio": -3.934121923521161e-05,
"logits/chosen": -1.7964673042297363,
"logits/rejected": -5.379039764404297,
"logps/chosen": -1.6427710056304932,
"logps/rejected": -19.057796478271484,
"loss": 1.3497,
"nll_loss": 1.349698781967163,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1642771065235138,
"rewards/margins": 1.7415024042129517,
"rewards/rejected": -1.9057796001434326,
"step": 439
},
{
"epoch": 1.6603773584905661,
"grad_norm": 0.41751906275749207,
"learning_rate": 8.49056603773585e-06,
"log_odds_chosen": 17.8037109375,
"log_odds_ratio": -0.0002321999636478722,
"logits/chosen": -1.1646761894226074,
"logits/rejected": -5.253706455230713,
"logps/chosen": -1.5499441623687744,
"logps/rejected": -18.97268295288086,
"loss": 1.4037,
"nll_loss": 1.403627872467041,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1549944281578064,
"rewards/margins": 1.7422740459442139,
"rewards/rejected": -1.897268533706665,
"step": 440
},
{
"epoch": 1.6641509433962263,
"grad_norm": 0.3891851007938385,
"learning_rate": 8.396226415094339e-06,
"log_odds_chosen": 15.996545791625977,
"log_odds_ratio": -0.0381770022213459,
"logits/chosen": -0.9191466569900513,
"logits/rejected": -4.70245361328125,
"logps/chosen": -1.4315601587295532,
"logps/rejected": -17.08255386352539,
"loss": 1.4039,
"nll_loss": 1.4000357389450073,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1431560218334198,
"rewards/margins": 1.5650992393493652,
"rewards/rejected": -1.7082552909851074,
"step": 441
},
{
"epoch": 1.6679245283018869,
"grad_norm": 0.3916653096675873,
"learning_rate": 8.30188679245283e-06,
"log_odds_chosen": 16.95732879638672,
"log_odds_ratio": -0.040689367800951004,
"logits/chosen": -1.9281044006347656,
"logits/rejected": -6.584669589996338,
"logps/chosen": -1.4779269695281982,
"logps/rejected": -18.117380142211914,
"loss": 1.3331,
"nll_loss": 1.3289951086044312,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14779271185398102,
"rewards/margins": 1.6639453172683716,
"rewards/rejected": -1.8117380142211914,
"step": 442
},
{
"epoch": 1.671698113207547,
"grad_norm": 0.4993988275527954,
"learning_rate": 8.207547169811321e-06,
"log_odds_chosen": 12.434319496154785,
"log_odds_ratio": -0.25445252656936646,
"logits/chosen": -1.8510253429412842,
"logits/rejected": -4.986669540405273,
"logps/chosen": -1.862165927886963,
"logps/rejected": -14.14360237121582,
"loss": 1.7,
"nll_loss": 1.674567461013794,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1862165927886963,
"rewards/margins": 1.2281436920166016,
"rewards/rejected": -1.4143602848052979,
"step": 443
},
{
"epoch": 1.6754716981132076,
"grad_norm": 0.37089723348617554,
"learning_rate": 8.11320754716981e-06,
"log_odds_chosen": 18.173118591308594,
"log_odds_ratio": -1.9893154785677325e-06,
"logits/chosen": -2.578376054763794,
"logits/rejected": -6.764125823974609,
"logps/chosen": -1.4696475267410278,
"logps/rejected": -19.300800323486328,
"loss": 1.4572,
"nll_loss": 1.4572360515594482,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14696475863456726,
"rewards/margins": 1.783115267753601,
"rewards/rejected": -1.9300800561904907,
"step": 444
},
{
"epoch": 1.6792452830188678,
"grad_norm": 0.35598012804985046,
"learning_rate": 8.018867924528302e-06,
"log_odds_chosen": 16.349313735961914,
"log_odds_ratio": -4.495953908190131e-05,
"logits/chosen": -1.1268545389175415,
"logits/rejected": -6.885201454162598,
"logps/chosen": -1.6435253620147705,
"logps/rejected": -17.746475219726562,
"loss": 1.5433,
"nll_loss": 1.5432648658752441,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16435253620147705,
"rewards/margins": 1.610295057296753,
"rewards/rejected": -1.77464759349823,
"step": 445
},
{
"epoch": 1.6830188679245284,
"grad_norm": 0.413074791431427,
"learning_rate": 7.924528301886793e-06,
"log_odds_chosen": 17.380868911743164,
"log_odds_ratio": -0.023090748116374016,
"logits/chosen": -0.9112629890441895,
"logits/rejected": -4.611825942993164,
"logps/chosen": -1.3334120512008667,
"logps/rejected": -18.372407913208008,
"loss": 1.3489,
"nll_loss": 1.3465766906738281,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13334119319915771,
"rewards/margins": 1.703899621963501,
"rewards/rejected": -1.8372408151626587,
"step": 446
},
{
"epoch": 1.6867924528301885,
"grad_norm": 0.3995283246040344,
"learning_rate": 7.830188679245282e-06,
"log_odds_chosen": 19.585247039794922,
"log_odds_ratio": -8.114101547107566e-06,
"logits/chosen": -0.4880484342575073,
"logits/rejected": -4.196340084075928,
"logps/chosen": -1.2747726440429688,
"logps/rejected": -20.202425003051758,
"loss": 1.3193,
"nll_loss": 1.3193085193634033,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1274772584438324,
"rewards/margins": 1.8927652835845947,
"rewards/rejected": -2.020242691040039,
"step": 447
},
{
"epoch": 1.6905660377358491,
"grad_norm": 0.3784193992614746,
"learning_rate": 7.735849056603773e-06,
"log_odds_chosen": 18.052352905273438,
"log_odds_ratio": -3.911668954970082e-06,
"logits/chosen": -1.333806037902832,
"logits/rejected": -6.3108344078063965,
"logps/chosen": -1.7495572566986084,
"logps/rejected": -19.464826583862305,
"loss": 1.5705,
"nll_loss": 1.5704904794692993,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17495572566986084,
"rewards/margins": 1.7715270519256592,
"rewards/rejected": -1.94648277759552,
"step": 448
},
{
"epoch": 1.6943396226415093,
"grad_norm": 0.37193769216537476,
"learning_rate": 7.641509433962264e-06,
"log_odds_chosen": 15.382280349731445,
"log_odds_ratio": -0.00010735265095718205,
"logits/chosen": -1.019675374031067,
"logits/rejected": -5.034616470336914,
"logps/chosen": -1.6141021251678467,
"logps/rejected": -16.724472045898438,
"loss": 1.5671,
"nll_loss": 1.5670870542526245,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16141021251678467,
"rewards/margins": 1.511036992073059,
"rewards/rejected": -1.6724472045898438,
"step": 449
},
{
"epoch": 1.6981132075471699,
"grad_norm": 0.3708736300468445,
"learning_rate": 7.547169811320755e-06,
"log_odds_chosen": 17.476974487304688,
"log_odds_ratio": -0.018685974180698395,
"logits/chosen": -1.8253635168075562,
"logits/rejected": -5.652876853942871,
"logps/chosen": -1.1619367599487305,
"logps/rejected": -18.148921966552734,
"loss": 1.23,
"nll_loss": 1.228140115737915,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.11619366705417633,
