gemma-2-9b-it-SimPO / trainer_state.json
xiamengzhou
update
dfefcd9
raw
history blame
50.3 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9989258861439313,
"eval_steps": 100000,
"global_step": 465,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.010741138560687433,
"grad_norm": 94.5147817778946,
"learning_rate": 8.51063829787234e-08,
"logits/chosen": -10.583702087402344,
"logits/rejected": -10.455877304077148,
"logps/chosen": -0.9049979448318481,
"logps/rejected": -0.8784100413322449,
"loss": 6.1451,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -9.04997730255127,
"rewards/margins": -0.2658771872520447,
"rewards/rejected": -8.784101486206055,
"step": 5
},
{
"epoch": 0.021482277121374866,
"grad_norm": 128.5515421485228,
"learning_rate": 1.702127659574468e-07,
"logits/chosen": -10.710015296936035,
"logits/rejected": -10.85377311706543,
"logps/chosen": -1.0046945810317993,
"logps/rejected": -0.8850045204162598,
"loss": 5.8491,
"rewards/accuracies": 0.512499988079071,
"rewards/chosen": -10.04694652557373,
"rewards/margins": -1.196901559829712,
"rewards/rejected": -8.850046157836914,
"step": 10
},
{
"epoch": 0.0322234156820623,
"grad_norm": 58.802331595913145,
"learning_rate": 2.553191489361702e-07,
"logits/chosen": -10.312850952148438,
"logits/rejected": -10.239133834838867,
"logps/chosen": -1.0889472961425781,
"logps/rejected": -1.1543949842453003,
"loss": 6.2505,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -10.889472961425781,
"rewards/margins": 0.6544777154922485,
"rewards/rejected": -11.543951034545898,
"step": 15
},
{
"epoch": 0.04296455424274973,
"grad_norm": 151.21348799898024,
"learning_rate": 3.404255319148936e-07,
"logits/chosen": -9.954164505004883,
"logits/rejected": -10.053568840026855,
"logps/chosen": -0.9611791372299194,
"logps/rejected": -1.1332345008850098,
"loss": 5.5619,
"rewards/accuracies": 0.44999998807907104,
"rewards/chosen": -9.611791610717773,
"rewards/margins": 1.7205528020858765,
"rewards/rejected": -11.332345008850098,
"step": 20
},
{
"epoch": 0.05370569280343716,
"grad_norm": 99.92865956439873,
"learning_rate": 4.25531914893617e-07,
"logits/chosen": -10.326103210449219,
"logits/rejected": -10.055009841918945,
"logps/chosen": -0.8260948061943054,
"logps/rejected": -1.1549828052520752,
"loss": 5.2635,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": -8.260948181152344,
"rewards/margins": 3.2888808250427246,
"rewards/rejected": -11.54982852935791,
"step": 25
},
{
"epoch": 0.0644468313641246,
"grad_norm": 81.6600880763309,
"learning_rate": 5.106382978723404e-07,
"logits/chosen": -9.319940567016602,
"logits/rejected": -9.13192081451416,
"logps/chosen": -0.6618553996086121,
"logps/rejected": -0.6553267240524292,
"loss": 5.0518,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -6.61855411529541,
"rewards/margins": -0.06528709828853607,
"rewards/rejected": -6.553267002105713,
"step": 30
},
{
"epoch": 0.07518796992481203,
"grad_norm": 92.79131753018717,
"learning_rate": 5.957446808510638e-07,
"logits/chosen": -8.877812385559082,
"logits/rejected": -8.929550170898438,
"logps/chosen": -0.717892050743103,
"logps/rejected": -0.6935927867889404,
"loss": 5.224,
"rewards/accuracies": 0.512499988079071,
"rewards/chosen": -7.178920745849609,
"rewards/margins": -0.24299363791942596,
"rewards/rejected": -6.935927391052246,
"step": 35
},
{
"epoch": 0.08592910848549946,
"grad_norm": 89.40050001716304,
"learning_rate": 6.808510638297872e-07,
"logits/chosen": -7.988096714019775,
"logits/rejected": -7.907191276550293,
"logps/chosen": -0.7402302622795105,
"logps/rejected": -0.7434382438659668,
"loss": 4.8885,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": -7.4023027420043945,
"rewards/margins": 0.032080501317977905,
"rewards/rejected": -7.434383392333984,
"step": 40
},
{
"epoch": 0.0966702470461869,
"grad_norm": 59.92644904113339,
"learning_rate": 7.659574468085107e-07,
"logits/chosen": -8.71805477142334,
"logits/rejected": -8.232014656066895,
"logps/chosen": -0.5317873954772949,
"logps/rejected": -0.6050616502761841,
"loss": 4.5879,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -5.317873477935791,
"rewards/margins": 0.7327424883842468,
"rewards/rejected": -6.0506157875061035,
"step": 45
},
{
"epoch": 0.10741138560687433,
"grad_norm": 60.25452206880678,
"learning_rate": 7.998983280184396e-07,
"logits/chosen": -8.83049488067627,
"logits/rejected": -8.585375785827637,
"logps/chosen": -0.5144228339195251,
"logps/rejected": -0.5809676647186279,
"loss": 4.6549,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -5.144228458404541,
"rewards/margins": 0.6654484868049622,
"rewards/rejected": -5.8096771240234375,
"step": 50
},
{
"epoch": 0.11815252416756176,
"grad_norm": 46.997605934809734,
"learning_rate": 7.992771864078597e-07,
"logits/chosen": -8.163946151733398,
"logits/rejected": -8.180994033813477,
"logps/chosen": -0.5956984758377075,
"logps/rejected": -0.7000880837440491,
"loss": 4.6606,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -5.956984519958496,
"rewards/margins": 1.0438958406448364,
"rewards/rejected": -7.000881195068359,
"step": 55
},
{
"epoch": 0.1288936627282492,
"grad_norm": 59.25357465703395,
