zephyr-7b-dpo-full / trainer_state.json
yihang7's picture
Model save
64b43a7
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"eval_steps": 100,
"global_step": 2907,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.7182130584192438e-09,
"logits/chosen": -2.7645790576934814,
"logits/rejected": -2.8125059604644775,
"logps/chosen": -113.67314910888672,
"logps/rejected": -132.0498504638672,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 1.718213058419244e-08,
"logits/chosen": -2.9990971088409424,
"logits/rejected": -3.0227837562561035,
"logps/chosen": -281.044921875,
"logps/rejected": -247.3936309814453,
"loss": 0.6906,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.004584211856126785,
"rewards/margins": 0.00850469246506691,
"rewards/rejected": -0.0039204806089401245,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 3.436426116838488e-08,
"logits/chosen": -2.88598895072937,
"logits/rejected": -2.917177200317383,
"logps/chosen": -359.26177978515625,
"logps/rejected": -298.42877197265625,
"loss": 0.6845,
"rewards/accuracies": 0.512499988079071,
"rewards/chosen": 0.05283154919743538,
"rewards/margins": 0.018162177875638008,
"rewards/rejected": 0.03466937318444252,
"step": 20
},
{
"epoch": 0.03,
"learning_rate": 5.154639175257731e-08,
"logits/chosen": -2.9570868015289307,
"logits/rejected": -2.9609949588775635,
"logps/chosen": -326.2544860839844,
"logps/rejected": -289.9393615722656,
"loss": 0.6564,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.1515582799911499,
"rewards/margins": 0.05830109864473343,
"rewards/rejected": 0.09325718879699707,
"step": 30
},
{
"epoch": 0.04,
"learning_rate": 6.872852233676976e-08,
"logits/chosen": -2.9531846046447754,
"logits/rejected": -2.955566883087158,
"logps/chosen": -376.5739440917969,
"logps/rejected": -331.3490295410156,
"loss": 0.6444,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": 0.3529122769832611,
"rewards/margins": 0.1296483278274536,
"rewards/rejected": 0.2232639044523239,
"step": 40
},
{
"epoch": 0.05,
"learning_rate": 8.59106529209622e-08,
"logits/chosen": -2.8878796100616455,
"logits/rejected": -2.9229512214660645,
"logps/chosen": -427.5284118652344,
"logps/rejected": -266.94415283203125,
"loss": 0.6204,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": 0.5041142702102661,
"rewards/margins": 0.29676300287246704,
"rewards/rejected": 0.20735123753547668,
"step": 50
},
{
"epoch": 0.06,
"learning_rate": 1.0309278350515462e-07,
"logits/chosen": -2.92777943611145,
"logits/rejected": -2.9371728897094727,
"logps/chosen": -318.0414123535156,
"logps/rejected": -265.1334228515625,
"loss": 0.6105,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.38802462816238403,
"rewards/margins": 0.31535086035728455,
"rewards/rejected": 0.0726737454533577,
"step": 60
},
{
"epoch": 0.07,
"learning_rate": 1.202749140893471e-07,
"logits/chosen": -2.9851737022399902,
"logits/rejected": -3.0005269050598145,
"logps/chosen": -404.5143127441406,
"logps/rejected": -300.8736572265625,
"loss": 0.6069,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": 0.5573440790176392,
"rewards/margins": 0.41277560591697693,
"rewards/rejected": 0.14456847310066223,
"step": 70
},
{
"epoch": 0.08,
"learning_rate": 1.3745704467353952e-07,
"logits/chosen": -2.96441912651062,
"logits/rejected": -2.968987464904785,
"logps/chosen": -314.7437438964844,
"logps/rejected": -254.7586669921875,
"loss": 0.5569,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 0.45461219549179077,
"rewards/margins": 0.5670984983444214,
"rewards/rejected": -0.1124863252043724,
"step": 80
},
{
"epoch": 0.09,
"learning_rate": 1.5463917525773197e-07,
"logits/chosen": -3.021219491958618,
"logits/rejected": -3.0178027153015137,
"logps/chosen": -308.31585693359375,
"logps/rejected": -257.63250732421875,
"loss": 0.5296,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": 0.5257282257080078,
"rewards/margins": 0.7057730555534363,
"rewards/rejected": -0.18004484474658966,
"step": 90
},
{
"epoch": 0.1,
"learning_rate": 1.718213058419244e-07,
"logits/chosen": -2.8890886306762695,
"logits/rejected": -2.896449327468872,
"logps/chosen": -375.84564208984375,
"logps/rejected": -241.34219360351562,
"loss": 0.5504,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": 0.6213200092315674,
"rewards/margins": 0.8095133900642395,
"rewards/rejected": -0.18819323182106018,
"step": 100
},
{
"epoch": 0.1,
"eval_logits/chosen": -2.936641216278076,
"eval_logits/rejected": -2.935973882675171,
"eval_logps/chosen": -361.9043884277344,
"eval_logps/rejected": -293.7761535644531,
"eval_loss": 0.5406630635261536,
"eval_rewards/accuracies": 0.7579365372657776,
"eval_rewards/chosen": 0.5287383794784546,
"eval_rewards/margins": 0.7097563743591309,
"eval_rewards/rejected": -0.18101799488067627,
"eval_runtime": 163.7175,
"eval_samples_per_second": 12.216,
"eval_steps_per_second": 0.385,
"step": 100
},
{
"epoch": 0.11,
"learning_rate": 1.8900343642611682e-07,
"logits/chosen": -2.917739152908325,
"logits/rejected": -2.8890061378479004,
"logps/chosen": -334.1250305175781,
"logps/rejected": -331.29571533203125,
"loss": 0.5741,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": 0.4206802248954773,
"rewards/margins": 0.6734089851379395,
"rewards/rejected": -0.25272876024246216,
"step": 110
},
{
"epoch": 0.12,
"learning_rate": 2.0618556701030925e-07,
"logits/chosen": -2.9665865898132324,
"logits/rejected": -2.970818519592285,
"logps/chosen": -386.2568664550781,
"logps/rejected": -280.7279357910156,
"loss": 0.5533,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": 0.3308308720588684,
"rewards/margins": 0.6611676216125488,
"rewards/rejected": -0.33033671975135803,
"step": 120
},
{
"epoch": 0.13,
"learning_rate": 2.2336769759450173e-07,
"logits/chosen": -3.012545347213745,
"logits/rejected": -2.9925591945648193,
"logps/chosen": -353.75469970703125,
"logps/rejected": -290.1478576660156,
"loss": 0.5447,
"rewards/accuracies": 0.6875,
"rewards/chosen": 0.47041910886764526,
"rewards/margins": 0.7254467010498047,
"rewards/rejected": -0.2550275921821594,
"step": 130
},
{
"epoch": 0.14,
"learning_rate": 2.405498281786942e-07,
"logits/chosen": -2.9241251945495605,
"logits/rejected": -2.9576869010925293,
"logps/chosen": -329.7611389160156,
"logps/rejected": -265.63006591796875,
"loss": 0.5113,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.4418914318084717,
"rewards/margins": 0.7908871173858643,
"rewards/rejected": -0.3489956259727478,
"step": 140
},
{
"epoch": 0.15,
"learning_rate": 2.5773195876288655e-07,
"logits/chosen": -2.9687321186065674,
"logits/rejected": -2.9832406044006348,
"logps/chosen": -331.42669677734375,
"logps/rejected": -269.3779296875,
"loss": 0.5387,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": 0.38882407546043396,
"rewards/margins": 0.8327142000198364,
"rewards/rejected": -0.4438902735710144,
"step": 150
},
{
"epoch": 0.17,
"learning_rate": 2.7491408934707903e-07,
"logits/chosen": -2.9920172691345215,
"logits/rejected": -3.013425827026367,
"logps/chosen": -383.51934814453125,
"logps/rejected": -297.9476318359375,
"loss": 0.5083,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": 0.6878620982170105,
"rewards/margins": 1.0685365200042725,
"rewards/rejected": -0.3806745111942291,
"step": 160
},
{
"epoch": 0.18,
"learning_rate": 2.9209621993127146e-07,
"logits/chosen": -2.9233837127685547,
"logits/rejected": -2.9321510791778564,
"logps/chosen": -339.95745849609375,
"logps/rejected": -280.793701171875,
"loss": 0.5131,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": 0.5272036194801331,
"rewards/margins": 0.8315987586975098,
"rewards/rejected": -0.3043951690196991,
"step": 170
},
{
"epoch": 0.19,
"learning_rate": 3.0927835051546394e-07,
"logits/chosen": -2.9920477867126465,
"logits/rejected": -2.9811954498291016,
"logps/chosen": -265.2094421386719,
"logps/rejected": -254.6926727294922,
"loss": 0.504,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 0.177840456366539,
"rewards/margins": 0.8774341344833374,
"rewards/rejected": -0.699593722820282,
"step": 180
},
{
"epoch": 0.2,
"learning_rate": 3.2646048109965636e-07,
"logits/chosen": -3.0027570724487305,
"logits/rejected": -2.987896203994751,
"logps/chosen": -330.7102966308594,
"logps/rejected": -239.6572723388672,
"loss": 0.5611,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": 0.5821124315261841,
"rewards/margins": 1.141722559928894,
"rewards/rejected": -0.5596100687980652,
"step": 190
},
{
"epoch": 0.21,
"learning_rate": 3.436426116838488e-07,
"logits/chosen": -3.048879384994507,
"logits/rejected": -2.9993340969085693,
"logps/chosen": -266.72430419921875,
"logps/rejected": -187.27467346191406,
"loss": 0.541,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": 0.2961394786834717,
"rewards/margins": 0.8773609399795532,
"rewards/rejected": -0.5812214612960815,
"step": 200
},
{
"epoch": 0.21,
"eval_logits/chosen": -2.980220317840576,
"eval_logits/rejected": -2.9785656929016113,
"eval_logps/chosen": -360.50030517578125,
"eval_logps/rejected": -297.53515625,
"eval_loss": 0.5220658779144287,
"eval_rewards/accuracies": 0.7698412537574768,
"eval_rewards/chosen": 0.6691505908966064,
"eval_rewards/margins": 1.2260682582855225,
"eval_rewards/rejected": -0.5569177269935608,
"eval_runtime": 163.6147,
"eval_samples_per_second": 12.224,
"eval_steps_per_second": 0.385,
"step": 200
},
{
"epoch": 0.22,
"learning_rate": 3.608247422680412e-07,
"logits/chosen": -2.9797844886779785,
"logits/rejected": -2.9449918270111084,
"logps/chosen": -364.26287841796875,
"logps/rejected": -251.58901977539062,
"loss": 0.4772,
"rewards/accuracies": 0.8125,
"rewards/chosen": 0.5583639740943909,
"rewards/margins": 1.3936102390289307,
"rewards/rejected": -0.835246205329895,
"step": 210
},
{
"epoch": 0.23,
"learning_rate": 3.7800687285223364e-07,
"logits/chosen": -2.9557044506073,
"logits/rejected": -2.9637341499328613,
"logps/chosen": -261.53216552734375,
"logps/rejected": -271.5208740234375,
"loss": 0.5707,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.367598295211792,
"rewards/margins": 1.1545054912567139,
"rewards/rejected": -0.7869071960449219,
"step": 220
},
{
"epoch": 0.24,
"learning_rate": 3.9518900343642607e-07,
"logits/chosen": -3.0034899711608887,
"logits/rejected": -2.991698980331421,
"logps/chosen": -308.8106689453125,
"logps/rejected": -278.55950927734375,
"loss": 0.5827,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": 0.4888441562652588,
"rewards/margins": 1.2474777698516846,
"rewards/rejected": -0.7586336731910706,
"step": 230
},
{
"epoch": 0.25,
"learning_rate": 4.123711340206185e-07,
"logits/chosen": -3.0844597816467285,
"logits/rejected": -3.0464837551116943,
"logps/chosen": -385.8021545410156,
"logps/rejected": -253.19869995117188,
"loss": 0.4898,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.5026682019233704,
"rewards/margins": 1.1020526885986328,
"rewards/rejected": -0.5993844270706177,
"step": 240
},
{
"epoch": 0.26,
"learning_rate": 4.2955326460481097e-07,
"logits/chosen": -3.0472395420074463,
"logits/rejected": -3.0599236488342285,
"logps/chosen": -341.8814697265625,
"logps/rejected": -295.29437255859375,
"loss": 0.5395,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": 0.24311120808124542,
"rewards/margins": 0.9385444521903992,
"rewards/rejected": -0.6954333186149597,
"step": 250
},
{
"epoch": 0.27,
"learning_rate": 4.4673539518900345e-07,
"logits/chosen": -3.0285518169403076,
"logits/rejected": -3.0690500736236572,
"logps/chosen": -353.20074462890625,
"logps/rejected": -244.77041625976562,
"loss": 0.6312,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.22555696964263916,
"rewards/margins": 0.8821722269058228,
"rewards/rejected": -1.107729196548462,
"step": 260
},
{
"epoch": 0.28,
"learning_rate": 4.639175257731959e-07,
"logits/chosen": -3.0879526138305664,
"logits/rejected": -3.0506978034973145,
"logps/chosen": -354.5426025390625,
"logps/rejected": -279.86773681640625,
"loss": 0.571,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": 0.26413029432296753,
"rewards/margins": 0.9725528955459595,
"rewards/rejected": -0.7084226012229919,
"step": 270
},
{
"epoch": 0.29,
"learning_rate": 4.810996563573884e-07,
"logits/chosen": -3.0588438510894775,
"logits/rejected": -3.0481762886047363,
"logps/chosen": -339.85675048828125,
"logps/rejected": -285.8063049316406,
"loss": 0.6383,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": 0.30532822012901306,
"rewards/margins": 1.10079026222229,
"rewards/rejected": -0.7954620122909546,
"step": 280
},
{
"epoch": 0.3,
"learning_rate": 4.982817869415807e-07,
"logits/chosen": -3.0889270305633545,
"logits/rejected": -3.0666940212249756,
"logps/chosen": -335.5870666503906,
"logps/rejected": -258.51641845703125,
"loss": 0.5611,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": 0.7204562425613403,
"rewards/margins": 1.0493910312652588,
"rewards/rejected": -0.32893460988998413,
"step": 290
},
{
"epoch": 0.31,
"learning_rate": 4.982798165137615e-07,
"logits/chosen": -3.0340983867645264,
"logits/rejected": -3.0090713500976562,
"logps/chosen": -281.38751220703125,
"logps/rejected": -289.4985656738281,
"loss": 0.6034,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.268043577671051,
"rewards/margins": 0.5827728509902954,
"rewards/rejected": -0.3147292733192444,
"step": 300
},
{
"epoch": 0.31,
"eval_logits/chosen": -3.035973072052002,
"eval_logits/rejected": -3.0234200954437256,
"eval_logps/chosen": -359.8170166015625,
"eval_logps/rejected": -296.5441589355469,
"eval_loss": 0.5459412932395935,
"eval_rewards/accuracies": 0.761904776096344,
"eval_rewards/chosen": 0.737476110458374,
"eval_rewards/margins": 1.1952924728393555,
"eval_rewards/rejected": -0.45781639218330383,
"eval_runtime": 164.3219,
"eval_samples_per_second": 12.171,
"eval_steps_per_second": 0.383,
"step": 300
},
{
"epoch": 0.32,
"learning_rate": 4.963685015290519e-07,
"logits/chosen": -3.1333563327789307,
"logits/rejected": -3.0529465675354004,
"logps/chosen": -394.2475280761719,
"logps/rejected": -328.84796142578125,
"loss": 0.5995,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": 0.5251134634017944,
"rewards/margins": 0.8031543493270874,
"rewards/rejected": -0.27804094552993774,
"step": 310
},
{
"epoch": 0.33,
"learning_rate": 4.944571865443424e-07,
"logits/chosen": -3.115387201309204,
"logits/rejected": -3.104794502258301,
"logps/chosen": -299.5379943847656,
"logps/rejected": -227.14413452148438,
"loss": 0.5504,
"rewards/accuracies": 0.8125,
"rewards/chosen": 0.5498681664466858,
"rewards/margins": 1.550806999206543,
"rewards/rejected": -1.0009387731552124,
"step": 320
},
{
"epoch": 0.34,
"learning_rate": 4.92545871559633e-07,
"logits/chosen": -3.1059436798095703,
"logits/rejected": -3.110661029815674,
"logps/chosen": -405.8400573730469,
"logps/rejected": -290.01934814453125,
"loss": 0.5355,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.5642995834350586,
"rewards/margins": 1.1246757507324219,
"rewards/rejected": -0.5603762269020081,
"step": 330
},
{
"epoch": 0.35,
"learning_rate": 4.906345565749235e-07,
"logits/chosen": -3.0694103240966797,
"logits/rejected": -3.075610876083374,
"logps/chosen": -301.7900695800781,
"logps/rejected": -273.09100341796875,
"loss": 0.6496,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": 0.26571425795555115,
"rewards/margins": 0.967176079750061,
"rewards/rejected": -0.701461672782898,
"step": 340
},
{
"epoch": 0.36,
"learning_rate": 4.88723241590214e-07,
"logits/chosen": -3.078815460205078,
"logits/rejected": -3.097691059112549,
"logps/chosen": -373.6755065917969,
"logps/rejected": -278.1918640136719,
"loss": 0.5251,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.5501624941825867,
"rewards/margins": 1.1470292806625366,
"rewards/rejected": -0.5968667268753052,
"step": 350
},
{
"epoch": 0.37,
"learning_rate": 4.868119266055046e-07,
"logits/chosen": -3.101353168487549,
"logits/rejected": -3.1290316581726074,
"logps/chosen": -370.21112060546875,
"logps/rejected": -328.2227783203125,
"loss": 0.5218,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": 0.4693407118320465,
"rewards/margins": 1.2798802852630615,
"rewards/rejected": -0.8105396032333374,
"step": 360
},
{