"rewards/margins": 1.69869863986969,
"rewards/rejected": -1.814892292022705,
"step": 450
},
{
"epoch": 1.70188679245283,
"grad_norm": 0.4171108603477478,
"learning_rate": 7.452830188679246e-06,
"log_odds_chosen": 18.037120819091797,
"log_odds_ratio": -4.321344704294461e-07,
"logits/chosen": -2.2186176776885986,
"logits/rejected": -7.5618438720703125,
"logps/chosen": -1.7550156116485596,
"logps/rejected": -19.57200050354004,
"loss": 1.4835,
"nll_loss": 1.4834994077682495,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.17550158500671387,
"rewards/margins": 1.7816987037658691,
"rewards/rejected": -1.957200050354004,
"step": 451
},
{
"epoch": 1.7056603773584906,
"grad_norm": 0.40649089217185974,
"learning_rate": 7.358490566037736e-06,
"log_odds_chosen": 17.768571853637695,
"log_odds_ratio": -3.509271664370317e-06,
"logits/chosen": -1.5206046104431152,
"logits/rejected": -6.697060585021973,
"logps/chosen": -1.5520222187042236,
"logps/rejected": -19.050880432128906,
"loss": 1.4503,
"nll_loss": 1.4502997398376465,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1552022248506546,
"rewards/margins": 1.7498859167099,
"rewards/rejected": -1.9050880670547485,
"step": 452
},
{
"epoch": 1.7094339622641508,
"grad_norm": 0.379649817943573,
"learning_rate": 7.264150943396226e-06,
"log_odds_chosen": 18.14077377319336,
"log_odds_ratio": -0.002329126000404358,
"logits/chosen": -1.208418846130371,
"logits/rejected": -4.804601669311523,
"logps/chosen": -1.673492431640625,
"logps/rejected": -19.51561737060547,
"loss": 1.5746,
"nll_loss": 1.5743178129196167,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16734924912452698,
"rewards/margins": 1.7842124700546265,
"rewards/rejected": -1.9515619277954102,
"step": 453
},
{
"epoch": 1.7132075471698114,
"grad_norm": 0.37834683060646057,
"learning_rate": 7.169811320754717e-06,
"log_odds_chosen": 17.196395874023438,
"log_odds_ratio": -8.45340036903508e-05,
"logits/chosen": -1.6690289974212646,
"logits/rejected": -5.095861434936523,
"logps/chosen": -1.4087400436401367,
"logps/rejected": -18.211074829101562,
"loss": 1.4106,
"nll_loss": 1.410545825958252,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1408740133047104,
"rewards/margins": 1.6802334785461426,
"rewards/rejected": -1.8211076259613037,
"step": 454
},
{
"epoch": 1.7169811320754715,
"grad_norm": 0.3911416530609131,
"learning_rate": 7.0754716981132075e-06,
"log_odds_chosen": 14.482208251953125,
"log_odds_ratio": -0.18219685554504395,
"logits/chosen": -1.772078514099121,
"logits/rejected": -4.952263832092285,
"logps/chosen": -1.774475336074829,
"logps/rejected": -16.061283111572266,
"loss": 1.6118,
"nll_loss": 1.5935897827148438,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.17744752764701843,
"rewards/margins": 1.4286808967590332,
"rewards/rejected": -1.6061283349990845,
"step": 455
},
{
"epoch": 1.7207547169811321,
"grad_norm": 0.3706720471382141,
"learning_rate": 6.981132075471699e-06,
"log_odds_chosen": 15.722382545471191,
"log_odds_ratio": -0.003935753367841244,
"logits/chosen": -2.7362916469573975,
"logits/rejected": -6.182605743408203,
"logps/chosen": -1.4580714702606201,
"logps/rejected": -16.79517364501953,
"loss": 1.4383,
"nll_loss": 1.4378933906555176,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.145807147026062,
"rewards/margins": 1.533710241317749,
"rewards/rejected": -1.6795175075531006,
"step": 456
},
{
"epoch": 1.7245283018867923,
"grad_norm": 0.40554314851760864,
"learning_rate": 6.886792452830189e-06,
"log_odds_chosen": 18.29325294494629,
"log_odds_ratio": -8.650338713778183e-06,
"logits/chosen": -0.5920654535293579,
"logits/rejected": -4.844081401824951,
"logps/chosen": -1.4798316955566406,
"logps/rejected": -19.450267791748047,
"loss": 1.2259,
"nll_loss": 1.2258515357971191,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14798316359519958,
"rewards/margins": 1.7970435619354248,
"rewards/rejected": -1.9450266361236572,
"step": 457
},
{
"epoch": 1.728301886792453,
"grad_norm": 0.42811644077301025,
"learning_rate": 6.792452830188679e-06,
"log_odds_chosen": 17.65870475769043,
"log_odds_ratio": -1.3024438885622658e-05,
"logits/chosen": -1.6862386465072632,
"logits/rejected": -4.607665061950684,
"logps/chosen": -1.463801383972168,
"logps/rejected": -18.766820907592773,
"loss": 1.473,
"nll_loss": 1.472968578338623,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14638014137744904,
"rewards/margins": 1.7303019762039185,
"rewards/rejected": -1.8766820430755615,
"step": 458
},
{
"epoch": 1.732075471698113,
"grad_norm": 0.34931331872940063,
"learning_rate": 6.69811320754717e-06,
"log_odds_chosen": 18.853530883789062,
"log_odds_ratio": -1.6465862699988065e-06,
"logits/chosen": -1.1972850561141968,
"logits/rejected": -5.578787326812744,
"logps/chosen": -1.192124366760254,
"logps/rejected": -19.51374626159668,
"loss": 1.2305,
"nll_loss": 1.2305028438568115,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.11921243369579315,
"rewards/margins": 1.8321621417999268,
"rewards/rejected": -1.9513745307922363,
"step": 459
},
{
"epoch": 1.7358490566037736,
"grad_norm": 0.38165482878685,
"learning_rate": 6.60377358490566e-06,
"log_odds_chosen": 17.29033851623535,
"log_odds_ratio": -0.003331738291308284,
"logits/chosen": -0.4877992570400238,
"logits/rejected": -6.1247406005859375,
"logps/chosen": -1.6295177936553955,
"logps/rejected": -18.68689727783203,
"loss": 1.559,
"nll_loss": 1.5586189031600952,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1629517823457718,
"rewards/margins": 1.7057380676269531,
"rewards/rejected": -1.8686898946762085,
"step": 460
},
{
"epoch": 1.7396226415094338,
"grad_norm": 0.372586727142334,
"learning_rate": 6.5094339622641515e-06,
"log_odds_chosen": 16.901809692382812,
"log_odds_ratio": -0.0549037791788578,
"logits/chosen": -0.6773020625114441,
"logits/rejected": -4.9583740234375,
"logps/chosen": -1.572051763534546,
"logps/rejected": -18.240394592285156,
"loss": 1.5153,
"nll_loss": 1.509831428527832,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15720517933368683,
"rewards/margins": 1.6668341159820557,
"rewards/rejected": -1.8240394592285156,
"step": 461
},
{
"epoch": 1.7433962264150944,
"grad_norm": 0.3676230013370514,
"learning_rate": 6.415094339622642e-06,
"log_odds_chosen": 17.052555084228516,
