Llama-3.2-1B-SPIN-iter1 / trainer_state.json
DavieLion's picture
Upload 10 files
2355347 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 5.988571428571428,
"eval_steps": 100,
"global_step": 1572,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 3.1645569620253163e-09,
"logits/generated": 3.512105703353882,
"logits/real": 3.202550172805786,
"logps/generated": -465.1200256347656,
"logps/real": -1523.068603515625,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/generated": 0.0,
"rewards/margins": 0.0,
"rewards/real": 0.0,
"step": 1
},
{
"epoch": 0.04,
"learning_rate": 3.1645569620253166e-08,
"logits/generated": 3.52996826171875,
"logits/real": 3.1474177837371826,
"logps/generated": -455.0594177246094,
"logps/real": -1564.237548828125,
"loss": 0.6843,
"rewards/accuracies": 0.4722222089767456,
"rewards/generated": 0.00841978657990694,
"rewards/margins": 0.014310520142316818,
"rewards/real": 0.022730309516191483,
"step": 10
},
{
"epoch": 0.08,
"learning_rate": 6.329113924050633e-08,
"logits/generated": 3.4747085571289062,
"logits/real": 2.9607560634613037,
"logps/generated": -404.8710632324219,
"logps/real": -1600.1513671875,
"loss": 0.5693,
"rewards/accuracies": 0.9125000238418579,
"rewards/generated": -0.09713525325059891,
"rewards/margins": 0.30425113439559937,
"rewards/real": 0.20711591839790344,
"step": 20
},
{
"epoch": 0.11,
"learning_rate": 9.49367088607595e-08,
"logits/generated": 3.6858959197998047,
"logits/real": 3.0412940979003906,
"logps/generated": -421.63653564453125,
"logps/real": -1533.8973388671875,
"loss": 0.3593,
"rewards/accuracies": 0.9624999761581421,
"rewards/generated": -0.3924674689769745,
"rewards/margins": 0.9940061569213867,
"rewards/real": 0.6015387177467346,
"step": 30
},
{
"epoch": 0.15,
"learning_rate": 1.2658227848101266e-07,
"logits/generated": 3.7982068061828613,
"logits/real": 2.8881797790527344,
"logps/generated": -403.89447021484375,
"logps/real": -2011.547607421875,
"loss": 0.2263,
"rewards/accuracies": 1.0,
"rewards/generated": -0.6449347734451294,
"rewards/margins": 1.951049566268921,
"rewards/real": 1.3061145544052124,
"step": 40
},
{
"epoch": 0.19,
"learning_rate": 1.5822784810126582e-07,
"logits/generated": 3.733821153640747,
"logits/real": 2.91530442237854,
"logps/generated": -424.7178649902344,
"logps/real": -1627.3607177734375,
"loss": 0.1737,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -1.025488257408142,
"rewards/margins": 2.2531561851501465,
"rewards/real": 1.2276678085327148,
"step": 50
},
{
"epoch": 0.23,
"learning_rate": 1.89873417721519e-07,
"logits/generated": 3.667829990386963,
"logits/real": 2.897120237350464,
"logps/generated": -397.2341003417969,
"logps/real": -1588.1214599609375,
"loss": 0.1079,
"rewards/accuracies": 1.0,
"rewards/generated": -1.396870493888855,
"rewards/margins": 3.004943370819092,
"rewards/real": 1.6080729961395264,
"step": 60
},
{
"epoch": 0.27,
"learning_rate": 2.2151898734177212e-07,
"logits/generated": 3.703988552093506,
"logits/real": 3.3058135509490967,
"logps/generated": -450.5978088378906,
"logps/real": -1583.788818359375,
"loss": 0.0638,
"rewards/accuracies": 1.0,
"rewards/generated": -2.1847541332244873,
"rewards/margins": 4.125738620758057,
"rewards/real": 1.9409840106964111,
"step": 70
},
{
"epoch": 0.3,
"learning_rate": 2.5316455696202533e-07,
"logits/generated": 3.5381362438201904,
"logits/real": 3.0336380004882812,
"logps/generated": -426.459716796875,
"logps/real": -1585.524658203125,
"loss": 0.0623,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -1.9288753271102905,
"rewards/margins": 4.461968898773193,
"rewards/real": 2.533093214035034,
"step": 80
},
{
"epoch": 0.34,
"learning_rate": 2.848101265822785e-07,
"logits/generated": 3.9699149131774902,
"logits/real": 2.959277868270874,
"logps/generated": -462.84478759765625,
"logps/real": -1669.855712890625,
"loss": 0.0539,
"rewards/accuracies": 1.0,
"rewards/generated": -2.831137180328369,
"rewards/margins": 5.640450477600098,
"rewards/real": 2.8093132972717285,
"step": 90
},
{
"epoch": 0.38,
"learning_rate": 3.1645569620253163e-07,
"logits/generated": 3.649505615234375,
"logits/real": 3.228405475616455,
"logps/generated": -463.5885314941406,
"logps/real": -1659.1978759765625,
"loss": 0.0399,
"rewards/accuracies": 1.0,
"rewards/generated": -2.8427340984344482,
"rewards/margins": 6.116082668304443,
"rewards/real": 3.273348331451416,
"step": 100
},
{
"epoch": 0.42,
"learning_rate": 3.481012658227848e-07,
"logits/generated": 3.5634586811065674,
"logits/real": 3.0387678146362305,
"logps/generated": -475.7396545410156,
"logps/real": -1772.5689697265625,
"loss": 0.043,
"rewards/accuracies": 1.0,
"rewards/generated": -2.7554657459259033,
"rewards/margins": 6.204745292663574,
"rewards/real": 3.449280261993408,
"step": 110
},
{
"epoch": 0.46,
"learning_rate": 3.79746835443038e-07,
"logits/generated": 3.531033992767334,
"logits/real": 3.0381414890289307,
"logps/generated": -435.5464782714844,
"logps/real": -1628.22119140625,
"loss": 0.0403,
"rewards/accuracies": 1.0,
"rewards/generated": -2.647515296936035,
"rewards/margins": 5.9579033851623535,
"rewards/real": 3.3103878498077393,
"step": 120
},
{
"epoch": 0.5,
"learning_rate": 4.1139240506329117e-07,
"logits/generated": 3.513434648513794,
"logits/real": 2.99871826171875,
"logps/generated": -438.3785095214844,
"logps/real": -1705.972412109375,
"loss": 0.0341,
"rewards/accuracies": 1.0,
"rewards/generated": -3.1811232566833496,
"rewards/margins": 7.164037227630615,
"rewards/real": 3.982914447784424,
"step": 130
},
{
"epoch": 0.53,
"learning_rate": 4.4303797468354424e-07,
"logits/generated": 3.7991561889648438,
"logits/real": 3.105447292327881,
"logps/generated": -479.6163635253906,
"logps/real": -1677.7916259765625,
"loss": 0.0325,
"rewards/accuracies": 1.0,
"rewards/generated": -3.3546764850616455,
"rewards/margins": 7.303455352783203,
"rewards/real": 3.9487788677215576,
"step": 140
},
{
"epoch": 0.57,
"learning_rate": 4.746835443037974e-07,
"logits/generated": 3.594744920730591,
"logits/real": 2.9824934005737305,
"logps/generated": -436.2431640625,
"logps/real": -1721.3759765625,
"loss": 0.0232,
"rewards/accuracies": 1.0,
"rewards/generated": -3.755667209625244,
"rewards/margins": 7.839377403259277,
"rewards/real": 4.083710670471191,
"step": 150
},
{
"epoch": 0.61,
"learning_rate": 4.992927864214993e-07,
"logits/generated": 3.6612861156463623,
"logits/real": 2.9343557357788086,
"logps/generated": -483.659912109375,
"logps/real": -1582.573486328125,
"loss": 0.0205,
"rewards/accuracies": 1.0,
"rewards/generated": -3.5960216522216797,
"rewards/margins": 7.640145301818848,
"rewards/real": 4.04412317276001,
"step": 160
},
{
"epoch": 0.65,
"learning_rate": 4.957567185289958e-07,
"logits/generated": 3.589632034301758,
"logits/real": 3.0019402503967285,
"logps/generated": -425.794677734375,
"logps/real": -1487.097412109375,
"loss": 0.0221,
"rewards/accuracies": 1.0,
"rewards/generated": -3.8624930381774902,
"rewards/margins": 8.045273780822754,
"rewards/real": 4.1827802658081055,
"step": 170
},
{
"epoch": 0.69,
"learning_rate": 4.922206506364922e-07,
"logits/generated": 3.754439115524292,
"logits/real": 2.9574403762817383,
"logps/generated": -503.462646484375,
"logps/real": -1799.0537109375,
"loss": 0.0181,
"rewards/accuracies": 1.0,
"rewards/generated": -4.270089149475098,
"rewards/margins": 9.556356430053711,
"rewards/real": 5.286266326904297,
"step": 180
},
{
"epoch": 0.72,
"learning_rate": 4.886845827439887e-07,
"logits/generated": 3.9537768363952637,
"logits/real": 2.8257808685302734,