"epoch": 0.38,
"learning_rate": 4.849006116207951e-07,
"logits/chosen": -3.108405113220215,
"logits/rejected": -3.108668804168701,
"logps/chosen": -357.5787048339844,
"logps/rejected": -308.5846252441406,
"loss": 0.5781,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": 0.5667105317115784,
"rewards/margins": 1.561586618423462,
"rewards/rejected": -0.9948760271072388,
"step": 370
},
{
"epoch": 0.39,
"learning_rate": 4.829892966360856e-07,
"logits/chosen": -3.1076834201812744,
"logits/rejected": -3.139901638031006,
"logps/chosen": -372.7229919433594,
"logps/rejected": -321.50347900390625,
"loss": 0.5748,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": 0.595112681388855,
"rewards/margins": 1.7188622951507568,
"rewards/rejected": -1.1237497329711914,
"step": 380
},
{
"epoch": 0.4,
"learning_rate": 4.810779816513762e-07,
"logits/chosen": -3.0231597423553467,
"logits/rejected": -3.055475950241089,
"logps/chosen": -308.81109619140625,
"logps/rejected": -280.67572021484375,
"loss": 0.594,
"rewards/accuracies": 0.6875,
"rewards/chosen": 0.022154245525598526,
"rewards/margins": 1.1715147495269775,
"rewards/rejected": -1.1493604183197021,
"step": 390
},
{
"epoch": 0.41,
"learning_rate": 4.791666666666667e-07,
"logits/chosen": -2.9821434020996094,
"logits/rejected": -2.990657329559326,
"logps/chosen": -350.4073791503906,
"logps/rejected": -234.08291625976562,
"loss": 0.5944,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 0.6456303000450134,
"rewards/margins": 1.8281257152557373,
"rewards/rejected": -1.182495355606079,
"step": 400
},
{
"epoch": 0.41,
"eval_logits/chosen": -2.96209979057312,
"eval_logits/rejected": -2.963911294937134,
"eval_logps/chosen": -362.2125549316406,
"eval_logps/rejected": -300.90362548828125,
"eval_loss": 0.5573462247848511,
"eval_rewards/accuracies": 0.7698412537574768,
"eval_rewards/chosen": 0.49792128801345825,
"eval_rewards/margins": 1.391687273979187,
"eval_rewards/rejected": -0.8937660455703735,
"eval_runtime": 163.7646,
"eval_samples_per_second": 12.213,
"eval_steps_per_second": 0.385,
"step": 400
},
{
"epoch": 0.42,
"learning_rate": 4.772553516819572e-07,
"logits/chosen": -2.9685988426208496,
"logits/rejected": -2.9469170570373535,
"logps/chosen": -359.9443054199219,
"logps/rejected": -339.13482666015625,
"loss": 0.7753,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.5800348520278931,
"rewards/margins": 1.4965015649795532,
"rewards/rejected": -0.9164667129516602,
"step": 410
},
{
"epoch": 0.43,
"learning_rate": 4.753440366972477e-07,
"logits/chosen": -3.039097785949707,
"logits/rejected": -3.0352489948272705,
"logps/chosen": -279.19451904296875,
"logps/rejected": -275.61077880859375,
"loss": 0.5719,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": 0.08171078562736511,
"rewards/margins": 0.8535135388374329,
"rewards/rejected": -0.7718027234077454,
"step": 420
},
{
"epoch": 0.44,
"learning_rate": 4.7343272171253825e-07,
"logits/chosen": -3.0542099475860596,
"logits/rejected": -3.048107624053955,
"logps/chosen": -304.2041015625,
"logps/rejected": -275.24664306640625,
"loss": 0.5521,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 0.6358417272567749,
"rewards/margins": 1.4337527751922607,
"rewards/rejected": -0.7979112863540649,
"step": 430
},
{
"epoch": 0.45,
"learning_rate": 4.715214067278288e-07,
"logits/chosen": -2.9832911491394043,
"logits/rejected": -2.9696083068847656,
"logps/chosen": -351.0896911621094,
"logps/rejected": -278.2879333496094,
"loss": 0.5257,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": 0.4420256018638611,
"rewards/margins": 1.773047685623169,
"rewards/rejected": -1.331022024154663,
"step": 440
},
{
"epoch": 0.46,
"learning_rate": 4.696100917431192e-07,
"logits/chosen": -3.115874767303467,
"logits/rejected": -3.0773837566375732,
"logps/chosen": -392.2452392578125,
"logps/rejected": -324.62640380859375,
"loss": 0.5536,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": 0.2861310839653015,
"rewards/margins": 0.8551927804946899,
"rewards/rejected": -0.5690616369247437,
"step": 450
},
{
"epoch": 0.47,
"learning_rate": 4.6769877675840974e-07,
"logits/chosen": -3.0585522651672363,
"logits/rejected": -3.089534282684326,
"logps/chosen": -310.84967041015625,
"logps/rejected": -287.9058532714844,
"loss": 0.5614,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 0.2514137625694275,
"rewards/margins": 1.2147700786590576,
"rewards/rejected": -0.9633563160896301,
"step": 460
},
{
"epoch": 0.49,
"learning_rate": 4.6578746177370027e-07,
"logits/chosen": -3.0050368309020996,
"logits/rejected": -3.0113613605499268,
"logps/chosen": -243.838623046875,
"logps/rejected": -224.61404418945312,
"loss": 0.5769,
"rewards/accuracies": 0.8125,
"rewards/chosen": 0.10031839460134506,
"rewards/margins": 1.2319433689117432,
"rewards/rejected": -1.1316249370574951,
"step": 470
},
{
"epoch": 0.5,
"learning_rate": 4.638761467889908e-07,
"logits/chosen": -3.01200795173645,
"logits/rejected": -2.9829325675964355,
"logps/chosen": -353.6679992675781,
"logps/rejected": -299.7701416015625,
"loss": 0.5141,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": 0.6005850434303284,
"rewards/margins": 2.408433437347412,
"rewards/rejected": -1.807848334312439,
"step": 480
},
{
"epoch": 0.51,
"learning_rate": 4.6196483180428133e-07,
"logits/chosen": -3.038440227508545,
"logits/rejected": -3.0429458618164062,
"logps/chosen": -330.0135192871094,
"logps/rejected": -262.1318359375,
"loss": 0.5292,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": 0.4290197491645813,
"rewards/margins": 1.5280876159667969,
"rewards/rejected": -1.0990678071975708,
"step": 490
},
{
"epoch": 0.52,
"learning_rate": 4.600535168195718e-07,
"logits/chosen": -3.0223565101623535,
"logits/rejected": -3.0170624256134033,
"logps/chosen": -259.1560363769531,
"logps/rejected": -268.68365478515625,
"loss": 0.5512,
"rewards/accuracies": 0.6875,
"rewards/chosen": 0.08852599561214447,
"rewards/margins": 1.076027750968933,
"rewards/rejected": -0.9875017404556274,
"step": 500
},
{
"epoch": 0.52,
"eval_logits/chosen": -3.0406343936920166,
"eval_logits/rejected": -3.0485074520111084,
"eval_logps/chosen": -362.83642578125,
"eval_logps/rejected": -302.1329650878906,
"eval_loss": 0.5256651043891907,
"eval_rewards/accuracies": 0.7579365372657776,
"eval_rewards/chosen": 0.4355368912220001,
"eval_rewards/margins": 1.452234148979187,
"eval_rewards/rejected": -1.0166972875595093,
"eval_runtime": 164.1914,
"eval_samples_per_second": 12.181,
"eval_steps_per_second": 0.384,
"step": 500
},
{
"epoch": 0.53,
"learning_rate": 4.5814220183486234e-07,
"logits/chosen": -2.971991777420044,
"logits/rejected": -2.9626731872558594,
"logps/chosen": -387.75872802734375,
"logps/rejected": -341.24224853515625,
"loss": 0.5611,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.0966944545507431,
"rewards/margins": 0.9770743250846863,
"rewards/rejected": -1.0737688541412354,
"step": 510
},
{
"epoch": 0.54,
"learning_rate": 4.562308868501529e-07,
"logits/chosen": -2.97809100151062,
"logits/rejected": -3.0156943798065186,
"logps/chosen": -325.83837890625,
"logps/rejected": -321.0384826660156,
"loss": 0.5693,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": 0.27487924695014954,
"rewards/margins": 1.646512746810913,
"rewards/rejected": -1.371633529663086,
"step": 520
},
{
"epoch": 0.55,
"learning_rate": 4.543195718654434e-07,
"logits/chosen": -3.0082881450653076,
"logits/rejected": -3.003408193588257,
"logps/chosen": -274.6020812988281,
"logps/rejected": -240.13998413085938,
"loss": 0.5953,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.0502743124961853,
"rewards/margins": 0.5772665739059448,
"rewards/rejected": -0.6275408864021301,
"step": 530
},
{
"epoch": 0.56,
"learning_rate": 4.5240825688073394e-07,
"logits/chosen": -3.0475857257843018,
"logits/rejected": -3.0587058067321777,
"logps/chosen": -345.28802490234375,
"logps/rejected": -276.25018310546875,
"loss": 0.559,
"rewards/accuracies": 0.875,
"rewards/chosen": 0.6526178121566772,
"rewards/margins": 1.6864182949066162,
"rewards/rejected": -1.033800482749939,
"step": 540
},
{
"epoch": 0.57,
"learning_rate": 4.504969418960244e-07,
"logits/chosen": -2.9780993461608887,
"logits/rejected": -3.0339550971984863,
"logps/chosen": -318.60699462890625,
"logps/rejected": -363.83966064453125,
"loss": 0.5182,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": 0.3249278664588928,
"rewards/margins": 1.6138547658920288,
"rewards/rejected": -1.2889269590377808,
"step": 550
},
{
"epoch": 0.58,
"learning_rate": 4.4858562691131495e-07,
"logits/chosen": -3.0293617248535156,
"logits/rejected": -3.0541815757751465,
"logps/chosen": -355.3965759277344,
"logps/rejected": -341.19097900390625,
"loss": 0.5655,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.15244658291339874,
"rewards/margins": 1.1955846548080444,
"rewards/rejected": -1.043138027191162,
"step": 560
},
{
"epoch": 0.59,
"learning_rate": 4.466743119266055e-07,
"logits/chosen": -2.955909252166748,
"logits/rejected": -2.966557502746582,
"logps/chosen": -339.918701171875,
"logps/rejected": -312.85992431640625,
"loss": 0.5342,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.24385514855384827,
"rewards/margins": 1.2561490535736084,
"rewards/rejected": -1.5000044107437134,
"step": 570
},
{
"epoch": 0.6,
"learning_rate": 4.44762996941896e-07,
"logits/chosen": -2.9457859992980957,
"logits/rejected": -2.921659231185913,
"logps/chosen": -361.46905517578125,
"logps/rejected": -314.6666259765625,
"loss": 0.5347,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": 0.31199535727500916,
"rewards/margins": 1.5148388147354126,
"rewards/rejected": -1.202843427658081,
"step": 580
},
{
"epoch": 0.61,
"learning_rate": 4.4285168195718655e-07,
"logits/chosen": -2.9674103260040283,
"logits/rejected": -2.9832658767700195,
"logps/chosen": -279.5147705078125,
"logps/rejected": -283.4952697753906,
"loss": 0.5475,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.18109655380249023,
"rewards/margins": 1.081386685371399,
"rewards/rejected": -1.2624832391738892,
"step": 590
},
{
"epoch": 0.62,
"learning_rate": 4.40940366972477e-07,
"logits/chosen": -3.038327932357788,
"logits/rejected": -3.081512928009033,
"logps/chosen": -282.9052429199219,
"logps/rejected": -260.5687255859375,
"loss": 0.5879,
"rewards/accuracies": 0.6875,
"rewards/chosen": 0.0032925487030297518,
"rewards/margins": 0.903986930847168,
"rewards/rejected": -0.9006943702697754,
"step": 600
},
{
"epoch": 0.62,
"eval_logits/chosen": -2.9869041442871094,
"eval_logits/rejected": -2.991122007369995,
"eval_logps/chosen": -362.4848327636719,
"eval_logps/rejected": -301.2572021484375,
"eval_loss": 0.5287741422653198,
"eval_rewards/accuracies": 0.7579365372657776,
"eval_rewards/chosen": 0.47069627046585083,
"eval_rewards/margins": 1.3998188972473145,
"eval_rewards/rejected": -0.9291225075721741,
"eval_runtime": 164.0279,
"eval_samples_per_second": 12.193,
"eval_steps_per_second": 0.384,
"step": 600
},
{
"epoch": 0.63,
"learning_rate": 4.3902905198776756e-07,
"logits/chosen": -3.0266683101654053,
"logits/rejected": -3.0426414012908936,
"logps/chosen": -345.2246398925781,
"logps/rejected": -280.60711669921875,
"loss": 0.5172,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": 0.3867380917072296,
"rewards/margins": 1.5311682224273682,
"rewards/rejected": -1.14443039894104,
"step": 610
},
{
"epoch": 0.64,
"learning_rate": 4.371177370030581e-07,
"logits/chosen": -3.0024008750915527,
"logits/rejected": -3.0336501598358154,
"logps/chosen": -340.01483154296875,
"logps/rejected": -288.4037170410156,
"loss": 0.5674,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": 0.09759467095136642,
"rewards/margins": 1.4280248880386353,
"rewards/rejected": -1.330430030822754,
"step": 620
},
{
"epoch": 0.65,
"learning_rate": 4.352064220183486e-07,
"logits/chosen": -3.073171377182007,
"logits/rejected": -3.0693984031677246,
"logps/chosen": -268.47442626953125,
"logps/rejected": -253.87173461914062,
"loss": 0.6129,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": 0.08323542028665543,
"rewards/margins": 1.263319969177246,
"rewards/rejected": -1.180084466934204,
"step": 630
},
{
"epoch": 0.66,
"learning_rate": 4.3329510703363915e-07,
"logits/chosen": -3.1394124031066895,
"logits/rejected": -3.147449493408203,
"logps/chosen": -316.50323486328125,
"logps/rejected": -256.6443786621094,
"loss": 0.5405,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": 0.0077073900029063225,
"rewards/margins": 1.4150127172470093,
"rewards/rejected": -1.407305359840393,
"step": 640
},
{
"epoch": 0.67,
"learning_rate": 4.313837920489297e-07,
"logits/chosen": -3.015110731124878,
"logits/rejected": -3.0439746379852295,
"logps/chosen": -309.4215087890625,
"logps/rejected": -278.88934326171875,
"loss": 0.5239,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": 0.3628634512424469,
"rewards/margins": 1.4821045398712158,
"rewards/rejected": -1.1192409992218018,
"step": 650
},
{
"epoch": 0.68,
"learning_rate": 4.2947247706422016e-07,
"logits/chosen": -3.0345845222473145,
"logits/rejected": -2.997607469558716,
"logps/chosen": -311.18719482421875,
"logps/rejected": -289.7060852050781,
"loss": 0.5288,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.4103531241416931,
"rewards/margins": 1.0822376012802124,
"rewards/rejected": -1.4925907850265503,
"step": 660
},
{
"epoch": 0.69,
"learning_rate": 4.275611620795107e-07,
"logits/chosen": -3.0080935955047607,
"logits/rejected": -3.015535593032837,
"logps/chosen": -377.9685974121094,
"logps/rejected": -297.92169189453125,
"loss": 0.5683,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": 0.14489376544952393,
"rewards/margins": 1.1130046844482422,
"rewards/rejected": -0.9681110382080078,
"step": 670
},
{
"epoch": 0.7,
"learning_rate": 4.2564984709480123e-07,
"logits/chosen": -3.030597448348999,
"logits/rejected": -3.059508800506592,
"logps/chosen": -368.32635498046875,
"logps/rejected": -274.53619384765625,
"loss": 0.6557,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.3344075083732605,
"rewards/margins": 1.3792588710784912,
"rewards/rejected": -1.0448510646820068,
"step": 680
},
{
"epoch": 0.71,
"learning_rate": 4.2373853211009176e-07,
"logits/chosen": -3.0355846881866455,
"logits/rejected": -3.0617101192474365,
"logps/chosen": -310.24530029296875,
"logps/rejected": -280.7437438964844,
"loss": 0.5629,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": 0.31564414501190186,
"rewards/margins": 1.532622218132019,
"rewards/rejected": -1.2169779539108276,
"step": 690
},
{
"epoch": 0.72,
"learning_rate": 4.2182721712538224e-07,
"logits/chosen": -3.029533863067627,
"logits/rejected": -3.05369234085083,
"logps/chosen": -370.49945068359375,
"logps/rejected": -285.1793212890625,
"loss": 0.6773,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.3230968117713928,
"rewards/margins": 1.0617311000823975,
"rewards/rejected": -0.7386342287063599,
"step": 700
},
{
"epoch": 0.72,
"eval_logits/chosen": -3.041776180267334,
"eval_logits/rejected": -3.0563852787017822,
"eval_logps/chosen": -366.7193603515625,
"eval_logps/rejected": -301.1505432128906,
"eval_loss": 0.585310697555542,
"eval_rewards/accuracies": 0.7460317611694336,
"eval_rewards/chosen": 0.047242674976587296,
"eval_rewards/margins": 0.9657005667686462,
"eval_rewards/rejected": -0.9184578657150269,
"eval_runtime": 164.258,
"eval_samples_per_second": 12.176,
"eval_steps_per_second": 0.384,
"step": 700
},
{
"epoch": 0.73,
"learning_rate": 4.199159021406727e-07,
"logits/chosen": -2.963630199432373,
"logits/rejected": -3.031212329864502,
"logps/chosen": -297.58990478515625,
"logps/rejected": -283.17572021484375,
"loss": 0.6067,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 0.13306304812431335,
"rewards/margins": 1.0036863088607788,
"rewards/rejected": -0.8706234097480774,
"step": 710
},
{
"epoch": 0.74,
"learning_rate": 4.1800458715596325e-07,
"logits/chosen": -3.0382869243621826,
"logits/rejected": -3.0224924087524414,
"logps/chosen": -373.01947021484375,
"logps/rejected": -315.932861328125,
"loss": 0.6166,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": 0.028559958562254906,
"rewards/margins": 0.9609702825546265,