"log_odds_ratio": -0.0029449001885950565,
"logits/chosen": -1.5178229808807373,
"logits/rejected": -6.080841541290283,
"logps/chosen": -1.308215618133545,
"logps/rejected": -17.94374656677246,
"loss": 1.4127,
"nll_loss": 1.4124376773834229,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1308215707540512,
"rewards/margins": 1.663553237915039,
"rewards/rejected": -1.794374704360962,
"step": 462
},
{
"epoch": 1.7471698113207546,
"grad_norm": 0.3529452085494995,
"learning_rate": 6.320754716981132e-06,
"log_odds_chosen": 17.932321548461914,
"log_odds_ratio": -0.034649647772312164,
"logits/chosen": -0.5845901966094971,
"logits/rejected": -5.674325466156006,
"logps/chosen": -1.6330788135528564,
"logps/rejected": -19.286094665527344,
"loss": 1.5689,
"nll_loss": 1.5654515027999878,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16330787539482117,
"rewards/margins": 1.7653017044067383,
"rewards/rejected": -1.9286096096038818,
"step": 463
},
{
"epoch": 1.7509433962264151,
"grad_norm": 0.5019151568412781,
"learning_rate": 6.226415094339623e-06,
"log_odds_chosen": 14.161150932312012,
"log_odds_ratio": -0.0002686173829715699,
"logits/chosen": -1.4875158071517944,
"logits/rejected": -5.882111549377441,
"logps/chosen": -1.6882565021514893,
"logps/rejected": -15.569530487060547,
"loss": 1.6722,
"nll_loss": 1.6721980571746826,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1688256412744522,
"rewards/margins": 1.388127326965332,
"rewards/rejected": -1.556952953338623,
"step": 464
},
{
"epoch": 1.7547169811320755,
"grad_norm": 0.32335349917411804,
"learning_rate": 6.132075471698113e-06,
"log_odds_chosen": 19.670560836791992,
"log_odds_ratio": -9.08325964701362e-05,
"logits/chosen": -1.4163541793823242,
"logits/rejected": -6.159734725952148,
"logps/chosen": -1.548288106918335,
"logps/rejected": -20.88932991027832,
"loss": 1.4723,
"nll_loss": 1.4722576141357422,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15482881665229797,
"rewards/margins": 1.9341042041778564,
"rewards/rejected": -2.088933229446411,
"step": 465
},
{
"epoch": 1.758490566037736,
"grad_norm": 0.4526292681694031,
"learning_rate": 6.037735849056604e-06,
"log_odds_chosen": 15.761448860168457,
"log_odds_ratio": -0.07157308608293533,
"logits/chosen": -2.5339674949645996,
"logits/rejected": -6.691962242126465,
"logps/chosen": -1.6193218231201172,
"logps/rejected": -17.078277587890625,
"loss": 1.5394,
"nll_loss": 1.5322078466415405,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16193217039108276,
"rewards/margins": 1.5458955764770508,
"rewards/rejected": -1.7078275680541992,
"step": 466
},
{
"epoch": 1.7622641509433963,
"grad_norm": 0.3781883120536804,
"learning_rate": 5.943396226415095e-06,
"log_odds_chosen": 17.7440128326416,
"log_odds_ratio": -2.7567359666136326e-06,
"logits/chosen": -0.3988521695137024,
"logits/rejected": -3.945779800415039,
"logps/chosen": -1.4811797142028809,
"logps/rejected": -18.924360275268555,
"loss": 1.2401,
"nll_loss": 1.2400920391082764,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14811795949935913,
"rewards/margins": 1.7443180084228516,
"rewards/rejected": -1.8924360275268555,
"step": 467
},
{
"epoch": 1.7660377358490567,
"grad_norm": 0.33292269706726074,
"learning_rate": 5.849056603773585e-06,
"log_odds_chosen": 17.42066192626953,
"log_odds_ratio": -0.016988694667816162,
"logits/chosen": -0.5932101011276245,
"logits/rejected": -6.109551429748535,
"logps/chosen": -1.484053611755371,
"logps/rejected": -18.59859275817871,
"loss": 1.4429,
"nll_loss": 1.4412463903427124,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14840535819530487,
"rewards/margins": 1.711454153060913,
"rewards/rejected": -1.8598594665527344,
"step": 468
},
{
"epoch": 1.769811320754717,
"grad_norm": 0.33539214730262756,
"learning_rate": 5.754716981132076e-06,
"log_odds_chosen": 18.23859977722168,
"log_odds_ratio": -0.058486614376306534,
"logits/chosen": -0.5683273077011108,
"logits/rejected": -6.111762046813965,
"logps/chosen": -1.6735715866088867,
"logps/rejected": -19.64827537536621,
"loss": 1.5931,
"nll_loss": 1.5872840881347656,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1673571765422821,
"rewards/margins": 1.7974703311920166,
"rewards/rejected": -1.9648274183273315,
"step": 469
},
{
"epoch": 1.7735849056603774,
"grad_norm": 0.3669726252555847,
"learning_rate": 5.660377358490566e-06,
"log_odds_chosen": 16.35857391357422,
"log_odds_ratio": -0.0035181858111172915,
"logits/chosen": -2.055553913116455,
"logits/rejected": -6.248283386230469,
"logps/chosen": -1.3732891082763672,
"logps/rejected": -17.36093521118164,
"loss": 1.3191,
"nll_loss": 1.318752408027649,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13732890784740448,
"rewards/margins": 1.5987646579742432,
"rewards/rejected": -1.736093521118164,
"step": 470
},
{
"epoch": 1.7773584905660378,
"grad_norm": 0.3617085814476013,
"learning_rate": 5.566037735849057e-06,
"log_odds_chosen": 16.787891387939453,
"log_odds_ratio": -9.142341696133371e-06,
"logits/chosen": -1.4523394107818604,
"logits/rejected": -7.092175006866455,
"logps/chosen": -1.681269884109497,
"logps/rejected": -18.210691452026367,
"loss": 1.5766,
"nll_loss": 1.5765758752822876,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16812700033187866,
"rewards/margins": 1.652942180633545,
"rewards/rejected": -1.8210691213607788,
"step": 471
},
{
"epoch": 1.7811320754716982,
"grad_norm": 0.3479359447956085,
"learning_rate": 5.4716981132075475e-06,
"log_odds_chosen": 16.96417808532715,
"log_odds_ratio": -0.00023999404220376164,
"logits/chosen": -1.1192617416381836,
"logits/rejected": -5.284356117248535,
"logps/chosen": -1.3302626609802246,
"logps/rejected": -17.81631088256836,
"loss": 1.3527,
"nll_loss": 1.3526952266693115,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13302627205848694,
"rewards/margins": 1.6486049890518188,
"rewards/rejected": -1.7816312313079834,
"step": 472
},
{
"epoch": 1.7849056603773585,
"grad_norm": 0.41884541511535645,
"learning_rate": 5.377358490566039e-06,
"log_odds_chosen": 18.43604278564453,
"log_odds_ratio": -3.442222805460915e-06,
"logits/chosen": -1.2943966388702393,
"logits/rejected": -5.607792377471924,
"logps/chosen": -1.663355827331543,
"logps/rejected": -19.86046600341797,
"loss": 1.4899,
"nll_loss": 1.4899128675460815,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1663355678319931,