"learning_rate": 7.980922636120897e-07,
"logits/chosen": -8.718216896057129,
"logits/rejected": -8.35698127746582,
"logps/chosen": -0.5706155896186829,
"logps/rejected": -0.6969493627548218,
"loss": 4.4885,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -5.706155776977539,
"rewards/margins": 1.2633379697799683,
"rewards/rejected": -6.969493865966797,
"step": 60
},
{
"epoch": 0.13963480128893663,
"grad_norm": 50.79780612404402,
"learning_rate": 7.963452327474534e-07,
"logits/chosen": -9.234804153442383,
"logits/rejected": -9.1095609664917,
"logps/chosen": -0.6090906858444214,
"logps/rejected": -0.7208055257797241,
"loss": 4.666,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -6.090908050537109,
"rewards/margins": 1.1171473264694214,
"rewards/rejected": -7.208055019378662,
"step": 65
},
{
"epoch": 0.15037593984962405,
"grad_norm": 56.36564641791114,
"learning_rate": 7.940385606293987e-07,
"logits/chosen": -8.946883201599121,
"logits/rejected": -8.716778755187988,
"logps/chosen": -0.6818052530288696,
"logps/rejected": -0.7961267828941345,
"loss": 4.577,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -6.818052768707275,
"rewards/margins": 1.1432150602340698,
"rewards/rejected": -7.961267948150635,
"step": 70
},
{
"epoch": 0.1611170784103115,
"grad_norm": 61.76930510948969,
"learning_rate": 7.911755042893434e-07,
"logits/chosen": -9.067525863647461,
"logits/rejected": -8.9346923828125,
"logps/chosen": -0.6832990646362305,
"logps/rejected": -0.7763570547103882,
"loss": 4.4179,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": -6.8329901695251465,
"rewards/margins": 0.9305804371833801,
"rewards/rejected": -7.763571262359619,
"step": 75
},
{
"epoch": 0.17185821697099893,
"grad_norm": 63.602633074853536,
"learning_rate": 7.877601063757321e-07,
"logits/chosen": -9.461370468139648,
"logits/rejected": -8.981520652770996,
"logps/chosen": -0.6881433129310608,
"logps/rejected": -0.8508684039115906,
"loss": 4.3763,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -6.881433010101318,
"rewards/margins": 1.6272509098052979,
"rewards/rejected": -8.508684158325195,
"step": 80
},
{
"epoch": 0.18259935553168635,
"grad_norm": 50.67613033462041,
"learning_rate": 7.837971894457989e-07,
"logits/chosen": -9.557887077331543,
"logits/rejected": -9.17081069946289,
"logps/chosen": -0.6830392479896545,
"logps/rejected": -0.799291729927063,
"loss": 4.6499,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -6.830392360687256,
"rewards/margins": 1.1625245809555054,
"rewards/rejected": -7.992917060852051,
"step": 85
},
{
"epoch": 0.1933404940923738,
"grad_norm": 57.58777625940833,
"learning_rate": 7.792923491560942e-07,
"logits/chosen": -8.579484939575195,
"logits/rejected": -8.546136856079102,
"logps/chosen": -0.6667743921279907,
"logps/rejected": -0.7407978177070618,
"loss": 4.4492,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -6.667743682861328,
"rewards/margins": 0.7402342557907104,
"rewards/rejected": -7.407977104187012,
"step": 90
},
{
"epoch": 0.20408163265306123,
"grad_norm": 68.91179001810359,
"learning_rate": 7.742519463613926e-07,
"logits/chosen": -9.336307525634766,
"logits/rejected": -9.128133773803711,
"logps/chosen": -0.706219494342804,
"logps/rejected": -0.7757526636123657,
"loss": 4.2763,
"rewards/accuracies": 0.5,
"rewards/chosen": -7.062193870544434,
"rewards/margins": 0.6953321099281311,
"rewards/rejected": -7.7575273513793945,
"step": 95
},
{
"epoch": 0.21482277121374865,
"grad_norm": 140.43504047927686,
"learning_rate": 7.68683098133138e-07,
"logits/chosen": -8.939419746398926,
"logits/rejected": -8.681028366088867,
"logps/chosen": -0.7093919515609741,
"logps/rejected": -0.8932901620864868,
"loss": 4.4002,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -7.093919277191162,
"rewards/margins": 1.838982343673706,
"rewards/rejected": -8.932901382446289,
"step": 100
},
{
"epoch": 0.22556390977443608,
"grad_norm": 105.52949175741638,
"learning_rate": 7.625936677101051e-07,
"logits/chosen": -8.601816177368164,
"logits/rejected": -8.625459671020508,
"logps/chosen": -0.8767679333686829,
"logps/rejected": -0.8515819311141968,
"loss": 4.4644,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -8.767679214477539,
"rewards/margins": -0.25185948610305786,
"rewards/rejected": -8.51581859588623,
"step": 105
},
{
"epoch": 0.23630504833512353,
"grad_norm": 56.44051818244315,
"learning_rate": 7.559922533954731e-07,
"logits/chosen": -9.58240795135498,
"logits/rejected": -9.501542091369629,
"logps/chosen": -0.7655607461929321,
"logps/rejected": -0.8916142582893372,
"loss": 4.2797,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": -7.6556077003479,
"rewards/margins": 1.2605348825454712,
"rewards/rejected": -8.916143417358398,
"step": 110
},
{
"epoch": 0.24704618689581095,
"grad_norm": 68.9497539150874,
"learning_rate": 7.488881764159808e-07,
"logits/chosen": -9.756335258483887,
"logits/rejected": -9.543218612670898,
"logps/chosen": -0.7038711309432983,
"logps/rejected": -0.7986757159233093,
"loss": 4.154,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -7.0387115478515625,
"rewards/margins": 0.9480463862419128,
"rewards/rejected": -7.986758232116699,
"step": 115
},
{