"rewards/rejected": -0.9324103593826294,
"step": 720
},
{
"epoch": 0.75,
"learning_rate": 4.160932721712538e-07,
"logits/chosen": -3.0072388648986816,
"logits/rejected": -3.0005228519439697,
"logps/chosen": -340.4766540527344,
"logps/rejected": -306.3741149902344,
"loss": 0.6079,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.09662823379039764,
"rewards/margins": 1.1530828475952148,
"rewards/rejected": -1.0564546585083008,
"step": 730
},
{
"epoch": 0.76,
"learning_rate": 4.141819571865443e-07,
"logits/chosen": -2.9518847465515137,
"logits/rejected": -2.9550204277038574,
"logps/chosen": -325.9070739746094,
"logps/rejected": -244.12588500976562,
"loss": 0.564,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": 0.1386883705854416,
"rewards/margins": 1.7188549041748047,
"rewards/rejected": -1.5801665782928467,
"step": 740
},
{
"epoch": 0.77,
"learning_rate": 4.1227064220183485e-07,
"logits/chosen": -2.9738943576812744,
"logits/rejected": -3.009288787841797,
"logps/chosen": -306.73614501953125,
"logps/rejected": -284.35089111328125,
"loss": 0.5213,
"rewards/accuracies": 0.8125,
"rewards/chosen": 0.16552898287773132,
"rewards/margins": 1.8094953298568726,
"rewards/rejected": -1.6439664363861084,
"step": 750
},
{
"epoch": 0.78,
"learning_rate": 4.103593272171253e-07,
"logits/chosen": -2.9576098918914795,
"logits/rejected": -2.9751369953155518,
"logps/chosen": -336.5853576660156,
"logps/rejected": -326.5455017089844,
"loss": 0.5703,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.07825515419244766,
"rewards/margins": 1.1398742198944092,
"rewards/rejected": -1.0616191625595093,
"step": 760
},
{
"epoch": 0.79,
"learning_rate": 4.0844801223241586e-07,
"logits/chosen": -3.0122196674346924,
"logits/rejected": -2.9879307746887207,
"logps/chosen": -350.8817138671875,
"logps/rejected": -298.84307861328125,
"loss": 0.5197,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.03767753392457962,
"rewards/margins": 1.09770929813385,
"rewards/rejected": -1.0600318908691406,
"step": 770
},
{
"epoch": 0.8,
"learning_rate": 4.065366972477064e-07,
"logits/chosen": -2.9043805599212646,
"logits/rejected": -2.9711837768554688,
"logps/chosen": -379.1385803222656,
"logps/rejected": -296.9505920410156,
"loss": 0.5669,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": 0.3532111942768097,
"rewards/margins": 1.7610466480255127,
"rewards/rejected": -1.407835602760315,
"step": 780
},
{
"epoch": 0.82,
"learning_rate": 4.046253822629969e-07,
"logits/chosen": -2.9510416984558105,
"logits/rejected": -2.961275100708008,
"logps/chosen": -330.33673095703125,
"logps/rejected": -288.71173095703125,
"loss": 0.5177,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.301142156124115,
"rewards/margins": 1.2936238050460815,
"rewards/rejected": -0.9924817085266113,
"step": 790
},
{
"epoch": 0.83,
"learning_rate": 4.0271406727828745e-07,
"logits/chosen": -2.991361141204834,
"logits/rejected": -2.974353790283203,
"logps/chosen": -322.8855895996094,
"logps/rejected": -271.7654113769531,
"loss": 0.5263,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.06158037111163139,
"rewards/margins": 1.1184431314468384,
"rewards/rejected": -1.056862711906433,
"step": 800
},
{
"epoch": 0.83,
"eval_logits/chosen": -2.963681936264038,
"eval_logits/rejected": -2.966184139251709,
"eval_logps/chosen": -364.9457702636719,
"eval_logps/rejected": -303.87957763671875,
"eval_loss": 0.5150811076164246,
"eval_rewards/accuracies": 0.761904776096344,
"eval_rewards/chosen": 0.22460374236106873,
"eval_rewards/margins": 1.4159626960754395,
"eval_rewards/rejected": -1.191359043121338,
"eval_runtime": 163.931,
"eval_samples_per_second": 12.2,
"eval_steps_per_second": 0.384,
"step": 800
},
{
"epoch": 0.84,
"learning_rate": 4.00802752293578e-07,
"logits/chosen": -2.974116802215576,
"logits/rejected": -2.9998645782470703,
"logps/chosen": -330.64910888671875,
"logps/rejected": -294.6690368652344,
"loss": 0.5031,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 0.18794824182987213,
"rewards/margins": 1.2928552627563477,
"rewards/rejected": -1.1049071550369263,
"step": 810
},
{
"epoch": 0.85,
"learning_rate": 3.9889143730886847e-07,
"logits/chosen": -2.9930388927459717,
"logits/rejected": -2.983773946762085,
"logps/chosen": -376.36212158203125,
"logps/rejected": -305.14111328125,
"loss": 0.5357,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.27564454078674316,
"rewards/margins": 1.5530188083648682,
"rewards/rejected": -1.277374267578125,
"step": 820
},
{
"epoch": 0.86,
"learning_rate": 3.96980122324159e-07,
"logits/chosen": -3.0382747650146484,
"logits/rejected": -3.0700857639312744,
"logps/chosen": -313.2106018066406,
"logps/rejected": -256.5130310058594,
"loss": 0.556,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.03291673585772514,
"rewards/margins": 1.2974836826324463,
"rewards/rejected": -1.3304002285003662,
"step": 830
},
{
"epoch": 0.87,
"learning_rate": 3.9506880733944953e-07,
"logits/chosen": -3.0589098930358887,
"logits/rejected": -3.058842897415161,
"logps/chosen": -304.68658447265625,
"logps/rejected": -276.25177001953125,
"loss": 0.5578,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.22373457252979279,
"rewards/margins": 1.489611268043518,
"rewards/rejected": -1.7133458852767944,
"step": 840
},
{
"epoch": 0.88,
"learning_rate": 3.9315749235474006e-07,
"logits/chosen": -3.037079334259033,
"logits/rejected": -3.0386836528778076,
"logps/chosen": -347.38897705078125,
"logps/rejected": -334.3331298828125,
"loss": 0.5433,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -0.15273378789424896,
"rewards/margins": 1.5688612461090088,
"rewards/rejected": -1.7215951681137085,
"step": 850
},
{
"epoch": 0.89,
"learning_rate": 3.912461773700306e-07,
"logits/chosen": -2.9914333820343018,
"logits/rejected": -3.013286828994751,
"logps/chosen": -361.6410217285156,
"logps/rejected": -342.3985900878906,
"loss": 0.5464,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.03575097769498825,
"rewards/margins": 1.2531265020370483,
"rewards/rejected": -1.2888776063919067,
"step": 860
},
{
"epoch": 0.9,
"learning_rate": 3.8933486238532107e-07,
"logits/chosen": -2.9541945457458496,
"logits/rejected": -2.979830265045166,
"logps/chosen": -424.258544921875,
"logps/rejected": -299.7648620605469,
"loss": 0.581,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": 0.013498688116669655,
"rewards/margins": 1.3451616764068604,
"rewards/rejected": -1.331662893295288,
"step": 870
},
{
"epoch": 0.91,
"learning_rate": 3.874235474006116e-07,
"logits/chosen": -2.9790916442871094,
"logits/rejected": -2.987037181854248,
"logps/chosen": -364.68048095703125,
"logps/rejected": -290.4891052246094,
"loss": 0.58,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.007329714484512806,
"rewards/margins": 1.4278209209442139,
"rewards/rejected": -1.4351506233215332,
"step": 880
},
{
"epoch": 0.92,
"learning_rate": 3.8551223241590214e-07,
"logits/chosen": -2.986210823059082,
"logits/rejected": -2.9739222526550293,
"logps/chosen": -300.7494812011719,
"logps/rejected": -278.2732849121094,
"loss": 0.5741,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.04843021556735039,
"rewards/margins": 1.3019744157791138,
"rewards/rejected": -1.3504045009613037,
"step": 890
},
{
"epoch": 0.93,
"learning_rate": 3.8360091743119267e-07,
"logits/chosen": -2.9838929176330566,
"logits/rejected": -2.9902117252349854,
"logps/chosen": -306.20025634765625,
"logps/rejected": -289.5735168457031,
"loss": 0.5366,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": 0.058762937784194946,
"rewards/margins": 1.2602014541625977,
"rewards/rejected": -1.2014386653900146,
"step": 900
},
{
"epoch": 0.93,
"eval_logits/chosen": -2.9907381534576416,
"eval_logits/rejected": -2.982360363006592,
"eval_logps/chosen": -364.6807556152344,
"eval_logps/rejected": -302.8385314941406,
"eval_loss": 0.5133689641952515,
"eval_rewards/accuracies": 0.75,
"eval_rewards/chosen": 0.25110283493995667,
"eval_rewards/margins": 1.338356614112854,
"eval_rewards/rejected": -1.0872538089752197,
"eval_runtime": 164.3114,
"eval_samples_per_second": 12.172,
"eval_steps_per_second": 0.383,
"step": 900
},
{
"epoch": 0.94,
"learning_rate": 3.816896024464832e-07,
"logits/chosen": -3.0253748893737793,
"logits/rejected": -2.9562289714813232,
"logps/chosen": -281.73016357421875,
"logps/rejected": -248.2506866455078,
"loss": 0.5377,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": 0.28038138151168823,
"rewards/margins": 1.607208251953125,
"rewards/rejected": -1.326826810836792,
"step": 910
},
{
"epoch": 0.95,
"learning_rate": 3.797782874617737e-07,
"logits/chosen": -2.9797048568725586,
"logits/rejected": -2.932326555252075,
"logps/chosen": -333.2131042480469,
"logps/rejected": -267.63128662109375,
"loss": 0.4959,
"rewards/accuracies": 0.6875,
"rewards/chosen": 0.26507893204689026,
"rewards/margins": 1.226858139038086,
"rewards/rejected": -0.9617794156074524,
"step": 920
},
{
"epoch": 0.96,
"learning_rate": 3.778669724770642e-07,
"logits/chosen": -2.9677836894989014,
"logits/rejected": -2.9711012840270996,
"logps/chosen": -301.1932067871094,
"logps/rejected": -239.915771484375,
"loss": 0.5646,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": 0.06536471843719482,
"rewards/margins": 1.3896596431732178,
"rewards/rejected": -1.3242948055267334,
"step": 930
},
{
"epoch": 0.97,
"learning_rate": 3.7595565749235474e-07,
"logits/chosen": -3.003399133682251,
"logits/rejected": -2.9879281520843506,
"logps/chosen": -351.9979553222656,
"logps/rejected": -264.519775390625,
"loss": 0.585,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": 0.3593365252017975,
"rewards/margins": 1.1577335596084595,
"rewards/rejected": -0.7983969449996948,
"step": 940
},
{
"epoch": 0.98,
"learning_rate": 3.740443425076452e-07,
"logits/chosen": -2.9658942222595215,
"logits/rejected": -2.982341766357422,
"logps/chosen": -336.6238708496094,
"logps/rejected": -299.588134765625,
"loss": 0.5176,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 0.2032889872789383,
"rewards/margins": 1.3534172773361206,
"rewards/rejected": -1.1501282453536987,
"step": 950
},
{
"epoch": 0.99,
"learning_rate": 3.7213302752293575e-07,
"logits/chosen": -2.951019763946533,
"logits/rejected": -2.985151767730713,
"logps/chosen": -338.8179931640625,
"logps/rejected": -283.80328369140625,
"loss": 0.538,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": 0.049173761159181595,
"rewards/margins": 0.8662222623825073,
"rewards/rejected": -0.8170484304428101,
"step": 960
},
{
"epoch": 1.0,
"learning_rate": 3.702217125382263e-07,
"logits/chosen": -2.9252991676330566,
"logits/rejected": -2.937505006790161,
"logps/chosen": -354.7286682128906,
"logps/rejected": -312.35333251953125,
"loss": 0.4865,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 0.2835424542427063,
"rewards/margins": 1.7692314386367798,
"rewards/rejected": -1.4856891632080078,
"step": 970
},
{
"epoch": 1.01,
"learning_rate": 3.6831039755351677e-07,
"logits/chosen": -2.975984573364258,
"logits/rejected": -2.9734318256378174,
"logps/chosen": -319.2844543457031,
"logps/rejected": -303.3651428222656,
"loss": 0.1133,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 1.7135050296783447,
"rewards/margins": 4.7504682540893555,
"rewards/rejected": -3.0369625091552734,
"step": 980
},
{
"epoch": 1.02,
"learning_rate": 3.663990825688073e-07,
"logits/chosen": -2.899402141571045,
"logits/rejected": -2.89802885055542,
"logps/chosen": -288.45123291015625,
"logps/rejected": -316.5885314941406,
"loss": 0.1405,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 1.7678325176239014,
"rewards/margins": 5.483719348907471,
"rewards/rejected": -3.715886354446411,
"step": 990
},
{
"epoch": 1.03,
"learning_rate": 3.6448776758409783e-07,
"logits/chosen": -2.8675971031188965,
"logits/rejected": -2.89615797996521,
"logps/chosen": -325.03863525390625,
"logps/rejected": -333.436767578125,
"loss": 0.1034,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.79555344581604,
"rewards/margins": 4.759924411773682,
"rewards/rejected": -2.9643709659576416,
"step": 1000
},
{
"epoch": 1.03,
"eval_logits/chosen": -2.92020583152771,
"eval_logits/rejected": -2.9095799922943115,
"eval_logps/chosen": -364.1185302734375,
"eval_logps/rejected": -306.2866516113281,
"eval_loss": 0.5107486248016357,
"eval_rewards/accuracies": 0.761904776096344,
"eval_rewards/chosen": 0.307327002286911,
"eval_rewards/margins": 1.739391803741455,
"eval_rewards/rejected": -1.4320647716522217,
"eval_runtime": 164.3142,
"eval_samples_per_second": 12.172,
"eval_steps_per_second": 0.383,
"step": 1000
},
{
"epoch": 1.04,
"learning_rate": 3.6257645259938836e-07,
"logits/chosen": -2.8848228454589844,
"logits/rejected": -2.9434664249420166,
"logps/chosen": -304.1281433105469,
"logps/rejected": -323.9388732910156,
"loss": 0.0912,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.5237512588500977,
"rewards/margins": 5.300021171569824,
"rewards/rejected": -3.7762699127197266,
"step": 1010
},
{
"epoch": 1.05,
"learning_rate": 3.606651376146789e-07,
"logits/chosen": -2.818145275115967,
"logits/rejected": -2.773864269256592,
"logps/chosen": -315.73687744140625,
"logps/rejected": -252.3991241455078,
"loss": 0.1072,
"rewards/accuracies": 0.9375,
"rewards/chosen": 1.2391575574874878,
"rewards/margins": 4.052863121032715,
"rewards/rejected": -2.8137052059173584,
"step": 1020
},
{
"epoch": 1.06,
"learning_rate": 3.5875382262996937e-07,
"logits/chosen": -2.8956587314605713,
"logits/rejected": -2.88509202003479,
"logps/chosen": -332.889404296875,
"logps/rejected": -375.0550231933594,
"loss": 0.0854,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 1.7886276245117188,
"rewards/margins": 4.777144432067871,
"rewards/rejected": -2.9885172843933105,
"step": 1030
},
{
"epoch": 1.07,
"learning_rate": 3.568425076452599e-07,
"logits/chosen": -2.9382426738739014,
"logits/rejected": -2.9390716552734375,
"logps/chosen": -339.12451171875,
"logps/rejected": -315.15625,
"loss": 0.099,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": 1.3939841985702515,
"rewards/margins": 4.764640808105469,
"rewards/rejected": -3.370656967163086,
"step": 1040
},
{
"epoch": 1.08,
"learning_rate": 3.5493119266055044e-07,
"logits/chosen": -2.8407671451568604,
"logits/rejected": -2.821763753890991,
"logps/chosen": -336.37298583984375,
"logps/rejected": -257.6861267089844,
"loss": 0.1132,
"rewards/accuracies": 0.9375,
"rewards/chosen": 1.3384716510772705,
"rewards/margins": 4.965681076049805,
"rewards/rejected": -3.627209424972534,
"step": 1050
},
{
"epoch": 1.09,
"learning_rate": 3.5301987767584097e-07,
"logits/chosen": -2.8167824745178223,
"logits/rejected": -2.810854434967041,
"logps/chosen": -323.439208984375,
"logps/rejected": -342.47991943359375,
"loss": 0.2041,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 0.7954254150390625,
"rewards/margins": 4.722014427185059,
"rewards/rejected": -3.9265894889831543,
"step": 1060
},
{
"epoch": 1.1,
"learning_rate": 3.511085626911315e-07,
"logits/chosen": -2.940957546234131,
"logits/rejected": -3.0021321773529053,
"logps/chosen": -366.2899475097656,
"logps/rejected": -343.2218933105469,
"loss": 0.3299,
"rewards/accuracies": 0.9375,
"rewards/chosen": 1.4886529445648193,
"rewards/margins": 5.359461307525635,
"rewards/rejected": -3.870807647705078,
"step": 1070
},
{
"epoch": 1.11,
"learning_rate": 3.49197247706422e-07,
"logits/chosen": -2.933786392211914,
"logits/rejected": -2.906247615814209,
"logps/chosen": -261.5579833984375,
"logps/rejected": -276.83026123046875,
"loss": 0.149,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 0.48736995458602905,
"rewards/margins": 4.608451843261719,
"rewards/rejected": -4.121081829071045,
"step": 1080
},
{
"epoch": 1.12,
"learning_rate": 3.472859327217125e-07,
"logits/chosen": -2.8349316120147705,
"logits/rejected": -2.9043667316436768,
"logps/chosen": -364.6941833496094,
"logps/rejected": -376.5315856933594,
"loss": 0.1092,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 1.2117881774902344,
"rewards/margins": 5.32895565032959,
"rewards/rejected": -4.117166996002197,
"step": 1090
},
{
"epoch": 1.14,
"learning_rate": 3.4537461773700304e-07,