"rewards/margins": 1.8197109699249268,
"rewards/rejected": -1.986046552658081,
"step": 473
},
{
"epoch": 1.788679245283019,
"grad_norm": 0.39448869228363037,
"learning_rate": 5.283018867924529e-06,
"log_odds_chosen": 16.650920867919922,
"log_odds_ratio": -0.0011353573063388467,
"logits/chosen": -3.2340540885925293,
"logits/rejected": -6.999566078186035,
"logps/chosen": -1.4379664659500122,
"logps/rejected": -17.618440628051758,
"loss": 1.4977,
"nll_loss": 1.4976294040679932,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1437966525554657,
"rewards/margins": 1.6180473566055298,
"rewards/rejected": -1.7618440389633179,
"step": 474
},
{
"epoch": 1.7924528301886793,
"grad_norm": 0.4289974570274353,
"learning_rate": 5.188679245283019e-06,
"log_odds_chosen": 17.761314392089844,
"log_odds_ratio": -0.03405510634183884,
"logits/chosen": -1.8382487297058105,
"logits/rejected": -6.273689270019531,
"logps/chosen": -1.5025469064712524,
"logps/rejected": -18.807750701904297,
"loss": 1.3993,
"nll_loss": 1.3959349393844604,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15025469660758972,
"rewards/margins": 1.7305206060409546,
"rewards/rejected": -1.8807752132415771,
"step": 475
},
{
"epoch": 1.7962264150943397,
"grad_norm": 0.3701404333114624,
"learning_rate": 5.09433962264151e-06,
"log_odds_chosen": 18.288116455078125,
"log_odds_ratio": -0.08911100029945374,
"logits/chosen": -2.354094982147217,
"logits/rejected": -4.444386959075928,
"logps/chosen": -1.5296881198883057,
"logps/rejected": -19.461259841918945,
"loss": 1.4101,
"nll_loss": 1.401210069656372,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.15296879410743713,
"rewards/margins": 1.7931572198867798,
"rewards/rejected": -1.946125864982605,
"step": 476
},
{
"epoch": 1.8,
"grad_norm": 0.3996965289115906,
"learning_rate": 5e-06,
"log_odds_chosen": 18.797988891601562,
"log_odds_ratio": -3.426304101594724e-05,
"logits/chosen": -0.7937726378440857,
"logits/rejected": -4.634369373321533,
"logps/chosen": -1.5706990957260132,
"logps/rejected": -20.071020126342773,
"loss": 1.4157,
"nll_loss": 1.4157121181488037,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15706990659236908,
"rewards/margins": 1.850032091140747,
"rewards/rejected": -2.0071020126342773,
"step": 477
},
{
"epoch": 1.8037735849056604,
"grad_norm": 0.35529616475105286,
"learning_rate": 4.9056603773584915e-06,
"log_odds_chosen": 18.047014236450195,
"log_odds_ratio": -0.0028157131746411324,
"logits/chosen": -2.256852149963379,
"logits/rejected": -5.895351886749268,
"logps/chosen": -1.6656599044799805,
"logps/rejected": -19.412460327148438,
"loss": 1.448,
"nll_loss": 1.4476892948150635,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16656598448753357,
"rewards/margins": 1.7746801376342773,
"rewards/rejected": -1.9412461519241333,
"step": 478
},
{
"epoch": 1.8075471698113208,
"grad_norm": 0.4262011647224426,
"learning_rate": 4.811320754716982e-06,
"log_odds_chosen": 18.934986114501953,
"log_odds_ratio": -2.272450956297689e-06,
"logits/chosen": -3.147275924682617,
"logits/rejected": -7.059483528137207,
"logps/chosen": -1.4018633365631104,
"logps/rejected": -19.886085510253906,
"loss": 1.3952,
"nll_loss": 1.395203948020935,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14018632471561432,
"rewards/margins": 1.8484222888946533,
"rewards/rejected": -1.988608479499817,
"step": 479
},
{
"epoch": 1.8113207547169812,
"grad_norm": 0.47693830728530884,
"learning_rate": 4.716981132075472e-06,
"log_odds_chosen": 19.023094177246094,
"log_odds_ratio": -9.76030378296855e-07,
"logits/chosen": -2.5737247467041016,
"logits/rejected": -6.724173545837402,
"logps/chosen": -1.2855720520019531,
"logps/rejected": -19.879941940307617,
"loss": 1.3567,
"nll_loss": 1.3567023277282715,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1285572052001953,
"rewards/margins": 1.8594369888305664,
"rewards/rejected": -1.9879940748214722,
"step": 480
},
{
"epoch": 1.8150943396226416,
"grad_norm": 0.33889421820640564,
"learning_rate": 4.622641509433962e-06,
"log_odds_chosen": 18.900848388671875,
"log_odds_ratio": -2.5331991082566674e-07,
"logits/chosen": -1.0899468660354614,
"logits/rejected": -6.098740100860596,
"logps/chosen": -1.4805465936660767,
"logps/rejected": -20.047718048095703,
"loss": 1.3192,
"nll_loss": 1.3191674947738647,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14805465936660767,
"rewards/margins": 1.8567172288894653,
"rewards/rejected": -2.0047719478607178,
"step": 481
},
{
"epoch": 1.818867924528302,
"grad_norm": 0.4076652526855469,
"learning_rate": 4.5283018867924524e-06,
"log_odds_chosen": 17.234874725341797,
"log_odds_ratio": -2.1215018932707608e-05,
"logits/chosen": -0.29048505425453186,
"logits/rejected": -5.432765007019043,
"logps/chosen": -1.6638712882995605,
"logps/rejected": -18.638235092163086,
"loss": 1.631,
"nll_loss": 1.631001591682434,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.166387140750885,
"rewards/margins": 1.6974364519119263,
"rewards/rejected": -1.863823413848877,
"step": 482
},
{
"epoch": 1.8226415094339623,
"grad_norm": 0.35861098766326904,
"learning_rate": 4.4339622641509435e-06,
"log_odds_chosen": 18.069866180419922,
"log_odds_ratio": -2.9603910661535338e-05,
"logits/chosen": -1.9246575832366943,
"logits/rejected": -4.993512153625488,
"logps/chosen": -1.2558917999267578,
"logps/rejected": -18.758874893188477,
"loss": 1.5623,
"nll_loss": 1.5623277425765991,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.12558917701244354,
"rewards/margins": 1.750298261642456,
"rewards/rejected": -1.8758875131607056,
"step": 483
},
{
"epoch": 1.8264150943396227,
"grad_norm": 0.374891459941864,
"learning_rate": 4.339622641509434e-06,
"log_odds_chosen": 17.780948638916016,
"log_odds_ratio": -7.957565685501322e-06,
"logits/chosen": -2.0383100509643555,
"logits/rejected": -5.80636739730835,
"logps/chosen": -1.346254825592041,
"logps/rejected": -18.59038543701172,
"loss": 1.2654,
"nll_loss": 1.2653898000717163,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13462547957897186,
"rewards/margins": 1.7244129180908203,
"rewards/rejected": -1.8590385913848877,
"step": 484
},
{
"epoch": 1.830188679245283,
"grad_norm": 0.37461966276168823,
"learning_rate": 4.245283018867925e-06,
"log_odds_chosen": 16.527099609375,
"log_odds_ratio": -0.003529702080413699,
