deepseek-8b-dpo-full2 / trainer_state.json
zfz1's picture
Model save
4ba1d38 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 10000,
"global_step": 2500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0004,
"grad_norm": 28.533812348552967,
"learning_rate": 2e-09,
"logits/chosen": 64.40786743164062,
"logits/rejected": 45.84376525878906,
"logps/chosen": -100.49942779541016,
"logps/rejected": -65.1117172241211,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.004,
"grad_norm": 30.859731079131965,
"learning_rate": 2e-08,
"logits/chosen": 57.91438674926758,
"logits/rejected": 56.28672790527344,
"logps/chosen": -55.96101760864258,
"logps/rejected": -68.71676635742188,
"loss": 0.6919,
"rewards/accuracies": 0.3888888955116272,
"rewards/chosen": -0.007306650280952454,
"rewards/margins": -0.014283686876296997,
"rewards/rejected": 0.006977038457989693,
"step": 10
},
{
"epoch": 0.008,
"grad_norm": 27.85742234963813,
"learning_rate": 4e-08,
"logits/chosen": 58.18536376953125,
"logits/rejected": 55.3450927734375,
"logps/chosen": -54.76381301879883,
"logps/rejected": -60.58549880981445,
"loss": 0.6953,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.011082855053246021,
"rewards/margins": 0.0007482476648874581,
"rewards/rejected": 0.010334606282413006,
"step": 20
},
{
"epoch": 0.012,
"grad_norm": 23.673728380473765,
"learning_rate": 6e-08,
"logits/chosen": 58.1522102355957,
"logits/rejected": 57.0158576965332,
"logps/chosen": -57.9251823425293,
"logps/rejected": -69.07908630371094,
"loss": 0.6949,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": 0.008253919892013073,
"rewards/margins": 0.016202187165617943,
"rewards/rejected": -0.007948270067572594,
"step": 30
},
{
"epoch": 0.016,
"grad_norm": 28.67344451856464,
"learning_rate": 8e-08,
"logits/chosen": 58.062538146972656,
"logits/rejected": 57.1534309387207,
"logps/chosen": -61.0693473815918,
"logps/rejected": -73.38080596923828,
"loss": 0.6963,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.019384900107979774,
"rewards/margins": 0.0025020535103976727,
"rewards/rejected": 0.016882847994565964,
"step": 40
},
{
"epoch": 0.02,
"grad_norm": 31.647118654717907,
"learning_rate": 1e-07,
"logits/chosen": 59.191986083984375,
"logits/rejected": 53.941322326660156,
"logps/chosen": -56.7202033996582,
"logps/rejected": -66.81363677978516,
"loss": 0.6931,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": 0.009077167138457298,
"rewards/margins": 0.010364162735641003,
"rewards/rejected": -0.0012869939673691988,
"step": 50
},
{
"epoch": 0.024,
"grad_norm": 24.669256574908005,
"learning_rate": 1.2e-07,
"logits/chosen": 56.26934051513672,
"logits/rejected": 56.80645751953125,
"logps/chosen": -58.25422286987305,
"logps/rejected": -62.55634689331055,
"loss": 0.6951,
"rewards/accuracies": 0.44999998807907104,
"rewards/chosen": -0.00721455505117774,
"rewards/margins": -0.0019339373102411628,
"rewards/rejected": -0.005280619021505117,
"step": 60
},
{
"epoch": 0.028,
"grad_norm": 26.74384143919655,
"learning_rate": 1.4e-07,
"logits/chosen": 58.523521423339844,
"logits/rejected": 55.622764587402344,
"logps/chosen": -54.381553649902344,
"logps/rejected": -60.544677734375,
"loss": 0.6959,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.024661114439368248,
"rewards/margins": -0.0074427248910069466,
"rewards/rejected": -0.017218390479683876,
"step": 70
},
{
"epoch": 0.032,
"grad_norm": 31.840440623804586,
"learning_rate": 1.6e-07,
"logits/chosen": 54.36933135986328,
"logits/rejected": 59.20940017700195,
"logps/chosen": -58.11821365356445,
"logps/rejected": -70.71514892578125,
"loss": 0.7036,
"rewards/accuracies": 0.4749999940395355,
"rewards/chosen": -0.01636091247200966,
"rewards/margins": -0.008592168800532818,
"rewards/rejected": -0.007768744137138128,
"step": 80
},
{
"epoch": 0.036,
"grad_norm": 34.438677762408695,
"learning_rate": 1.8e-07,
"logits/chosen": 57.025726318359375,
"logits/rejected": 58.043739318847656,
"logps/chosen": -54.304481506347656,
"logps/rejected": -68.1335678100586,
"loss": 0.6923,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.009257683530449867,
"rewards/margins": -0.007113204337656498,
"rewards/rejected": -0.002144479425624013,
"step": 90
},
{
"epoch": 0.04,
"grad_norm": 27.55087060204859,
"learning_rate": 2e-07,
"logits/chosen": 58.55786895751953,
"logits/rejected": 55.395263671875,
"logps/chosen": -63.9180793762207,
"logps/rejected": -65.21315002441406,
"loss": 0.6993,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.010401496663689613,
"rewards/margins": 0.014750251546502113,
"rewards/rejected": -0.025151750072836876,
"step": 100
},
{
"epoch": 0.044,
"grad_norm": 34.47770001073974,
"learning_rate": 2.1999999999999998e-07,
"logits/chosen": 56.813232421875,
"logits/rejected": 56.03766632080078,
"logps/chosen": -58.6301155090332,
"logps/rejected": -74.81730651855469,
"loss": 0.6944,
"rewards/accuracies": 0.4749999940395355,
"rewards/chosen": -0.02511642314493656,
"rewards/margins": 0.00451306626200676,
"rewards/rejected": -0.02962948940694332,
"step": 110
},
{
"epoch": 0.048,
"grad_norm": 30.953399050687445,
"learning_rate": 2.4e-07,
"logits/chosen": 57.8872184753418,
"logits/rejected": 57.22182083129883,
"logps/chosen": -59.11281204223633,
"logps/rejected": -70.58454895019531,
"loss": 0.6964,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.04789043590426445,
"rewards/margins": 0.009716681204736233,
"rewards/rejected": -0.057607125490903854,
"step": 120
},
{
"epoch": 0.052,
"grad_norm": 28.466689861082365,
"learning_rate": 2.6e-07,
"logits/chosen": 57.37153244018555,
"logits/rejected": 56.422645568847656,
"logps/chosen": -62.57783889770508,
"logps/rejected": -73.04331970214844,
"loss": 0.6871,
"rewards/accuracies": 0.4749999940395355,
"rewards/chosen": -0.023358307778835297,
"rewards/margins": 0.020040828734636307,
"rewards/rejected": -0.043399132788181305,
"step": 130
},
{
"epoch": 0.056,
"grad_norm": 28.47666983148757,
"learning_rate": 2.8e-07,
"logits/chosen": 56.3998908996582,
"logits/rejected": 56.23912811279297,