"logits/chosen": -2.985729694366455,
"logits/rejected": -2.8761606216430664,
"logps/chosen": -258.05841064453125,
"logps/rejected": -241.39053344726562,
"loss": 0.1114,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.6233962774276733,
"rewards/margins": 4.293813228607178,
"rewards/rejected": -3.6704165935516357,
"step": 1100
},
{
"epoch": 1.14,
"eval_logits/chosen": -2.9666378498077393,
"eval_logits/rejected": -2.95609450340271,
"eval_logps/chosen": -365.8598327636719,
"eval_logps/rejected": -310.414794921875,
"eval_loss": 0.534447431564331,
"eval_rewards/accuracies": 0.7460317611694336,
"eval_rewards/chosen": 0.1331927627325058,
"eval_rewards/margins": 1.9780747890472412,
"eval_rewards/rejected": -1.8448821306228638,
"eval_runtime": 164.1399,
"eval_samples_per_second": 12.185,
"eval_steps_per_second": 0.384,
"step": 1100
},
{
"epoch": 1.15,
"learning_rate": 3.434633027522936e-07,
"logits/chosen": -2.9507124423980713,
"logits/rejected": -2.9483211040496826,
"logps/chosen": -338.0868835449219,
"logps/rejected": -325.01483154296875,
"loss": 0.1007,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.6764055490493774,
"rewards/margins": 5.684920310974121,
"rewards/rejected": -4.008514404296875,
"step": 1110
},
{
"epoch": 1.16,
"learning_rate": 3.415519877675841e-07,
"logits/chosen": -2.976590156555176,
"logits/rejected": -3.025784730911255,
"logps/chosen": -277.34710693359375,
"logps/rejected": -323.576171875,
"loss": 0.1131,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 0.841607928276062,
"rewards/margins": 4.658609867095947,
"rewards/rejected": -3.8170018196105957,
"step": 1120
},
{
"epoch": 1.17,
"learning_rate": 3.3964067278287464e-07,
"logits/chosen": -3.0445570945739746,
"logits/rejected": -3.0413312911987305,
"logps/chosen": -337.9605407714844,
"logps/rejected": -288.26666259765625,
"loss": 0.1463,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.5894230604171753,
"rewards/margins": 5.3310723304748535,
"rewards/rejected": -3.7416489124298096,
"step": 1130
},
{
"epoch": 1.18,
"learning_rate": 3.377293577981651e-07,
"logits/chosen": -2.9471421241760254,
"logits/rejected": -2.9865708351135254,
"logps/chosen": -288.2189025878906,
"logps/rejected": -309.2388610839844,
"loss": 0.0934,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.4304211139678955,
"rewards/margins": 5.832246780395508,
"rewards/rejected": -4.401825428009033,
"step": 1140
},
{
"epoch": 1.19,
"learning_rate": 3.3581804281345565e-07,
"logits/chosen": -2.9803059101104736,
"logits/rejected": -2.9711978435516357,
"logps/chosen": -337.70697021484375,
"logps/rejected": -298.4077453613281,
"loss": 0.0967,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 1.4117494821548462,
"rewards/margins": 5.474527359008789,
"rewards/rejected": -4.062777519226074,
"step": 1150
},
{
"epoch": 1.2,
"learning_rate": 3.339067278287462e-07,
"logits/chosen": -2.8603241443634033,
"logits/rejected": -2.8709046840667725,
"logps/chosen": -312.73504638671875,
"logps/rejected": -306.9026794433594,
"loss": 0.0785,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.3839662075042725,
"rewards/margins": 5.900813102722168,
"rewards/rejected": -4.516847133636475,
"step": 1160
},
{
"epoch": 1.21,
"learning_rate": 3.319954128440367e-07,
"logits/chosen": -2.975525379180908,
"logits/rejected": -2.9611260890960693,
"logps/chosen": -325.86163330078125,
"logps/rejected": -285.2755432128906,
"loss": 0.0992,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.0602104663848877,
"rewards/margins": 4.849926948547363,
"rewards/rejected": -3.7897167205810547,
"step": 1170
},
{
"epoch": 1.22,
"learning_rate": 3.3008409785932725e-07,
"logits/chosen": -2.877586841583252,
"logits/rejected": -2.821748971939087,
"logps/chosen": -324.6281433105469,
"logps/rejected": -323.02301025390625,
"loss": 0.0892,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.1957648992538452,
"rewards/margins": 5.952631950378418,
"rewards/rejected": -4.756867408752441,
"step": 1180
},
{
"epoch": 1.23,
"learning_rate": 3.2817278287461773e-07,
"logits/chosen": -2.9451098442077637,
"logits/rejected": -2.9684863090515137,
"logps/chosen": -279.90216064453125,
"logps/rejected": -338.3842468261719,
"loss": 0.1045,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.2665460109710693,
"rewards/margins": 5.084068298339844,
"rewards/rejected": -3.8175220489501953,
"step": 1190
},
{
"epoch": 1.24,
"learning_rate": 3.262614678899082e-07,
"logits/chosen": -2.889819383621216,
"logits/rejected": -2.9235751628875732,
"logps/chosen": -303.02838134765625,
"logps/rejected": -356.177734375,
"loss": 0.1338,
"rewards/accuracies": 0.9375,
"rewards/chosen": 0.8252049684524536,
"rewards/margins": 5.507418155670166,
"rewards/rejected": -4.6822123527526855,
"step": 1200
},
{
"epoch": 1.24,
"eval_logits/chosen": -2.9508416652679443,
"eval_logits/rejected": -2.9459922313690186,
"eval_logps/chosen": -368.0057678222656,
"eval_logps/rejected": -313.3835144042969,
"eval_loss": 0.534950315952301,
"eval_rewards/accuracies": 0.773809552192688,
"eval_rewards/chosen": -0.0813969075679779,
"eval_rewards/margins": 2.0603599548339844,
"eval_rewards/rejected": -2.141756772994995,
"eval_runtime": 164.0736,
"eval_samples_per_second": 12.19,
"eval_steps_per_second": 0.384,
"step": 1200
},
{
"epoch": 1.25,
"learning_rate": 3.2435015290519874e-07,
"logits/chosen": -2.916611671447754,
"logits/rejected": -2.927777051925659,
"logps/chosen": -283.2217712402344,
"logps/rejected": -297.02850341796875,
"loss": 0.0893,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 1.0804212093353271,
"rewards/margins": 5.549715518951416,
"rewards/rejected": -4.469293594360352,
"step": 1210
},
{
"epoch": 1.26,
"learning_rate": 3.2243883792048927e-07,
"logits/chosen": -2.977875232696533,
"logits/rejected": -2.986704111099243,
"logps/chosen": -335.274658203125,
"logps/rejected": -380.4412536621094,
"loss": 0.1303,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.4366910457611084,
"rewards/margins": 5.666425704956055,
"rewards/rejected": -4.229735374450684,
"step": 1220
},
{
"epoch": 1.27,
"learning_rate": 3.205275229357798e-07,
"logits/chosen": -2.868638753890991,
"logits/rejected": -2.8948395252227783,
"logps/chosen": -387.9947204589844,
"logps/rejected": -389.3511657714844,
"loss": 0.1117,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.3278411626815796,
"rewards/margins": 5.698910236358643,
"rewards/rejected": -4.371068954467773,
"step": 1230
},
{
"epoch": 1.28,
"learning_rate": 3.186162079510703e-07,
"logits/chosen": -2.9128642082214355,
"logits/rejected": -2.91692852973938,
"logps/chosen": -351.5616149902344,
"logps/rejected": -373.9852600097656,
"loss": 0.1466,
"rewards/accuracies": 0.9375,
"rewards/chosen": 1.3745101690292358,
"rewards/margins": 5.025930404663086,
"rewards/rejected": -3.6514201164245605,
"step": 1240
},
{
"epoch": 1.29,
"learning_rate": 3.167048929663608e-07,
"logits/chosen": -2.955967426300049,
"logits/rejected": -2.923954486846924,
"logps/chosen": -278.7707824707031,
"logps/rejected": -281.9942321777344,
"loss": 0.1003,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": 1.0149275064468384,
"rewards/margins": 5.51505184173584,
"rewards/rejected": -4.500124931335449,
"step": 1250
},
{
"epoch": 1.3,
"learning_rate": 3.1479357798165134e-07,
"logits/chosen": -2.9661002159118652,
"logits/rejected": -2.948564052581787,
"logps/chosen": -339.5476989746094,
"logps/rejected": -321.3616638183594,
"loss": 0.0984,
"rewards/accuracies": 0.9375,
"rewards/chosen": 1.038356065750122,
"rewards/margins": 5.16934061050415,
"rewards/rejected": -4.130984306335449,
"step": 1260
},
{
"epoch": 1.31,
"learning_rate": 3.128822629969419e-07,
"logits/chosen": -2.862750291824341,
"logits/rejected": -2.8853306770324707,
"logps/chosen": -350.9757995605469,
"logps/rejected": -333.6067199707031,
"loss": 0.1195,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.147101640701294,
"rewards/margins": 5.550149917602539,
"rewards/rejected": -4.403048038482666,
"step": 1270
},
{
"epoch": 1.32,
"learning_rate": 3.109709480122324e-07,
"logits/chosen": -2.9103734493255615,
"logits/rejected": -2.9115426540374756,
"logps/chosen": -286.4703063964844,
"logps/rejected": -298.028076171875,
"loss": 0.1039,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 1.1303044557571411,
"rewards/margins": 5.467093467712402,
"rewards/rejected": -4.336789131164551,
"step": 1280
},
{
"epoch": 1.33,
"learning_rate": 3.0905963302752294e-07,
"logits/chosen": -2.7934536933898926,
"logits/rejected": -2.880432605743408,
"logps/chosen": -312.3811340332031,
"logps/rejected": -316.52215576171875,
"loss": 0.1168,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 0.9917739629745483,
"rewards/margins": 5.529503345489502,
"rewards/rejected": -4.537729263305664,
"step": 1290
},
{
"epoch": 1.34,
"learning_rate": 3.071483180428134e-07,
"logits/chosen": -2.9836788177490234,
"logits/rejected": -2.9340128898620605,
"logps/chosen": -382.5011291503906,
"logps/rejected": -295.3705749511719,
"loss": 0.0979,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 1.5801069736480713,
"rewards/margins": 5.381975173950195,
"rewards/rejected": -3.801867723464966,
"step": 1300
},
{
"epoch": 1.34,
"eval_logits/chosen": -2.9200918674468994,
"eval_logits/rejected": -2.9171833992004395,
"eval_logps/chosen": -368.1370849609375,
"eval_logps/rejected": -314.4656982421875,
"eval_loss": 0.5474238991737366,
"eval_rewards/accuracies": 0.7658730149269104,
"eval_rewards/chosen": -0.09453116357326508,
"eval_rewards/margins": 2.1554412841796875,
"eval_rewards/rejected": -2.249972343444824,
"eval_runtime": 164.7724,
"eval_samples_per_second": 12.138,
"eval_steps_per_second": 0.382,
"step": 1300
},
{
"epoch": 1.35,
"learning_rate": 3.0523700305810395e-07,
"logits/chosen": -2.9245269298553467,
"logits/rejected": -2.9436841011047363,
"logps/chosen": -330.574951171875,
"logps/rejected": -342.8641662597656,
"loss": 0.1043,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.3053131103515625,
"rewards/margins": 5.542339324951172,
"rewards/rejected": -4.237026214599609,
"step": 1310
},
{
"epoch": 1.36,
"learning_rate": 3.033256880733945e-07,
"logits/chosen": -2.9248242378234863,
"logits/rejected": -2.935176372528076,
"logps/chosen": -286.57171630859375,
"logps/rejected": -278.96746826171875,
"loss": 0.104,
"rewards/accuracies": 0.9375,
"rewards/chosen": 0.7352786660194397,
"rewards/margins": 5.138430118560791,
"rewards/rejected": -4.403151035308838,
"step": 1320
},
{
"epoch": 1.37,
"learning_rate": 3.01414373088685e-07,
"logits/chosen": -2.8516454696655273,
"logits/rejected": -2.7985987663269043,
"logps/chosen": -344.3554382324219,
"logps/rejected": -341.85986328125,
"loss": 0.1138,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 1.1177431344985962,
"rewards/margins": 5.806307792663574,
"rewards/rejected": -4.688565254211426,
"step": 1330
},
{
"epoch": 1.38,
"learning_rate": 2.9950305810397555e-07,
"logits/chosen": -2.906580686569214,
"logits/rejected": -2.97481369972229,
"logps/chosen": -335.2439880371094,
"logps/rejected": -320.96929931640625,
"loss": 0.1256,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 0.8055500984191895,
"rewards/margins": 6.211544990539551,
"rewards/rejected": -5.405994892120361,
"step": 1340
},
{
"epoch": 1.39,
"learning_rate": 2.9759174311926603e-07,
"logits/chosen": -2.9623026847839355,
"logits/rejected": -2.9445879459381104,
"logps/chosen": -323.3135070800781,
"logps/rejected": -329.90496826171875,
"loss": 0.1101,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.9566561579704285,
"rewards/margins": 7.040016174316406,
"rewards/rejected": -6.083359718322754,
"step": 1350
},
{
"epoch": 1.4,
"learning_rate": 2.9568042813455656e-07,
"logits/chosen": -2.8847999572753906,
"logits/rejected": -2.8803889751434326,
"logps/chosen": -342.5070495605469,
"logps/rejected": -269.67431640625,
"loss": 0.1087,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 0.4027617573738098,
"rewards/margins": 4.224934101104736,
"rewards/rejected": -3.8221726417541504,
"step": 1360
},
{
"epoch": 1.41,
"learning_rate": 2.937691131498471e-07,
"logits/chosen": -2.949441909790039,
"logits/rejected": -2.9045028686523438,
"logps/chosen": -338.4786376953125,
"logps/rejected": -334.69189453125,
"loss": 0.1096,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 0.9296592473983765,
"rewards/margins": 5.884530067443848,
"rewards/rejected": -4.954870700836182,
"step": 1370
},
{
"epoch": 1.42,
"learning_rate": 2.918577981651376e-07,
"logits/chosen": -2.932290554046631,
"logits/rejected": -2.9427378177642822,
"logps/chosen": -280.7291564941406,
"logps/rejected": -314.51953125,
"loss": 0.1177,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": 0.7406275868415833,
"rewards/margins": 5.4294962882995605,
"rewards/rejected": -4.688868522644043,
"step": 1380
},
{
"epoch": 1.43,
"learning_rate": 2.8994648318042816e-07,
"logits/chosen": -2.9206976890563965,
"logits/rejected": -2.9712460041046143,
"logps/chosen": -358.7654113769531,
"logps/rejected": -313.7131652832031,
"loss": 0.112,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 0.8611236810684204,
"rewards/margins": 5.0603203773498535,
"rewards/rejected": -4.199196815490723,
"step": 1390
},
{
"epoch": 1.44,
"learning_rate": 2.8803516819571863e-07,
"logits/chosen": -2.979775905609131,
"logits/rejected": -2.9886953830718994,
"logps/chosen": -379.1478576660156,
"logps/rejected": -358.28179931640625,
"loss": 0.1366,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.4315125942230225,
"rewards/margins": 6.2576189041137695,
"rewards/rejected": -4.826106071472168,
"step": 1400
},
{
"epoch": 1.44,
"eval_logits/chosen": -2.9143617153167725,
"eval_logits/rejected": -2.9134304523468018,
"eval_logps/chosen": -371.9402770996094,
"eval_logps/rejected": -315.933837890625,
"eval_loss": 0.5439518094062805,
"eval_rewards/accuracies": 0.7579365372657776,
"eval_rewards/chosen": -0.47485068440437317,
"eval_rewards/margins": 1.921934962272644,
"eval_rewards/rejected": -2.3967857360839844,
"eval_runtime": 165.1605,
"eval_samples_per_second": 12.109,
"eval_steps_per_second": 0.381,
"step": 1400
},
{
"epoch": 1.46,
"learning_rate": 2.8612385321100917e-07,
"logits/chosen": -2.8306632041931152,
"logits/rejected": -2.9071240425109863,
"logps/chosen": -294.634033203125,
"logps/rejected": -327.87896728515625,
"loss": 0.1281,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": 0.9478427171707153,
"rewards/margins": 5.661940097808838,
"rewards/rejected": -4.714097499847412,
"step": 1410
},
{
"epoch": 1.47,
"learning_rate": 2.842125382262997e-07,
"logits/chosen": -2.9503073692321777,
"logits/rejected": -2.9379420280456543,
"logps/chosen": -308.3216247558594,
"logps/rejected": -308.57574462890625,
"loss": 0.1361,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.2659790515899658,
"rewards/margins": 5.583965301513672,
"rewards/rejected": -4.317985534667969,
"step": 1420
},
{
"epoch": 1.48,
"learning_rate": 2.8230122324159023e-07,
"logits/chosen": -2.8611526489257812,
"logits/rejected": -2.9008944034576416,
"logps/chosen": -375.9707946777344,
"logps/rejected": -374.29913330078125,
"loss": 0.1194,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.3691354990005493,
"rewards/margins": 6.1049299240112305,
"rewards/rejected": -4.735795021057129,
"step": 1430
},
{
"epoch": 1.49,
"learning_rate": 2.8038990825688076e-07,
"logits/chosen": -2.994868516921997,
"logits/rejected": -2.9603443145751953,
"logps/chosen": -263.36474609375,
"logps/rejected": -250.1201934814453,
"loss": 0.1098,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": 0.8430646657943726,
"rewards/margins": 4.541081428527832,
"rewards/rejected": -3.698017120361328,
"step": 1440
},
{
"epoch": 1.5,
"learning_rate": 2.784785932721712e-07,
"logits/chosen": -2.9225330352783203,
"logits/rejected": -2.925787925720215,
"logps/chosen": -319.07574462890625,
"logps/rejected": -327.4895324707031,
"loss": 0.1336,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.1656725406646729,
"rewards/margins": 5.621832847595215,
"rewards/rejected": -4.456160068511963,
"step": 1450
},
{
"epoch": 1.51,
"learning_rate": 2.765672782874617e-07,