"logits/chosen": -3.2636947631835938,
"logits/rejected": -5.3443403244018555,
"logps/chosen": -1.5424690246582031,
"logps/rejected": -17.72702407836914,
"loss": 1.532,
"nll_loss": 1.5316277742385864,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15424689650535583,
"rewards/margins": 1.6184556484222412,
"rewards/rejected": -1.7727024555206299,
"step": 485
},
{
"epoch": 1.8339622641509434,
"grad_norm": 0.42353299260139465,
"learning_rate": 4.150943396226415e-06,
"log_odds_chosen": 17.351669311523438,
"log_odds_ratio": -0.02633853070437908,
"logits/chosen": -1.5371365547180176,
"logits/rejected": -6.403836250305176,
"logps/chosen": -1.605021357536316,
"logps/rejected": -18.7138671875,
"loss": 1.4473,
"nll_loss": 1.4446651935577393,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1605021357536316,
"rewards/margins": 1.7108845710754395,
"rewards/rejected": -1.8713867664337158,
"step": 486
},
{
"epoch": 1.8377358490566038,
"grad_norm": 0.34952792525291443,
"learning_rate": 4.056603773584905e-06,
"log_odds_chosen": 15.732725143432617,
"log_odds_ratio": -0.01928078942000866,
"logits/chosen": -3.1891517639160156,
"logits/rejected": -6.666206359863281,
"logps/chosen": -1.8744854927062988,
"logps/rejected": -17.41217041015625,
"loss": 1.664,
"nll_loss": 1.6620932817459106,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18744854629039764,
"rewards/margins": 1.553768515586853,
"rewards/rejected": -1.741217017173767,
"step": 487
},
{
"epoch": 1.8415094339622642,
"grad_norm": 0.36892521381378174,
"learning_rate": 3.962264150943396e-06,
"log_odds_chosen": 19.160884857177734,
"log_odds_ratio": -9.086857608053833e-05,
"logits/chosen": -3.0095064640045166,
"logits/rejected": -5.858531475067139,
"logps/chosen": -1.4025964736938477,
"logps/rejected": -20.25691032409668,
"loss": 1.5139,
"nll_loss": 1.513940691947937,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14025965332984924,
"rewards/margins": 1.8854316473007202,
"rewards/rejected": -2.025691270828247,
"step": 488
},
{
"epoch": 1.8452830188679246,
"grad_norm": 0.4228150546550751,
"learning_rate": 3.867924528301887e-06,
"log_odds_chosen": 12.711286544799805,
"log_odds_ratio": -0.03767332434654236,
"logits/chosen": -0.7741556763648987,
"logits/rejected": -4.395710468292236,
"logps/chosen": -1.6399070024490356,
"logps/rejected": -14.126274108886719,
"loss": 1.4961,
"nll_loss": 1.492356538772583,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16399070620536804,
"rewards/margins": 1.2486367225646973,
"rewards/rejected": -1.4126274585723877,
"step": 489
},
{
"epoch": 1.849056603773585,
"grad_norm": 0.3612087666988373,
"learning_rate": 3.7735849056603773e-06,
"log_odds_chosen": 20.139719009399414,
"log_odds_ratio": -2.384188633186568e-07,
"logits/chosen": -1.845539927482605,
"logits/rejected": -6.139976978302002,
"logps/chosen": -1.3304532766342163,
"logps/rejected": -21.026758193969727,
"loss": 1.3543,
"nll_loss": 1.3543174266815186,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13304531574249268,
"rewards/margins": 1.969630479812622,
"rewards/rejected": -2.1026759147644043,
"step": 490
},
{
"epoch": 1.8528301886792453,
"grad_norm": 0.33791664242744446,
"learning_rate": 3.679245283018868e-06,
"log_odds_chosen": 17.015119552612305,
"log_odds_ratio": -0.00011756140156649053,
"logits/chosen": -0.6114660501480103,
"logits/rejected": -5.934682369232178,
"logps/chosen": -1.4735393524169922,
"logps/rejected": -18.142919540405273,
"loss": 1.493,
"nll_loss": 1.4929838180541992,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14735394716262817,
"rewards/margins": 1.666938066482544,
"rewards/rejected": -1.814292073249817,
"step": 491
},
{
"epoch": 1.8566037735849057,
"grad_norm": 0.42551860213279724,
"learning_rate": 3.5849056603773586e-06,
"log_odds_chosen": 15.526764869689941,
"log_odds_ratio": -0.06459427624940872,
"logits/chosen": -0.3648894727230072,
"logits/rejected": -5.722745418548584,
"logps/chosen": -1.61395263671875,
"logps/rejected": -16.911762237548828,
"loss": 1.4662,
"nll_loss": 1.4597229957580566,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16139525175094604,
"rewards/margins": 1.5297811031341553,
"rewards/rejected": -1.691176414489746,
"step": 492
},
{
"epoch": 1.860377358490566,
"grad_norm": 0.3618764579296112,
"learning_rate": 3.4905660377358493e-06,
"log_odds_chosen": 17.239789962768555,
"log_odds_ratio": -1.4826684946456226e-06,
"logits/chosen": -0.21888774633407593,
"logits/rejected": -4.190219879150391,
"logps/chosen": -1.3131389617919922,
"logps/rejected": -18.119144439697266,
"loss": 1.271,
"nll_loss": 1.270983099937439,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13131389021873474,
"rewards/margins": 1.6806005239486694,
"rewards/rejected": -1.8119144439697266,
"step": 493
},
{
"epoch": 1.8641509433962264,
"grad_norm": 0.4203263223171234,
"learning_rate": 3.3962264150943395e-06,
"log_odds_chosen": 17.666906356811523,
"log_odds_ratio": -4.238808833179064e-05,
"logits/chosen": -2.081303119659424,
"logits/rejected": -6.445936679840088,
"logps/chosen": -1.6142175197601318,
"logps/rejected": -18.978389739990234,
"loss": 1.6097,
"nll_loss": 1.6096843481063843,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1614217460155487,
"rewards/margins": 1.736417293548584,
"rewards/rejected": -1.897839069366455,
"step": 494
},
{
"epoch": 1.8679245283018868,
"grad_norm": 0.32720765471458435,
"learning_rate": 3.30188679245283e-06,
"log_odds_chosen": 18.920677185058594,
"log_odds_ratio": -1.1772006018873071e-06,
"logits/chosen": 0.03428873419761658,
"logits/rejected": -4.708007335662842,
"logps/chosen": -1.2015572786331177,
"logps/rejected": -19.660449981689453,
"loss": 1.3305,
"nll_loss": 1.3305293321609497,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.12015574425458908,
"rewards/margins": 1.8458893299102783,
"rewards/rejected": -1.9660451412200928,
"step": 495
},
{
"epoch": 1.8716981132075472,
"grad_norm": 0.4030371308326721,
"learning_rate": 3.207547169811321e-06,
"log_odds_chosen": 17.407262802124023,
"log_odds_ratio": -9.635530295781791e-05,
"logits/chosen": -1.2139384746551514,
"logits/rejected": -5.052234172821045,
"logps/chosen": -1.58628511428833,
"logps/rejected": -18.708999633789062,
"loss": 1.4232,
"nll_loss": 1.4231928586959839,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15862850844860077,