"epoch": 0.2577873254564984,
"grad_norm": 90.20504733911939,
"learning_rate": 7.412914677603135e-07,
"logits/chosen": -9.883420944213867,
"logits/rejected": -9.735390663146973,
"logps/chosen": -0.9017173647880554,
"logps/rejected": -0.9996153116226196,
"loss": 4.2168,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -9.017173767089844,
"rewards/margins": 0.9789786338806152,
"rewards/rejected": -9.996152877807617,
"step": 120
},
{
"epoch": 0.26852846401718583,
"grad_norm": 63.31580116626401,
"learning_rate": 7.332128540153017e-07,
"logits/chosen": -10.71928596496582,
"logits/rejected": -10.555776596069336,
"logps/chosen": -0.788918673992157,
"logps/rejected": -0.9437187910079956,
"loss": 4.1085,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -7.889185905456543,
"rewards/margins": 1.5480016469955444,
"rewards/rejected": -9.437189102172852,
"step": 125
},
{
"epoch": 0.27926960257787325,
"grad_norm": 81.18347953083857,
"learning_rate": 7.246637422199322e-07,
"logits/chosen": -10.676037788391113,
"logits/rejected": -10.630210876464844,
"logps/chosen": -0.8381876945495605,
"logps/rejected": -1.0221493244171143,
"loss": 4.0857,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -8.381875991821289,
"rewards/margins": 1.8396151065826416,
"rewards/rejected": -10.221491813659668,
"step": 130
},
{
"epoch": 0.2900107411385607,
"grad_norm": 80.90445716094146,
"learning_rate": 7.156562037585574e-07,
"logits/chosen": -11.714326858520508,
"logits/rejected": -11.1636323928833,
"logps/chosen": -0.8452903628349304,
"logps/rejected": -1.1077954769134521,
"loss": 3.9406,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -8.452905654907227,
"rewards/margins": 2.6250510215759277,
"rewards/rejected": -11.07795524597168,
"step": 135
},
{
"epoch": 0.3007518796992481,
"grad_norm": 88.8085364984583,
"learning_rate": 7.062029573160467e-07,
"logits/chosen": -11.935297012329102,
"logits/rejected": -11.792046546936035,
"logps/chosen": -0.9109989404678345,
"logps/rejected": -1.1446388959884644,
"loss": 3.6921,
"rewards/accuracies": 0.8125,
"rewards/chosen": -9.109989166259766,
"rewards/margins": 2.3363993167877197,
"rewards/rejected": -11.446390151977539,
"step": 140
},
{
"epoch": 0.31149301825993553,
"grad_norm": 110.78828702618637,
"learning_rate": 6.963173509189455e-07,
"logits/chosen": -13.552042007446289,
"logits/rejected": -13.324705123901367,
"logps/chosen": -1.0502710342407227,
"logps/rejected": -1.2693006992340088,
"loss": 3.8236,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": -10.502711296081543,
"rewards/margins": 2.190295696258545,
"rewards/rejected": -12.69300651550293,
"step": 145
},
{
"epoch": 0.322234156820623,
"grad_norm": 111.15417892636452,
"learning_rate": 6.860133430880024e-07,
"logits/chosen": -14.586761474609375,
"logits/rejected": -14.23077392578125,
"logps/chosen": -1.2444875240325928,
"logps/rejected": -1.4712624549865723,
"loss": 3.7527,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -12.444875717163086,
"rewards/margins": 2.2677478790283203,
"rewards/rejected": -14.712623596191406,
"step": 150
},
{
"epoch": 0.33297529538131043,
"grad_norm": 119.28100769496392,
"learning_rate": 6.753054831286747e-07,
"logits/chosen": -14.823234558105469,
"logits/rejected": -14.770757675170898,
"logps/chosen": -1.3944904804229736,
"logps/rejected": -1.6155385971069336,
"loss": 3.893,
"rewards/accuracies": 0.6875,
"rewards/chosen": -13.944903373718262,
"rewards/margins": 2.2104804515838623,
"rewards/rejected": -16.155384063720703,
"step": 155
},
{
"epoch": 0.34371643394199786,
"grad_norm": 103.61310969210844,
"learning_rate": 6.642088905874433e-07,
"logits/chosen": -14.195696830749512,
"logits/rejected": -14.07690143585205,
"logps/chosen": -1.3757129907608032,
"logps/rejected": -1.5891997814178467,
"loss": 3.7363,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -13.75713062286377,
"rewards/margins": 2.134868860244751,
"rewards/rejected": -15.891998291015625,
"step": 160
},
{
"epoch": 0.3544575725026853,
"grad_norm": 95.46872943421629,
"learning_rate": 6.527392339029455e-07,
"logits/chosen": -14.401777267456055,
"logits/rejected": -14.322749137878418,
"logps/chosen": -1.2530758380889893,
"logps/rejected": -1.5336121320724487,
"loss": 3.4763,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -12.530759811401367,
"rewards/margins": 2.8053627014160156,
"rewards/rejected": -15.336122512817383,
"step": 165
},
{
"epoch": 0.3651987110633727,
"grad_norm": 119.23452912777815,
"learning_rate": 6.409127082820689e-07,
"logits/chosen": -14.566454887390137,
"logits/rejected": -14.484842300415039,
"logps/chosen": -1.531166434288025,
"logps/rejected": -1.7996248006820679,
"loss": 3.6554,
"rewards/accuracies": 0.6875,
"rewards/chosen": -15.311663627624512,
"rewards/margins": 2.6845829486846924,
"rewards/rejected": -17.996248245239258,
"step": 170
},
{
"epoch": 0.37593984962406013,
"grad_norm": 123.93953756068933,
"learning_rate": 6.287460128322457e-07,
"logits/chosen": -14.157377243041992,
"logits/rejected": -14.0371675491333,
"logps/chosen": -1.5080561637878418,
"logps/rejected": -1.9248558282852173,
"loss": 3.4374,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -15.080561637878418,