"logits/chosen": -2.896669864654541,
"logits/rejected": -2.895914316177368,
"logps/chosen": -331.0616760253906,
"logps/rejected": -286.6056213378906,
"loss": 0.1367,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.3613170385360718,
"rewards/margins": 5.62969970703125,
"rewards/rejected": -4.268383026123047,
"step": 1460
},
{
"epoch": 1.52,
"learning_rate": 2.7465596330275225e-07,
"logits/chosen": -2.955178737640381,
"logits/rejected": -2.9608724117279053,
"logps/chosen": -350.20703125,
"logps/rejected": -255.1401824951172,
"loss": 0.104,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": 1.233802080154419,
"rewards/margins": 5.684638023376465,
"rewards/rejected": -4.450836658477783,
"step": 1470
},
{
"epoch": 1.53,
"learning_rate": 2.727446483180428e-07,
"logits/chosen": -2.9283223152160645,
"logits/rejected": -2.952641010284424,
"logps/chosen": -313.20306396484375,
"logps/rejected": -316.35333251953125,
"loss": 0.1125,
"rewards/accuracies": 0.9375,
"rewards/chosen": 0.9838566780090332,
"rewards/margins": 5.377806186676025,
"rewards/rejected": -4.393948554992676,
"step": 1480
},
{
"epoch": 1.54,
"learning_rate": 2.708333333333333e-07,
"logits/chosen": -2.850559949874878,
"logits/rejected": -2.833322048187256,
"logps/chosen": -341.3831481933594,
"logps/rejected": -314.9398498535156,
"loss": 0.0943,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 0.7250410914421082,
"rewards/margins": 5.533560276031494,
"rewards/rejected": -4.808518409729004,
"step": 1490
},
{
"epoch": 1.55,
"learning_rate": 2.6892201834862385e-07,
"logits/chosen": -2.937903642654419,
"logits/rejected": -2.9050183296203613,
"logps/chosen": -326.52691650390625,
"logps/rejected": -302.30694580078125,
"loss": 0.1042,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 0.6643081903457642,
"rewards/margins": 5.094948768615723,
"rewards/rejected": -4.43064022064209,
"step": 1500
},
{
"epoch": 1.55,
"eval_logits/chosen": -2.930583953857422,
"eval_logits/rejected": -2.936053991317749,
"eval_logps/chosen": -372.2054138183594,
"eval_logps/rejected": -318.7686462402344,
"eval_loss": 0.552377462387085,
"eval_rewards/accuracies": 0.7698412537574768,
"eval_rewards/chosen": -0.5013648867607117,
"eval_rewards/margins": 2.178898334503174,
"eval_rewards/rejected": -2.6802632808685303,
"eval_runtime": 167.7329,
"eval_samples_per_second": 11.924,
"eval_steps_per_second": 0.376,
"step": 1500
},
{
"epoch": 1.56,
"learning_rate": 2.6701070336391433e-07,
"logits/chosen": -2.90950345993042,
"logits/rejected": -2.87695050239563,
"logps/chosen": -378.1885070800781,
"logps/rejected": -322.77337646484375,
"loss": 0.0924,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.4327830076217651,
"rewards/margins": 5.619394779205322,
"rewards/rejected": -4.186612129211426,
"step": 1510
},
{
"epoch": 1.57,
"learning_rate": 2.6509938837920486e-07,
"logits/chosen": -2.849907398223877,
"logits/rejected": -2.8833765983581543,
"logps/chosen": -289.51605224609375,
"logps/rejected": -320.0068054199219,
"loss": 0.0975,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 0.8192381858825684,
"rewards/margins": 4.99267053604126,
"rewards/rejected": -4.173432350158691,
"step": 1520
},
{
"epoch": 1.58,
"learning_rate": 2.631880733944954e-07,
"logits/chosen": -2.964118719100952,
"logits/rejected": -2.984459400177002,
"logps/chosen": -303.44866943359375,
"logps/rejected": -299.94482421875,
"loss": 0.105,
"rewards/accuracies": 0.9375,
"rewards/chosen": 0.37635737657546997,
"rewards/margins": 4.797235488891602,
"rewards/rejected": -4.420877933502197,
"step": 1530
},
{
"epoch": 1.59,
"learning_rate": 2.612767584097859e-07,
"logits/chosen": -2.9242002964019775,
"logits/rejected": -2.9575366973876953,
"logps/chosen": -308.75616455078125,
"logps/rejected": -282.21380615234375,
"loss": 0.1016,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.4492263793945312,
"rewards/margins": 5.21218204498291,
"rewards/rejected": -3.7629554271698,
"step": 1540
},
{
"epoch": 1.6,
"learning_rate": 2.5936544342507646e-07,
"logits/chosen": -2.902669668197632,
"logits/rejected": -2.932953357696533,
"logps/chosen": -306.1797790527344,
"logps/rejected": -315.36700439453125,
"loss": 0.1412,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.0359132289886475,
"rewards/margins": 5.480694770812988,
"rewards/rejected": -4.444781303405762,
"step": 1550
},
{
"epoch": 1.61,
"learning_rate": 2.5745412844036693e-07,
"logits/chosen": -2.862687110900879,
"logits/rejected": -2.9322876930236816,
"logps/chosen": -361.41583251953125,
"logps/rejected": -309.0520935058594,
"loss": 0.1228,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.3139979839324951,
"rewards/margins": 5.389029026031494,
"rewards/rejected": -4.07503080368042,
"step": 1560
},
{
"epoch": 1.62,
"learning_rate": 2.5554281345565747e-07,
"logits/chosen": -2.937886953353882,
"logits/rejected": -2.9431166648864746,
"logps/chosen": -312.0148620605469,
"logps/rejected": -329.90863037109375,
"loss": 0.1005,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 0.7804427146911621,
"rewards/margins": 5.692571640014648,
"rewards/rejected": -4.912128925323486,
"step": 1570
},
{
"epoch": 1.63,
"learning_rate": 2.53631498470948e-07,
"logits/chosen": -2.972658634185791,
"logits/rejected": -2.9699690341949463,
"logps/chosen": -347.6422424316406,
"logps/rejected": -312.10858154296875,
"loss": 0.0857,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 1.3729878664016724,
"rewards/margins": 5.7400736808776855,
"rewards/rejected": -4.367085933685303,
"step": 1580
},
{
"epoch": 1.64,
"learning_rate": 2.5172018348623853e-07,
"logits/chosen": -2.8885016441345215,
"logits/rejected": -2.9170756340026855,
"logps/chosen": -355.39813232421875,
"logps/rejected": -336.2825927734375,
"loss": 0.0852,
"rewards/accuracies": 0.9375,
"rewards/chosen": 1.1573688983917236,
"rewards/margins": 5.676226615905762,
"rewards/rejected": -4.518857479095459,
"step": 1590
},
{
"epoch": 1.65,
"learning_rate": 2.4980886850152906e-07,
"logits/chosen": -2.895519256591797,
"logits/rejected": -2.851107120513916,
"logps/chosen": -341.70904541015625,
"logps/rejected": -305.86480712890625,
"loss": 0.1313,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.0455483198165894,
"rewards/margins": 5.084899425506592,
"rewards/rejected": -4.039351463317871,
"step": 1600
},
{
"epoch": 1.65,
"eval_logits/chosen": -2.8998661041259766,
"eval_logits/rejected": -2.9059910774230957,
"eval_logps/chosen": -369.42547607421875,
"eval_logps/rejected": -313.8333435058594,
"eval_loss": 0.5333446264266968,
"eval_rewards/accuracies": 0.75,
"eval_rewards/chosen": -0.22337232530117035,
"eval_rewards/margins": 1.9633642435073853,
"eval_rewards/rejected": -2.1867363452911377,
"eval_runtime": 167.9925,
"eval_samples_per_second": 11.905,
"eval_steps_per_second": 0.375,
"step": 1600
},
{
"epoch": 1.66,
"learning_rate": 2.478975535168196e-07,
"logits/chosen": -2.892448663711548,
"logits/rejected": -2.8929343223571777,
"logps/chosen": -341.67431640625,
"logps/rejected": -287.61383056640625,
"loss": 0.1044,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.8163628578186035,
"rewards/margins": 4.475451469421387,
"rewards/rejected": -3.659088611602783,
"step": 1610
},
{
"epoch": 1.67,
"learning_rate": 2.459862385321101e-07,
"logits/chosen": -2.9265084266662598,
"logits/rejected": -2.9548892974853516,
"logps/chosen": -382.5340576171875,
"logps/rejected": -347.4888916015625,
"loss": 0.1018,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.6915686130523682,
"rewards/margins": 6.067580223083496,
"rewards/rejected": -4.376010894775391,
"step": 1620
},
{
"epoch": 1.68,
"learning_rate": 2.440749235474006e-07,
"logits/chosen": -2.8851680755615234,
"logits/rejected": -2.903552532196045,
"logps/chosen": -342.8496398925781,
"logps/rejected": -317.72845458984375,
"loss": 0.1005,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.0220366716384888,
"rewards/margins": 5.634666442871094,
"rewards/rejected": -4.6126298904418945,
"step": 1630
},
{
"epoch": 1.69,
"learning_rate": 2.421636085626911e-07,
"logits/chosen": -2.8142755031585693,
"logits/rejected": -2.8399770259857178,
"logps/chosen": -344.5146484375,
"logps/rejected": -301.95928955078125,
"loss": 0.0981,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.6624820232391357,
"rewards/margins": 6.2719035148620605,
"rewards/rejected": -4.6094207763671875,
"step": 1640
},
{
"epoch": 1.7,
"learning_rate": 2.402522935779816e-07,
"logits/chosen": -2.8734793663024902,
"logits/rejected": -2.876209259033203,
"logps/chosen": -339.1289367675781,
"logps/rejected": -351.3002014160156,
"loss": 0.1207,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 1.1800963878631592,
"rewards/margins": 5.746790409088135,
"rewards/rejected": -4.566694736480713,
"step": 1650
},
{
"epoch": 1.71,
"learning_rate": 2.3834097859327215e-07,
"logits/chosen": -2.922632932662964,
"logits/rejected": -2.973679780960083,
"logps/chosen": -285.7434997558594,
"logps/rejected": -304.81536865234375,
"loss": 0.1184,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 0.9965537786483765,
"rewards/margins": 5.077877998352051,
"rewards/rejected": -4.081325054168701,
"step": 1660
},
{
"epoch": 1.72,
"learning_rate": 2.3642966360856268e-07,
"logits/chosen": -2.9541144371032715,
"logits/rejected": -2.929344654083252,
"logps/chosen": -326.902587890625,
"logps/rejected": -306.6372985839844,
"loss": 0.0972,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.7346267700195312,
"rewards/margins": 5.666862964630127,
"rewards/rejected": -3.9322357177734375,
"step": 1670
},
{
"epoch": 1.73,
"learning_rate": 2.345183486238532e-07,
"logits/chosen": -2.857109785079956,
"logits/rejected": -2.8801960945129395,
"logps/chosen": -308.47369384765625,
"logps/rejected": -375.78692626953125,
"loss": 0.0899,
"rewards/accuracies": 0.9375,
"rewards/chosen": 1.088503122329712,
"rewards/margins": 5.432967185974121,
"rewards/rejected": -4.344464302062988,
"step": 1680
},
{
"epoch": 1.74,
"learning_rate": 2.3260703363914372e-07,
"logits/chosen": -2.9087703227996826,
"logits/rejected": -2.9551265239715576,
"logps/chosen": -381.01959228515625,
"logps/rejected": -338.1856994628906,
"loss": 0.105,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.9901115894317627,
"rewards/margins": 5.308182716369629,
"rewards/rejected": -3.318070888519287,
"step": 1690
},
{
"epoch": 1.75,
"learning_rate": 2.3069571865443425e-07,
"logits/chosen": -2.9682905673980713,
"logits/rejected": -2.9819796085357666,
"logps/chosen": -277.3031921386719,
"logps/rejected": -325.71649169921875,
"loss": 0.1629,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 0.5376863479614258,
"rewards/margins": 5.757152557373047,
"rewards/rejected": -5.219466209411621,
"step": 1700
},
{
"epoch": 1.75,
"eval_logits/chosen": -2.909576654434204,
"eval_logits/rejected": -2.9181904792785645,
"eval_logps/chosen": -371.09588623046875,
"eval_logps/rejected": -319.5571594238281,
"eval_loss": 0.5655122399330139,
"eval_rewards/accuracies": 0.75,
"eval_rewards/chosen": -0.3904118835926056,
"eval_rewards/margins": 2.3687071800231934,
"eval_rewards/rejected": -2.7591187953948975,
"eval_runtime": 164.0305,
"eval_samples_per_second": 12.193,
"eval_steps_per_second": 0.384,
"step": 1700
},
{
"epoch": 1.76,
"learning_rate": 2.2878440366972476e-07,
"logits/chosen": -2.9808902740478516,
"logits/rejected": -2.9869067668914795,
"logps/chosen": -326.5906677246094,
"logps/rejected": -384.11944580078125,
"loss": 0.0867,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 0.5069032311439514,
"rewards/margins": 5.799986362457275,
"rewards/rejected": -5.2930827140808105,
"step": 1710
},
{
"epoch": 1.78,
"learning_rate": 2.268730886850153e-07,
"logits/chosen": -2.8385868072509766,
"logits/rejected": -2.9057114124298096,
"logps/chosen": -325.4120178222656,
"logps/rejected": -309.59136962890625,
"loss": 0.0989,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": 1.1605224609375,
"rewards/margins": 5.591654300689697,
"rewards/rejected": -4.431131362915039,
"step": 1720
},
{
"epoch": 1.79,
"learning_rate": 2.249617737003058e-07,
"logits/chosen": -2.8839237689971924,
"logits/rejected": -2.8796088695526123,
"logps/chosen": -352.14886474609375,
"logps/rejected": -371.3978576660156,
"loss": 0.1089,
"rewards/accuracies": 0.9375,
"rewards/chosen": 1.0480362176895142,
"rewards/margins": 5.741724967956543,
"rewards/rejected": -4.693687915802002,
"step": 1730
},
{
"epoch": 1.8,
"learning_rate": 2.2305045871559633e-07,
"logits/chosen": -2.883430242538452,
"logits/rejected": -2.8605690002441406,
"logps/chosen": -345.23272705078125,
"logps/rejected": -331.52325439453125,
"loss": 0.1311,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.180057168006897,
"rewards/margins": 5.705449104309082,
"rewards/rejected": -4.525391578674316,
"step": 1740
},
{
"epoch": 1.81,
"learning_rate": 2.2113914373088686e-07,
"logits/chosen": -2.962857484817505,
"logits/rejected": -2.9696333408355713,
"logps/chosen": -377.6351623535156,
"logps/rejected": -362.8825378417969,
"loss": 0.1067,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.8095831871032715,
"rewards/margins": 5.408170700073242,
"rewards/rejected": -4.598587512969971,
"step": 1750
},
{
"epoch": 1.82,
"learning_rate": 2.1922782874617736e-07,
"logits/chosen": -2.9029316902160645,
"logits/rejected": -2.9539952278137207,
"logps/chosen": -289.267822265625,
"logps/rejected": -366.2077941894531,
"loss": 0.0929,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 0.47237950563430786,
"rewards/margins": 6.199611186981201,
"rewards/rejected": -5.727231502532959,
"step": 1760
},
{
"epoch": 1.83,
"learning_rate": 2.1731651376146787e-07,
"logits/chosen": -2.8582608699798584,
"logits/rejected": -2.8988289833068848,
"logps/chosen": -277.98406982421875,
"logps/rejected": -328.0066833496094,
"loss": 0.1048,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 0.7530109286308289,
"rewards/margins": 5.7497334480285645,
"rewards/rejected": -4.99672269821167,
"step": 1770
},
{
"epoch": 1.84,
"learning_rate": 2.154051987767584e-07,
"logits/chosen": -2.9023542404174805,
"logits/rejected": -2.9242827892303467,
"logps/chosen": -314.0538330078125,
"logps/rejected": -299.74420166015625,
"loss": 0.12,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 0.7526110410690308,
"rewards/margins": 5.213059425354004,
"rewards/rejected": -4.460447311401367,
"step": 1780
},
{
"epoch": 1.85,
"learning_rate": 2.134938837920489e-07,
"logits/chosen": -2.8436591625213623,
"logits/rejected": -2.8463809490203857,
"logps/chosen": -255.3565673828125,
"logps/rejected": -273.94464111328125,
"loss": 0.0818,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.06976697593927383,
"rewards/margins": 4.849638938903809,
"rewards/rejected": -4.919405937194824,
"step": 1790
},
{
"epoch": 1.86,
"learning_rate": 2.1158256880733944e-07,
"logits/chosen": -2.9030632972717285,
"logits/rejected": -2.8941729068756104,
"logps/chosen": -367.6248779296875,
"logps/rejected": -302.12249755859375,
"loss": 0.0993,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": 0.4400938153266907,
"rewards/margins": 5.627293586730957,
"rewards/rejected": -5.187199592590332,
"step": 1800
},
{
"epoch": 1.86,
"eval_logits/chosen": -2.8477160930633545,
"eval_logits/rejected": -2.8601999282836914,
"eval_logps/chosen": -374.3083801269531,
"eval_logps/rejected": -321.6667785644531,
"eval_loss": 0.5605445504188538,
"eval_rewards/accuracies": 0.7460317611694336,
"eval_rewards/chosen": -0.711660623550415,
"eval_rewards/margins": 2.2584221363067627,
"eval_rewards/rejected": -2.9700827598571777,
"eval_runtime": 164.7388,
"eval_samples_per_second": 12.14,
"eval_steps_per_second": 0.382,
"step": 1800
},
{
"epoch": 1.87,
"learning_rate": 2.0967125382262994e-07,
"logits/chosen": -2.8688273429870605,
"logits/rejected": -2.868739366531372,
"logps/chosen": -337.7546081542969,
"logps/rejected": -312.27569580078125,
"loss": 0.1163,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.7751233577728271,
"rewards/margins": 5.873146057128906,
"rewards/rejected": -5.098022937774658,
"step": 1810
},
{
"epoch": 1.88,
"learning_rate": 2.0775993883792048e-07,
"logits/chosen": -2.8374381065368652,