"logps/generated": -414.7232971191406,
"logps/real": -1848.5228271484375,
"loss": 0.0182,
"rewards/accuracies": 1.0,
"rewards/generated": -3.797616958618164,
"rewards/margins": 9.304213523864746,
"rewards/real": 5.506596565246582,
"step": 190
},
{
"epoch": 0.76,
"learning_rate": 4.851485148514851e-07,
"logits/generated": 3.7061352729797363,
"logits/real": 2.828097105026245,
"logps/generated": -478.8387145996094,
"logps/real": -1567.3736572265625,
"loss": 0.0162,
"rewards/accuracies": 1.0,
"rewards/generated": -4.287292003631592,
"rewards/margins": 9.111185073852539,
"rewards/real": 4.823891639709473,
"step": 200
},
{
"epoch": 0.8,
"learning_rate": 4.816124469589816e-07,
"logits/generated": 3.8071022033691406,
"logits/real": 2.786395788192749,
"logps/generated": -497.02044677734375,
"logps/real": -1951.7906494140625,
"loss": 0.0116,
"rewards/accuracies": 1.0,
"rewards/generated": -5.036887168884277,
"rewards/margins": 11.377471923828125,
"rewards/real": 6.340585231781006,
"step": 210
},
{
"epoch": 0.84,
"learning_rate": 4.780763790664781e-07,
"logits/generated": 3.5943808555603027,
"logits/real": 3.0639989376068115,
"logps/generated": -466.7726135253906,
"logps/real": -1364.220458984375,
"loss": 0.0138,
"rewards/accuracies": 1.0,
"rewards/generated": -4.49011754989624,
"rewards/margins": 9.089958190917969,
"rewards/real": 4.599841594696045,
"step": 220
},
{
"epoch": 0.88,
"learning_rate": 4.745403111739745e-07,
"logits/generated": 3.7227370738983154,
"logits/real": 2.9175028800964355,
"logps/generated": -474.84954833984375,
"logps/real": -1725.422119140625,
"loss": 0.0109,
"rewards/accuracies": 1.0,
"rewards/generated": -4.81961727142334,
"rewards/margins": 10.596981048583984,
"rewards/real": 5.777363300323486,
"step": 230
},
{
"epoch": 0.91,
"learning_rate": 4.7100424328147096e-07,
"logits/generated": 3.338122844696045,
"logits/real": 3.0875296592712402,
"logps/generated": -435.6485290527344,
"logps/real": -1685.8828125,
"loss": 0.0103,
"rewards/accuracies": 1.0,
"rewards/generated": -4.066224098205566,
"rewards/margins": 9.890088081359863,
"rewards/real": 5.823863506317139,
"step": 240
},
{
"epoch": 0.95,
"learning_rate": 4.674681753889674e-07,
"logits/generated": 3.593212127685547,
"logits/real": 2.7413556575775146,
"logps/generated": -402.1564025878906,
"logps/real": -1605.6131591796875,
"loss": 0.0143,
"rewards/accuracies": 1.0,
"rewards/generated": -4.504222869873047,
"rewards/margins": 10.564704895019531,
"rewards/real": 6.060481071472168,
"step": 250
},
{
"epoch": 0.99,
"learning_rate": 4.6393210749646387e-07,
"logits/generated": 3.5109028816223145,
"logits/real": 2.8503494262695312,
"logps/generated": -526.8668823242188,
"logps/real": -1490.2159423828125,
"loss": 0.013,
"rewards/accuracies": 1.0,
"rewards/generated": -5.489691257476807,
"rewards/margins": 11.207723617553711,
"rewards/real": 5.718031883239746,
"step": 260
},
{
"epoch": 1.03,
"learning_rate": 4.603960396039604e-07,
"logits/generated": 3.4985408782958984,
"logits/real": 2.8388476371765137,
"logps/generated": -510.9580993652344,
"logps/real": -1608.238525390625,
"loss": 0.0104,
"rewards/accuracies": 1.0,
"rewards/generated": -5.455083847045898,
"rewards/margins": 11.62474250793457,
"rewards/real": 6.1696577072143555,
"step": 270
},
{
"epoch": 1.07,
"learning_rate": 4.5685997171145683e-07,
"logits/generated": 3.408998966217041,
"logits/real": 2.9803500175476074,
"logps/generated": -467.8755798339844,
"logps/real": -1577.332763671875,
"loss": 0.0086,
"rewards/accuracies": 1.0,
"rewards/generated": -4.607177734375,
"rewards/margins": 10.38925552368164,
"rewards/real": 5.782077312469482,
"step": 280
},
{
"epoch": 1.1,
"learning_rate": 4.533239038189533e-07,
"logits/generated": 3.7580649852752686,
"logits/real": 3.202885866165161,
"logps/generated": -476.220458984375,
"logps/real": -1511.313720703125,
"loss": 0.0071,
"rewards/accuracies": 1.0,
"rewards/generated": -5.54539155960083,
"rewards/margins": 11.575006484985352,
"rewards/real": 6.0296149253845215,
"step": 290
},
{
"epoch": 1.14,
"learning_rate": 4.4978783592644974e-07,
"logits/generated": 3.447097063064575,
"logits/real": 2.6315269470214844,
"logps/generated": -475.2987365722656,
"logps/real": -1564.4840087890625,
"loss": 0.0076,
"rewards/accuracies": 1.0,
"rewards/generated": -5.546289443969727,
"rewards/margins": 11.75828742980957,
"rewards/real": 6.211997032165527,
"step": 300
},
{
"epoch": 1.18,
"learning_rate": 4.462517680339462e-07,
"logits/generated": 3.4015917778015137,
"logits/real": 2.9662914276123047,
"logps/generated": -437.78826904296875,
"logps/real": -1598.526123046875,
"loss": 0.0096,
"rewards/accuracies": 1.0,
"rewards/generated": -4.93220329284668,
"rewards/margins": 12.351015090942383,
"rewards/real": 7.4188127517700195,
"step": 310
},
{
"epoch": 1.22,
"learning_rate": 4.427157001414427e-07,
"logits/generated": 3.510097026824951,
"logits/real": 2.913818836212158,
"logps/generated": -456.273681640625,
"logps/real": -1717.4222412109375,
"loss": 0.0076,
"rewards/accuracies": 1.0,
"rewards/generated": -5.022789478302002,
"rewards/margins": 13.268719673156738,
"rewards/real": 8.245931625366211,
"step": 320
},
{
"epoch": 1.26,
"learning_rate": 4.3917963224893915e-07,
"logits/generated": 3.4145877361297607,
"logits/real": 2.8273959159851074,
"logps/generated": -465.119873046875,
"logps/real": -1708.9052734375,
"loss": 0.0066,
"rewards/accuracies": 1.0,
"rewards/generated": -5.375365257263184,
"rewards/margins": 12.598322868347168,
"rewards/real": 7.222956657409668,
"step": 330
},
{
"epoch": 1.3,
"learning_rate": 4.356435643564356e-07,
"logits/generated": 3.495450973510742,
"logits/real": 2.7503676414489746,
"logps/generated": -445.3560485839844,
"logps/real": -1563.655517578125,
"loss": 0.0071,
"rewards/accuracies": 1.0,
"rewards/generated": -5.360494136810303,
"rewards/margins": 12.658773422241211,
"rewards/real": 7.29827880859375,
"step": 340
},
{
"epoch": 1.33,
"learning_rate": 4.3210749646393206e-07,
"logits/generated": 3.579958438873291,
"logits/real": 2.7801036834716797,
"logps/generated": -436.92364501953125,
"logps/real": -1706.568359375,
"loss": 0.0072,
"rewards/accuracies": 1.0,
"rewards/generated": -5.444735527038574,
"rewards/margins": 12.98694133758545,
"rewards/real": 7.5422043800354,
"step": 350
},
{
"epoch": 1.37,
"learning_rate": 4.285714285714285e-07,
"logits/generated": 3.5498404502868652,
"logits/real": 2.8857247829437256,
"logps/generated": -467.14898681640625,
"logps/real": -1606.5517578125,
"loss": 0.0073,
"rewards/accuracies": 1.0,
"rewards/generated": -4.917501926422119,
"rewards/margins": 12.757287979125977,
"rewards/real": 7.839786529541016,
"step": 360
},
{
"epoch": 1.41,
"learning_rate": 4.25035360678925e-07,
"logits/generated": 3.588599681854248,
"logits/real": 2.7910194396972656,
"logps/generated": -484.51446533203125,
"logps/real": -1753.3662109375,
"loss": 0.0095,
"rewards/accuracies": 1.0,
"rewards/generated": -5.856254577636719,
"rewards/margins": 14.548685073852539,
"rewards/real": 8.69243049621582,
"step": 370
},
{
"epoch": 1.45,
"learning_rate": 4.214992927864215e-07,
"logits/generated": 3.4956557750701904,
"logits/real": 2.6511971950531006,
"logps/generated": -466.9720764160156,
"logps/real": -1695.6744384765625,
"loss": 0.0032,
"rewards/accuracies": 1.0,
"rewards/generated": -5.525470733642578,
"rewards/margins": 14.2725191116333,
"rewards/real": 8.747048377990723,
"step": 380
},
{
"epoch": 1.49,
"learning_rate": 4.1796322489391793e-07,
"logits/generated": 3.2205729484558105,
"logits/real": 2.746440887451172,
"logps/generated": -445.98779296875,