"rewards/margins": 4.167994022369385,
"rewards/rejected": -19.24855613708496,
"step": 175
},
{
"epoch": 0.3866809881847476,
"grad_norm": 223.710665367037,
"learning_rate": 6.16256326982239e-07,
"logits/chosen": -16.03777313232422,
"logits/rejected": -16.100345611572266,
"logps/chosen": -1.5300877094268799,
"logps/rejected": -1.8575336933135986,
"loss": 3.4606,
"rewards/accuracies": 0.6875,
"rewards/chosen": -15.300875663757324,
"rewards/margins": 3.2744598388671875,
"rewards/rejected": -18.575336456298828,
"step": 180
},
{
"epoch": 0.39742212674543503,
"grad_norm": 294.4348098673441,
"learning_rate": 6.034612862247114e-07,
"logits/chosen": -14.142799377441406,
"logits/rejected": -13.796422958374023,
"logps/chosen": -1.5025275945663452,
"logps/rejected": -1.7810484170913696,
"loss": 3.1185,
"rewards/accuracies": 0.6875,
"rewards/chosen": -15.025274276733398,
"rewards/margins": 2.7852089405059814,
"rewards/rejected": -17.810483932495117,
"step": 185
},
{
"epoch": 0.40816326530612246,
"grad_norm": 112.37773385751002,
"learning_rate": 5.903789572148295e-07,
"logits/chosen": -14.8009614944458,
"logits/rejected": -14.250249862670898,
"logps/chosen": -1.5931546688079834,
"logps/rejected": -2.022378444671631,
"loss": 3.3847,
"rewards/accuracies": 0.75,
"rewards/chosen": -15.931546211242676,
"rewards/margins": 4.292238712310791,
"rewards/rejected": -20.223783493041992,
"step": 190
},
{
"epoch": 0.4189044038668099,
"grad_norm": 125.77761635765717,
"learning_rate": 5.770278122600662e-07,
"logits/chosen": -14.832977294921875,
"logits/rejected": -14.608530044555664,
"logps/chosen": -1.6177564859390259,
"logps/rejected": -1.9727256298065186,
"loss": 3.6009,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -16.177562713623047,
"rewards/margins": 3.549692153930664,
"rewards/rejected": -19.72725486755371,
"step": 195
},
{
"epoch": 0.4296455424274973,
"grad_norm": 101.70153964682511,
"learning_rate": 5.634267032372192e-07,
"logits/chosen": -14.803668022155762,
"logits/rejected": -14.786203384399414,
"logps/chosen": -1.6423594951629639,
"logps/rejected": -1.9881032705307007,
"loss": 3.3904,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -16.423595428466797,
"rewards/margins": 3.457437515258789,
"rewards/rejected": -19.881032943725586,
"step": 200
},
{
"epoch": 0.44038668098818473,
"grad_norm": 123.22102599092351,
"learning_rate": 5.495948349734758e-07,
"logits/chosen": -14.582061767578125,
"logits/rejected": -14.51270580291748,
"logps/chosen": -1.747982382774353,
"logps/rejected": -2.051506757736206,
"loss": 3.1521,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -17.47982406616211,
"rewards/margins": 3.035243034362793,
"rewards/rejected": -20.51506805419922,
"step": 205
},
{
"epoch": 0.45112781954887216,
"grad_norm": 106.53206062736002,
"learning_rate": 5.355517381291105e-07,
"logits/chosen": -15.856142044067383,
"logits/rejected": -15.048059463500977,
"logps/chosen": -1.9086406230926514,
"logps/rejected": -2.3940463066101074,
"loss": 3.1856,
"rewards/accuracies": 0.75,
"rewards/chosen": -19.086406707763672,
"rewards/margins": 4.854057312011719,
"rewards/rejected": -23.940462112426758,
"step": 210
},
{
"epoch": 0.46186895810955964,
"grad_norm": 135.01895835142213,
"learning_rate": 5.21317241620105e-07,
"logits/chosen": -17.223520278930664,
"logits/rejected": -16.822795867919922,
"logps/chosen": -1.9180253744125366,
"logps/rejected": -2.2895708084106445,
"loss": 3.4171,
"rewards/accuracies": 0.75,
"rewards/chosen": -19.180253982543945,
"rewards/margins": 3.7154533863067627,
"rewards/rejected": -22.895706176757812,
"step": 215
},
{
"epoch": 0.47261009667024706,
"grad_norm": 116.88786548811646,
"learning_rate": 5.069114446196291e-07,
"logits/chosen": -14.405430793762207,
"logits/rejected": -14.03125,
"logps/chosen": -1.8195463418960571,
"logps/rejected": -2.373373508453369,
"loss": 3.1087,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -18.195463180541992,
"rewards/margins": 5.538268566131592,
"rewards/rejected": -23.73373031616211,
"step": 220
},
{
"epoch": 0.4833512352309345,
"grad_norm": 117.33800643418842,
"learning_rate": 4.923546881779183e-07,
"logits/chosen": -15.34239387512207,
"logits/rejected": -15.118896484375,
"logps/chosen": -1.5427398681640625,
"logps/rejected": -1.9821481704711914,
"loss": 3.0885,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -15.427398681640625,
"rewards/margins": 4.394083023071289,
"rewards/rejected": -19.821481704711914,
"step": 225
},
{
"epoch": 0.4940923737916219,
"grad_norm": 116.49865117484065,
"learning_rate": 4.776675265006186e-07,
"logits/chosen": -14.630195617675781,
"logits/rejected": -14.59937858581543,
"logps/chosen": -1.6563146114349365,
"logps/rejected": -2.0928232669830322,
"loss": 3.2032,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -16.56314468383789,
"rewards/margins": 4.365086555480957,
"rewards/rejected": -20.928232192993164,
"step": 230
},
{
"epoch": 0.5048335123523093,
"grad_norm": 203.29377037587201,
"learning_rate": 4.62870697926156e-07,
"logits/chosen": -14.8600435256958,
"logits/rejected": -15.338279724121094,
"logps/chosen": -1.7831714153289795,
"logps/rejected": -2.145383358001709,