"logps/chosen": -54.915016174316406,
"logps/rejected": -70.63322448730469,
"loss": 0.6795,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.04837559536099434,
"rewards/margins": 0.03748052567243576,
"rewards/rejected": -0.0858561173081398,
"step": 140
},
{
"epoch": 0.06,
"grad_norm": 31.41163403514937,
"learning_rate": 3e-07,
"logits/chosen": 58.12103271484375,
"logits/rejected": 56.27421188354492,
"logps/chosen": -63.10313034057617,
"logps/rejected": -66.35491943359375,
"loss": 0.6908,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.08965396881103516,
"rewards/margins": -0.004495841450989246,
"rewards/rejected": -0.08515812456607819,
"step": 150
},
{
"epoch": 0.064,
"grad_norm": 26.840900506278157,
"learning_rate": 3.2e-07,
"logits/chosen": 55.839454650878906,
"logits/rejected": 58.34368133544922,
"logps/chosen": -50.59211349487305,
"logps/rejected": -73.50721740722656,
"loss": 0.6767,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.06941857188940048,
"rewards/margins": 0.04276493936777115,
"rewards/rejected": -0.11218351125717163,
"step": 160
},
{
"epoch": 0.068,
"grad_norm": 27.292091726963864,
"learning_rate": 3.4000000000000003e-07,
"logits/chosen": 57.70301055908203,
"logits/rejected": 56.25288009643555,
"logps/chosen": -63.054412841796875,
"logps/rejected": -68.32513427734375,
"loss": 0.6661,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.07662694901227951,
"rewards/margins": 0.05604536086320877,
"rewards/rejected": -0.13267230987548828,
"step": 170
},
{
"epoch": 0.072,
"grad_norm": 30.73299858141547,
"learning_rate": 3.6e-07,
"logits/chosen": 56.193214416503906,
"logits/rejected": 58.65944290161133,
"logps/chosen": -59.230201721191406,
"logps/rejected": -72.97864532470703,
"loss": 0.6541,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.11543898284435272,
"rewards/margins": 0.10617627948522568,
"rewards/rejected": -0.221615269780159,
"step": 180
},
{
"epoch": 0.076,
"grad_norm": 27.159221184778563,
"learning_rate": 3.7999999999999996e-07,
"logits/chosen": 57.91032791137695,
"logits/rejected": 55.99603271484375,
"logps/chosen": -59.06760787963867,
"logps/rejected": -74.1017837524414,
"loss": 0.642,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.12510164082050323,
"rewards/margins": 0.09499747306108475,
"rewards/rejected": -0.22009912133216858,
"step": 190
},
{
"epoch": 0.08,
"grad_norm": 27.221778714068638,
"learning_rate": 4e-07,
"logits/chosen": 58.486793518066406,
"logits/rejected": 55.91951370239258,
"logps/chosen": -58.47349166870117,
"logps/rejected": -62.717193603515625,
"loss": 0.6393,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.12836897373199463,
"rewards/margins": 0.1229625716805458,
"rewards/rejected": -0.25133153796195984,
"step": 200
},
{
"epoch": 0.084,
"grad_norm": 29.03653153648935,
"learning_rate": 4.1999999999999995e-07,
"logits/chosen": 57.81336212158203,
"logits/rejected": 58.15123748779297,
"logps/chosen": -61.33382034301758,
"logps/rejected": -74.703125,
"loss": 0.6376,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.23162539303302765,
"rewards/margins": 0.11328538507223129,
"rewards/rejected": -0.34491077065467834,
"step": 210
},
{
"epoch": 0.088,
"grad_norm": 26.95404534183593,
"learning_rate": 4.3999999999999997e-07,
"logits/chosen": 57.90010452270508,
"logits/rejected": 54.545555114746094,
"logps/chosen": -63.3861198425293,
"logps/rejected": -69.0276107788086,
"loss": 0.6285,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.22368836402893066,
"rewards/margins": 0.12568101286888123,
"rewards/rejected": -0.3493694067001343,
"step": 220
},
{
"epoch": 0.092,
"grad_norm": 26.883373858805115,
"learning_rate": 4.6e-07,
"logits/chosen": 56.64574432373047,
"logits/rejected": 55.44426345825195,
"logps/chosen": -54.02278518676758,
"logps/rejected": -70.68146514892578,
"loss": 0.608,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.3271670937538147,
"rewards/margins": 0.13781145215034485,
"rewards/rejected": -0.46497854590415955,
"step": 230
},
{
"epoch": 0.096,
"grad_norm": 26.103399727840255,
"learning_rate": 4.8e-07,
"logits/chosen": 56.06645965576172,
"logits/rejected": 53.25556182861328,
"logps/chosen": -59.9603271484375,
"logps/rejected": -73.2823257446289,
"loss": 0.6089,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.3069990277290344,
"rewards/margins": 0.23678632080554962,
"rewards/rejected": -0.5437853336334229,
"step": 240
},
{
"epoch": 0.1,
"grad_norm": 23.722260965256044,
"learning_rate": 5e-07,
"logits/chosen": 56.92375946044922,
"logits/rejected": 55.26447677612305,
"logps/chosen": -69.1079330444336,
"logps/rejected": -73.78802490234375,
"loss": 0.6297,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.4947397708892822,
"rewards/margins": 0.14389055967330933,
"rewards/rejected": -0.6386303305625916,
"step": 250
},
{
"epoch": 0.104,
"grad_norm": 25.422708939657227,
"learning_rate": 4.99975631002326e-07,
"logits/chosen": 54.239952087402344,
"logits/rejected": 54.537017822265625,
"logps/chosen": -58.73674392700195,
"logps/rejected": -69.70175170898438,
"loss": 0.5727,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.4754973351955414,
"rewards/margins": 0.28132364153862,
"rewards/rejected": -0.7568209767341614,
"step": 260
},
{
"epoch": 0.108,
"grad_norm": 26.17681951761786,
"learning_rate": 4.999025287600885e-07,
"logits/chosen": 56.4760627746582,
"logits/rejected": 54.777679443359375,
"logps/chosen": -68.2262954711914,
"logps/rejected": -77.97975158691406,
"loss": 0.558,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.5877800583839417,
"rewards/margins": 0.398207426071167,
"rewards/rejected": -0.9859874844551086,
"step": 270
},
{
"epoch": 0.112,
"grad_norm": 20.299221772759086,
"learning_rate": 4.997807075247145e-07,
"logits/chosen": 55.838172912597656,
"logits/rejected": 55.217002868652344,
"logps/chosen": -58.478309631347656,
"logps/rejected": -74.65936279296875,
"loss": 0.4985,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.5783756971359253,
"rewards/margins": 0.5172783136367798,