"logps/real": -1526.604736328125,
"loss": 0.006,
"rewards/accuracies": 1.0,
"rewards/generated": -5.600844383239746,
"rewards/margins": 13.368339538574219,
"rewards/real": 7.7674970626831055,
"step": 390
},
{
"epoch": 1.52,
"learning_rate": 4.144271570014144e-07,
"logits/generated": 3.5135910511016846,
"logits/real": 2.751537799835205,
"logps/generated": -441.55523681640625,
"logps/real": -1792.7021484375,
"loss": 0.0051,
"rewards/accuracies": 1.0,
"rewards/generated": -5.988682746887207,
"rewards/margins": 15.011758804321289,
"rewards/real": 9.023075103759766,
"step": 400
},
{
"epoch": 1.56,
"learning_rate": 4.1089108910891084e-07,
"logits/generated": 3.4969635009765625,
"logits/real": 2.874459981918335,
"logps/generated": -480.62884521484375,
"logps/real": -1726.501953125,
"loss": 0.0069,
"rewards/accuracies": 1.0,
"rewards/generated": -6.103398323059082,
"rewards/margins": 14.921289443969727,
"rewards/real": 8.817892074584961,
"step": 410
},
{
"epoch": 1.6,
"learning_rate": 4.0735502121640734e-07,
"logits/generated": 3.732043743133545,
"logits/real": 2.9122977256774902,
"logps/generated": -471.2142639160156,
"logps/real": -1475.9720458984375,
"loss": 0.006,
"rewards/accuracies": 1.0,
"rewards/generated": -5.9775614738464355,
"rewards/margins": 14.218009948730469,
"rewards/real": 8.240448951721191,
"step": 420
},
{
"epoch": 1.64,
"learning_rate": 4.038189533239038e-07,
"logits/generated": 3.5188686847686768,
"logits/real": 2.9900317192077637,
"logps/generated": -560.73486328125,
"logps/real": -1395.8021240234375,
"loss": 0.02,
"rewards/accuracies": 0.987500011920929,
"rewards/generated": -7.657801628112793,
"rewards/margins": 15.55529499053955,
"rewards/real": 7.897493839263916,
"step": 430
},
{
"epoch": 1.68,
"learning_rate": 4.0028288543140025e-07,
"logits/generated": 3.3781158924102783,
"logits/real": 2.629373073577881,
"logps/generated": -485.42156982421875,
"logps/real": -1796.0904541015625,
"loss": 0.0067,
"rewards/accuracies": 1.0,
"rewards/generated": -7.0816969871521,
"rewards/margins": 16.453693389892578,
"rewards/real": 9.371994972229004,
"step": 440
},
{
"epoch": 1.71,
"learning_rate": 3.967468175388967e-07,
"logits/generated": 3.460228443145752,
"logits/real": 2.9703357219696045,
"logps/generated": -460.6011657714844,
"logps/real": -1839.8109130859375,
"loss": 0.0032,
"rewards/accuracies": 1.0,
"rewards/generated": -6.300644397735596,
"rewards/margins": 15.726030349731445,
"rewards/real": 9.425384521484375,
"step": 450
},
{
"epoch": 1.75,
"learning_rate": 3.9321074964639316e-07,
"logits/generated": 3.0205512046813965,
"logits/real": 2.8035550117492676,
"logps/generated": -502.0563049316406,
"logps/real": -1389.227783203125,
"loss": 0.0048,
"rewards/accuracies": 1.0,
"rewards/generated": -7.311892032623291,
"rewards/margins": 15.14806842803955,
"rewards/real": 7.836176872253418,
"step": 460
},
{
"epoch": 1.79,
"learning_rate": 3.8967468175388967e-07,
"logits/generated": 3.6069533824920654,
"logits/real": 2.726623773574829,
"logps/generated": -505.29437255859375,
"logps/real": -1689.9912109375,
"loss": 0.0036,
"rewards/accuracies": 1.0,
"rewards/generated": -7.359295845031738,
"rewards/margins": 16.689287185668945,
"rewards/real": 9.32999324798584,
"step": 470
},
{
"epoch": 1.83,
"learning_rate": 3.861386138613861e-07,
"logits/generated": 3.4885826110839844,
"logits/real": 2.5398058891296387,
"logps/generated": -468.04608154296875,
"logps/real": -1690.434326171875,
"loss": 0.0029,
"rewards/accuracies": 1.0,
"rewards/generated": -7.317099571228027,
"rewards/margins": 15.526530265808105,
"rewards/real": 8.209431648254395,
"step": 480
},
{
"epoch": 1.87,
"learning_rate": 3.8260254596888257e-07,
"logits/generated": 3.4325199127197266,
"logits/real": 2.7468791007995605,
"logps/generated": -485.3282165527344,
"logps/real": -1564.2491455078125,
"loss": 0.0038,
"rewards/accuracies": 1.0,
"rewards/generated": -8.328545570373535,
"rewards/margins": 16.984691619873047,
"rewards/real": 8.656146049499512,
"step": 490
},
{
"epoch": 1.9,
"learning_rate": 3.7906647807637903e-07,
"logits/generated": 3.2969658374786377,
"logits/real": 2.644705295562744,
"logps/generated": -463.7992248535156,
"logps/real": -1622.1402587890625,
"loss": 0.0026,
"rewards/accuracies": 1.0,
"rewards/generated": -7.6635894775390625,
"rewards/margins": 16.660659790039062,
"rewards/real": 8.9970703125,
"step": 500
},
{
"epoch": 1.94,
"learning_rate": 3.755304101838755e-07,
"logits/generated": 3.4488253593444824,
"logits/real": 2.9401419162750244,
"logps/generated": -500.5553283691406,
"logps/real": -1580.5606689453125,
"loss": 0.0033,
"rewards/accuracies": 1.0,
"rewards/generated": -7.708639621734619,
"rewards/margins": 16.773143768310547,
"rewards/real": 9.064504623413086,
"step": 510
},
{
"epoch": 1.98,
"learning_rate": 3.71994342291372e-07,
"logits/generated": 3.3167099952697754,
"logits/real": 3.012571334838867,
"logps/generated": -521.168701171875,
"logps/real": -1524.885986328125,
"loss": 0.0039,
"rewards/accuracies": 1.0,
"rewards/generated": -7.435615539550781,
"rewards/margins": 16.3592472076416,
"rewards/real": 8.923630714416504,
"step": 520
},
{
"epoch": 2.02,
"learning_rate": 3.6845827439886844e-07,
"logits/generated": 3.354684352874756,
"logits/real": 2.7756412029266357,
"logps/generated": -468.66552734375,
"logps/real": -1526.357666015625,
"loss": 0.0048,
"rewards/accuracies": 1.0,
"rewards/generated": -8.224092483520508,
"rewards/margins": 16.959665298461914,
"rewards/real": 8.735573768615723,
"step": 530
},
{
"epoch": 2.06,
"learning_rate": 3.649222065063649e-07,
"logits/generated": 3.5977184772491455,
"logits/real": 2.68579363822937,
"logps/generated": -544.5694580078125,
"logps/real": -1571.634033203125,
"loss": 0.0018,
"rewards/accuracies": 1.0,
"rewards/generated": -8.363500595092773,
"rewards/margins": 17.53289794921875,
"rewards/real": 9.16939640045166,
"step": 540
},
{
"epoch": 2.1,
"learning_rate": 3.6138613861386135e-07,
"logits/generated": 3.4918408393859863,
"logits/real": 2.729196071624756,
"logps/generated": -518.4525146484375,
"logps/real": -1421.8359375,
"loss": 0.0036,
"rewards/accuracies": 1.0,
"rewards/generated": -8.855493545532227,
"rewards/margins": 17.412639617919922,
"rewards/real": 8.557147979736328,
"step": 550
},
{
"epoch": 2.13,
"learning_rate": 3.578500707213578e-07,
"logits/generated": 3.2267754077911377,
"logits/real": 2.8264071941375732,
"logps/generated": -458.85308837890625,
"logps/real": -1482.390625,
"loss": 0.0039,
"rewards/accuracies": 1.0,
"rewards/generated": -7.917211055755615,
"rewards/margins": 17.222782135009766,
"rewards/real": 9.305570602416992,
"step": 560
},
{
"epoch": 2.17,
"learning_rate": 3.543140028288543e-07,
"logits/generated": 3.449310302734375,
"logits/real": 2.881521701812744,
"logps/generated": -479.64202880859375,
"logps/real": -1629.423095703125,
"loss": 0.002,
"rewards/accuracies": 1.0,
"rewards/generated": -7.995371341705322,
"rewards/margins": 18.3244686126709,
"rewards/real": 10.329096794128418,
"step": 570
},
{
"epoch": 2.21,
"learning_rate": 3.5077793493635076e-07,
"logits/generated": 3.4941279888153076,
"logits/real": 2.9073030948638916,
"logps/generated": -487.73846435546875,
"logps/real": -1576.749267578125,
"loss": 0.0037,
"rewards/accuracies": 1.0,
"rewards/generated": -7.9393134117126465,
"rewards/margins": 17.776561737060547,
"rewards/real": 9.837248802185059,
"step": 580
},
{
"epoch": 2.25,
"learning_rate": 3.472418670438472e-07,
"logits/generated": 3.173529863357544,
"logits/real": 2.8594255447387695,
"logps/generated": -487.2096252441406,