"loss": 3.2633,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -17.831714630126953,
"rewards/margins": 3.622117519378662,
"rewards/rejected": -21.45383071899414,
"step": 235
},
{
"epoch": 0.5155746509129968,
"grad_norm": 116.16395762640381,
"learning_rate": 4.479850956431092e-07,
"logits/chosen": -14.476922988891602,
"logits/rejected": -14.742956161499023,
"logps/chosen": -1.6810489892959595,
"logps/rejected": -2.0232789516448975,
"loss": 3.0869,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -16.810489654541016,
"rewards/margins": 3.4223015308380127,
"rewards/rejected": -20.232791900634766,
"step": 240
},
{
"epoch": 0.5263157894736842,
"grad_norm": 98.59837238595348,
"learning_rate": 4.33031738188933e-07,
"logits/chosen": -15.394210815429688,
"logits/rejected": -14.864255905151367,
"logps/chosen": -1.6588356494903564,
"logps/rejected": -2.0340933799743652,
"loss": 3.2303,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -16.58835792541504,
"rewards/margins": 3.752579927444458,
"rewards/rejected": -20.3409366607666,
"step": 245
},
{
"epoch": 0.5370569280343717,
"grad_norm": 122.58746074195308,
"learning_rate": 4.180317397716889e-07,
"logits/chosen": -15.588345527648926,
"logits/rejected": -15.222723007202148,
"logps/chosen": -1.6328115463256836,
"logps/rejected": -2.2505877017974854,
"loss": 3.0874,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -16.328113555908203,
"rewards/margins": 6.177763938903809,
"rewards/rejected": -22.505878448486328,
"step": 250
},
{
"epoch": 0.547798066595059,
"grad_norm": 120.49234048427142,
"learning_rate": 4.030062804566888e-07,
"logits/chosen": -15.462881088256836,
"logits/rejected": -15.3878173828125,
"logps/chosen": -1.6617505550384521,
"logps/rejected": -1.9518375396728516,
"loss": 3.0471,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -16.61750602722168,
"rewards/margins": 2.9008688926696777,
"rewards/rejected": -19.518375396728516,
"step": 255
},
{
"epoch": 0.5585392051557465,
"grad_norm": 102.36282696415171,
"learning_rate": 3.8797657626014614e-07,
"logits/chosen": -15.693799018859863,
"logits/rejected": -15.475125312805176,
"logps/chosen": -1.7303409576416016,
"logps/rejected": -2.111295223236084,
"loss": 3.1486,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -17.303409576416016,
"rewards/margins": 3.809544801712036,
"rewards/rejected": -21.112953186035156,
"step": 260
},
{
"epoch": 0.569280343716434,
"grad_norm": 120.30352416855824,
"learning_rate": 3.729638491920669e-07,
"logits/chosen": -14.018827438354492,
"logits/rejected": -14.128240585327148,
"logps/chosen": -1.5650581121444702,
"logps/rejected": -1.8844165802001953,
"loss": 3.0723,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -15.650581359863281,
"rewards/margins": 3.193586826324463,
"rewards/rejected": -18.84417152404785,
"step": 265
},
{
"epoch": 0.5800214822771214,
"grad_norm": 91.11946128351651,
"learning_rate": 3.5798929729067464e-07,
"logits/chosen": -15.980966567993164,
"logits/rejected": -15.59577465057373,
"logps/chosen": -1.759478211402893,
"logps/rejected": -2.1487960815429688,
"loss": 2.7323,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -17.59478187561035,
"rewards/margins": 3.8931777477264404,
"rewards/rejected": -21.487960815429688,
"step": 270
},
{
"epoch": 0.5907626208378088,
"grad_norm": 111.96074414088407,
"learning_rate": 3.4307406469068595e-07,
"logits/chosen": -15.691810607910156,
"logits/rejected": -15.631698608398438,
"logps/chosen": -1.7383606433868408,
"logps/rejected": -2.2635440826416016,
"loss": 2.9053,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -17.38360595703125,
"rewards/margins": 5.251835823059082,
"rewards/rejected": -22.635440826416016,
"step": 275
},
{
"epoch": 0.6015037593984962,
"grad_norm": 122.95732420434838,
"learning_rate": 3.282392117676968e-07,
"logits/chosen": -15.389913558959961,
"logits/rejected": -15.556841850280762,
"logps/chosen": -1.9042613506317139,
"logps/rejected": -2.476677417755127,
"loss": 2.9112,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -19.04261589050293,
"rewards/margins": 5.72415828704834,
"rewards/rejected": -24.766775131225586,
"step": 280
},
{
"epoch": 0.6122448979591837,
"grad_norm": 104.08928006990811,
"learning_rate": 3.135056854008371e-07,
"logits/chosen": -16.152729034423828,
"logits/rejected": -16.002233505249023,
"logps/chosen": -1.8163830041885376,
"logps/rejected": -2.228738784790039,
"loss": 2.8591,
"rewards/accuracies": 0.75,
"rewards/chosen": -18.163829803466797,
"rewards/margins": 4.123559474945068,
"rewards/rejected": -22.28738784790039,
"step": 285
},
{
"epoch": 0.6229860365198711,
"grad_norm": 113.24003069429473,
"learning_rate": 2.988942893956833e-07,
"logits/chosen": -15.338768005371094,
"logits/rejected": -15.249606132507324,
"logps/chosen": -1.8918412923812866,
"logps/rejected": -2.32609224319458,
"loss": 3.0229,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -18.918415069580078,
"rewards/margins": 4.342508792877197,
"rewards/rejected": -23.260921478271484,
"step": 290
},
{
"epoch": 0.6337271750805585,
"grad_norm": 151.8876206935079,
"learning_rate": 2.844256551091911e-07,
"logits/chosen": -16.8232421875,
"logits/rejected": -16.842761993408203,