"logits/rejected": -2.8085215091705322,
"logps/chosen": -352.53192138671875,
"logps/rejected": -316.3230895996094,
"loss": 0.0933,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": 0.9547752141952515,
"rewards/margins": 5.203994274139404,
"rewards/rejected": -4.249218940734863,
"step": 1820
},
{
"epoch": 1.89,
"learning_rate": 2.05848623853211e-07,
"logits/chosen": -2.874891757965088,
"logits/rejected": -2.839573621749878,
"logps/chosen": -366.4833679199219,
"logps/rejected": -319.9959411621094,
"loss": 0.0966,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 1.212294101715088,
"rewards/margins": 5.914790630340576,
"rewards/rejected": -4.702496528625488,
"step": 1830
},
{
"epoch": 1.9,
"learning_rate": 2.0393730886850151e-07,
"logits/chosen": -2.8277204036712646,
"logits/rejected": -2.878105640411377,
"logps/chosen": -378.3955383300781,
"logps/rejected": -314.2088623046875,
"loss": 0.0863,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 0.7747803926467896,
"rewards/margins": 5.967954158782959,
"rewards/rejected": -5.193174362182617,
"step": 1840
},
{
"epoch": 1.91,
"learning_rate": 2.0202599388379205e-07,
"logits/chosen": -2.8658251762390137,
"logits/rejected": -2.8985071182250977,
"logps/chosen": -339.0852355957031,
"logps/rejected": -384.46112060546875,
"loss": 0.0786,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 0.9408707618713379,
"rewards/margins": 5.965841770172119,
"rewards/rejected": -5.024971008300781,
"step": 1850
},
{
"epoch": 1.92,
"learning_rate": 2.0011467889908258e-07,
"logits/chosen": -2.829246997833252,
"logits/rejected": -2.8732194900512695,
"logps/chosen": -381.65655517578125,
"logps/rejected": -284.0471496582031,
"loss": 0.0945,
"rewards/accuracies": 0.9375,
"rewards/chosen": 0.5882245898246765,
"rewards/margins": 5.461816787719727,
"rewards/rejected": -4.873592376708984,
"step": 1860
},
{
"epoch": 1.93,
"learning_rate": 1.9820336391437308e-07,
"logits/chosen": -2.8398656845092773,
"logits/rejected": -2.8620615005493164,
"logps/chosen": -309.2004089355469,
"logps/rejected": -296.1297302246094,
"loss": 0.093,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 0.8088359832763672,
"rewards/margins": 5.4316887855529785,
"rewards/rejected": -4.6228532791137695,
"step": 1870
},
{
"epoch": 1.94,
"learning_rate": 1.9629204892966362e-07,
"logits/chosen": -2.941847324371338,
"logits/rejected": -2.950911283493042,
"logps/chosen": -329.76617431640625,
"logps/rejected": -295.0538635253906,
"loss": 0.1113,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 1.2022342681884766,
"rewards/margins": 5.600251197814941,
"rewards/rejected": -4.398016929626465,
"step": 1880
},
{
"epoch": 1.95,
"learning_rate": 1.943807339449541e-07,
"logits/chosen": -2.8697052001953125,
"logits/rejected": -2.901094913482666,
"logps/chosen": -311.559326171875,
"logps/rejected": -333.4175720214844,
"loss": 0.0948,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 0.7903985977172852,
"rewards/margins": 5.597433090209961,
"rewards/rejected": -4.807034015655518,
"step": 1890
},
{
"epoch": 1.96,
"learning_rate": 1.9246941896024463e-07,
"logits/chosen": -2.939120054244995,
"logits/rejected": -2.9861233234405518,
"logps/chosen": -320.5481262207031,
"logps/rejected": -347.7875061035156,
"loss": 0.1116,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.6647524833679199,
"rewards/margins": 6.036587238311768,
"rewards/rejected": -5.3718342781066895,
"step": 1900
},
{
"epoch": 1.96,
"eval_logits/chosen": -2.914954900741577,
"eval_logits/rejected": -2.927724599838257,
"eval_logps/chosen": -373.57073974609375,
"eval_logps/rejected": -319.2250061035156,
"eval_loss": 0.5649252533912659,
"eval_rewards/accuracies": 0.7539682388305664,
"eval_rewards/chosen": -0.6378985047340393,
"eval_rewards/margins": 2.088006019592285,
"eval_rewards/rejected": -2.7259042263031006,
"eval_runtime": 164.2377,
"eval_samples_per_second": 12.177,
"eval_steps_per_second": 0.384,
"step": 1900
},
{
"epoch": 1.97,
"learning_rate": 1.9055810397553516e-07,
"logits/chosen": -2.9238085746765137,
"logits/rejected": -2.9308090209960938,
"logps/chosen": -313.63665771484375,
"logps/rejected": -304.2153625488281,
"loss": 0.1214,
"rewards/accuracies": 0.9375,
"rewards/chosen": 0.6549821496009827,
"rewards/margins": 4.487866401672363,
"rewards/rejected": -3.8328843116760254,
"step": 1910
},
{
"epoch": 1.98,
"learning_rate": 1.8864678899082566e-07,
"logits/chosen": -2.8484818935394287,
"logits/rejected": -2.866534948348999,
"logps/chosen": -347.75689697265625,
"logps/rejected": -279.4710693359375,
"loss": 0.1082,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": 0.3496394753456116,
"rewards/margins": 4.312170505523682,
"rewards/rejected": -3.9625308513641357,
"step": 1920
},
{
"epoch": 1.99,
"learning_rate": 1.867354740061162e-07,
"logits/chosen": -2.9492716789245605,
"logits/rejected": -2.956796169281006,
"logps/chosen": -307.85845947265625,
"logps/rejected": -332.1622619628906,
"loss": 0.1061,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.4475575387477875,
"rewards/margins": 5.942025184631348,
"rewards/rejected": -5.494467735290527,
"step": 1930
},
{
"epoch": 2.0,
"learning_rate": 1.8482415902140673e-07,
"logits/chosen": -2.923053503036499,
"logits/rejected": -2.920959949493408,
"logps/chosen": -331.311767578125,
"logps/rejected": -320.19586181640625,
"loss": 0.0801,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 1.0848143100738525,
"rewards/margins": 5.605216979980469,
"rewards/rejected": -4.520401954650879,
"step": 1940
},
{
"epoch": 2.01,
"learning_rate": 1.8291284403669723e-07,
"logits/chosen": -2.881058692932129,
"logits/rejected": -2.93363618850708,
"logps/chosen": -309.11212158203125,
"logps/rejected": -322.50665283203125,
"loss": 0.0254,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 1.7099040746688843,
"rewards/margins": 6.735787868499756,
"rewards/rejected": -5.025883674621582,
"step": 1950
},
{
"epoch": 2.02,
"learning_rate": 1.8100152905198777e-07,
"logits/chosen": -2.7668607234954834,
"logits/rejected": -2.7822773456573486,
"logps/chosen": -351.9031677246094,
"logps/rejected": -415.9180603027344,
"loss": 0.0185,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.090543508529663,
"rewards/margins": 7.8234100341796875,
"rewards/rejected": -6.7328667640686035,
"step": 1960
},
{
"epoch": 2.03,
"learning_rate": 1.7909021406727827e-07,
"logits/chosen": -2.9600331783294678,
"logits/rejected": -2.8843834400177,
"logps/chosen": -309.39642333984375,
"logps/rejected": -306.4966735839844,
"loss": 0.0282,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.9620206952095032,
"rewards/margins": 6.899697303771973,
"rewards/rejected": -5.937676429748535,
"step": 1970
},
{
"epoch": 2.04,
"learning_rate": 1.771788990825688e-07,
"logits/chosen": -2.923687696456909,
"logits/rejected": -2.9661598205566406,
"logps/chosen": -330.7653503417969,
"logps/rejected": -352.5653076171875,
"loss": 0.0243,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.7405881881713867,
"rewards/margins": 8.00406265258789,
"rewards/rejected": -6.263474941253662,
"step": 1980
},
{
"epoch": 2.05,
"learning_rate": 1.7526758409785934e-07,
"logits/chosen": -2.9299581050872803,
"logits/rejected": -2.8949360847473145,
"logps/chosen": -362.6274719238281,
"logps/rejected": -363.09149169921875,
"loss": 0.0148,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.0967247486114502,
"rewards/margins": 7.466977119445801,
"rewards/rejected": -6.37025260925293,
"step": 1990
},
{
"epoch": 2.06,
"learning_rate": 1.7335626911314984e-07,
"logits/chosen": -2.845986843109131,
"logits/rejected": -2.8671188354492188,
"logps/chosen": -274.60870361328125,
"logps/rejected": -295.59478759765625,
"loss": 0.0193,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -0.014614641666412354,
"rewards/margins": 6.743406772613525,
"rewards/rejected": -6.758021354675293,
"step": 2000
},
{
"epoch": 2.06,
"eval_logits/chosen": -2.882474422454834,
"eval_logits/rejected": -2.8919453620910645,
"eval_logps/chosen": -376.60406494140625,
"eval_logps/rejected": -329.82745361328125,
"eval_loss": 0.6121558547019958,
"eval_rewards/accuracies": 0.761904776096344,
"eval_rewards/chosen": -0.941230058670044,
"eval_rewards/margins": 2.8449153900146484,
"eval_rewards/rejected": -3.7861454486846924,
"eval_runtime": 164.9655,
"eval_samples_per_second": 12.124,
"eval_steps_per_second": 0.382,
"step": 2000
},
{
"epoch": 2.07,
"learning_rate": 1.7144495412844037e-07,
"logits/chosen": -2.9446756839752197,
"logits/rejected": -2.953831911087036,
"logps/chosen": -353.67376708984375,
"logps/rejected": -347.7017822265625,
"loss": 0.0214,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 0.4430414140224457,
"rewards/margins": 7.570870399475098,
"rewards/rejected": -7.127829074859619,
"step": 2010
},
{
"epoch": 2.08,
"learning_rate": 1.6953363914373088e-07,
"logits/chosen": -2.940734386444092,
"logits/rejected": -2.9746463298797607,
"logps/chosen": -348.05328369140625,
"logps/rejected": -333.2148742675781,
"loss": 0.012,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.2261013984680176,
"rewards/margins": 8.234363555908203,
"rewards/rejected": -7.008261680603027,
"step": 2020
},
{
"epoch": 2.09,
"learning_rate": 1.6762232415902138e-07,
"logits/chosen": -2.875319719314575,
"logits/rejected": -2.855180263519287,
"logps/chosen": -306.70050048828125,
"logps/rejected": -349.5177917480469,
"loss": 0.0203,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.3287349343299866,
"rewards/margins": 7.007230281829834,
"rewards/rejected": -6.678494930267334,
"step": 2030
},
{
"epoch": 2.11,
"learning_rate": 1.6571100917431192e-07,
"logits/chosen": -2.9315755367279053,
"logits/rejected": -2.930187702178955,
"logps/chosen": -306.041259765625,
"logps/rejected": -305.6824951171875,
"loss": 0.0151,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.5696347951889038,
"rewards/margins": 7.535808563232422,
"rewards/rejected": -6.9661736488342285,
"step": 2040
},
{
"epoch": 2.12,
"learning_rate": 1.6379969418960242e-07,
"logits/chosen": -2.8848538398742676,
"logits/rejected": -2.905867338180542,
"logps/chosen": -389.7286682128906,
"logps/rejected": -386.9409484863281,
"loss": 0.0148,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.3652772903442383,
"rewards/margins": 8.069811820983887,
"rewards/rejected": -6.704535484313965,
"step": 2050
},
{
"epoch": 2.13,
"learning_rate": 1.6188837920489295e-07,
"logits/chosen": -2.8828773498535156,
"logits/rejected": -2.8783280849456787,
"logps/chosen": -359.57666015625,
"logps/rejected": -339.35345458984375,
"loss": 0.0154,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.8870366811752319,
"rewards/margins": 8.021839141845703,
"rewards/rejected": -7.13480281829834,
"step": 2060
},
{
"epoch": 2.14,
"learning_rate": 1.5997706422018349e-07,
"logits/chosen": -2.88275408744812,
"logits/rejected": -2.929903984069824,
"logps/chosen": -322.4759216308594,
"logps/rejected": -423.30682373046875,
"loss": 0.0171,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 1.5736116170883179,
"rewards/margins": 8.667892456054688,
"rewards/rejected": -7.0942816734313965,
"step": 2070
},
{
"epoch": 2.15,
"learning_rate": 1.58065749235474e-07,
"logits/chosen": -2.90217661857605,
"logits/rejected": -2.8910233974456787,
"logps/chosen": -413.54522705078125,
"logps/rejected": -385.43341064453125,
"loss": 0.0127,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.9447764158248901,
"rewards/margins": 7.688788414001465,
"rewards/rejected": -6.744012355804443,
"step": 2080
},
{
"epoch": 2.16,
"learning_rate": 1.5615443425076452e-07,
"logits/chosen": -2.7480947971343994,
"logits/rejected": -2.747185230255127,
"logps/chosen": -354.87493896484375,
"logps/rejected": -351.3457946777344,
"loss": 0.0178,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.513919472694397,
"rewards/margins": 8.175249099731445,
"rewards/rejected": -7.661329746246338,
"step": 2090
},
{
"epoch": 2.17,
"learning_rate": 1.5424311926605506e-07,
"logits/chosen": -2.8367042541503906,
"logits/rejected": -2.8408215045928955,
"logps/chosen": -298.4134826660156,
"logps/rejected": -300.4717712402344,
"loss": 0.0175,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.2804573178291321,
"rewards/margins": 7.449028968811035,
"rewards/rejected": -7.729485511779785,
"step": 2100
},
{
"epoch": 2.17,
"eval_logits/chosen": -2.839301586151123,
"eval_logits/rejected": -2.8474462032318115,
"eval_logps/chosen": -383.2186279296875,
"eval_logps/rejected": -338.7977294921875,
"eval_loss": 0.6523212790489197,
"eval_rewards/accuracies": 0.7658730149269104,
"eval_rewards/chosen": -1.6026798486709595,
"eval_rewards/margins": 3.080495834350586,
"eval_rewards/rejected": -4.683175563812256,
"eval_runtime": 165.5125,
"eval_samples_per_second": 12.084,
"eval_steps_per_second": 0.381,
"step": 2100
},
{
"epoch": 2.18,
"learning_rate": 1.5233180428134556e-07,
"logits/chosen": -2.9054439067840576,
"logits/rejected": -2.913278102874756,
"logps/chosen": -323.6388244628906,
"logps/rejected": -323.73419189453125,
"loss": 0.0138,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.8671594858169556,
"rewards/margins": 7.387481689453125,
"rewards/rejected": -6.520320892333984,
"step": 2110
},
{
"epoch": 2.19,
"learning_rate": 1.504204892966361e-07,
"logits/chosen": -2.800830364227295,
"logits/rejected": -2.8197312355041504,
"logps/chosen": -359.3259582519531,
"logps/rejected": -394.8112487792969,
"loss": 0.0173,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.4949137568473816,
"rewards/margins": 9.693056106567383,
"rewards/rejected": -9.198141098022461,
"step": 2120
},
{
"epoch": 2.2,
"learning_rate": 1.485091743119266e-07,
"logits/chosen": -2.890476942062378,
"logits/rejected": -2.925356388092041,
"logps/chosen": -315.17742919921875,
"logps/rejected": -378.8518371582031,
"loss": 0.0188,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.27211472392082214,
"rewards/margins": 8.301239967346191,
"rewards/rejected": -8.02912425994873,
"step": 2130
},
{
"epoch": 2.21,
"learning_rate": 1.465978593272171e-07,
"logits/chosen": -2.820862054824829,
"logits/rejected": -2.8192131519317627,
"logps/chosen": -233.12344360351562,
"logps/rejected": -238.68014526367188,
"loss": 0.0187,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.23832440376281738,
"rewards/margins": 7.234049320220947,
"rewards/rejected": -7.472373008728027,
"step": 2140
},
{
"epoch": 2.22,
"learning_rate": 1.4468654434250764e-07,
"logits/chosen": -2.7812180519104004,
"logits/rejected": -2.839566946029663,
"logps/chosen": -400.56396484375,
"logps/rejected": -418.9078063964844,
"loss": 0.0183,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 1.306671380996704,
"rewards/margins": 9.309846878051758,
"rewards/rejected": -8.003175735473633,
"step": 2150
},
{
"epoch": 2.23,
"learning_rate": 1.4277522935779814e-07,
"logits/chosen": -2.857119083404541,
"logits/rejected": -2.8069121837615967,
"logps/chosen": -346.87091064453125,
"logps/rejected": -364.4837341308594,
"loss": 0.0185,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.42278409004211426,
"rewards/margins": 7.95212459564209,
"rewards/rejected": -7.5293402671813965,
"step": 2160
},
{
"epoch": 2.24,
"learning_rate": 1.4086391437308867e-07,
"logits/chosen": -2.828322172164917,
"logits/rejected": -2.872556209564209,
"logps/chosen": -371.8916015625,
"logps/rejected": -404.73162841796875,
"loss": 0.0183,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 0.5918793082237244,
"rewards/margins": 8.069284439086914,
"rewards/rejected": -7.477405548095703,
"step": 2170
},
{
"epoch": 2.25,
"learning_rate": 1.389525993883792e-07,
"logits/chosen": -2.8225388526916504,
"logits/rejected": -2.8491692543029785,
"logps/chosen": -293.30047607421875,
"logps/rejected": -313.1904296875,
"loss": 0.018,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.3963487148284912,
"rewards/margins": 7.192727565765381,
"rewards/rejected": -7.589076042175293,
"step": 2180
},
{
"epoch": 2.26,
"learning_rate": 1.370412844036697e-07,
"logits/chosen": -2.8727283477783203,
"logits/rejected": -2.850238561630249,
"logps/chosen": -306.93695068359375,
"logps/rejected": -345.2283020019531,