"logps/real": -1416.3604736328125,
"loss": 0.002,
"rewards/accuracies": 1.0,
"rewards/generated": -7.219686031341553,
"rewards/margins": 15.569984436035156,
"rewards/real": 8.350296974182129,
"step": 590
},
{
"epoch": 2.29,
"learning_rate": 3.4370579915134367e-07,
"logits/generated": 3.167581081390381,
"logits/real": 2.622349500656128,
"logps/generated": -450.83447265625,
"logps/real": -1434.6671142578125,
"loss": 0.0023,
"rewards/accuracies": 1.0,
"rewards/generated": -7.358783721923828,
"rewards/margins": 16.64669418334961,
"rewards/real": 9.287908554077148,
"step": 600
},
{
"epoch": 2.32,
"learning_rate": 3.401697312588401e-07,
"logits/generated": 3.5104973316192627,
"logits/real": 2.580019474029541,
"logps/generated": -495.33746337890625,
"logps/real": -1505.913818359375,
"loss": 0.0037,
"rewards/accuracies": 1.0,
"rewards/generated": -7.833502292633057,
"rewards/margins": 16.95751953125,
"rewards/real": 9.124017715454102,
"step": 610
},
{
"epoch": 2.36,
"learning_rate": 3.3663366336633663e-07,
"logits/generated": 3.2511913776397705,
"logits/real": 2.7187373638153076,
"logps/generated": -495.783447265625,
"logps/real": -1531.77294921875,
"loss": 0.0045,
"rewards/accuracies": 1.0,
"rewards/generated": -8.153390884399414,
"rewards/margins": 17.530384063720703,
"rewards/real": 9.376993179321289,
"step": 620
},
{
"epoch": 2.4,
"learning_rate": 3.330975954738331e-07,
"logits/generated": 3.2725577354431152,
"logits/real": 2.7447569370269775,
"logps/generated": -491.76397705078125,
"logps/real": -1607.060791015625,
"loss": 0.0027,
"rewards/accuracies": 1.0,
"rewards/generated": -7.7982048988342285,
"rewards/margins": 17.868427276611328,
"rewards/real": 10.07022476196289,
"step": 630
},
{
"epoch": 2.44,
"learning_rate": 3.2956152758132954e-07,
"logits/generated": 3.2170231342315674,
"logits/real": 2.760910749435425,
"logps/generated": -479.82989501953125,
"logps/real": -1328.6890869140625,
"loss": 0.0024,
"rewards/accuracies": 1.0,
"rewards/generated": -7.747633934020996,
"rewards/margins": 16.90505027770996,
"rewards/real": 9.157418251037598,
"step": 640
},
{
"epoch": 2.48,
"learning_rate": 3.26025459688826e-07,
"logits/generated": 3.2980682849884033,
"logits/real": 2.6026058197021484,
"logps/generated": -484.3726501464844,
"logps/real": -1501.859619140625,
"loss": 0.0012,
"rewards/accuracies": 1.0,
"rewards/generated": -7.929776668548584,
"rewards/margins": 18.278278350830078,
"rewards/real": 10.348502159118652,
"step": 650
},
{
"epoch": 2.51,
"learning_rate": 3.2248939179632245e-07,
"logits/generated": 3.410261631011963,
"logits/real": 2.8707950115203857,
"logps/generated": -550.3458251953125,
"logps/real": -1484.9326171875,
"loss": 0.0009,
"rewards/accuracies": 1.0,
"rewards/generated": -8.744172096252441,
"rewards/margins": 18.63943862915039,
"rewards/real": 9.895263671875,
"step": 660
},
{
"epoch": 2.55,
"learning_rate": 3.1895332390381895e-07,
"logits/generated": 3.624340772628784,
"logits/real": 2.644595146179199,
"logps/generated": -510.18096923828125,
"logps/real": -1679.375,
"loss": 0.0024,
"rewards/accuracies": 1.0,
"rewards/generated": -8.56155014038086,
"rewards/margins": 20.045724868774414,
"rewards/real": 11.484173774719238,
"step": 670
},
{
"epoch": 2.59,
"learning_rate": 3.154172560113154e-07,
"logits/generated": 3.5800070762634277,
"logits/real": 2.905097246170044,
"logps/generated": -514.9693603515625,
"logps/real": -1498.704345703125,
"loss": 0.0023,
"rewards/accuracies": 1.0,
"rewards/generated": -8.549358367919922,
"rewards/margins": 18.856287002563477,
"rewards/real": 10.306925773620605,
"step": 680
},
{
"epoch": 2.63,
"learning_rate": 3.1188118811881186e-07,
"logits/generated": 3.3421058654785156,
"logits/real": 2.714266538619995,
"logps/generated": -497.5679626464844,
"logps/real": -1502.726806640625,
"loss": 0.0023,
"rewards/accuracies": 1.0,
"rewards/generated": -8.259305953979492,
"rewards/margins": 18.61314582824707,
"rewards/real": 10.353840827941895,
"step": 690
},
{
"epoch": 2.67,
"learning_rate": 3.083451202263083e-07,
"logits/generated": 3.3471736907958984,
"logits/real": 2.7787163257598877,
"logps/generated": -517.1281127929688,
"logps/real": -1472.466064453125,
"loss": 0.0027,
"rewards/accuracies": 1.0,
"rewards/generated": -8.416834831237793,
"rewards/margins": 18.77850914001465,
"rewards/real": 10.361673355102539,
"step": 700
},
{
"epoch": 2.7,
"learning_rate": 3.0480905233380477e-07,
"logits/generated": 3.2355308532714844,
"logits/real": 2.852546215057373,
"logps/generated": -500.0751953125,
"logps/real": -1651.260986328125,
"loss": 0.0014,
"rewards/accuracies": 1.0,
"rewards/generated": -8.010438919067383,
"rewards/margins": 19.449283599853516,
"rewards/real": 11.438843727111816,
"step": 710
},
{
"epoch": 2.74,
"learning_rate": 3.012729844413013e-07,
"logits/generated": 3.437945604324341,
"logits/real": 2.749911308288574,
"logps/generated": -542.2654418945312,
"logps/real": -1487.933837890625,
"loss": 0.002,
"rewards/accuracies": 1.0,
"rewards/generated": -8.703554153442383,
"rewards/margins": 19.010265350341797,
"rewards/real": 10.306710243225098,
"step": 720
},
{
"epoch": 2.78,
"learning_rate": 2.9773691654879773e-07,
"logits/generated": 3.234304428100586,
"logits/real": 2.689356803894043,
"logps/generated": -459.7379455566406,
"logps/real": -1269.926025390625,
"loss": 0.0021,
"rewards/accuracies": 1.0,
"rewards/generated": -7.5548601150512695,
"rewards/margins": 16.455080032348633,
"rewards/real": 8.900218963623047,
"step": 730
},
{
"epoch": 2.82,
"learning_rate": 2.942008486562942e-07,
"logits/generated": 3.4382824897766113,
"logits/real": 2.6454930305480957,
"logps/generated": -517.6546630859375,
"logps/real": -1617.267333984375,
"loss": 0.0014,
"rewards/accuracies": 1.0,
"rewards/generated": -8.587407112121582,
"rewards/margins": 19.053064346313477,
"rewards/real": 10.465655326843262,
"step": 740
},
{
"epoch": 2.86,
"learning_rate": 2.9066478076379064e-07,
"logits/generated": 3.0672454833984375,
"logits/real": 2.67156720161438,
"logps/generated": -459.13262939453125,
"logps/real": -1313.4677734375,
"loss": 0.0022,
"rewards/accuracies": 1.0,
"rewards/generated": -8.137083053588867,
"rewards/margins": 17.47877311706543,
"rewards/real": 9.341691970825195,
"step": 750
},
{
"epoch": 2.9,
"learning_rate": 2.871287128712871e-07,
"logits/generated": 3.4013969898223877,
"logits/real": 2.417377233505249,
"logps/generated": -492.2867736816406,
"logps/real": -1638.200927734375,
"loss": 0.0019,
"rewards/accuracies": 1.0,
"rewards/generated": -8.533792495727539,
"rewards/margins": 19.663969039916992,
"rewards/real": 11.13017749786377,
"step": 760
},
{
"epoch": 2.93,
"learning_rate": 2.835926449787836e-07,
"logits/generated": 3.388955593109131,
"logits/real": 2.9141921997070312,
"logps/generated": -545.1387939453125,
"logps/real": -1538.4222412109375,
"loss": 0.0014,
"rewards/accuracies": 1.0,
"rewards/generated": -9.785470962524414,
"rewards/margins": 21.193279266357422,
"rewards/real": 11.407808303833008,
"step": 770
},
{
"epoch": 2.97,
"learning_rate": 2.8005657708628005e-07,
"logits/generated": 3.622591733932495,
"logits/real": 2.7291853427886963,
"logps/generated": -547.1531982421875,
"logps/real": -1743.8941650390625,
"loss": 0.0018,
"rewards/accuracies": 1.0,
"rewards/generated": -10.029909133911133,
"rewards/margins": 21.701557159423828,
"rewards/real": 11.671648025512695,
"step": 780
},
{
"epoch": 3.01,
"learning_rate": 2.765205091937765e-07,
"logits/generated": 3.347827196121216,
"logits/real": 2.813518524169922,
"logps/generated": -518.5848388671875,