"logps/chosen": -1.9518108367919922,
"logps/rejected": -2.4939913749694824,
"loss": 2.833,
"rewards/accuracies": 0.8125,
"rewards/chosen": -19.518108367919922,
"rewards/margins": 5.421802997589111,
"rewards/rejected": -24.939910888671875,
"step": 295
},
{
"epoch": 0.644468313641246,
"grad_norm": 273.6603281637076,
"learning_rate": 2.7012021231812664e-07,
"logits/chosen": -16.766956329345703,
"logits/rejected": -16.424999237060547,
"logps/chosen": -1.969435453414917,
"logps/rejected": -2.308202028274536,
"loss": 3.2442,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -19.694355010986328,
"rewards/margins": 3.387664318084717,
"rewards/rejected": -23.082019805908203,
"step": 300
},
{
"epoch": 0.6552094522019334,
"grad_norm": 187.01720889425104,
"learning_rate": 2.5599816037212954e-07,
"logits/chosen": -14.743069648742676,
"logits/rejected": -14.623723983764648,
"logps/chosen": -1.8649146556854248,
"logps/rejected": -2.4030163288116455,
"loss": 2.9136,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -18.64914894104004,
"rewards/margins": 5.381015777587891,
"rewards/rejected": -24.03016471862793,
"step": 305
},
{
"epoch": 0.6659505907626209,
"grad_norm": 142.83830462137144,
"learning_rate": 2.4207943967214064e-07,
"logits/chosen": -16.09463119506836,
"logits/rejected": -15.806689262390137,
"logps/chosen": -2.033447742462158,
"logps/rejected": -2.524364471435547,
"loss": 3.1207,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -20.334476470947266,
"rewards/margins": 4.909164905548096,
"rewards/rejected": -25.24364471435547,
"step": 310
},
{
"epoch": 0.6766917293233082,
"grad_norm": 106.6308691579763,
"learning_rate": 2.2838370351446547e-07,
"logits/chosen": -15.870585441589355,
"logits/rejected": -15.55876350402832,
"logps/chosen": -1.8023881912231445,
"logps/rejected": -2.277968168258667,
"loss": 2.7641,
"rewards/accuracies": 0.6875,
"rewards/chosen": -18.023881912231445,
"rewards/margins": 4.755801200866699,
"rewards/rejected": -22.779682159423828,
"step": 315
},
{
"epoch": 0.6874328678839957,
"grad_norm": 108.7814729394269,
"learning_rate": 2.1493029034023188e-07,
"logits/chosen": -15.210580825805664,
"logits/rejected": -15.057415962219238,
"logps/chosen": -1.789072036743164,
"logps/rejected": -2.2775633335113525,
"loss": 2.8863,
"rewards/accuracies": 0.75,
"rewards/chosen": -17.89072036743164,
"rewards/margins": 4.884912014007568,
"rewards/rejected": -22.775630950927734,
"step": 320
},
{
"epoch": 0.6981740064446831,
"grad_norm": 180.86259023584043,
"learning_rate": 2.0173819642942376e-07,
"logits/chosen": -14.378689765930176,
"logits/rejected": -14.201916694641113,
"logps/chosen": -1.9404065608978271,
"logps/rejected": -2.5850563049316406,
"loss": 2.9848,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -19.404064178466797,
"rewards/margins": 6.446499824523926,
"rewards/rejected": -25.85056495666504,
"step": 325
},
{
"epoch": 0.7089151450053706,
"grad_norm": 132.13798361660807,
"learning_rate": 1.888260490780485e-07,
"logits/chosen": -14.281087875366211,
"logits/rejected": -14.154438972473145,
"logps/chosen": -1.7613548040390015,
"logps/rejected": -2.238495349884033,
"loss": 3.0495,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -17.61355209350586,
"rewards/margins": 4.771404266357422,
"rewards/rejected": -22.384952545166016,
"step": 330
},
{
"epoch": 0.719656283566058,
"grad_norm": 127.44733826541191,
"learning_rate": 1.7621208029631078e-07,
"logits/chosen": -14.766406059265137,
"logits/rejected": -14.667470932006836,
"logps/chosen": -1.9043552875518799,
"logps/rejected": -2.4970054626464844,
"loss": 2.9109,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -19.04355239868164,
"rewards/margins": 5.9265007972717285,
"rewards/rejected": -24.970054626464844,
"step": 335
},
{
"epoch": 0.7303974221267454,
"grad_norm": 114.16195914051926,
"learning_rate": 1.6391410106493227e-07,
"logits/chosen": -14.881872177124023,
"logits/rejected": -14.627456665039062,
"logps/chosen": -1.9786325693130493,
"logps/rejected": -2.5424978733062744,
"loss": 2.8817,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -19.786325454711914,
"rewards/margins": 5.63865327835083,
"rewards/rejected": -25.424976348876953,
"step": 340
},
{
"epoch": 0.7411385606874329,
"grad_norm": 99.76652109445614,
"learning_rate": 1.5194947618596673e-07,
"logits/chosen": -15.026782035827637,
"logits/rejected": -14.58587646484375,
"logps/chosen": -1.9211170673370361,
"logps/rejected": -2.394774913787842,
"loss": 3.0269,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -19.211170196533203,
"rewards/margins": 4.736577033996582,
"rewards/rejected": -23.9477481842041,
"step": 345
},
{
"epoch": 0.7518796992481203,
"grad_norm": 107.2380444923777,
"learning_rate": 1.4033509976362083e-07,
"logits/chosen": -15.670697212219238,
"logits/rejected": -15.579752922058105,
"logps/chosen": -1.9338979721069336,
"logps/rejected": -2.3520257472991943,
"loss": 2.9892,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -19.338979721069336,
"rewards/margins": 4.181277751922607,
"rewards/rejected": -23.52025604248047,
"step": 350
},
{
"epoch": 0.7626208378088077,
"grad_norm": 108.07049767560719,