"rewards/margins": 1.7122715711593628,
"rewards/rejected": -1.87090003490448,
"step": 496
},
{
"epoch": 1.8754716981132076,
"grad_norm": 0.3452160954475403,
"learning_rate": 3.1132075471698115e-06,
"log_odds_chosen": 19.35751724243164,
"log_odds_ratio": -2.719517624427681e-06,
"logits/chosen": -1.754660725593567,
"logits/rejected": -6.674939155578613,
"logps/chosen": -1.4923593997955322,
"logps/rejected": -20.52996826171875,
"loss": 1.3139,
"nll_loss": 1.3138580322265625,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14923594892024994,
"rewards/margins": 1.9037607908248901,
"rewards/rejected": -2.0529966354370117,
"step": 497
},
{
"epoch": 1.879245283018868,
"grad_norm": 0.31435737013816833,
"learning_rate": 3.018867924528302e-06,
"log_odds_chosen": 17.633760452270508,
"log_odds_ratio": -0.07248206436634064,
"logits/chosen": -1.433829665184021,
"logits/rejected": -6.44128942489624,
"logps/chosen": -1.4513802528381348,
"logps/rejected": -18.749958038330078,
"loss": 1.5574,
"nll_loss": 1.5501039028167725,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.14513802528381348,
"rewards/margins": 1.7298578023910522,
"rewards/rejected": -1.8749958276748657,
"step": 498
},
{
"epoch": 1.8830188679245283,
"grad_norm": 0.415445476770401,
"learning_rate": 2.9245283018867924e-06,
"log_odds_chosen": 16.431304931640625,
"log_odds_ratio": -0.002142214449122548,
"logits/chosen": -0.9190161824226379,
"logits/rejected": -5.215124130249023,
"logps/chosen": -1.5988054275512695,
"logps/rejected": -17.76313018798828,
"loss": 1.5031,
"nll_loss": 1.5028951168060303,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15988054871559143,
"rewards/margins": 1.6164326667785645,
"rewards/rejected": -1.776313066482544,
"step": 499
},
{
"epoch": 1.8867924528301887,
"grad_norm": 0.34781375527381897,
"learning_rate": 2.830188679245283e-06,
"log_odds_chosen": 16.619857788085938,
"log_odds_ratio": -0.0440763421356678,
"logits/chosen": -1.902965784072876,
"logits/rejected": -7.038515090942383,
"logps/chosen": -1.5125619173049927,
"logps/rejected": -17.874229431152344,
"loss": 1.45,
"nll_loss": 1.4455456733703613,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15125618875026703,
"rewards/margins": 1.6361666917800903,
"rewards/rejected": -1.7874228954315186,
"step": 500
},
{
"epoch": 1.890566037735849,
"grad_norm": 0.4033926725387573,
"learning_rate": 2.7358490566037738e-06,
"log_odds_chosen": 18.919782638549805,
"log_odds_ratio": -0.04001227393746376,
"logits/chosen": -1.2150888442993164,
"logits/rejected": -4.071748733520508,
"logps/chosen": -1.1845972537994385,
"logps/rejected": -19.4738712310791,
"loss": 1.2712,
"nll_loss": 1.2671959400177002,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.11845972388982773,
"rewards/margins": 1.8289273977279663,
"rewards/rejected": -1.9473869800567627,
"step": 501
},
{
"epoch": 1.8943396226415095,
"grad_norm": 0.4240841865539551,
"learning_rate": 2.6415094339622644e-06,
"log_odds_chosen": 18.493478775024414,
"log_odds_ratio": -5.438935772872355e-07,
"logits/chosen": -1.6389120817184448,
"logits/rejected": -5.454832077026367,
"logps/chosen": -1.660611629486084,
"logps/rejected": -19.89701271057129,
"loss": 1.6124,
"nll_loss": 1.6123967170715332,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1660611629486084,
"rewards/margins": 1.8236401081085205,
"rewards/rejected": -1.989701271057129,
"step": 502
},
{
"epoch": 1.8981132075471698,
"grad_norm": 0.379930317401886,
"learning_rate": 2.547169811320755e-06,
"log_odds_chosen": 18.02582550048828,
"log_odds_ratio": -5.4391039157053456e-06,
"logits/chosen": -0.07657934725284576,
"logits/rejected": -4.238829135894775,
"logps/chosen": -1.2943615913391113,
"logps/rejected": -18.893966674804688,
"loss": 1.2938,
"nll_loss": 1.2937697172164917,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.12943615019321442,
"rewards/margins": 1.759960412979126,
"rewards/rejected": -1.8893966674804688,
"step": 503
},
{
"epoch": 1.9018867924528302,
"grad_norm": 0.4327671527862549,
"learning_rate": 2.4528301886792457e-06,
"log_odds_chosen": 17.240550994873047,
"log_odds_ratio": -0.08509234338998795,
"logits/chosen": -0.8051078915596008,
"logits/rejected": -4.25758695602417,
"logps/chosen": -1.6059653759002686,
"logps/rejected": -18.553573608398438,
"loss": 1.4932,
"nll_loss": 1.4846444129943848,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16059653460979462,
"rewards/margins": 1.694760799407959,
"rewards/rejected": -1.85535728931427,
"step": 504
},
{
"epoch": 1.9056603773584906,
"grad_norm": 0.3554707169532776,
"learning_rate": 2.358490566037736e-06,
"log_odds_chosen": 15.833232879638672,
"log_odds_ratio": -0.008360692299902439,
"logits/chosen": -2.8660082817077637,
"logits/rejected": -5.381879806518555,
"logps/chosen": -1.5323365926742554,
"logps/rejected": -17.071136474609375,
"loss": 1.5256,
"nll_loss": 1.5248125791549683,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15323364734649658,
"rewards/margins": 1.553879976272583,
"rewards/rejected": -1.70711350440979,
"step": 505
},
{
"epoch": 1.909433962264151,
"grad_norm": 0.40750032663345337,
"learning_rate": 2.2641509433962262e-06,
"log_odds_chosen": 18.917707443237305,
"log_odds_ratio": -1.3187664080760442e-06,
"logits/chosen": -2.4501075744628906,
"logits/rejected": -4.550923824310303,
"logps/chosen": -1.6083648204803467,
"logps/rejected": -20.180192947387695,
"loss": 1.422,
"nll_loss": 1.4219931364059448,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16083648800849915,
"rewards/margins": 1.8571827411651611,
"rewards/rejected": -2.018019199371338,
"step": 506
},
{
"epoch": 1.9132075471698113,
"grad_norm": 0.3912452459335327,
"learning_rate": 2.169811320754717e-06,
"log_odds_chosen": 16.282840728759766,
"log_odds_ratio": -0.0001460638886783272,
"logits/chosen": -1.1887915134429932,
"logits/rejected": -5.86929988861084,
"logps/chosen": -1.3933725357055664,
"logps/rejected": -17.320636749267578,
"loss": 1.3639,
"nll_loss": 1.3638955354690552,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13933727145195007,
"rewards/margins": 1.5927265882492065,
"rewards/rejected": -1.7320637702941895,
"step": 507
},
{
"epoch": 1.9169811320754717,
"grad_norm": 0.3855321705341339,
"learning_rate": 2.0754716981132075e-06,
"log_odds_chosen": 17.623600006103516,