"rewards/rejected": -1.095654010772705,
"step": 280
},
{
"epoch": 0.116,
"grad_norm": 28.36770834985096,
"learning_rate": 4.996101910454953e-07,
"logits/chosen": 56.4325065612793,
"logits/rejected": 54.16033172607422,
"logps/chosen": -69.69315338134766,
"logps/rejected": -79.90055847167969,
"loss": 0.5267,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.7261027097702026,
"rewards/margins": 0.5981040596961975,
"rewards/rejected": -1.324206829071045,
"step": 290
},
{
"epoch": 0.12,
"grad_norm": 25.02377762418046,
"learning_rate": 4.99391012564956e-07,
"logits/chosen": 53.800933837890625,
"logits/rejected": 55.656532287597656,
"logps/chosen": -66.37216186523438,
"logps/rejected": -82.07975769042969,
"loss": 0.4772,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.7666959762573242,
"rewards/margins": 0.5859344005584717,
"rewards/rejected": -1.352630376815796,
"step": 300
},
{
"epoch": 0.124,
"grad_norm": 29.446782373035884,
"learning_rate": 4.991232148123761e-07,
"logits/chosen": 57.31342697143555,
"logits/rejected": 54.00205612182617,
"logps/chosen": -72.55195617675781,
"logps/rejected": -85.0060806274414,
"loss": 0.4842,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -1.089627742767334,
"rewards/margins": 0.5522868633270264,
"rewards/rejected": -1.64191472530365,
"step": 310
},
{
"epoch": 0.128,
"grad_norm": 31.47681515659833,
"learning_rate": 4.988068499954577e-07,
"logits/chosen": 55.69438934326172,
"logits/rejected": 54.8391227722168,
"logps/chosen": -70.90913391113281,
"logps/rejected": -82.49961853027344,
"loss": 0.5177,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.8803223371505737,
"rewards/margins": 0.687156081199646,
"rewards/rejected": -1.5674786567687988,
"step": 320
},
{
"epoch": 0.132,
"grad_norm": 29.597337732185277,
"learning_rate": 4.984419797901491e-07,
"logits/chosen": 53.4282112121582,
"logits/rejected": 54.79206466674805,
"logps/chosen": -69.86516571044922,
"logps/rejected": -90.27011108398438,
"loss": 0.4469,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -0.9048943519592285,
"rewards/margins": 1.1756442785263062,
"rewards/rejected": -2.080538511276245,
"step": 330
},
{
"epoch": 0.136,
"grad_norm": 17.009359229526957,
"learning_rate": 4.980286753286194e-07,
"logits/chosen": 52.42332077026367,
"logits/rejected": 57.04686737060547,
"logps/chosen": -64.6177749633789,
"logps/rejected": -89.59310150146484,
"loss": 0.4411,
"rewards/accuracies": 0.875,
"rewards/chosen": -1.1452090740203857,
"rewards/margins": 0.9245441555976868,
"rewards/rejected": -2.0697531700134277,
"step": 340
},
{
"epoch": 0.14,
"grad_norm": 25.909018906834667,
"learning_rate": 4.975670171853925e-07,
"logits/chosen": 56.42598342895508,
"logits/rejected": 53.42926025390625,
"logps/chosen": -69.07212829589844,
"logps/rejected": -84.19108581542969,
"loss": 0.4175,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -1.2340478897094727,
"rewards/margins": 1.0201036930084229,
"rewards/rejected": -2.2541518211364746,
"step": 350
},
{
"epoch": 0.144,
"grad_norm": 27.75848608177815,
"learning_rate": 4.970570953616382e-07,
"logits/chosen": 53.7008056640625,
"logits/rejected": 55.94395065307617,
"logps/chosen": -66.99665832519531,
"logps/rejected": -87.76789855957031,
"loss": 0.4435,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -1.3015058040618896,
"rewards/margins": 0.9406806826591492,
"rewards/rejected": -2.2421865463256836,
"step": 360
},
{
"epoch": 0.148,
"grad_norm": 24.273863705867576,
"learning_rate": 4.964990092676262e-07,
"logits/chosen": 49.97655487060547,
"logits/rejected": 58.69891357421875,
"logps/chosen": -68.9350814819336,
"logps/rejected": -98.9703369140625,
"loss": 0.3817,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -1.4265937805175781,
"rewards/margins": 1.1566370725631714,
"rewards/rejected": -2.583230495452881,
"step": 370
},
{
"epoch": 0.152,
"grad_norm": 37.00653796376565,
"learning_rate": 4.958928677033465e-07,
"logits/chosen": 55.870933532714844,
"logits/rejected": 54.31379318237305,
"logps/chosen": -77.0854263305664,
"logps/rejected": -90.93013000488281,
"loss": 0.4856,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -1.7595773935317993,
"rewards/margins": 0.8833327293395996,
"rewards/rejected": -2.6429102420806885,
"step": 380
},
{
"epoch": 0.156,
"grad_norm": 28.256248306216513,
"learning_rate": 4.952387888372978e-07,
"logits/chosen": 53.375587463378906,
"logits/rejected": 52.61153030395508,
"logps/chosen": -78.0414810180664,
"logps/rejected": -98.44496154785156,
"loss": 0.38,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -1.7100744247436523,
"rewards/margins": 1.1639991998672485,
"rewards/rejected": -2.8740735054016113,
"step": 390
},
{
"epoch": 0.16,
"grad_norm": 29.2072173920723,
"learning_rate": 4.945369001834514e-07,
"logits/chosen": 51.90997314453125,
"logits/rejected": 53.769920349121094,
"logps/chosen": -66.5736083984375,
"logps/rejected": -94.8015365600586,
"loss": 0.4308,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -1.6438815593719482,
"rewards/margins": 1.1793160438537598,
"rewards/rejected": -2.823197841644287,
"step": 400
},
{
"epoch": 0.164,
"grad_norm": 20.246428107882906,
"learning_rate": 4.937873385763907e-07,
"logits/chosen": 51.662147521972656,
"logits/rejected": 52.028175354003906,
"logps/chosen": -79.66545867919922,
"logps/rejected": -107.46966552734375,
"loss": 0.394,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -1.8286956548690796,
"rewards/margins": 1.288888931274414,
"rewards/rejected": -3.117584705352783,
"step": 410
},
{
"epoch": 0.168,
"grad_norm": 27.307865717439135,
"learning_rate": 4.929902501446366e-07,
"logits/chosen": 50.78424835205078,
"logits/rejected": 50.39312744140625,
"logps/chosen": -75.19126892089844,
"logps/rejected": -95.62612915039062,
"loss": 0.4315,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -2.0071020126342773,
"rewards/margins": 1.3031747341156006,
"rewards/rejected": -3.310276746749878,
"step": 420
},
{
"epoch": 0.172,
"grad_norm": 20.03898118642927,
"learning_rate": 4.921457902821578e-07,