"learning_rate": 1.2908737134970363e-07,
"logits/chosen": -14.5513334274292,
"logits/rejected": -14.516873359680176,
"logps/chosen": -1.8658726215362549,
"logps/rejected": -2.476576566696167,
"loss": 3.1196,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -18.65872573852539,
"rewards/margins": 6.1070404052734375,
"rewards/rejected": -24.765766143798828,
"step": 355
},
{
"epoch": 0.7733619763694952,
"grad_norm": 103.96338647941381,
"learning_rate": 1.1822217278738515e-07,
"logits/chosen": -15.559527397155762,
"logits/rejected": -15.545167922973633,
"logps/chosen": -1.9091074466705322,
"logps/rejected": -2.4284536838531494,
"loss": 3.0461,
"rewards/accuracies": 0.75,
"rewards/chosen": -19.091075897216797,
"rewards/margins": 5.193462371826172,
"rewards/rejected": -24.284536361694336,
"step": 360
},
{
"epoch": 0.7841031149301826,
"grad_norm": 138.78303459839336,
"learning_rate": 1.0775484578596241e-07,
"logits/chosen": -15.669352531433105,
"logits/rejected": -15.537897109985352,
"logps/chosen": -1.94468092918396,
"logps/rejected": -2.5436809062957764,
"loss": 2.7974,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -19.44681167602539,
"rewards/margins": 5.989997386932373,
"rewards/rejected": -25.436809539794922,
"step": 365
},
{
"epoch": 0.7948442534908701,
"grad_norm": 122.14734643501443,
"learning_rate": 9.770017025829673e-08,
"logits/chosen": -15.961019515991211,
"logits/rejected": -15.95417308807373,
"logps/chosen": -2.1812241077423096,
"logps/rejected": -2.6908135414123535,
"loss": 2.6374,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -21.812240600585938,
"rewards/margins": 5.095890998840332,
"rewards/rejected": -26.908132553100586,
"step": 370
},
{
"epoch": 0.8055853920515574,
"grad_norm": 114.19485857094502,
"learning_rate": 8.807234345151027e-08,
"logits/chosen": -14.920249938964844,
"logits/rejected": -14.890344619750977,
"logps/chosen": -2.0413661003112793,
"logps/rejected": -2.6543760299682617,
"loss": 2.8891,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -20.41366195678711,
"rewards/margins": 6.13009786605835,
"rewards/rejected": -26.543758392333984,
"step": 375
},
{
"epoch": 0.8163265306122449,
"grad_norm": 184.4289644042616,
"learning_rate": 7.888495990040924e-08,
"logits/chosen": -13.656982421875,
"logits/rejected": -13.701431274414062,
"logps/chosen": -2.011772871017456,
"logps/rejected": -2.7216084003448486,
"loss": 2.9801,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": -20.117727279663086,
"rewards/margins": 7.098354339599609,
"rewards/rejected": -27.21608543395996,
"step": 380
},
{
"epoch": 0.8270676691729323,
"grad_norm": 132.46593111411627,
"learning_rate": 7.015099223193943e-08,
"logits/chosen": -15.658330917358398,
"logits/rejected": -15.693890571594238,
"logps/chosen": -1.9581212997436523,
"logps/rejected": -2.5001060962677,
"loss": 2.8611,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -19.58121681213379,
"rewards/margins": 5.419846534729004,
"rewards/rejected": -25.001062393188477,
"step": 385
},
{
"epoch": 0.8378088077336198,
"grad_norm": 109.73562851103766,
"learning_rate": 6.188277284777857e-08,
"logits/chosen": -14.48884391784668,
"logits/rejected": -13.800481796264648,
"logps/chosen": -1.9877732992172241,
"logps/rejected": -2.5746169090270996,
"loss": 2.8295,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": -19.877731323242188,
"rewards/margins": 5.868436813354492,
"rewards/rejected": -25.746166229248047,
"step": 390
},
{
"epoch": 0.8485499462943072,
"grad_norm": 119.04235166242496,
"learning_rate": 5.409197651092965e-08,
"logits/chosen": -15.729510307312012,
"logits/rejected": -15.620780944824219,
"logps/chosen": -2.1660141944885254,
"logps/rejected": -2.667978286743164,
"loss": 2.7298,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -21.660140991210938,
"rewards/margins": 5.019640922546387,
"rewards/rejected": -26.679784774780273,
"step": 395
},
{
"epoch": 0.8592910848549946,
"grad_norm": 105.88618147362118,
"learning_rate": 4.678960386090298e-08,
"logits/chosen": -15.191770553588867,
"logits/rejected": -15.158576965332031,
"logps/chosen": -1.9015194177627563,
"logps/rejected": -2.5145716667175293,
"loss": 2.7402,
"rewards/accuracies": 0.8125,
"rewards/chosen": -19.015193939208984,
"rewards/margins": 6.130521297454834,
"rewards/rejected": -25.145715713500977,
"step": 400
},
{
"epoch": 0.8700322234156821,
"grad_norm": 114.47737417347113,
"learning_rate": 3.998596588076366e-08,
"logits/chosen": -13.7559814453125,
"logits/rejected": -13.483953475952148,
"logps/chosen": -1.9659799337387085,
"logps/rejected": -2.390488862991333,
"loss": 3.1028,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -19.659801483154297,
"rewards/margins": 4.245090484619141,
"rewards/rejected": -23.904891967773438,
"step": 405
},
{
"epoch": 0.8807733619763695,
"grad_norm": 124.82158209754382,
"learning_rate": 3.3690669337976996e-08,
"logits/chosen": -15.061103820800781,
"logits/rejected": -14.891242980957031,
"logps/chosen": -1.8727645874023438,
"logps/rejected": -2.3343753814697266,
"loss": 2.7051,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -18.727645874023438,
"rewards/margins": 4.6161088943481445,
"rewards/rejected": -23.343753814697266,