"logps/real": -1608.0418701171875,
"loss": 0.0024,
"rewards/accuracies": 1.0,
"rewards/generated": -9.312664985656738,
"rewards/margins": 21.00381851196289,
"rewards/real": 11.69115161895752,
"step": 790
},
{
"epoch": 3.05,
"learning_rate": 2.7298444130127296e-07,
"logits/generated": 3.2987990379333496,
"logits/real": 2.9453186988830566,
"logps/generated": -522.5285034179688,
"logps/real": -1608.2236328125,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/generated": -9.82221794128418,
"rewards/margins": 22.511926651000977,
"rewards/real": 12.689709663391113,
"step": 800
},
{
"epoch": 3.09,
"learning_rate": 2.694483734087694e-07,
"logits/generated": 3.4462966918945312,
"logits/real": 2.625929117202759,
"logps/generated": -545.0180053710938,
"logps/real": -1427.6646728515625,
"loss": 0.0022,
"rewards/accuracies": 1.0,
"rewards/generated": -11.479304313659668,
"rewards/margins": 22.244796752929688,
"rewards/real": 10.765493392944336,
"step": 810
},
{
"epoch": 3.12,
"learning_rate": 2.659123055162659e-07,
"logits/generated": 3.2480366230010986,
"logits/real": 2.8531947135925293,
"logps/generated": -530.7342529296875,
"logps/real": -1727.2086181640625,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/generated": -9.427630424499512,
"rewards/margins": 22.206254959106445,
"rewards/real": 12.778624534606934,
"step": 820
},
{
"epoch": 3.16,
"learning_rate": 2.623762376237624e-07,
"logits/generated": 3.5078930854797363,
"logits/real": 2.7171075344085693,
"logps/generated": -507.3907775878906,
"logps/real": -1634.822998046875,
"loss": 0.0011,
"rewards/accuracies": 1.0,
"rewards/generated": -8.803173065185547,
"rewards/margins": 20.158788681030273,
"rewards/real": 11.355618476867676,
"step": 830
},
{
"epoch": 3.2,
"learning_rate": 2.5884016973125883e-07,
"logits/generated": 3.389615535736084,
"logits/real": 2.6584954261779785,
"logps/generated": -519.1912841796875,
"logps/real": -1511.683349609375,
"loss": 0.0012,
"rewards/accuracies": 1.0,
"rewards/generated": -9.256002426147461,
"rewards/margins": 19.75589370727539,
"rewards/real": 10.499893188476562,
"step": 840
},
{
"epoch": 3.24,
"learning_rate": 2.553041018387553e-07,
"logits/generated": 3.1219124794006348,
"logits/real": 2.69732928276062,
"logps/generated": -513.461181640625,
"logps/real": -1559.185302734375,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/generated": -9.307373046875,
"rewards/margins": 20.579065322875977,
"rewards/real": 11.271692276000977,
"step": 850
},
{
"epoch": 3.28,
"learning_rate": 2.5176803394625174e-07,
"logits/generated": 3.169861316680908,
"logits/real": 2.5963492393493652,
"logps/generated": -531.0164184570312,
"logps/real": -1578.7210693359375,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/generated": -9.622026443481445,
"rewards/margins": 21.500263214111328,
"rewards/real": 11.878233909606934,
"step": 860
},
{
"epoch": 3.31,
"learning_rate": 2.482319660537482e-07,
"logits/generated": 3.3711845874786377,
"logits/real": 2.759648323059082,
"logps/generated": -523.5821533203125,
"logps/real": -1462.2935791015625,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/generated": -10.834948539733887,
"rewards/margins": 21.79311180114746,
"rewards/real": 10.958162307739258,
"step": 870
},
{
"epoch": 3.35,
"learning_rate": 2.446958981612447e-07,
"logits/generated": 3.282954454421997,
"logits/real": 2.7643918991088867,
"logps/generated": -473.2938537597656,
"logps/real": -1457.6839599609375,
"loss": 0.0013,
"rewards/accuracies": 1.0,
"rewards/generated": -8.40355110168457,
"rewards/margins": 19.281604766845703,
"rewards/real": 10.878053665161133,
"step": 880
},
{
"epoch": 3.39,
"learning_rate": 2.4115983026874115e-07,
"logits/generated": 3.4354591369628906,
"logits/real": 2.6684250831604004,
"logps/generated": -536.3674926757812,
"logps/real": -1729.115966796875,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/generated": -10.493001937866211,
"rewards/margins": 23.19765853881836,
"rewards/real": 12.704656600952148,
"step": 890
},
{
"epoch": 3.43,
"learning_rate": 2.376237623762376e-07,
"logits/generated": 3.266200304031372,
"logits/real": 2.749070644378662,
"logps/generated": -506.13970947265625,
"logps/real": -1660.871337890625,
"loss": 0.0006,
"rewards/accuracies": 1.0,
"rewards/generated": -9.553144454956055,
"rewards/margins": 22.273033142089844,
"rewards/real": 12.719888687133789,
"step": 900
},
{
"epoch": 3.47,
"learning_rate": 2.3408769448373408e-07,
"logits/generated": 3.4035401344299316,
"logits/real": 2.6919925212860107,
"logps/generated": -545.0686645507812,
"logps/real": -1454.680419921875,
"loss": 0.0011,
"rewards/accuracies": 1.0,
"rewards/generated": -9.882238388061523,
"rewards/margins": 21.769603729248047,
"rewards/real": 11.88736629486084,
"step": 910
},
{
"epoch": 3.5,
"learning_rate": 2.3055162659123054e-07,
"logits/generated": 3.244786024093628,
"logits/real": 2.7183642387390137,
"logps/generated": -538.6968383789062,
"logps/real": -1382.13330078125,
"loss": 0.001,
"rewards/accuracies": 1.0,
"rewards/generated": -10.824942588806152,
"rewards/margins": 22.102800369262695,
"rewards/real": 11.277857780456543,
"step": 920
},
{
"epoch": 3.54,
"learning_rate": 2.2701555869872702e-07,
"logits/generated": 3.3680167198181152,
"logits/real": 2.5044057369232178,
"logps/generated": -547.4840087890625,
"logps/real": -1493.070556640625,
"loss": 0.0013,
"rewards/accuracies": 1.0,
"rewards/generated": -10.53519344329834,
"rewards/margins": 22.55548667907715,
"rewards/real": 12.020292282104492,
"step": 930
},
{
"epoch": 3.58,
"learning_rate": 2.2347949080622347e-07,
"logits/generated": 3.4719784259796143,
"logits/real": 2.6997435092926025,
"logps/generated": -531.0570678710938,
"logps/real": -1509.5135498046875,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/generated": -10.211065292358398,
"rewards/margins": 22.259336471557617,
"rewards/real": 12.048271179199219,
"step": 940
},
{
"epoch": 3.62,
"learning_rate": 2.1994342291371993e-07,
"logits/generated": 3.3083972930908203,
"logits/real": 2.4025731086730957,
"logps/generated": -575.9044189453125,
"logps/real": -1759.0048828125,
"loss": 0.002,
"rewards/accuracies": 1.0,
"rewards/generated": -11.274171829223633,
"rewards/margins": 24.654094696044922,
"rewards/real": 13.379923820495605,
"step": 950
},
{
"epoch": 3.66,
"learning_rate": 2.164073550212164e-07,
"logits/generated": 3.4808990955352783,
"logits/real": 2.577362060546875,
"logps/generated": -463.35638427734375,
"logps/real": -1399.489013671875,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/generated": -8.753828048706055,
"rewards/margins": 19.656505584716797,
"rewards/real": 10.902679443359375,
"step": 960
},
{
"epoch": 3.7,
"learning_rate": 2.1287128712871286e-07,
"logits/generated": 3.4557316303253174,
"logits/real": 2.766918659210205,
"logps/generated": -553.6686401367188,
"logps/real": -1662.569091796875,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/generated": -10.936054229736328,
"rewards/margins": 24.027223587036133,
"rewards/real": 13.091169357299805,
"step": 970
},
{
"epoch": 3.73,
"learning_rate": 2.0933521923620934e-07,
"logits/generated": 3.14534592628479,
"logits/real": 2.6436448097229004,
"logps/generated": -510.17950439453125,
"logps/real": -1495.9129638671875,
"loss": 0.0006,
"rewards/accuracies": 1.0,
"rewards/generated": -9.579751968383789,
"rewards/margins": 21.640422821044922,
"rewards/real": 12.06067180633545,
"step": 980
},
{
"epoch": 3.77,
"learning_rate": 2.057991513437058e-07,
"logits/generated": 3.3893933296203613,
"logits/real": 2.515793800354004,