"loss": 0.0144,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.09160284698009491,
"rewards/margins": 7.823256492614746,
"rewards/rejected": -7.731653690338135,
"step": 2190
},
{
"epoch": 2.27,
"learning_rate": 1.3512996941896024e-07,
"logits/chosen": -2.8500583171844482,
"logits/rejected": -2.8594961166381836,
"logps/chosen": -283.15771484375,
"logps/rejected": -311.3097839355469,
"loss": 0.0131,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.03272407129406929,
"rewards/margins": 8.4783353805542,
"rewards/rejected": -8.445611953735352,
"step": 2200
},
{
"epoch": 2.27,
"eval_logits/chosen": -2.806851387023926,
"eval_logits/rejected": -2.812812328338623,
"eval_logps/chosen": -386.09039306640625,
"eval_logps/rejected": -342.2704162597656,
"eval_loss": 0.6702452898025513,
"eval_rewards/accuracies": 0.7420634627342224,
"eval_rewards/chosen": -1.8898613452911377,
"eval_rewards/margins": 3.1405844688415527,
"eval_rewards/rejected": -5.0304460525512695,
"eval_runtime": 165.1336,
"eval_samples_per_second": 12.111,
"eval_steps_per_second": 0.382,
"step": 2200
},
{
"epoch": 2.28,
"learning_rate": 1.3321865443425075e-07,
"logits/chosen": -2.7931952476501465,
"logits/rejected": -2.8073198795318604,
"logps/chosen": -338.2393493652344,
"logps/rejected": -352.142333984375,
"loss": 0.015,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.13923540711402893,
"rewards/margins": 8.090972900390625,
"rewards/rejected": -8.230208396911621,
"step": 2210
},
{
"epoch": 2.29,
"learning_rate": 1.3130733944954128e-07,
"logits/chosen": -2.74983811378479,
"logits/rejected": -2.748617649078369,
"logps/chosen": -358.42401123046875,
"logps/rejected": -402.30328369140625,
"loss": 0.0129,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 1.6026802062988281,
"rewards/margins": 9.088810920715332,
"rewards/rejected": -7.4861297607421875,
"step": 2220
},
{
"epoch": 2.3,
"learning_rate": 1.293960244648318e-07,
"logits/chosen": -2.8457603454589844,
"logits/rejected": -2.8344614505767822,
"logps/chosen": -365.7544860839844,
"logps/rejected": -347.2682189941406,
"loss": 0.0163,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.001628613448701799,
"rewards/margins": 8.202213287353516,
"rewards/rejected": -8.203841209411621,
"step": 2230
},
{
"epoch": 2.31,
"learning_rate": 1.2748470948012232e-07,
"logits/chosen": -2.8093724250793457,
"logits/rejected": -2.81803822517395,
"logps/chosen": -340.55352783203125,
"logps/rejected": -407.7304992675781,
"loss": 0.0119,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.47526517510414124,
"rewards/margins": 8.480849266052246,
"rewards/rejected": -8.005583763122559,
"step": 2240
},
{
"epoch": 2.32,
"learning_rate": 1.2557339449541285e-07,
"logits/chosen": -2.8672242164611816,
"logits/rejected": -2.855675220489502,
"logps/chosen": -343.7786865234375,
"logps/rejected": -365.4543151855469,
"loss": 0.0212,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 0.35550713539123535,
"rewards/margins": 8.239429473876953,
"rewards/rejected": -7.8839240074157715,
"step": 2250
},
{
"epoch": 2.33,
"learning_rate": 1.2366207951070336e-07,
"logits/chosen": -2.8647074699401855,
"logits/rejected": -2.8598999977111816,
"logps/chosen": -335.4911193847656,
"logps/rejected": -369.7025146484375,
"loss": 0.0148,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.1035120040178299,
"rewards/margins": 8.297709465026855,
"rewards/rejected": -8.401222229003906,
"step": 2260
},
{
"epoch": 2.34,
"learning_rate": 1.217507645259939e-07,
"logits/chosen": -2.8038744926452637,
"logits/rejected": -2.8534445762634277,
"logps/chosen": -327.49005126953125,
"logps/rejected": -348.63116455078125,
"loss": 0.0103,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.261627733707428,
"rewards/margins": 8.056116104125977,
"rewards/rejected": -7.794488430023193,
"step": 2270
},
{
"epoch": 2.35,
"learning_rate": 1.198394495412844e-07,
"logits/chosen": -2.879183053970337,
"logits/rejected": -2.9233028888702393,
"logps/chosen": -337.91790771484375,
"logps/rejected": -346.1882019042969,
"loss": 0.0176,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.12253670394420624,
"rewards/margins": 7.606545925140381,
"rewards/rejected": -7.4840087890625,
"step": 2280
},
{
"epoch": 2.36,
"learning_rate": 1.1792813455657493e-07,
"logits/chosen": -2.7606253623962402,
"logits/rejected": -2.8114898204803467,
"logps/chosen": -337.6861877441406,
"logps/rejected": -345.7854309082031,
"loss": 0.0258,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.3526005446910858,
"rewards/margins": 8.258612632751465,
"rewards/rejected": -7.906012058258057,
"step": 2290
},
{
"epoch": 2.37,
"learning_rate": 1.1601681957186543e-07,
"logits/chosen": -2.8984854221343994,
"logits/rejected": -2.912468910217285,
"logps/chosen": -334.9092102050781,
"logps/rejected": -334.67669677734375,
"loss": 0.0243,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.7173303961753845,
"rewards/margins": 7.802558898925781,
"rewards/rejected": -7.085227966308594,
"step": 2300
},
{
"epoch": 2.37,
"eval_logits/chosen": -2.8489737510681152,
"eval_logits/rejected": -2.854724407196045,
"eval_logps/chosen": -383.90655517578125,
"eval_logps/rejected": -339.3347473144531,
"eval_loss": 0.6559089422225952,
"eval_rewards/accuracies": 0.7698412537574768,
"eval_rewards/chosen": -1.6714773178100586,
"eval_rewards/margins": 3.0653984546661377,
"eval_rewards/rejected": -4.736875534057617,
"eval_runtime": 164.8339,
"eval_samples_per_second": 12.133,
"eval_steps_per_second": 0.382,
"step": 2300
},
{
"epoch": 2.38,
"learning_rate": 1.1410550458715595e-07,
"logits/chosen": -2.8347411155700684,
"logits/rejected": -2.851090908050537,
"logps/chosen": -329.1361999511719,
"logps/rejected": -359.9030456542969,
"loss": 0.0278,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 0.2210700958967209,
"rewards/margins": 7.436942100524902,
"rewards/rejected": -7.215872287750244,
"step": 2310
},
{
"epoch": 2.39,
"learning_rate": 1.1219418960244648e-07,
"logits/chosen": -2.8228423595428467,
"logits/rejected": -2.841404438018799,
"logps/chosen": -282.3636169433594,
"logps/rejected": -409.47979736328125,
"loss": 0.0135,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.2855607867240906,
"rewards/margins": 8.16025161743164,
"rewards/rejected": -8.445813179016113,
"step": 2320
},
{
"epoch": 2.4,
"learning_rate": 1.10282874617737e-07,
"logits/chosen": -2.8471851348876953,
"logits/rejected": -2.8798093795776367,
"logps/chosen": -295.41900634765625,
"logps/rejected": -340.5544738769531,
"loss": 0.0184,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.3259337544441223,
"rewards/margins": 8.084188461303711,
"rewards/rejected": -8.410122871398926,
"step": 2330
},
{
"epoch": 2.41,
"learning_rate": 1.0837155963302752e-07,
"logits/chosen": -2.636784076690674,
"logits/rejected": -2.740302562713623,
"logps/chosen": -287.13702392578125,
"logps/rejected": -391.1552429199219,
"loss": 0.0185,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.528149425983429,
"rewards/margins": 9.060527801513672,
"rewards/rejected": -8.532378196716309,
"step": 2340
},
{
"epoch": 2.43,
"learning_rate": 1.0646024464831804e-07,
"logits/chosen": -2.753213882446289,
"logits/rejected": -2.822252035140991,
"logps/chosen": -369.473388671875,
"logps/rejected": -360.38983154296875,
"loss": 0.0221,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.3261668086051941,
"rewards/margins": 8.502967834472656,
"rewards/rejected": -8.829133033752441,
"step": 2350
},
{
"epoch": 2.44,
"learning_rate": 1.0454892966360856e-07,
"logits/chosen": -2.782691717147827,
"logits/rejected": -2.868027448654175,
"logps/chosen": -333.0803527832031,
"logps/rejected": -355.0961608886719,
"loss": 0.0144,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.5834169983863831,
"rewards/margins": 8.312009811401367,
"rewards/rejected": -8.895425796508789,
"step": 2360
},
{
"epoch": 2.45,
"learning_rate": 1.0263761467889908e-07,
"logits/chosen": -2.770711898803711,
"logits/rejected": -2.796137809753418,
"logps/chosen": -336.739990234375,
"logps/rejected": -372.0965576171875,
"loss": 0.0151,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.35953274369239807,
"rewards/margins": 8.809865951538086,
"rewards/rejected": -8.450332641601562,
"step": 2370
},
{
"epoch": 2.46,
"learning_rate": 1.007262996941896e-07,
"logits/chosen": -2.7547390460968018,
"logits/rejected": -2.7793593406677246,
"logps/chosen": -335.936279296875,
"logps/rejected": -330.6647033691406,
"loss": 0.0167,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.02794322930276394,
"rewards/margins": 8.719170570373535,
"rewards/rejected": -8.747113227844238,
"step": 2380
},
{
"epoch": 2.47,
"learning_rate": 9.881498470948011e-08,
"logits/chosen": -2.846524477005005,
"logits/rejected": -2.799567222595215,
"logps/chosen": -343.198486328125,
"logps/rejected": -335.6533508300781,
"loss": 0.0136,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.013063406571745872,
"rewards/margins": 8.952108383178711,
"rewards/rejected": -8.965171813964844,
"step": 2390
},
{
"epoch": 2.48,
"learning_rate": 9.690366972477065e-08,
"logits/chosen": -2.85577392578125,
"logits/rejected": -2.8093135356903076,
"logps/chosen": -333.2208251953125,
"logps/rejected": -358.0810241699219,
"loss": 0.0142,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.41089487075805664,
"rewards/margins": 8.724878311157227,
"rewards/rejected": -9.135773658752441,
"step": 2400
},
{
"epoch": 2.48,
"eval_logits/chosen": -2.835172653198242,
"eval_logits/rejected": -2.839359760284424,
"eval_logps/chosen": -386.6546936035156,
"eval_logps/rejected": -343.19000244140625,
"eval_loss": 0.6733575463294983,
"eval_rewards/accuracies": 0.7579365372657776,
"eval_rewards/chosen": -1.946290373802185,
"eval_rewards/margins": 3.1761116981506348,
"eval_rewards/rejected": -5.122402191162109,
"eval_runtime": 165.3843,
"eval_samples_per_second": 12.093,
"eval_steps_per_second": 0.381,
"step": 2400
},
{
"epoch": 2.49,
"learning_rate": 9.499235474006116e-08,
"logits/chosen": -2.846043109893799,
"logits/rejected": -2.8555102348327637,
"logps/chosen": -376.3670349121094,
"logps/rejected": -341.2032470703125,
"loss": 0.0246,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.03427610173821449,
"rewards/margins": 8.654411315917969,
"rewards/rejected": -8.688688278198242,
"step": 2410
},
{
"epoch": 2.5,
"learning_rate": 9.308103975535168e-08,
"logits/chosen": -2.8411316871643066,
"logits/rejected": -2.8570432662963867,
"logps/chosen": -373.59844970703125,
"logps/rejected": -401.067138671875,
"loss": 0.0176,
"rewards/accuracies": 1.0,
"rewards/chosen": 1.0549451112747192,
"rewards/margins": 8.842530250549316,
"rewards/rejected": -7.787585258483887,
"step": 2420
},
{
"epoch": 2.51,
"learning_rate": 9.116972477064219e-08,
"logits/chosen": -2.895292282104492,
"logits/rejected": -2.854443073272705,
"logps/chosen": -345.359375,
"logps/rejected": -408.4029846191406,
"loss": 0.0169,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.30679136514663696,
"rewards/margins": 7.920645713806152,
"rewards/rejected": -8.227437019348145,
"step": 2430
},
{
"epoch": 2.52,
"learning_rate": 8.925840978593272e-08,
"logits/chosen": -2.835501194000244,
"logits/rejected": -2.896915912628174,
"logps/chosen": -264.5487365722656,
"logps/rejected": -387.1824951171875,
"loss": 0.019,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.3389735221862793,
"rewards/margins": 9.347002029418945,
"rewards/rejected": -9.008028030395508,
"step": 2440
},
{
"epoch": 2.53,
"learning_rate": 8.734709480122324e-08,
"logits/chosen": -2.806790828704834,
"logits/rejected": -2.8148555755615234,
"logps/chosen": -308.4158630371094,
"logps/rejected": -376.0751953125,
"loss": 0.0166,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -0.38033682107925415,
"rewards/margins": 8.450287818908691,
"rewards/rejected": -8.8306245803833,
"step": 2450
},
{
"epoch": 2.54,
"learning_rate": 8.543577981651376e-08,
"logits/chosen": -2.7967381477355957,
"logits/rejected": -2.792023181915283,
"logps/chosen": -455.0721740722656,
"logps/rejected": -405.89501953125,
"loss": 0.0156,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.7061554789543152,
"rewards/margins": 9.382705688476562,
"rewards/rejected": -8.676549911499023,
"step": 2460
},
{
"epoch": 2.55,
"learning_rate": 8.352446483180428e-08,
"logits/chosen": -2.8607754707336426,
"logits/rejected": -2.8268520832061768,
"logps/chosen": -331.96820068359375,
"logps/rejected": -321.39422607421875,
"loss": 0.0236,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.4299864172935486,
"rewards/margins": 8.559895515441895,
"rewards/rejected": -8.129908561706543,
"step": 2470
},
{
"epoch": 2.56,
"learning_rate": 8.161314984709481e-08,
"logits/chosen": -2.8822827339172363,
"logits/rejected": -2.893578052520752,
"logps/chosen": -339.42449951171875,
"logps/rejected": -356.1263427734375,
"loss": 0.0083,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.2752775549888611,
"rewards/margins": 8.704290390014648,
"rewards/rejected": -8.4290132522583,
"step": 2480
},
{
"epoch": 2.57,
"learning_rate": 7.970183486238531e-08,
"logits/chosen": -2.828721284866333,
"logits/rejected": -2.833087205886841,
"logps/chosen": -328.60418701171875,
"logps/rejected": -360.6470642089844,
"loss": 0.0176,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.30725544691085815,
"rewards/margins": 8.530462265014648,
"rewards/rejected": -8.223207473754883,
"step": 2490
},
{
"epoch": 2.58,
"learning_rate": 7.779051987767583e-08,
"logits/chosen": -2.8711142539978027,
"logits/rejected": -2.892519950866699,
"logps/chosen": -370.29339599609375,
"logps/rejected": -355.296875,
"loss": 0.0211,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.0712406188249588,
"rewards/margins": 8.155640602111816,
"rewards/rejected": -8.084399223327637,
"step": 2500
},
{
"epoch": 2.58,
"eval_logits/chosen": -2.8333258628845215,
"eval_logits/rejected": -2.8368897438049316,
"eval_logps/chosen": -388.3058776855469,
"eval_logps/rejected": -347.57440185546875,
"eval_loss": 0.6890397667884827,
"eval_rewards/accuracies": 0.7698412537574768,
"eval_rewards/chosen": -2.1114044189453125,
"eval_rewards/margins": 3.4494407176971436,
"eval_rewards/rejected": -5.560845375061035,
"eval_runtime": 164.7492,
"eval_samples_per_second": 12.14,
"eval_steps_per_second": 0.382,
"step": 2500
},
{
"epoch": 2.59,
"learning_rate": 7.587920489296635e-08,
"logits/chosen": -2.855881690979004,
"logits/rejected": -2.8854427337646484,
"logps/chosen": -351.69769287109375,
"logps/rejected": -358.4553527832031,
"loss": 0.0164,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.2860868573188782,
"rewards/margins": 8.237478256225586,
"rewards/rejected": -8.523565292358398,
"step": 2510
},
{
"epoch": 2.6,
"learning_rate": 7.396788990825688e-08,
"logits/chosen": -2.8813681602478027,
"logits/rejected": -2.9079108238220215,
"logps/chosen": -322.7754821777344,
"logps/rejected": -327.5832824707031,
"loss": 0.0137,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.06775089353322983,
"rewards/margins": 8.004460334777832,
"rewards/rejected": -8.072211265563965,
"step": 2520
},
{
"epoch": 2.61,
"learning_rate": 7.20565749235474e-08,
"logits/chosen": -2.810084819793701,
"logits/rejected": -2.815389394760132,
"logps/chosen": -325.9468688964844,
"logps/rejected": -330.6631164550781,
"loss": 0.0133,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.12425418943166733,
"rewards/margins": 8.769124984741211,
"rewards/rejected": -8.893379211425781,
"step": 2530
},
{
"epoch": 2.62,
"learning_rate": 7.014525993883792e-08,
"logits/chosen": -2.7919540405273438,
"logits/rejected": -2.7934675216674805,
"logps/chosen": -353.1927185058594,
"logps/rejected": -365.3847351074219,
"loss": 0.0154,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.2072305679321289,
"rewards/margins": 8.390886306762695,
"rewards/rejected": -8.18365478515625,
"step": 2540
},
{
"epoch": 2.63,
"learning_rate": 6.823394495412843e-08,
"logits/chosen": -2.7678780555725098,
"logits/rejected": -2.765697479248047,
"logps/chosen": -358.8880615234375,
"logps/rejected": -376.55706787109375,
"loss": 0.02,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.11311036348342896,