"logits/chosen": 53.468177795410156,
"logits/rejected": 53.30914306640625,
"logps/chosen": -82.42778015136719,
"logps/rejected": -109.83222961425781,
"loss": 0.3376,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -1.6454699039459229,
"rewards/margins": 1.615443468093872,
"rewards/rejected": -3.260913133621216,
"step": 430
},
{
"epoch": 0.176,
"grad_norm": 30.325957397734495,
"learning_rate": 4.912541236180778e-07,
"logits/chosen": 51.10878372192383,
"logits/rejected": 53.10695266723633,
"logps/chosen": -74.79537200927734,
"logps/rejected": -100.06138610839844,
"loss": 0.3929,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -1.8177331686019897,
"rewards/margins": 1.257211446762085,
"rewards/rejected": -3.0749447345733643,
"step": 440
},
{
"epoch": 0.18,
"grad_norm": 14.262220601317786,
"learning_rate": 4.903154239845797e-07,
"logits/chosen": 51.69508743286133,
"logits/rejected": 52.68989181518555,
"logps/chosen": -68.33168029785156,
"logps/rejected": -107.64765930175781,
"loss": 0.3691,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -1.747357726097107,
"rewards/margins": 1.74077570438385,
"rewards/rejected": -3.488133192062378,
"step": 450
},
{
"epoch": 0.184,
"grad_norm": 22.979886521568336,
"learning_rate": 4.893298743830167e-07,
"logits/chosen": 52.4492301940918,
"logits/rejected": 53.06267547607422,
"logps/chosen": -81.17327117919922,
"logps/rejected": -111.0452651977539,
"loss": 0.3741,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -2.073838710784912,
"rewards/margins": 1.7042049169540405,
"rewards/rejected": -3.778043031692505,
"step": 460
},
{
"epoch": 0.188,
"grad_norm": 14.722797570982818,
"learning_rate": 4.882976669482367e-07,
"logits/chosen": 50.58995819091797,
"logits/rejected": 51.509681701660156,
"logps/chosen": -79.2942123413086,
"logps/rejected": -102.22583770751953,
"loss": 0.3359,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -1.9251492023468018,
"rewards/margins": 1.7899284362792969,
"rewards/rejected": -3.7150776386260986,
"step": 470
},
{
"epoch": 0.192,
"grad_norm": 14.245338009793953,
"learning_rate": 4.872190029111241e-07,
"logits/chosen": 50.895538330078125,
"logits/rejected": 54.43767166137695,
"logps/chosen": -79.74618530273438,
"logps/rejected": -120.28243255615234,
"loss": 0.2751,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -2.2277004718780518,
"rewards/margins": 1.9976272583007812,
"rewards/rejected": -4.225327491760254,
"step": 480
},
{
"epoch": 0.196,
"grad_norm": 21.050016087172274,
"learning_rate": 4.860940925593702e-07,
"logits/chosen": 48.723018646240234,
"logits/rejected": 50.939937591552734,
"logps/chosen": -77.64906311035156,
"logps/rejected": -111.71051025390625,
"loss": 0.2957,
"rewards/accuracies": 0.875,
"rewards/chosen": -2.465684652328491,
"rewards/margins": 2.0183236598968506,
"rewards/rejected": -4.484008312225342,
"step": 490
},
{
"epoch": 0.2,
"grad_norm": 50.320855504854826,
"learning_rate": 4.849231551964771e-07,
"logits/chosen": 51.236595153808594,
"logits/rejected": 50.26428985595703,
"logps/chosen": -92.86045837402344,
"logps/rejected": -112.11564636230469,
"loss": 0.3985,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -2.6628010272979736,
"rewards/margins": 1.7543842792510986,
"rewards/rejected": -4.417185306549072,
"step": 500
},
{
"epoch": 0.204,
"grad_norm": 14.73313528703832,
"learning_rate": 4.837064190990036e-07,
"logits/chosen": 55.05743408203125,
"logits/rejected": 50.69791793823242,
"logps/chosen": -98.44610595703125,
"logps/rejected": -122.69828796386719,
"loss": 0.3647,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -2.5547454357147217,
"rewards/margins": 2.1788697242736816,
"rewards/rejected": -4.733614921569824,
"step": 510
},
{
"epoch": 0.208,
"grad_norm": 27.57905277549734,
"learning_rate": 4.824441214720628e-07,
"logits/chosen": 50.13386535644531,
"logits/rejected": 50.777809143066406,
"logps/chosen": -86.73146057128906,
"logps/rejected": -112.5364990234375,
"loss": 0.3732,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -2.320972442626953,
"rewards/margins": 2.0467121601104736,
"rewards/rejected": -4.367684364318848,
"step": 520
},
{
"epoch": 0.212,
"grad_norm": 25.097188193390338,
"learning_rate": 4.811365084030783e-07,
"logits/chosen": 49.33405685424805,
"logits/rejected": 50.48732376098633,
"logps/chosen": -75.11846160888672,
"logps/rejected": -104.57139587402344,
"loss": 0.3868,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -2.4332354068756104,
"rewards/margins": 1.9999496936798096,
"rewards/rejected": -4.433184623718262,
"step": 530
},
{
"epoch": 0.216,
"grad_norm": 32.7903194698391,
"learning_rate": 4.797838348138086e-07,
"logits/chosen": 50.36452102661133,
"logits/rejected": 50.02265167236328,
"logps/chosen": -90.46897888183594,
"logps/rejected": -112.5868911743164,
"loss": 0.3413,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -2.6965606212615967,
"rewards/margins": 1.6849288940429688,
"rewards/rejected": -4.381489276885986,
"step": 540
},
{
"epoch": 0.22,
"grad_norm": 24.213068673102388,
"learning_rate": 4.783863644106502e-07,
"logits/chosen": 52.17128372192383,
"logits/rejected": 50.315284729003906,
"logps/chosen": -93.4994125366211,
"logps/rejected": -119.5452880859375,
"loss": 0.3351,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -2.50398325920105,
"rewards/margins": 2.50455904006958,
"rewards/rejected": -5.008542537689209,
"step": 550
},
{
"epoch": 0.224,
"grad_norm": 29.49589281477644,
"learning_rate": 4.769443696332272e-07,
"logits/chosen": 47.702674865722656,
"logits/rejected": 50.343238830566406,
"logps/chosen": -83.9803695678711,
"logps/rejected": -123.3960189819336,
"loss": 0.3181,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -2.564993381500244,
"rewards/margins": 2.2127933502197266,
"rewards/rejected": -4.7777862548828125,
"step": 560
},
{
"epoch": 0.228,
"grad_norm": 19.69678290974152,
"learning_rate": 4.7545813160127845e-07,
"logits/chosen": 50.214874267578125,
"logits/rejected": 48.206787109375,
"logps/chosen": -90.08515930175781,