"logps/generated": -507.2007751464844,
"logps/real": -1553.794189453125,
"loss": 0.0011,
"rewards/accuracies": 1.0,
"rewards/generated": -10.590211868286133,
"rewards/margins": 23.068220138549805,
"rewards/real": 12.478008270263672,
"step": 990
},
{
"epoch": 3.81,
"learning_rate": 2.0226308345120225e-07,
"logits/generated": 3.168189525604248,
"logits/real": 2.5998377799987793,
"logps/generated": -535.4756469726562,
"logps/real": -1227.1500244140625,
"loss": 0.0011,
"rewards/accuracies": 1.0,
"rewards/generated": -11.283498764038086,
"rewards/margins": 21.739755630493164,
"rewards/real": 10.456255912780762,
"step": 1000
},
{
"epoch": 3.85,
"learning_rate": 1.9872701555869873e-07,
"logits/generated": 3.0489237308502197,
"logits/real": 2.500216484069824,
"logps/generated": -534.8651123046875,
"logps/real": -1446.293701171875,
"loss": 0.0006,
"rewards/accuracies": 1.0,
"rewards/generated": -11.704988479614258,
"rewards/margins": 22.91129493713379,
"rewards/real": 11.206305503845215,
"step": 1010
},
{
"epoch": 3.89,
"learning_rate": 1.9519094766619518e-07,
"logits/generated": 3.4261863231658936,
"logits/real": 2.486532688140869,
"logps/generated": -550.0924072265625,
"logps/real": -1607.067138671875,
"loss": 0.0004,
"rewards/accuracies": 1.0,
"rewards/generated": -11.577852249145508,
"rewards/margins": 24.368900299072266,
"rewards/real": 12.791049003601074,
"step": 1020
},
{
"epoch": 3.92,
"learning_rate": 1.9165487977369166e-07,
"logits/generated": 3.2348227500915527,
"logits/real": 2.7686595916748047,
"logps/generated": -570.3921508789062,
"logps/real": -1670.5628662109375,
"loss": 0.0009,
"rewards/accuracies": 1.0,
"rewards/generated": -11.573652267456055,
"rewards/margins": 25.506744384765625,
"rewards/real": 13.93309211730957,
"step": 1030
},
{
"epoch": 3.96,
"learning_rate": 1.8811881188118812e-07,
"logits/generated": 3.115082263946533,
"logits/real": 2.5799102783203125,
"logps/generated": -503.86956787109375,
"logps/real": -1525.8857421875,
"loss": 0.0013,
"rewards/accuracies": 1.0,
"rewards/generated": -9.889609336853027,
"rewards/margins": 22.231243133544922,
"rewards/real": 12.341632843017578,
"step": 1040
},
{
"epoch": 4.0,
"learning_rate": 1.8458274398868457e-07,
"logits/generated": 3.2662482261657715,
"logits/real": 2.468737840652466,
"logps/generated": -522.85498046875,
"logps/real": -1517.8037109375,
"loss": 0.0004,
"rewards/accuracies": 1.0,
"rewards/generated": -11.054136276245117,
"rewards/margins": 24.148630142211914,
"rewards/real": 13.094491958618164,
"step": 1050
},
{
"epoch": 4.04,
"learning_rate": 1.8104667609618105e-07,
"logits/generated": 3.3418450355529785,
"logits/real": 2.6237406730651855,
"logps/generated": -586.31591796875,
"logps/real": -1651.44921875,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/generated": -13.165428161621094,
"rewards/margins": 27.307897567749023,
"rewards/real": 14.142468452453613,
"step": 1060
},
{
"epoch": 4.08,
"learning_rate": 1.775106082036775e-07,
"logits/generated": 3.2586989402770996,
"logits/real": 2.557404041290283,
"logps/generated": -502.6537170410156,
"logps/real": -1429.9771728515625,
"loss": 0.0009,
"rewards/accuracies": 1.0,
"rewards/generated": -11.023027420043945,
"rewards/margins": 23.140275955200195,
"rewards/real": 12.117250442504883,
"step": 1070
},
{
"epoch": 4.11,
"learning_rate": 1.7397454031117398e-07,
"logits/generated": 3.426239013671875,
"logits/real": 2.6774344444274902,
"logps/generated": -572.1189575195312,
"logps/real": -1397.4552001953125,
"loss": 0.0004,
"rewards/accuracies": 1.0,
"rewards/generated": -13.265462875366211,
"rewards/margins": 24.69881820678711,
"rewards/real": 11.433354377746582,
"step": 1080
},
{
"epoch": 4.15,
"learning_rate": 1.7043847241867044e-07,
"logits/generated": 3.0871334075927734,
"logits/real": 2.5882654190063477,
"logps/generated": -550.581298828125,
"logps/real": -1440.5426025390625,
"loss": 0.0005,
"rewards/accuracies": 1.0,
"rewards/generated": -11.109870910644531,
"rewards/margins": 25.10569190979004,
"rewards/real": 13.995819091796875,
"step": 1090
},
{
"epoch": 4.19,
"learning_rate": 1.669024045261669e-07,
"logits/generated": 3.4628639221191406,
"logits/real": 2.6520979404449463,
"logps/generated": -516.7633666992188,
"logps/real": -1631.560302734375,
"loss": 0.0004,
"rewards/accuracies": 1.0,
"rewards/generated": -11.816521644592285,
"rewards/margins": 25.77557373046875,
"rewards/real": 13.959050178527832,
"step": 1100
},
{
"epoch": 4.23,
"learning_rate": 1.6336633663366337e-07,
"logits/generated": 3.325192928314209,
"logits/real": 2.6561620235443115,
"logps/generated": -579.8635864257812,
"logps/real": -1503.974609375,
"loss": 0.0004,
"rewards/accuracies": 1.0,
"rewards/generated": -12.559514045715332,
"rewards/margins": 25.418582916259766,
"rewards/real": 12.85906982421875,
"step": 1110
},
{
"epoch": 4.27,
"learning_rate": 1.5983026874115983e-07,
"logits/generated": 3.2172133922576904,
"logits/real": 2.644768238067627,
"logps/generated": -523.1590576171875,
"logps/real": -1488.0220947265625,
"loss": 0.0005,
"rewards/accuracies": 1.0,
"rewards/generated": -11.404559135437012,
"rewards/margins": 23.979520797729492,
"rewards/real": 12.574960708618164,
"step": 1120
},
{
"epoch": 4.3,
"learning_rate": 1.562942008486563e-07,
"logits/generated": 3.0821359157562256,
"logits/real": 2.5439810752868652,
"logps/generated": -466.92694091796875,
"logps/real": -1334.6812744140625,
"loss": 0.0004,
"rewards/accuracies": 1.0,
"rewards/generated": -10.233351707458496,
"rewards/margins": 21.963369369506836,
"rewards/real": 11.73001766204834,
"step": 1130
},
{
"epoch": 4.34,
"learning_rate": 1.5275813295615276e-07,
"logits/generated": 3.4721286296844482,
"logits/real": 2.5629734992980957,
"logps/generated": -569.9315795898438,
"logps/real": -1553.0869140625,
"loss": 0.0003,
"rewards/accuracies": 1.0,
"rewards/generated": -11.754822731018066,
"rewards/margins": 25.470844268798828,
"rewards/real": 13.716024398803711,
"step": 1140
},
{
"epoch": 4.38,
"learning_rate": 1.4922206506364921e-07,
"logits/generated": 3.156252384185791,
"logits/real": 2.703667163848877,
"logps/generated": -512.1444091796875,
"logps/real": -1513.850830078125,
"loss": 0.0003,
"rewards/accuracies": 1.0,
"rewards/generated": -11.257915496826172,
"rewards/margins": 24.550378799438477,
"rewards/real": 13.292462348937988,
"step": 1150
},
{
"epoch": 4.42,
"learning_rate": 1.456859971711457e-07,
"logits/generated": 3.141005277633667,
"logits/real": 2.8431308269500732,
"logps/generated": -560.7545776367188,
"logps/real": -1514.60693359375,
"loss": 0.0001,
"rewards/accuracies": 1.0,
"rewards/generated": -12.668886184692383,
"rewards/margins": 25.635141372680664,
"rewards/real": 12.966257095336914,
"step": 1160
},
{
"epoch": 4.46,
"learning_rate": 1.4214992927864215e-07,
"logits/generated": 3.4051051139831543,
"logits/real": 2.6105704307556152,
"logps/generated": -574.8484497070312,
"logps/real": -1481.771484375,
"loss": 0.0002,
"rewards/accuracies": 1.0,
"rewards/generated": -12.965988159179688,
"rewards/margins": 26.073007583618164,
"rewards/real": 13.107017517089844,
"step": 1170
},
{
"epoch": 4.5,
"learning_rate": 1.3861386138613863e-07,
"logits/generated": 3.417850971221924,
"logits/real": 2.5431482791900635,
"logps/generated": -557.5526123046875,
"logps/real": -1832.751953125,
"loss": 0.0005,
"rewards/accuracies": 1.0,
"rewards/generated": -11.886221885681152,
"rewards/margins": 26.28768539428711,
"rewards/real": 14.401464462280273,
"step": 1180
},
{
"epoch": 4.53,
"learning_rate": 1.3507779349363508e-07,
"logits/generated": 3.3368937969207764,