"log_odds_ratio": -4.701381840277463e-06,
"logits/chosen": -0.25701072812080383,
"logits/rejected": -5.022630214691162,
"logps/chosen": -1.5275490283966064,
"logps/rejected": -18.818309783935547,
"loss": 1.3883,
"nll_loss": 1.3883423805236816,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15275491774082184,
"rewards/margins": 1.7290761470794678,
"rewards/rejected": -1.8818310499191284,
"step": 508
},
{
"epoch": 1.920754716981132,
"grad_norm": 0.4456048607826233,
"learning_rate": 1.981132075471698e-06,
"log_odds_chosen": 19.397144317626953,
"log_odds_ratio": -0.08647830039262772,
"logits/chosen": -2.1884102821350098,
"logits/rejected": -4.864468574523926,
"logps/chosen": -1.1446378231048584,
"logps/rejected": -19.82569694519043,
"loss": 1.1951,
"nll_loss": 1.1864542961120605,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.11446378380060196,
"rewards/margins": 1.8681057691574097,
"rewards/rejected": -1.982569694519043,
"step": 509
},
{
"epoch": 1.9245283018867925,
"grad_norm": 0.38309693336486816,
"learning_rate": 1.8867924528301887e-06,
"log_odds_chosen": 17.333568572998047,
"log_odds_ratio": -0.0018029811326414347,
"logits/chosen": -1.732748031616211,
"logits/rejected": -5.647907733917236,
"logps/chosen": -1.6717514991760254,
"logps/rejected": -18.71475601196289,
"loss": 1.6,
"nll_loss": 1.5998656749725342,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16717512905597687,
"rewards/margins": 1.7043005228042603,
"rewards/rejected": -1.8714755773544312,
"step": 510
},
{
"epoch": 1.9283018867924528,
"grad_norm": 0.42033860087394714,
"learning_rate": 1.7924528301886793e-06,
"log_odds_chosen": 18.408260345458984,
"log_odds_ratio": -0.07517862319946289,
"logits/chosen": -1.3036704063415527,
"logits/rejected": -6.073916435241699,
"logps/chosen": -1.500823736190796,
"logps/rejected": -19.579364776611328,
"loss": 1.3745,
"nll_loss": 1.3670070171356201,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.15008236467838287,
"rewards/margins": 1.807854175567627,
"rewards/rejected": -1.957936406135559,
"step": 511
},
{
"epoch": 1.9320754716981132,
"grad_norm": 0.43365299701690674,
"learning_rate": 1.6981132075471698e-06,
"log_odds_chosen": 16.9204158782959,
"log_odds_ratio": -0.010048388503491879,
"logits/chosen": -1.4353740215301514,
"logits/rejected": -4.844537734985352,
"logps/chosen": -1.4958455562591553,
"logps/rejected": -18.103038787841797,
"loss": 1.2225,
"nll_loss": 1.2214566469192505,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14958456158638,
"rewards/margins": 1.6607191562652588,
"rewards/rejected": -1.810303807258606,
"step": 512
},
{
"epoch": 1.9358490566037736,
"grad_norm": 0.41454991698265076,
"learning_rate": 1.6037735849056604e-06,
"log_odds_chosen": 17.356592178344727,
"log_odds_ratio": -0.02275564707815647,
"logits/chosen": -2.385758399963379,
"logits/rejected": -6.4152421951293945,
"logps/chosen": -1.799842119216919,
"logps/rejected": -18.904577255249023,
"loss": 1.6663,
"nll_loss": 1.664039134979248,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1799842119216919,
"rewards/margins": 1.7104735374450684,
"rewards/rejected": -1.8904577493667603,
"step": 513
},
{
"epoch": 1.939622641509434,
"grad_norm": 0.3880457878112793,
"learning_rate": 1.509433962264151e-06,
"log_odds_chosen": 15.238203048706055,
"log_odds_ratio": -0.01757655292749405,
"logits/chosen": -1.1916530132293701,
"logits/rejected": -5.214702606201172,
"logps/chosen": -1.4446446895599365,
"logps/rejected": -16.36115264892578,
"loss": 1.5477,
"nll_loss": 1.5459623336791992,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14446447789669037,
"rewards/margins": 1.4916508197784424,
"rewards/rejected": -1.636115312576294,
"step": 514
},
{
"epoch": 1.9433962264150944,
"grad_norm": 0.47617748379707336,
"learning_rate": 1.4150943396226415e-06,
"log_odds_chosen": 16.34947967529297,
"log_odds_ratio": -0.0004449410189408809,
"logits/chosen": -2.4440126419067383,
"logits/rejected": -5.788022994995117,
"logps/chosen": -1.6996843814849854,
"logps/rejected": -17.788955688476562,
"loss": 1.4768,
"nll_loss": 1.4767930507659912,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16996845602989197,
"rewards/margins": 1.6089271306991577,
"rewards/rejected": -1.778895616531372,
"step": 515
},
{
"epoch": 1.9471698113207547,
"grad_norm": 0.40238529443740845,
"learning_rate": 1.3207547169811322e-06,
"log_odds_chosen": 18.627809524536133,
"log_odds_ratio": -4.4853818508272525e-06,
"logits/chosen": -1.0439683198928833,
"logits/rejected": -5.652813911437988,
"logps/chosen": -1.4960782527923584,
"logps/rejected": -19.751792907714844,
"loss": 1.3031,
"nll_loss": 1.3030521869659424,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1496078073978424,
"rewards/margins": 1.8255715370178223,
"rewards/rejected": -1.9751791954040527,
"step": 516
},
{
"epoch": 1.950943396226415,
"grad_norm": 0.41020315885543823,
"learning_rate": 1.2264150943396229e-06,
"log_odds_chosen": 17.56089973449707,
"log_odds_ratio": -0.049375876784324646,
"logits/chosen": -0.404629647731781,
"logits/rejected": -5.214582443237305,
"logps/chosen": -1.3766056299209595,
"logps/rejected": -18.530118942260742,
"loss": 1.2775,
"nll_loss": 1.2725828886032104,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.13766056299209595,
"rewards/margins": 1.7153512239456177,
"rewards/rejected": -1.8530117273330688,
"step": 517
},
{
"epoch": 1.9547169811320755,
"grad_norm": 0.5101109147071838,
"learning_rate": 1.1320754716981131e-06,
"log_odds_chosen": 19.110702514648438,
"log_odds_ratio": -1.9727249309653416e-05,
"logits/chosen": -2.067298412322998,
"logits/rejected": -5.900681018829346,
"logps/chosen": -1.4211063385009766,
"logps/rejected": -20.093542098999023,
"loss": 1.5492,
"nll_loss": 1.5492104291915894,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14211063086986542,
"rewards/margins": 1.867243766784668,
"rewards/rejected": -2.00935435295105,
"step": 518
},
{
"epoch": 1.9584905660377359,
"grad_norm": 0.37587234377861023,
"learning_rate": 1.0377358490566038e-06,
"log_odds_chosen": 17.110137939453125,
"log_odds_ratio": -0.0024836428929120302,
"logits/chosen": -1.4974017143249512,
"logits/rejected": -4.80449914932251,
"logps/chosen": -1.6387498378753662,
"logps/rejected": -18.401721954345703,
"loss": 1.3765,
"nll_loss": 1.376255989074707,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16387496888637543,