"logps/rejected": -118.11824798583984,
"loss": 0.4009,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -2.8223087787628174,
"rewards/margins": 2.011645555496216,
"rewards/rejected": -4.833954811096191,
"step": 570
},
{
"epoch": 0.232,
"grad_norm": 13.246157822107891,
"learning_rate": 4.739279400598532e-07,
"logits/chosen": 51.175697326660156,
"logits/rejected": 48.75572967529297,
"logps/chosen": -82.0949935913086,
"logps/rejected": -111.55613708496094,
"loss": 0.3156,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -2.699707269668579,
"rewards/margins": 1.895646333694458,
"rewards/rejected": -4.595353126525879,
"step": 580
},
{
"epoch": 0.236,
"grad_norm": 22.333620027716158,
"learning_rate": 4.7235409332282436e-07,
"logits/chosen": 49.49880599975586,
"logits/rejected": 51.62006759643555,
"logps/chosen": -73.26454162597656,
"logps/rejected": -111.3040771484375,
"loss": 0.3451,
"rewards/accuracies": 0.875,
"rewards/chosen": -2.179260730743408,
"rewards/margins": 2.1387388706207275,
"rewards/rejected": -4.317999839782715,
"step": 590
},
{
"epoch": 0.24,
"grad_norm": 26.7757211390623,
"learning_rate": 4.707368982147317e-07,
"logits/chosen": 47.9814567565918,
"logits/rejected": 51.40210723876953,
"logps/chosen": -91.81172180175781,
"logps/rejected": -128.8383331298828,
"loss": 0.2713,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -2.9874844551086426,
"rewards/margins": 2.1201720237731934,
"rewards/rejected": -5.107656955718994,
"step": 600
},
{
"epoch": 0.244,
"grad_norm": 40.79432104607236,
"learning_rate": 4.6907667001096585e-07,
"logits/chosen": 48.35408401489258,
"logits/rejected": 47.354530334472656,
"logps/chosen": -90.21234893798828,
"logps/rejected": -118.80155944824219,
"loss": 0.2689,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -3.0216469764709473,
"rewards/margins": 2.228628635406494,
"rewards/rejected": -5.250275611877441,
"step": 610
},
{
"epoch": 0.248,
"grad_norm": 31.8455689759253,
"learning_rate": 4.6737373237630473e-07,
"logits/chosen": 48.34640121459961,
"logits/rejected": 51.8663444519043,
"logps/chosen": -85.09153747558594,
"logps/rejected": -127.3193130493164,
"loss": 0.3664,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -2.7675275802612305,
"rewards/margins": 2.4404561519622803,
"rewards/rejected": -5.207983493804932,
"step": 620
},
{
"epoch": 0.252,
"grad_norm": 19.397624289021156,
"learning_rate": 4.6562841730181435e-07,
"logits/chosen": 48.62126159667969,
"logits/rejected": 48.2672119140625,
"logps/chosen": -90.38846588134766,
"logps/rejected": -122.53373718261719,
"loss": 0.3227,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -3.1435444355010986,
"rewards/margins": 2.2223925590515137,
"rewards/rejected": -5.365937232971191,
"step": 630
},
{
"epoch": 0.256,
"grad_norm": 34.367583617826874,
"learning_rate": 4.6384106504012665e-07,
"logits/chosen": 48.072669982910156,
"logits/rejected": 49.07495880126953,
"logps/chosen": -93.60935974121094,
"logps/rejected": -128.94163513183594,
"loss": 0.2825,
"rewards/accuracies": 0.875,
"rewards/chosen": -3.1240580081939697,
"rewards/margins": 2.519449472427368,
"rewards/rejected": -5.64350700378418,
"step": 640
},
{
"epoch": 0.26,
"grad_norm": 27.938891622888274,
"learning_rate": 4.6201202403910643e-07,
"logits/chosen": 45.11116027832031,
"logits/rejected": 51.54563522338867,
"logps/chosen": -76.74453735351562,
"logps/rejected": -129.6508331298828,
"loss": 0.2858,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -2.9022960662841797,
"rewards/margins": 2.7343924045562744,
"rewards/rejected": -5.636688709259033,
"step": 650
},
{
"epoch": 0.264,
"grad_norm": 33.20013982791981,
"learning_rate": 4.6014165087392105e-07,
"logits/chosen": 47.67237854003906,
"logits/rejected": 46.87964630126953,
"logps/chosen": -87.0808334350586,
"logps/rejected": -122.21199035644531,
"loss": 0.2727,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -2.771348476409912,
"rewards/margins": 2.724174976348877,
"rewards/rejected": -5.495523452758789,
"step": 660
},
{
"epoch": 0.268,
"grad_norm": 29.256933473551793,
"learning_rate": 4.582303101775248e-07,
"logits/chosen": 49.69163513183594,
"logits/rejected": 45.95726013183594,
"logps/chosen": -93.39743041992188,
"logps/rejected": -120.0958023071289,
"loss": 0.2879,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -2.911816120147705,
"rewards/margins": 2.617412805557251,
"rewards/rejected": -5.529229164123535,
"step": 670
},
{
"epoch": 0.272,
"grad_norm": 43.80379099330064,
"learning_rate": 4.5627837456957374e-07,
"logits/chosen": 48.20374298095703,
"logits/rejected": 47.810935974121094,
"logps/chosen": -94.90364074707031,
"logps/rejected": -125.94508361816406,
"loss": 0.3458,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -3.2113685607910156,
"rewards/margins": 2.2549712657928467,
"rewards/rejected": -5.466340065002441,
"step": 680
},
{
"epoch": 0.276,
"grad_norm": 38.68916019817628,
"learning_rate": 4.542862245837821e-07,
"logits/chosen": 48.5870246887207,
"logits/rejected": 47.81504440307617,
"logps/chosen": -102.28499603271484,
"logps/rejected": -129.62196350097656,
"loss": 0.2492,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -3.5993125438690186,
"rewards/margins": 2.266263484954834,
"rewards/rejected": -5.865576267242432,
"step": 690
},
{
"epoch": 0.28,
"grad_norm": 21.578470273867236,
"learning_rate": 4.5225424859373684e-07,
"logits/chosen": 49.09315490722656,
"logits/rejected": 48.342491149902344,
"logps/chosen": -98.66539001464844,
"logps/rejected": -139.3084259033203,
"loss": 0.215,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -3.2259106636047363,
"rewards/margins": 3.0345213413238525,
"rewards/rejected": -6.260432243347168,
"step": 700
},
{
"epoch": 0.284,
"grad_norm": 16.808420798200792,
"learning_rate": 4.501828427371833e-07,
"logits/chosen": 46.62156295776367,
"logits/rejected": 48.56031036376953,
"logps/chosen": -92.45174407958984,
"logps/rejected": -137.70553588867188,
"loss": 0.3082,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -3.2984604835510254,