"logits/real": 2.492691993713379,
"logps/generated": -503.24652099609375,
"logps/real": -1523.6234130859375,
"loss": 0.0007,
"rewards/accuracies": 1.0,
"rewards/generated": -11.509819030761719,
"rewards/margins": 25.515331268310547,
"rewards/real": 14.005514144897461,
"step": 1190
},
{
"epoch": 4.57,
"learning_rate": 1.3154172560113154e-07,
"logits/generated": 3.381847858428955,
"logits/real": 2.6287689208984375,
"logps/generated": -530.4554443359375,
"logps/real": -1445.392333984375,
"loss": 0.0004,
"rewards/accuracies": 1.0,
"rewards/generated": -12.088003158569336,
"rewards/margins": 25.268281936645508,
"rewards/real": 13.180277824401855,
"step": 1200
},
{
"epoch": 4.61,
"learning_rate": 1.2800565770862802e-07,
"logits/generated": 3.0044195652008057,
"logits/real": 2.6941545009613037,
"logps/generated": -580.5851440429688,
"logps/real": -1566.2730712890625,
"loss": 0.0006,
"rewards/accuracies": 1.0,
"rewards/generated": -13.416638374328613,
"rewards/margins": 27.365131378173828,
"rewards/real": 13.948491096496582,
"step": 1210
},
{
"epoch": 4.65,
"learning_rate": 1.2446958981612447e-07,
"logits/generated": 3.0417532920837402,
"logits/real": 2.775116443634033,
"logps/generated": -564.0277099609375,
"logps/real": -1237.641357421875,
"loss": 0.0003,
"rewards/accuracies": 1.0,
"rewards/generated": -12.711797714233398,
"rewards/margins": 24.17923355102539,
"rewards/real": 11.467435836791992,
"step": 1220
},
{
"epoch": 4.69,
"learning_rate": 1.2093352192362092e-07,
"logits/generated": 3.1363422870635986,
"logits/real": 2.7018938064575195,
"logps/generated": -546.12109375,
"logps/real": -1475.1849365234375,
"loss": 0.0003,
"rewards/accuracies": 1.0,
"rewards/generated": -12.732572555541992,
"rewards/margins": 26.642467498779297,
"rewards/real": 13.90989875793457,
"step": 1230
},
{
"epoch": 4.72,
"learning_rate": 1.1739745403111739e-07,
"logits/generated": 3.2991225719451904,
"logits/real": 2.5671286582946777,
"logps/generated": -463.2415466308594,
"logps/real": -1702.0286865234375,
"loss": 0.0004,
"rewards/accuracies": 1.0,
"rewards/generated": -11.021214485168457,
"rewards/margins": 26.277912139892578,
"rewards/real": 15.256696701049805,
"step": 1240
},
{
"epoch": 4.76,
"learning_rate": 1.1386138613861386e-07,
"logits/generated": 3.336590528488159,
"logits/real": 2.4211459159851074,
"logps/generated": -574.0828857421875,
"logps/real": -1439.0128173828125,
"loss": 0.0003,
"rewards/accuracies": 1.0,
"rewards/generated": -13.705076217651367,
"rewards/margins": 25.831045150756836,
"rewards/real": 12.125969886779785,
"step": 1250
},
{
"epoch": 4.8,
"learning_rate": 1.1032531824611033e-07,
"logits/generated": 3.1198062896728516,
"logits/real": 2.5011868476867676,
"logps/generated": -568.57177734375,
"logps/real": -1326.2716064453125,
"loss": 0.0006,
"rewards/accuracies": 1.0,
"rewards/generated": -12.998083114624023,
"rewards/margins": 25.015640258789062,
"rewards/real": 12.017557144165039,
"step": 1260
},
{
"epoch": 4.84,
"learning_rate": 1.0678925035360678e-07,
"logits/generated": 3.2028965950012207,
"logits/real": 2.5122649669647217,
"logps/generated": -561.6862182617188,
"logps/real": -1607.349365234375,
"loss": 0.0005,
"rewards/accuracies": 1.0,
"rewards/generated": -13.135000228881836,
"rewards/margins": 26.576717376708984,
"rewards/real": 13.441716194152832,
"step": 1270
},
{
"epoch": 4.88,
"learning_rate": 1.0325318246110325e-07,
"logits/generated": 3.216439723968506,
"logits/real": 2.6768908500671387,
"logps/generated": -613.3177490234375,
"logps/real": -1402.431396484375,
"loss": 0.0003,
"rewards/accuracies": 1.0,
"rewards/generated": -15.124387741088867,
"rewards/margins": 28.185840606689453,
"rewards/real": 13.061452865600586,
"step": 1280
},
{
"epoch": 4.91,
"learning_rate": 9.971711456859971e-08,
"logits/generated": 3.2150936126708984,
"logits/real": 2.443791627883911,
"logps/generated": -570.3302001953125,
"logps/real": -1579.4129638671875,
"loss": 0.0003,
"rewards/accuracies": 1.0,
"rewards/generated": -13.96008586883545,
"rewards/margins": 27.931583404541016,
"rewards/real": 13.97149658203125,
"step": 1290
},
{
"epoch": 4.95,
"learning_rate": 9.618104667609618e-08,
"logits/generated": 3.172091007232666,
"logits/real": 2.476423978805542,
"logps/generated": -550.3682250976562,
"logps/real": -1797.6273193359375,
"loss": 0.0003,
"rewards/accuracies": 1.0,
"rewards/generated": -13.748133659362793,
"rewards/margins": 27.63075828552246,
"rewards/real": 13.882623672485352,
"step": 1300
},
{
"epoch": 4.99,
"learning_rate": 9.264497878359263e-08,
"logits/generated": 3.2214691638946533,
"logits/real": 2.508018732070923,
"logps/generated": -520.4338989257812,
"logps/real": -1491.5899658203125,
"loss": 0.0001,
"rewards/accuracies": 1.0,
"rewards/generated": -12.596437454223633,
"rewards/margins": 27.078861236572266,
"rewards/real": 14.482423782348633,
"step": 1310
},
{
"epoch": 5.03,
"learning_rate": 8.91089108910891e-08,
"logits/generated": 3.090144634246826,
"logits/real": 2.64501953125,
"logps/generated": -575.699462890625,
"logps/real": -1449.6285400390625,
"loss": 0.0003,
"rewards/accuracies": 1.0,
"rewards/generated": -13.906789779663086,
"rewards/margins": 27.012332916259766,
"rewards/real": 13.105542182922363,
"step": 1320
},
{
"epoch": 5.07,
"learning_rate": 8.557284299858557e-08,
"logits/generated": 3.216085910797119,
"logits/real": 2.628757953643799,
"logps/generated": -563.4234619140625,
"logps/real": -1506.6536865234375,
"loss": 0.0002,
"rewards/accuracies": 1.0,
"rewards/generated": -14.473469734191895,
"rewards/margins": 27.185466766357422,
"rewards/real": 12.711997032165527,
"step": 1330
},
{
"epoch": 5.1,
"learning_rate": 8.203677510608204e-08,
"logits/generated": 3.2252001762390137,
"logits/real": 2.5635406970977783,
"logps/generated": -554.8922119140625,
"logps/real": -1379.961669921875,
"loss": 0.0003,
"rewards/accuracies": 1.0,
"rewards/generated": -13.215415954589844,
"rewards/margins": 24.81317138671875,
"rewards/real": 11.597757339477539,
"step": 1340
},
{
"epoch": 5.14,
"learning_rate": 7.85007072135785e-08,
"logits/generated": 3.1512575149536133,
"logits/real": 2.3962998390197754,
"logps/generated": -574.1869506835938,
"logps/real": -1498.155517578125,
"loss": 0.0002,
"rewards/accuracies": 1.0,
"rewards/generated": -13.913461685180664,
"rewards/margins": 27.563989639282227,
"rewards/real": 13.650527954101562,
"step": 1350
},
{
"epoch": 5.18,
"learning_rate": 7.496463932107496e-08,
"logits/generated": 3.3186583518981934,
"logits/real": 2.5883545875549316,
"logps/generated": -550.5670166015625,
"logps/real": -1544.77392578125,
"loss": 0.0002,
"rewards/accuracies": 1.0,
"rewards/generated": -12.482427597045898,
"rewards/margins": 27.15085220336914,
"rewards/real": 14.668424606323242,
"step": 1360
},
{
"epoch": 5.22,
"learning_rate": 7.142857142857142e-08,
"logits/generated": 3.263967990875244,
"logits/real": 2.536264181137085,
"logps/generated": -589.7340698242188,
"logps/real": -1500.179931640625,
"loss": 0.0004,
"rewards/accuracies": 1.0,
"rewards/generated": -15.068285942077637,
"rewards/margins": 28.84073829650879,
"rewards/real": 13.772455215454102,
"step": 1370
},
{
"epoch": 5.26,
"learning_rate": 6.789250353606789e-08,
"logits/generated": 3.497816801071167,
"logits/real": 2.428816318511963,
"logps/generated": -596.6787109375,
"logps/real": -1555.1085205078125,
"loss": 0.0003,
"rewards/accuracies": 1.0,
"rewards/generated": -15.438451766967773,
"rewards/margins": 29.081518173217773,
"rewards/real": 13.643072128295898,
"step": 1380
},
{
"epoch": 5.3,