"rewards/margins": 1.6762971878051758,
"rewards/rejected": -1.8401721715927124,
"step": 519
},
{
"epoch": 1.9622641509433962,
"grad_norm": 0.35912734270095825,
"learning_rate": 9.433962264150943e-07,
"log_odds_chosen": 14.829183578491211,
"log_odds_ratio": -0.15696510672569275,
"logits/chosen": -1.3042566776275635,
"logits/rejected": -6.660241603851318,
"logps/chosen": -1.6939544677734375,
"logps/rejected": -16.27517318725586,
"loss": 1.5792,
"nll_loss": 1.5635175704956055,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.16939544677734375,
"rewards/margins": 1.4581220149993896,
"rewards/rejected": -1.6275173425674438,
"step": 520
},
{
"epoch": 1.9660377358490566,
"grad_norm": 0.35668516159057617,
"learning_rate": 8.490566037735849e-07,
"log_odds_chosen": 18.270919799804688,
"log_odds_ratio": -0.09477733820676804,
"logits/chosen": -1.378293752670288,
"logits/rejected": -4.495629787445068,
"logps/chosen": -1.3341569900512695,
"logps/rejected": -19.200576782226562,
"loss": 1.3796,
"nll_loss": 1.3701521158218384,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.13341571390628815,
"rewards/margins": 1.786642074584961,
"rewards/rejected": -1.9200576543807983,
"step": 521
},
{
"epoch": 1.969811320754717,
"grad_norm": 0.40886542201042175,
"learning_rate": 7.547169811320755e-07,
"log_odds_chosen": 17.687740325927734,
"log_odds_ratio": -0.02107013203203678,
"logits/chosen": -1.3524720668792725,
"logits/rejected": -5.940802574157715,
"logps/chosen": -1.4188027381896973,
"logps/rejected": -18.724929809570312,
"loss": 1.2466,
"nll_loss": 1.2445337772369385,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.14188028872013092,
"rewards/margins": 1.730612874031067,
"rewards/rejected": -1.8724932670593262,
"step": 522
},
{
"epoch": 1.9735849056603774,
"grad_norm": 0.4124062657356262,
"learning_rate": 6.603773584905661e-07,
"log_odds_chosen": 15.467195510864258,
"log_odds_ratio": -0.046547312289476395,
"logits/chosen": -1.658618450164795,
"logits/rejected": -6.218527793884277,
"logps/chosen": -1.8022215366363525,
"logps/rejected": -16.846336364746094,
"loss": 1.6649,
"nll_loss": 1.660265326499939,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.18022215366363525,
"rewards/margins": 1.5044115781784058,
"rewards/rejected": -1.684633731842041,
"step": 523
},
{
"epoch": 1.9773584905660377,
"grad_norm": 0.38052645325660706,
"learning_rate": 5.660377358490566e-07,
"log_odds_chosen": 19.262311935424805,
"log_odds_ratio": -5.662463422595465e-07,
"logits/chosen": -2.7431163787841797,
"logits/rejected": -7.38714075088501,
"logps/chosen": -1.8242018222808838,
"logps/rejected": -20.831439971923828,
"loss": 1.4274,
"nll_loss": 1.4274204969406128,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.18242016434669495,
"rewards/margins": 1.9007238149642944,
"rewards/rejected": -2.083144187927246,
"step": 524
},
{
"epoch": 1.9811320754716981,
"grad_norm": 0.37092652916908264,
"learning_rate": 4.7169811320754717e-07,
"log_odds_chosen": 18.2442626953125,
"log_odds_ratio": -0.06912226229906082,
"logits/chosen": -2.5067853927612305,
"logits/rejected": -7.028432369232178,
"logps/chosen": -1.737499475479126,
"logps/rejected": -19.720117568969727,
"loss": 1.5818,
"nll_loss": 1.5748425722122192,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.1737499237060547,
"rewards/margins": 1.7982620000839233,
"rewards/rejected": -1.9720118045806885,
"step": 525
},
{
"epoch": 1.9849056603773585,
"grad_norm": 0.39420637488365173,
"learning_rate": 3.773584905660378e-07,
"log_odds_chosen": 16.574909210205078,
"log_odds_ratio": -0.07863669842481613,
"logits/chosen": -1.2284739017486572,
"logits/rejected": -4.368472099304199,
"logps/chosen": -1.423435926437378,
"logps/rejected": -17.645301818847656,
"loss": 1.3318,
"nll_loss": 1.323962688446045,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.14234358072280884,
"rewards/margins": 1.6221866607666016,
"rewards/rejected": -1.7645303010940552,
"step": 526
},
{
"epoch": 1.9886792452830189,
"grad_norm": 0.39243245124816895,
"learning_rate": 2.830188679245283e-07,
"log_odds_chosen": 17.68594741821289,
"log_odds_ratio": -3.010060481756227e-06,
"logits/chosen": -2.5417604446411133,
"logits/rejected": -5.660012722015381,
"logps/chosen": -1.545115351676941,
"logps/rejected": -18.896028518676758,
"loss": 1.6222,
"nll_loss": 1.6222350597381592,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15451154112815857,
"rewards/margins": 1.735091209411621,
"rewards/rejected": -1.889602780342102,
"step": 527
},
{
"epoch": 1.9924528301886792,
"grad_norm": 0.37130141258239746,
"learning_rate": 1.886792452830189e-07,
"log_odds_chosen": 19.309329986572266,
"log_odds_ratio": -3.0204251743271016e-05,
"logits/chosen": -1.1784321069717407,
"logits/rejected": -4.860876083374023,
"logps/chosen": -1.4668664932250977,
"logps/rejected": -20.424339294433594,
"loss": 1.4702,
"nll_loss": 1.47021484375,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1466866433620453,
"rewards/margins": 1.895747184753418,
"rewards/rejected": -2.042433977127075,
"step": 528
},
{
"epoch": 1.9962264150943396,
"grad_norm": 0.3702877461910248,
"learning_rate": 9.433962264150944e-08,
"log_odds_chosen": 20.07143783569336,
"log_odds_ratio": -8.866221037351352e-07,
"logits/chosen": -0.7111749649047852,
"logits/rejected": -4.296523094177246,
"logps/chosen": -1.6536825895309448,
"logps/rejected": -21.461524963378906,
"loss": 1.5899,
"nll_loss": 1.5898981094360352,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.16536825895309448,
"rewards/margins": 1.9807841777801514,
"rewards/rejected": -2.1461524963378906,
"step": 529
},
{
"epoch": 2.0,
"grad_norm": 0.5285991430282593,
"learning_rate": 0.0,
"log_odds_chosen": 17.8989315032959,
"log_odds_ratio": -0.0035720046143978834,
"logits/chosen": -1.8316022157669067,
"logits/rejected": -4.168578147888184,
"logps/chosen": -1.5039995908737183,
"logps/rejected": -19.07382583618164,
"loss": 1.3711,
"nll_loss": 1.3707059621810913,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.15039995312690735,
"rewards/margins": 1.7569825649261475,
"rewards/rejected": -1.9073827266693115,
"step": 530
}
],
"logging_steps": 1,
"max_steps": 530,
"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": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}