"rewards/margins": 2.9785618782043457,
"rewards/rejected": -6.277022361755371,
"step": 710
},
{
"epoch": 0.288,
"grad_norm": 25.497403458788426,
"learning_rate": 4.4807241083879764e-07,
"logits/chosen": 44.841453552246094,
"logits/rejected": 48.752197265625,
"logps/chosen": -88.73442077636719,
"logps/rejected": -139.46563720703125,
"loss": 0.2662,
"rewards/accuracies": 0.875,
"rewards/chosen": -3.5202784538269043,
"rewards/margins": 2.9266552925109863,
"rewards/rejected": -6.446933746337891,
"step": 720
},
{
"epoch": 0.292,
"grad_norm": 26.661142376929647,
"learning_rate": 4.459233643314599e-07,
"logits/chosen": 48.446815490722656,
"logits/rejected": 48.538719177246094,
"logps/chosen": -84.2086410522461,
"logps/rejected": -123.36454772949219,
"loss": 0.2538,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -2.9666292667388916,
"rewards/margins": 2.7437281608581543,
"rewards/rejected": -5.710357189178467,
"step": 730
},
{
"epoch": 0.296,
"grad_norm": 8.139706838501583,
"learning_rate": 4.437361221760449e-07,
"logits/chosen": 48.00127410888672,
"logits/rejected": 48.78575897216797,
"logps/chosen": -96.60485076904297,
"logps/rejected": -124.87052917480469,
"loss": 0.2151,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -3.223308563232422,
"rewards/margins": 2.8362784385681152,
"rewards/rejected": -6.059587478637695,
"step": 740
},
{
"epoch": 0.3,
"grad_norm": 34.66768947372861,
"learning_rate": 4.415111107797445e-07,
"logits/chosen": 46.998199462890625,
"logits/rejected": 46.860145568847656,
"logps/chosen": -99.70954895019531,
"logps/rejected": -132.5911407470703,
"loss": 0.3284,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -3.904909133911133,
"rewards/margins": 2.286958932876587,
"rewards/rejected": -6.191867828369141,
"step": 750
},
{
"epoch": 0.304,
"grad_norm": 27.280956863429665,
"learning_rate": 4.392487639129391e-07,
"logits/chosen": 44.16685485839844,
"logits/rejected": 46.52399826049805,
"logps/chosen": -89.0407485961914,
"logps/rejected": -125.99409484863281,
"loss": 0.2755,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -3.4610557556152344,
"rewards/margins": 2.747231960296631,
"rewards/rejected": -6.208288669586182,
"step": 760
},
{
"epoch": 0.308,
"grad_norm": 45.42173052837384,
"learning_rate": 4.36949522624633e-07,
"logits/chosen": 46.8876838684082,
"logits/rejected": 45.308528900146484,
"logps/chosen": -98.10558319091797,
"logps/rejected": -131.274658203125,
"loss": 0.2579,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -3.8448243141174316,
"rewards/margins": 2.463934898376465,
"rewards/rejected": -6.3087592124938965,
"step": 770
},
{
"epoch": 0.312,
"grad_norm": 44.27869354906801,
"learning_rate": 4.34613835156471e-07,
"logits/chosen": 46.87074279785156,
"logits/rejected": 45.099308013916016,
"logps/chosen": -111.92747497558594,
"logps/rejected": -131.96014404296875,
"loss": 0.3277,
"rewards/accuracies": 0.75,
"rewards/chosen": -4.581133842468262,
"rewards/margins": 2.0176618099212646,
"rewards/rejected": -6.5987958908081055,
"step": 780
},
{
"epoch": 0.316,
"grad_norm": 22.111566033012554,
"learning_rate": 4.3224215685535287e-07,
"logits/chosen": 45.65962219238281,
"logits/rejected": 45.8907585144043,
"logps/chosen": -99.42485809326172,
"logps/rejected": -141.40390014648438,
"loss": 0.2526,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -4.320735454559326,
"rewards/margins": 2.8679370880126953,
"rewards/rejected": -7.188672065734863,
"step": 790
},
{
"epoch": 0.32,
"grad_norm": 52.35572333396793,
"learning_rate": 4.2983495008466273e-07,
"logits/chosen": 44.55757141113281,
"logits/rejected": 47.1688232421875,
"logps/chosen": -100.08128356933594,
"logps/rejected": -150.44851684570312,
"loss": 0.256,
"rewards/accuracies": 0.875,
"rewards/chosen": -4.098430633544922,
"rewards/margins": 3.171543598175049,
"rewards/rejected": -7.269974708557129,
"step": 800
},
{
"epoch": 0.324,
"grad_norm": 33.21188816300218,
"learning_rate": 4.273926841341302e-07,
"logits/chosen": 46.28192138671875,
"logits/rejected": 43.64000701904297,
"logps/chosen": -99.21965789794922,
"logps/rejected": -129.105712890625,
"loss": 0.2544,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -3.7107014656066895,
"rewards/margins": 3.0634143352508545,
"rewards/rejected": -6.774115085601807,
"step": 810
},
{
"epoch": 0.328,
"grad_norm": 12.094382688192468,
"learning_rate": 4.249158351283413e-07,
"logits/chosen": 44.60408020019531,
"logits/rejected": 46.49042510986328,
"logps/chosen": -90.05680847167969,
"logps/rejected": -142.6536102294922,
"loss": 0.1868,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -3.59674072265625,
"rewards/margins": 3.3266303539276123,
"rewards/rejected": -6.923371315002441,
"step": 820
},
{
"epoch": 0.332,
"grad_norm": 4.00271990880729,
"learning_rate": 4.224048859339174e-07,
"logits/chosen": 45.333580017089844,
"logits/rejected": 45.82084274291992,
"logps/chosen": -102.02961730957031,
"logps/rejected": -137.76678466796875,
"loss": 0.2172,
"rewards/accuracies": 0.875,
"rewards/chosen": -4.266281604766846,
"rewards/margins": 2.6827759742736816,
"rewards/rejected": -6.949057102203369,
"step": 830
},
{
"epoch": 0.336,
"grad_norm": 20.780824485200213,
"learning_rate": 4.1986032606537916e-07,
"logits/chosen": 46.636512756347656,
"logits/rejected": 44.504981994628906,
"logps/chosen": -97.64836120605469,
"logps/rejected": -135.35731506347656,
"loss": 0.2624,
"rewards/accuracies": 0.875,
"rewards/chosen": -3.672236680984497,
"rewards/margins": 3.0585567951202393,
"rewards/rejected": -6.730792999267578,
"step": 840
},
{
"epoch": 0.34,
"grad_norm": 42.59198624354184,
"learning_rate": 4.172826515897145e-07,
"logits/chosen": 43.75339126586914,
"logits/rejected": 47.46000289916992,
"logps/chosen": -104.68599700927734,
"logps/rejected": -141.9625701904297,
"loss": 0.2583,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -4.268365859985352,
"rewards/margins": 2.5187840461730957,
"rewards/rejected": -6.787149906158447,
"step": 850
},
{
"epoch": 0.344,
"grad_norm": 20.008127683146643,