"step": 410
},
{
"epoch": 0.8915145005370569,
"grad_norm": 127.84134494966216,
"learning_rate": 2.7912603219609798e-08,
"logits/chosen": -15.650156021118164,
"logits/rejected": -15.531412124633789,
"logps/chosen": -2.0558724403381348,
"logps/rejected": -2.456268787384033,
"loss": 2.7054,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -20.558725357055664,
"rewards/margins": 4.003961563110352,
"rewards/rejected": -24.562685012817383,
"step": 415
},
{
"epoch": 0.9022556390977443,
"grad_norm": 121.68417066288369,
"learning_rate": 2.265992618104029e-08,
"logits/chosen": -15.883665084838867,
"logits/rejected": -15.859227180480957,
"logps/chosen": -2.1571857929229736,
"logps/rejected": -2.7032248973846436,
"loss": 2.7706,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -21.571857452392578,
"rewards/margins": 5.460390090942383,
"rewards/rejected": -27.03224754333496,
"step": 420
},
{
"epoch": 0.9129967776584318,
"grad_norm": 205.5451817035243,
"learning_rate": 1.7940055025900304e-08,
"logits/chosen": -14.086555480957031,
"logits/rejected": -13.882139205932617,
"logps/chosen": -2.0481374263763428,
"logps/rejected": -2.428682804107666,
"loss": 3.0577,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -20.481372833251953,
"rewards/margins": 3.8054566383361816,
"rewards/rejected": -24.28683090209961,
"step": 425
},
{
"epoch": 0.9237379162191193,
"grad_norm": 119.46095967048043,
"learning_rate": 1.3759654233514817e-08,
"logits/chosen": -14.88987922668457,
"logits/rejected": -14.742823600769043,
"logps/chosen": -1.941057562828064,
"logps/rejected": -2.4432146549224854,
"loss": 2.8075,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -19.41057586669922,
"rewards/margins": 5.021571636199951,
"rewards/rejected": -24.432147979736328,
"step": 430
},
{
"epoch": 0.9344790547798066,
"grad_norm": 112.52416047730715,
"learning_rate": 1.0124626548627402e-08,
"logits/chosen": -15.55200481414795,
"logits/rejected": -15.57677936553955,
"logps/chosen": -2.0663094520568848,
"logps/rejected": -2.723595380783081,
"loss": 2.7934,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -20.663097381591797,
"rewards/margins": 6.5728559494018555,
"rewards/rejected": -27.235950469970703,
"step": 435
},
{
"epoch": 0.9452201933404941,
"grad_norm": 95.43516446345191,
"learning_rate": 7.040104646698042e-09,
"logits/chosen": -14.149500846862793,
"logits/rejected": -14.121160507202148,
"logps/chosen": -2.1648197174072266,
"logps/rejected": -2.803515911102295,
"loss": 2.7258,
"rewards/accuracies": 0.75,
"rewards/chosen": -21.648197174072266,
"rewards/margins": 6.386962890625,
"rewards/rejected": -28.035160064697266,
"step": 440
},
{
"epoch": 0.9559613319011815,
"grad_norm": 105.32519506523218,
"learning_rate": 4.510443886542114e-09,
"logits/chosen": -15.509679794311523,
"logits/rejected": -15.587133407592773,
"logps/chosen": -2.019160509109497,
"logps/rejected": -2.5319721698760986,
"loss": 2.8015,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -20.191600799560547,
"rewards/margins": 5.128118991851807,
"rewards/rejected": -25.319721221923828,
"step": 445
},
{
"epoch": 0.966702470461869,
"grad_norm": 142.71634790119586,
"learning_rate": 2.539216160544333e-09,
"logits/chosen": -15.480878829956055,
"logits/rejected": -15.102048873901367,
"logps/chosen": -2.1352379322052,
"logps/rejected": -2.589895486831665,
"loss": 2.889,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -21.35237693786621,
"rewards/margins": 4.546576023101807,
"rewards/rejected": -25.898956298828125,
"step": 450
},
{
"epoch": 0.9774436090225563,
"grad_norm": 123.94313579799326,
"learning_rate": 1.1292048511303054e-09,
"logits/chosen": -14.889852523803711,
"logits/rejected": -15.104809761047363,
"logps/chosen": -1.9709218740463257,
"logps/rejected": -2.4727888107299805,
"loss": 2.9813,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -19.709218978881836,
"rewards/margins": 5.018665313720703,
"rewards/rejected": -24.727886199951172,
"step": 455
},
{
"epoch": 0.9881847475832438,
"grad_norm": 100.41632245481235,
"learning_rate": 2.82400900618418e-10,
"logits/chosen": -15.215599060058594,
"logits/rejected": -15.19567584991455,
"logps/chosen": -1.937787652015686,
"logps/rejected": -2.593761920928955,
"loss": 2.5524,
"rewards/accuracies": 0.8125,
"rewards/chosen": -19.377878189086914,
"rewards/margins": 6.559741020202637,
"rewards/rejected": -25.937618255615234,
"step": 460
},
{
"epoch": 0.9989258861439313,
"grad_norm": 139.5358860215804,
"learning_rate": 0.0,
"logits/chosen": -15.462666511535645,
"logits/rejected": -15.356382369995117,
"logps/chosen": -2.155303716659546,
"logps/rejected": -2.6153688430786133,
"loss": 2.5621,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -21.55303955078125,
"rewards/margins": 4.600649356842041,
"rewards/rejected": -26.1536865234375,
"step": 465
},
{
"epoch": 0.9989258861439313,
"step": 465,
"total_flos": 0.0,
"train_loss": 3.5518602760889197,
"train_runtime": 6148.2375,
"train_samples_per_second": 9.689,
"train_steps_per_second": 0.076
}
],
"logging_steps": 5,
"max_steps": 465,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 1000000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}