"rewards/margins": 9.944357872009277,
"rewards/rejected": -9.83124828338623,
"step": 2550
},
{
"epoch": 2.64,
"learning_rate": 6.632262996941895e-08,
"logits/chosen": -2.834345817565918,
"logits/rejected": -2.7858288288116455,
"logps/chosen": -336.33648681640625,
"logps/rejected": -366.84991455078125,
"loss": 0.0164,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.12319626659154892,
"rewards/margins": 9.35567569732666,
"rewards/rejected": -9.478872299194336,
"step": 2560
},
{
"epoch": 2.65,
"learning_rate": 6.441131498470948e-08,
"logits/chosen": -2.847996950149536,
"logits/rejected": -2.863615036010742,
"logps/chosen": -304.58502197265625,
"logps/rejected": -352.5277404785156,
"loss": 0.0135,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.37131237983703613,
"rewards/margins": 8.316202163696289,
"rewards/rejected": -8.687514305114746,
"step": 2570
},
{
"epoch": 2.66,
"learning_rate": 6.25e-08,
"logits/chosen": -2.8685457706451416,
"logits/rejected": -2.876739501953125,
"logps/chosen": -394.1883850097656,
"logps/rejected": -382.19287109375,
"loss": 0.0135,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.5528956055641174,
"rewards/margins": 9.068865776062012,
"rewards/rejected": -8.515970230102539,
"step": 2580
},
{
"epoch": 2.67,
"learning_rate": 6.058868501529052e-08,
"logits/chosen": -2.9075653553009033,
"logits/rejected": -2.8715763092041016,
"logps/chosen": -366.0291442871094,
"logps/rejected": -358.59381103515625,
"loss": 0.0202,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.14853370189666748,
"rewards/margins": 8.612794876098633,
"rewards/rejected": -8.464262008666992,
"step": 2590
},
{
"epoch": 2.68,
"learning_rate": 5.8677370030581035e-08,
"logits/chosen": -2.797910213470459,
"logits/rejected": -2.840148687362671,
"logps/chosen": -331.3750305175781,
"logps/rejected": -344.34332275390625,
"loss": 0.011,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.22872698307037354,
"rewards/margins": 8.751152038574219,
"rewards/rejected": -8.979879379272461,
"step": 2600
},
{
"epoch": 2.68,
"eval_logits/chosen": -2.8258047103881836,
"eval_logits/rejected": -2.8298983573913574,
"eval_logps/chosen": -390.2113952636719,
"eval_logps/rejected": -350.03887939453125,
"eval_loss": 0.6998910307884216,
"eval_rewards/accuracies": 0.7658730149269104,
"eval_rewards/chosen": -2.301961660385132,
"eval_rewards/margins": 3.5053274631500244,
"eval_rewards/rejected": -5.807290077209473,
"eval_runtime": 164.7101,
"eval_samples_per_second": 12.143,
"eval_steps_per_second": 0.382,
"step": 2600
},
{
"epoch": 2.69,
"learning_rate": 5.6766055045871554e-08,
"logits/chosen": -2.837218761444092,
"logits/rejected": -2.8603646755218506,
"logps/chosen": -325.1515197753906,
"logps/rejected": -377.93707275390625,
"loss": 0.0122,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.7425957918167114,
"rewards/margins": 9.372137069702148,
"rewards/rejected": -8.629541397094727,
"step": 2610
},
{
"epoch": 2.7,
"learning_rate": 5.485474006116208e-08,
"logits/chosen": -2.859614372253418,
"logits/rejected": -2.907731294631958,
"logps/chosen": -323.7126159667969,
"logps/rejected": -337.7955627441406,
"loss": 0.0124,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.9508829116821289,
"rewards/margins": 9.375367164611816,
"rewards/rejected": -8.424482345581055,
"step": 2620
},
{
"epoch": 2.71,
"learning_rate": 5.294342507645259e-08,
"logits/chosen": -2.8355846405029297,
"logits/rejected": -2.8445563316345215,
"logps/chosen": -387.78021240234375,
"logps/rejected": -341.34332275390625,
"loss": 0.0122,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.06057599186897278,
"rewards/margins": 8.586808204650879,
"rewards/rejected": -8.647383689880371,
"step": 2630
},
{
"epoch": 2.72,
"learning_rate": 5.1032110091743117e-08,
"logits/chosen": -2.8416004180908203,
"logits/rejected": -2.8135132789611816,
"logps/chosen": -294.2474670410156,
"logps/rejected": -342.1490173339844,
"loss": 0.0112,
"rewards/accuracies": 1.0,
"rewards/chosen": -1.0905003547668457,
"rewards/margins": 7.913638114929199,
"rewards/rejected": -9.004137992858887,
"step": 2640
},
{
"epoch": 2.73,
"learning_rate": 4.9120795107033635e-08,
"logits/chosen": -2.808621406555176,
"logits/rejected": -2.8129184246063232,
"logps/chosen": -361.94146728515625,
"logps/rejected": -372.2251892089844,
"loss": 0.0108,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.09845595061779022,
"rewards/margins": 9.0397367477417,
"rewards/rejected": -9.138191223144531,
"step": 2650
},
{
"epoch": 2.75,
"learning_rate": 4.7209480122324154e-08,
"logits/chosen": -2.857626438140869,
"logits/rejected": -2.854701042175293,
"logps/chosen": -334.3382568359375,
"logps/rejected": -397.60504150390625,
"loss": 0.0135,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.4237436354160309,
"rewards/margins": 8.734308242797852,
"rewards/rejected": -9.158050537109375,
"step": 2660
},
{
"epoch": 2.76,
"learning_rate": 4.529816513761467e-08,
"logits/chosen": -2.833742618560791,
"logits/rejected": -2.848910093307495,
"logps/chosen": -376.8042297363281,
"logps/rejected": -431.098388671875,
"loss": 0.0185,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.09423612058162689,
"rewards/margins": 8.842924118041992,
"rewards/rejected": -8.937159538269043,
"step": 2670
},
{
"epoch": 2.77,
"learning_rate": 4.33868501529052e-08,
"logits/chosen": -2.8585665225982666,
"logits/rejected": -2.888023614883423,
"logps/chosen": -299.4255065917969,
"logps/rejected": -347.65032958984375,
"loss": 0.0227,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -0.39737778902053833,
"rewards/margins": 8.897387504577637,
"rewards/rejected": -9.294764518737793,
"step": 2680
},
{
"epoch": 2.78,
"learning_rate": 4.147553516819572e-08,
"logits/chosen": -2.7752485275268555,
"logits/rejected": -2.7679455280303955,
"logps/chosen": -288.93524169921875,
"logps/rejected": -362.3262634277344,
"loss": 0.0138,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.21344709396362305,
"rewards/margins": 8.969237327575684,
"rewards/rejected": -9.182684898376465,
"step": 2690
},
{
"epoch": 2.79,
"learning_rate": 3.9564220183486236e-08,
"logits/chosen": -2.771638870239258,
"logits/rejected": -2.7894372940063477,
"logps/chosen": -358.27276611328125,
"logps/rejected": -337.4700012207031,
"loss": 0.0114,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.08803452551364899,
"rewards/margins": 8.622003555297852,
"rewards/rejected": -8.710036277770996,
"step": 2700
},
{
"epoch": 2.79,
"eval_logits/chosen": -2.817159414291382,
"eval_logits/rejected": -2.820690393447876,
"eval_logps/chosen": -389.5738525390625,
"eval_logps/rejected": -348.8511962890625,
"eval_loss": 0.6951248645782471,
"eval_rewards/accuracies": 0.7698412537574768,
"eval_rewards/chosen": -2.238208055496216,
"eval_rewards/margins": 3.4503118991851807,
"eval_rewards/rejected": -5.6885199546813965,
"eval_runtime": 164.1407,
"eval_samples_per_second": 12.185,
"eval_steps_per_second": 0.384,
"step": 2700
},
{
"epoch": 2.8,
"learning_rate": 3.7652905198776755e-08,
"logits/chosen": -2.8291115760803223,
"logits/rejected": -2.812997817993164,
"logps/chosen": -361.16973876953125,
"logps/rejected": -371.3473205566406,
"loss": 0.0109,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.2673804759979248,
"rewards/margins": 8.214799880981445,
"rewards/rejected": -8.482179641723633,
"step": 2710
},
{
"epoch": 2.81,
"learning_rate": 3.574159021406728e-08,
"logits/chosen": -2.8169431686401367,
"logits/rejected": -2.780579090118408,
"logps/chosen": -340.25567626953125,
"logps/rejected": -452.1532287597656,
"loss": 0.0125,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.4266482889652252,
"rewards/margins": 8.901152610778809,
"rewards/rejected": -9.327801704406738,
"step": 2720
},
{
"epoch": 2.82,
"learning_rate": 3.383027522935779e-08,
"logits/chosen": -2.852733850479126,
"logits/rejected": -2.8627407550811768,
"logps/chosen": -345.10504150390625,
"logps/rejected": -381.2701416015625,
"loss": 0.0144,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.6709401607513428,
"rewards/margins": 9.368196487426758,
"rewards/rejected": -8.697256088256836,
"step": 2730
},
{
"epoch": 2.83,
"learning_rate": 3.191896024464832e-08,
"logits/chosen": -2.847033977508545,
"logits/rejected": -2.880303382873535,
"logps/chosen": -344.7592468261719,
"logps/rejected": -366.505615234375,
"loss": 0.0093,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.2561042606830597,
"rewards/margins": 9.626019477844238,
"rewards/rejected": -9.369915008544922,
"step": 2740
},
{
"epoch": 2.84,
"learning_rate": 3.0007645259938836e-08,
"logits/chosen": -2.829150676727295,
"logits/rejected": -2.8306522369384766,
"logps/chosen": -299.6269836425781,
"logps/rejected": -362.5341796875,
"loss": 0.0191,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.312942773103714,
"rewards/margins": 10.283103942871094,
"rewards/rejected": -9.970161437988281,
"step": 2750
},
{
"epoch": 2.85,
"learning_rate": 2.809633027522936e-08,
"logits/chosen": -2.815882444381714,
"logits/rejected": -2.7824299335479736,
"logps/chosen": -315.24993896484375,
"logps/rejected": -347.3058776855469,
"loss": 0.0207,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.3008100688457489,
"rewards/margins": 8.677629470825195,
"rewards/rejected": -8.978440284729004,
"step": 2760
},
{
"epoch": 2.86,
"learning_rate": 2.6185015290519877e-08,
"logits/chosen": -2.8008246421813965,
"logits/rejected": -2.7953882217407227,
"logps/chosen": -333.69329833984375,
"logps/rejected": -373.4231872558594,
"loss": 0.0131,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.08695399761199951,
"rewards/margins": 7.9680304527282715,
"rewards/rejected": -8.054986000061035,
"step": 2770
},
{
"epoch": 2.87,
"learning_rate": 2.4273700305810396e-08,
"logits/chosen": -2.790097951889038,
"logits/rejected": -2.827036142349243,
"logps/chosen": -378.98236083984375,
"logps/rejected": -420.4288635253906,
"loss": 0.0135,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.11630038917064667,
"rewards/margins": 10.843367576599121,
"rewards/rejected": -10.727069854736328,
"step": 2780
},
{
"epoch": 2.88,
"learning_rate": 2.2362385321100918e-08,
"logits/chosen": -2.7879481315612793,
"logits/rejected": -2.7845988273620605,
"logps/chosen": -350.80572509765625,
"logps/rejected": -345.9007873535156,
"loss": 0.018,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -0.001116895698942244,
"rewards/margins": 8.821355819702148,
"rewards/rejected": -8.822473526000977,
"step": 2790
},
{
"epoch": 2.89,
"learning_rate": 2.0451070336391437e-08,
"logits/chosen": -2.7627055644989014,
"logits/rejected": -2.7335832118988037,
"logps/chosen": -341.35662841796875,
"logps/rejected": -318.8877258300781,
"loss": 0.0437,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 0.1377846747636795,
"rewards/margins": 8.672611236572266,
"rewards/rejected": -8.534826278686523,
"step": 2800
},
{
"epoch": 2.89,
"eval_logits/chosen": -2.8117332458496094,
"eval_logits/rejected": -2.8151025772094727,
"eval_logps/chosen": -389.4859924316406,
"eval_logps/rejected": -348.1217346191406,
"eval_loss": 0.6910788416862488,
"eval_rewards/accuracies": 0.7658730149269104,
"eval_rewards/chosen": -2.229426622390747,
"eval_rewards/margins": 3.3861491680145264,
"eval_rewards/rejected": -5.615575313568115,
"eval_runtime": 165.138,
"eval_samples_per_second": 12.111,
"eval_steps_per_second": 0.381,
"step": 2800
},
{
"epoch": 2.9,
"learning_rate": 1.8539755351681956e-08,
"logits/chosen": -2.7365012168884277,
"logits/rejected": -2.788407325744629,
"logps/chosen": -330.33197021484375,
"logps/rejected": -380.91168212890625,
"loss": 0.0163,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -0.0645025223493576,
"rewards/margins": 8.317387580871582,
"rewards/rejected": -8.381890296936035,
"step": 2810
},
{
"epoch": 2.91,
"learning_rate": 1.6628440366972478e-08,
"logits/chosen": -2.8033618927001953,
"logits/rejected": -2.8255763053894043,
"logps/chosen": -373.3817443847656,
"logps/rejected": -359.75421142578125,
"loss": 0.0142,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 0.008309101685881615,
"rewards/margins": 8.045055389404297,
"rewards/rejected": -8.03674602508545,
"step": 2820
},
{
"epoch": 2.92,
"learning_rate": 1.4717125382262997e-08,
"logits/chosen": -2.854548692703247,
"logits/rejected": -2.8665812015533447,
"logps/chosen": -339.0101318359375,
"logps/rejected": -377.62847900390625,
"loss": 0.0123,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": -0.6132253408432007,
"rewards/margins": 8.178349494934082,
"rewards/rejected": -8.791574478149414,
"step": 2830
},
{
"epoch": 2.93,
"learning_rate": 1.2805810397553517e-08,
"logits/chosen": -2.8801310062408447,
"logits/rejected": -2.826385021209717,
"logps/chosen": -346.5010681152344,
"logps/rejected": -360.3970031738281,
"loss": 0.026,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -0.022508572787046432,
"rewards/margins": 8.316872596740723,
"rewards/rejected": -8.33938217163086,
"step": 2840
},
{
"epoch": 2.94,
"learning_rate": 1.0894495412844038e-08,
"logits/chosen": -2.758545160293579,
"logits/rejected": -2.7856967449188232,
"logps/chosen": -326.37896728515625,
"logps/rejected": -359.3761291503906,
"loss": 0.015,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.21567471325397491,
"rewards/margins": 8.950045585632324,
"rewards/rejected": -8.734369277954102,
"step": 2850
},
{
"epoch": 2.95,
"learning_rate": 8.983180428134555e-09,
"logits/chosen": -2.8466389179229736,
"logits/rejected": -2.8278822898864746,
"logps/chosen": -327.270751953125,
"logps/rejected": -307.3416748046875,
"loss": 0.0158,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -0.5247852802276611,
"rewards/margins": 8.214715957641602,
"rewards/rejected": -8.739501953125,
"step": 2860
},
{
"epoch": 2.96,
"learning_rate": 7.071865443425076e-09,
"logits/chosen": -2.7788777351379395,
"logits/rejected": -2.7975292205810547,
"logps/chosen": -361.37384033203125,
"logps/rejected": -391.8774719238281,
"loss": 0.0134,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.35571426153182983,
"rewards/margins": 8.981501579284668,
"rewards/rejected": -9.337217330932617,
"step": 2870
},
{
"epoch": 2.97,
"learning_rate": 5.1605504587155965e-09,
"logits/chosen": -2.8489837646484375,
"logits/rejected": -2.7892398834228516,
"logps/chosen": -342.7001953125,
"logps/rejected": -358.37469482421875,
"loss": 0.012,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.04198342561721802,
"rewards/margins": 8.752424240112305,
"rewards/rejected": -8.710439682006836,
"step": 2880
},
{
"epoch": 2.98,
"learning_rate": 3.249235474006116e-09,
"logits/chosen": -2.8391404151916504,
"logits/rejected": -2.862032175064087,
"logps/chosen": -332.72100830078125,
"logps/rejected": -362.6523132324219,
"loss": 0.0282,
"rewards/accuracies": 1.0,
"rewards/chosen": 0.023767167702317238,
"rewards/margins": 8.220497131347656,
"rewards/rejected": -8.19672966003418,
"step": 2890
},
{
"epoch": 2.99,
"learning_rate": 1.3379204892966359e-09,
"logits/chosen": -2.854654550552368,
"logits/rejected": -2.8116354942321777,
"logps/chosen": -330.1148986816406,
"logps/rejected": -358.40155029296875,
"loss": 0.0109,
"rewards/accuracies": 1.0,
"rewards/chosen": -0.3439770042896271,
"rewards/margins": 8.63255500793457,
"rewards/rejected": -8.976531028747559,
"step": 2900
},
{
"epoch": 2.99,
"eval_logits/chosen": -2.814802408218384,
"eval_logits/rejected": -2.818735122680664,
"eval_logps/chosen": -389.9677429199219,
"eval_logps/rejected": -348.89801025390625,
"eval_loss": 0.6909257769584656,
"eval_rewards/accuracies": 0.7658730149269104,
"eval_rewards/chosen": -2.2775967121124268,
"eval_rewards/margins": 3.415607452392578,
"eval_rewards/rejected": -5.693204402923584,
"eval_runtime": 164.8452,
"eval_samples_per_second": 12.133,
"eval_steps_per_second": 0.382,
"step": 2900
},
{
"epoch": 3.0,
"step": 2907,
"total_flos": 0.0,
"train_loss": 0.23139607249449978,
"train_runtime": 34004.0578,
"train_samples_per_second": 5.467,
"train_steps_per_second": 0.085
}
],
"logging_steps": 10,
"max_steps": 2907,
"num_train_epochs": 3,
"save_steps": 500,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}