"learning_rate": 6.435643564356436e-08,
"logits/generated": 3.010061264038086,
"logits/real": 2.4792978763580322,
"logps/generated": -526.48046875,
"logps/real": -1130.676513671875,
"loss": 0.0002,
"rewards/accuracies": 1.0,
"rewards/generated": -13.853174209594727,
"rewards/margins": 25.5134334564209,
"rewards/real": 11.660261154174805,
"step": 1390
},
{
"epoch": 5.33,
"learning_rate": 6.082036775106081e-08,
"logits/generated": 3.1093571186065674,
"logits/real": 2.4732513427734375,
"logps/generated": -529.5757446289062,
"logps/real": -1343.6026611328125,
"loss": 0.0001,
"rewards/accuracies": 1.0,
"rewards/generated": -12.401544570922852,
"rewards/margins": 25.137725830078125,
"rewards/real": 12.736178398132324,
"step": 1400
},
{
"epoch": 5.37,
"learning_rate": 5.7284299858557285e-08,
"logits/generated": 3.3876266479492188,
"logits/real": 2.5899083614349365,
"logps/generated": -588.9549560546875,
"logps/real": -1509.0943603515625,
"loss": 0.0002,
"rewards/accuracies": 1.0,
"rewards/generated": -13.996371269226074,
"rewards/margins": 27.71066665649414,
"rewards/real": 13.714296340942383,
"step": 1410
},
{
"epoch": 5.41,
"learning_rate": 5.374823196605375e-08,
"logits/generated": 3.112064838409424,
"logits/real": 2.6190438270568848,
"logps/generated": -551.2916870117188,
"logps/real": -1541.237060546875,
"loss": 0.0003,
"rewards/accuracies": 1.0,
"rewards/generated": -13.962061882019043,
"rewards/margins": 27.94354248046875,
"rewards/real": 13.981483459472656,
"step": 1420
},
{
"epoch": 5.45,
"learning_rate": 5.0212164073550206e-08,
"logits/generated": 3.4540042877197266,
"logits/real": 2.519503116607666,
"logps/generated": -547.7754516601562,
"logps/real": -1539.6217041015625,
"loss": 0.0002,
"rewards/accuracies": 1.0,
"rewards/generated": -13.80566692352295,
"rewards/margins": 27.109554290771484,
"rewards/real": 13.303888320922852,
"step": 1430
},
{
"epoch": 5.49,
"learning_rate": 4.667609618104667e-08,
"logits/generated": 3.2564361095428467,
"logits/real": 2.5632336139678955,
"logps/generated": -564.44287109375,
"logps/real": -1403.699951171875,
"loss": 0.0004,
"rewards/accuracies": 1.0,
"rewards/generated": -14.87768268585205,
"rewards/margins": 27.802310943603516,
"rewards/real": 12.924627304077148,
"step": 1440
},
{
"epoch": 5.52,
"learning_rate": 4.3140028288543134e-08,
"logits/generated": 3.235797166824341,
"logits/real": 2.559129238128662,
"logps/generated": -535.4729614257812,
"logps/real": -1371.961669921875,
"loss": 0.0001,
"rewards/accuracies": 1.0,
"rewards/generated": -13.979475021362305,
"rewards/margins": 26.749774932861328,
"rewards/real": 12.770299911499023,
"step": 1450
},
{
"epoch": 5.56,
"learning_rate": 3.96039603960396e-08,
"logits/generated": 3.105957508087158,
"logits/real": 2.7094485759735107,
"logps/generated": -624.28662109375,
"logps/real": -1619.6658935546875,
"loss": 0.0002,
"rewards/accuracies": 1.0,
"rewards/generated": -16.26133155822754,
"rewards/margins": 30.236867904663086,
"rewards/real": 13.97553825378418,
"step": 1460
},
{
"epoch": 5.6,
"learning_rate": 3.606789250353607e-08,
"logits/generated": 3.269968032836914,
"logits/real": 2.525156021118164,
"logps/generated": -547.2183837890625,
"logps/real": -1683.659423828125,
"loss": 0.0001,
"rewards/accuracies": 1.0,
"rewards/generated": -13.063116073608398,
"rewards/margins": 28.569652557373047,
"rewards/real": 15.506538391113281,
"step": 1470
},
{
"epoch": 5.64,
"learning_rate": 3.253182461103253e-08,
"logits/generated": 3.100898265838623,
"logits/real": 2.5007834434509277,
"logps/generated": -537.1228637695312,
"logps/real": -1622.945556640625,
"loss": 0.0003,
"rewards/accuracies": 1.0,
"rewards/generated": -13.297723770141602,
"rewards/margins": 27.18996810913086,
"rewards/real": 13.892248153686523,
"step": 1480
},
{
"epoch": 5.68,
"learning_rate": 2.8995756718528992e-08,
"logits/generated": 3.1697795391082764,
"logits/real": 2.515049457550049,
"logps/generated": -528.9276733398438,
"logps/real": -1365.165771484375,
"loss": 0.0001,
"rewards/accuracies": 1.0,
"rewards/generated": -12.908604621887207,
"rewards/margins": 25.516712188720703,
"rewards/real": 12.608107566833496,
"step": 1490
},
{
"epoch": 5.71,
"learning_rate": 2.545968882602546e-08,
"logits/generated": 3.247465133666992,
"logits/real": 2.738851308822632,
"logps/generated": -526.6806030273438,
"logps/real": -1680.171142578125,
"loss": 0.0002,
"rewards/accuracies": 1.0,
"rewards/generated": -13.784246444702148,
"rewards/margins": 29.31841468811035,
"rewards/real": 15.53416633605957,
"step": 1500
},
{
"epoch": 5.75,
"learning_rate": 2.1923620933521923e-08,
"logits/generated": 3.180821180343628,
"logits/real": 2.666006565093994,
"logps/generated": -635.9178466796875,
"logps/real": -1368.9986572265625,
"loss": 0.0001,
"rewards/accuracies": 1.0,
"rewards/generated": -14.650575637817383,
"rewards/margins": 27.650531768798828,
"rewards/real": 12.999956130981445,
"step": 1510
},
{
"epoch": 5.79,
"learning_rate": 1.8387553041018386e-08,
"logits/generated": 3.0839998722076416,
"logits/real": 2.581054210662842,
"logps/generated": -544.0230712890625,
"logps/real": -1359.1251220703125,
"loss": 0.0003,
"rewards/accuracies": 1.0,
"rewards/generated": -12.829536437988281,
"rewards/margins": 26.41360855102539,
"rewards/real": 13.584071159362793,
"step": 1520
},
{
"epoch": 5.83,
"learning_rate": 1.485148514851485e-08,
"logits/generated": 3.3319427967071533,
"logits/real": 2.6179628372192383,
"logps/generated": -590.0933837890625,
"logps/real": -1592.672119140625,
"loss": 0.0001,
"rewards/accuracies": 1.0,
"rewards/generated": -14.341830253601074,
"rewards/margins": 29.359546661376953,
"rewards/real": 15.017718315124512,
"step": 1530
},
{
"epoch": 5.87,
"learning_rate": 1.1315417256011316e-08,
"logits/generated": 3.4769108295440674,
"logits/real": 2.5944294929504395,
"logps/generated": -614.6587524414062,
"logps/real": -1540.0328369140625,
"loss": 0.0003,
"rewards/accuracies": 1.0,
"rewards/generated": -15.357019424438477,
"rewards/margins": 28.344806671142578,
"rewards/real": 12.987787246704102,
"step": 1540
},
{
"epoch": 5.9,
"learning_rate": 7.77934936350778e-09,
"logits/generated": 3.231849193572998,
"logits/real": 2.5611472129821777,
"logps/generated": -572.6622314453125,
"logps/real": -1665.4876708984375,
"loss": 0.0002,
"rewards/accuracies": 1.0,
"rewards/generated": -14.289937019348145,
"rewards/margins": 29.341079711914062,
"rewards/real": 15.051142692565918,
"step": 1550
},
{
"epoch": 5.94,
"learning_rate": 4.243281471004243e-09,
"logits/generated": 3.283064603805542,
"logits/real": 2.623944044113159,
"logps/generated": -635.8428955078125,
"logps/real": -1552.221923828125,
"loss": 0.0001,
"rewards/accuracies": 1.0,
"rewards/generated": -15.60615348815918,
"rewards/margins": 30.579345703125,
"rewards/real": 14.973190307617188,
"step": 1560
},
{
"epoch": 5.98,
"learning_rate": 7.072135785007072e-10,
"logits/generated": 3.2523193359375,
"logits/real": 2.4982213973999023,
"logps/generated": -530.0173950195312,
"logps/real": -1398.9669189453125,
"loss": 0.0001,
"rewards/accuracies": 1.0,
"rewards/generated": -12.438911437988281,
"rewards/margins": 25.966604232788086,
"rewards/real": 13.527694702148438,
"step": 1570
},
{
"epoch": 5.99,
"step": 1572,
"total_flos": 0.0,
"train_loss": 0.018845485343664946,
"train_runtime": 21431.5998,
"train_samples_per_second": 4.703,
"train_steps_per_second": 0.073
}
],
"logging_steps": 10,
"max_steps": 1572,
"num_input_tokens_seen": 0,
"num_train_epochs": 6,
"save_steps": 500,
"total_flos": 0.0,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}