"learning_rate": 4.146723650296701e-07,
"logits/chosen": 43.556148529052734,
"logits/rejected": 43.975608825683594,
"logps/chosen": -95.18064880371094,
"logps/rejected": -139.5054931640625,
"loss": 0.3563,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -3.966346025466919,
"rewards/margins": 2.8288350105285645,
"rewards/rejected": -6.795180320739746,
"step": 860
},
{
"epoch": 0.348,
"grad_norm": 20.334985858976758,
"learning_rate": 4.120299752657827e-07,
"logits/chosen": 43.38778305053711,
"logits/rejected": 45.5791015625,
"logps/chosen": -98.27373504638672,
"logps/rejected": -140.9309539794922,
"loss": 0.1811,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -4.264477252960205,
"rewards/margins": 3.044534206390381,
"rewards/rejected": -7.309010982513428,
"step": 870
},
{
"epoch": 0.352,
"grad_norm": 28.757062776427272,
"learning_rate": 4.0935599743717244e-07,
"logits/chosen": 44.6265983581543,
"logits/rejected": 44.702178955078125,
"logps/chosen": -97.5745620727539,
"logps/rejected": -126.3858413696289,
"loss": 0.2602,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -4.221047401428223,
"rewards/margins": 2.2703094482421875,
"rewards/rejected": -6.49135684967041,
"step": 880
},
{
"epoch": 0.356,
"grad_norm": 14.773459336240398,
"learning_rate": 4.066509528411151e-07,
"logits/chosen": 45.27838897705078,
"logits/rejected": 43.685638427734375,
"logps/chosen": -95.4419937133789,
"logps/rejected": -131.39169311523438,
"loss": 0.198,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -3.907078981399536,
"rewards/margins": 3.0583863258361816,
"rewards/rejected": -6.9654645919799805,
"step": 890
},
{
"epoch": 0.36,
"grad_norm": 52.381278515402784,
"learning_rate": 4.039153688314145e-07,
"logits/chosen": 41.99245834350586,
"logits/rejected": 46.48981857299805,
"logps/chosen": -99.83820343017578,
"logps/rejected": -151.70034790039062,
"loss": 0.2602,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -4.080926418304443,
"rewards/margins": 3.463571071624756,
"rewards/rejected": -7.544497489929199,
"step": 900
},
{
"epoch": 0.364,
"grad_norm": 18.192241188575945,
"learning_rate": 4.0114977871559377e-07,
"logits/chosen": 42.960872650146484,
"logits/rejected": 43.536293029785156,
"logps/chosen": -89.50424194335938,
"logps/rejected": -141.3973388671875,
"loss": 0.1604,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -3.7081432342529297,
"rewards/margins": 3.6759707927703857,
"rewards/rejected": -7.3841142654418945,
"step": 910
},
{
"epoch": 0.368,
"grad_norm": 40.31098563989808,
"learning_rate": 3.983547216509254e-07,
"logits/chosen": 42.86689758300781,
"logits/rejected": 45.748714447021484,
"logps/chosen": -103.40921783447266,
"logps/rejected": -158.37911987304688,
"loss": 0.2078,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -4.184783458709717,
"rewards/margins": 3.848146438598633,
"rewards/rejected": -8.032930374145508,
"step": 920
},
{
"epoch": 0.372,
"grad_norm": 10.07444543448383,
"learning_rate": 3.9553074253932233e-07,
"logits/chosen": 42.905662536621094,
"logits/rejected": 41.272789001464844,
"logps/chosen": -100.79463195800781,
"logps/rejected": -128.82476806640625,
"loss": 0.2664,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -4.691293239593506,
"rewards/margins": 2.189171314239502,
"rewards/rejected": -6.880465507507324,
"step": 930
},
{
"epoch": 0.376,
"grad_norm": 20.93691989678441,
"learning_rate": 3.9267839192110797e-07,
"logits/chosen": 43.836544036865234,
"logits/rejected": 42.157772064208984,
"logps/chosen": -106.8996810913086,
"logps/rejected": -141.3354034423828,
"loss": 0.2145,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -4.653143405914307,
"rewards/margins": 3.2738826274871826,
"rewards/rejected": -7.927026271820068,
"step": 940
},
{
"epoch": 0.38,
"grad_norm": 31.706598510789135,
"learning_rate": 3.8979822586768666e-07,
"logits/chosen": 41.932212829589844,
"logits/rejected": 42.14612579345703,
"logps/chosen": -99.3321304321289,
"logps/rejected": -132.33554077148438,
"loss": 0.2512,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -4.518923282623291,
"rewards/margins": 2.864084243774414,
"rewards/rejected": -7.383008003234863,
"step": 950
},
{
"epoch": 0.384,
"grad_norm": 37.02742534141979,
"learning_rate": 3.8689080587313755e-07,
"logits/chosen": 40.756717681884766,
"logits/rejected": 43.150856018066406,
"logps/chosen": -94.36783599853516,
"logps/rejected": -147.3682403564453,
"loss": 0.2317,
"rewards/accuracies": 0.875,
"rewards/chosen": -4.52658748626709,
"rewards/margins": 3.43182373046875,
"rewards/rejected": -7.95841121673584,
"step": 960
},
{
"epoch": 0.388,
"grad_norm": 46.22429207605269,
"learning_rate": 3.839566987447491e-07,
"logits/chosen": 40.86772918701172,
"logits/rejected": 43.2618408203125,
"logps/chosen": -96.85578918457031,
"logps/rejected": -158.2089080810547,
"loss": 0.2296,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -4.181971073150635,
"rewards/margins": 3.6908957958221436,
"rewards/rejected": -7.872866630554199,
"step": 970
},
{
"epoch": 0.392,
"grad_norm": 53.76947098726966,
"learning_rate": 3.809964764925198e-07,
"logits/chosen": 43.136131286621094,
"logits/rejected": 43.18577194213867,
"logps/chosen": -109.56819915771484,
"logps/rejected": -163.19302368164062,
"loss": 0.197,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -4.513330459594727,
"rewards/margins": 4.060441493988037,
"rewards/rejected": -8.573771476745605,
"step": 980
},
{
"epoch": 0.396,
"grad_norm": 32.036494164799386,
"learning_rate": 3.780107162176429e-07,
"logits/chosen": 42.74140167236328,
"logits/rejected": 44.60755157470703,
"logps/chosen": -110.55721282958984,
"logps/rejected": -153.0295867919922,
"loss": 0.2728,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -4.982661247253418,
"rewards/margins": 2.932508945465088,
"rewards/rejected": -7.915169715881348,
"step": 990
},
{
"epoch": 0.4,
"grad_norm": 52.99187462304637,
"learning_rate": 3.75e-07,
"logits/chosen": 40.694297790527344,
"logits/rejected": 43.3404426574707,
"logps/chosen": -113.76619720458984,
"logps/rejected": -155.80531311035156,
"loss": 0.2724,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -5.276650428771973,
"rewards/margins": 3.189429998397827,
"rewards/rejected": -8.466080665588379,
"step": 1000
},
{
"epoch": 0.404,
"grad_norm": 25.664982214116712,
"learning_rate": 3.7196491478468316e-07,
"logits/chosen": 41.09137725830078,
"logits/rejected": 44.35279083251953,
"logps/chosen": -101.16764068603516,
"logps/rejected": -150.22509765625,
"loss": 0.1726,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -4.6303510665893555,
"rewards/margins": 3.4471046924591064,
"rewards/rejected": -8.077455520629883,
"step": 1010
},
{
"epoch": 0.408,
"grad_norm": 4.601273961866918,
"learning_rate": 3.689060522675688e-07,
"logits/chosen": 42.17076873779297,
"logits/rejected": 43.80558776855469,
"logps/chosen": -114.15092468261719,
"logps/rejected": -155.05160522460938,
"loss": 0.1537,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -4.794250965118408,
"rewards/margins": 4.128665447235107,
"rewards/rejected": -8.922918319702148,
"step": 1020
},
{
"epoch": 0.412,
"grad_norm": 23.199840684684883,
"learning_rate": 3.658240087799654e-07,
"logits/chosen": 39.78450393676758,
"logits/rejected": 42.42991638183594,
"logps/chosen": -101.93121337890625,
"logps/rejected": -154.7505340576172,
"loss": 0.1553,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -4.455100059509277,
"rewards/margins": 4.042738914489746,
"rewards/rejected": -8.497838973999023,
"step": 1030
},
{
"epoch": 0.416,
"grad_norm": 31.277129565852135,
"learning_rate": 3.6271938517235765e-07,
"logits/chosen": 40.167823791503906,
"logits/rejected": 42.2595100402832,
"logps/chosen": -107.13337707519531,
"logps/rejected": -156.47689819335938,
"loss": 0.2185,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -5.04506778717041,
"rewards/margins": 3.721726655960083,
"rewards/rejected": -8.76679515838623,
"step": 1040
},
{
"epoch": 0.42,
"grad_norm": 27.680206880133927,
"learning_rate": 3.595927866972693e-07,
"logits/chosen": 38.985374450683594,
"logits/rejected": 39.77815628051758,
"logps/chosen": -106.81770324707031,
"logps/rejected": -149.25894165039062,
"loss": 0.2087,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -4.879289150238037,
"rewards/margins": 3.535437822341919,
"rewards/rejected": -8.414728164672852,
"step": 1050
},
{
"epoch": 0.424,
"grad_norm": 1.249926854778577,
"learning_rate": 3.5644482289126813e-07,
"logits/chosen": 39.94206237792969,
"logits/rejected": 43.77482604980469,
"logps/chosen": -106.6061782836914,
"logps/rejected": -151.3231658935547,
"loss": 0.2439,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -4.947367191314697,
"rewards/margins": 3.605046033859253,
"rewards/rejected": -8.552412986755371,
"step": 1060
},
{
"epoch": 0.428,
"grad_norm": 6.362016874170334,
"learning_rate": 3.5327610745613546e-07,
"logits/chosen": 40.91377258300781,
"logits/rejected": 41.30303192138672,
"logps/chosen": -115.51036071777344,
"logps/rejected": -159.98422241210938,
"loss": 0.1804,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -5.147997856140137,
"rewards/margins": 3.556520462036133,
"rewards/rejected": -8.70451831817627,
"step": 1070
},
{
"epoch": 0.432,
"grad_norm": 17.200244197099025,
"learning_rate": 3.500872581392238e-07,
"logits/chosen": 39.87400436401367,
"logits/rejected": 40.99711990356445,
"logps/chosen": -107.35628509521484,
"logps/rejected": -152.93698120117188,
"loss": 0.1926,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -5.61740255355835,
"rewards/margins": 3.3743271827697754,
"rewards/rejected": -8.991729736328125,
"step": 1080
},
{
"epoch": 0.436,
"grad_norm": 6.686230161730707,
"learning_rate": 3.468788966130257e-07,
"logits/chosen": 39.11838150024414,
"logits/rejected": 40.328468322753906,
"logps/chosen": -111.20975494384766,
"logps/rejected": -164.8342742919922,
"loss": 0.2234,
"rewards/accuracies": 0.875,
"rewards/chosen": -5.404995918273926,
"rewards/margins": 4.121036529541016,
"rewards/rejected": -9.526032447814941,
"step": 1090
},
{
"epoch": 0.44,
"grad_norm": 43.68985759612088,
"learning_rate": 3.43651648353978e-07,
"logits/chosen": 38.73166275024414,
"logits/rejected": 42.4930534362793,
"logps/chosen": -101.05915069580078,
"logps/rejected": -165.46095275878906,
"loss": 0.1615,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -5.2063822746276855,
"rewards/margins": 4.015776634216309,
"rewards/rejected": -9.222158432006836,
"step": 1100
},
{
"epoch": 0.444,
"grad_norm": 26.738862667956763,
"learning_rate": 3.40406142520523e-07,
"logits/chosen": 42.12168502807617,
"logits/rejected": 41.788394927978516,
"logps/chosen": -123.79948425292969,
"logps/rejected": -153.36776733398438,
"loss": 0.2421,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -5.871365547180176,
"rewards/margins": 2.8958840370178223,
"rewards/rejected": -8.76724910736084,
"step": 1110
},
{
"epoch": 0.448,
"grad_norm": 17.03770002319556,
"learning_rate": 3.371430118304538e-07,
"logits/chosen": 36.466461181640625,
"logits/rejected": 40.315452575683594,
"logps/chosen": -112.66862487792969,
"logps/rejected": -174.78466796875,
"loss": 0.2123,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -5.390850067138672,
"rewards/margins": 4.580835342407227,
"rewards/rejected": -9.971685409545898,
"step": 1120
},
{
"epoch": 0.452,
"grad_norm": 24.01278859618808,
"learning_rate": 3.338628924375638e-07,
"logits/chosen": 40.4681282043457,
"logits/rejected": 41.152469635009766,
"logps/chosen": -113.4747543334961,
"logps/rejected": -158.70449829101562,
"loss": 0.2367,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -5.428784370422363,
"rewards/margins": 3.6359362602233887,
"rewards/rejected": -9.06472110748291,
"step": 1130
},
{
"epoch": 0.456,
"grad_norm": 29.87651257208437,
"learning_rate": 3.305664238076278e-07,
"logits/chosen": 39.47797775268555,
"logits/rejected": 40.79694366455078,
"logps/chosen": -100.62152099609375,
"logps/rejected": -157.24269104003906,
"loss": 0.1456,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -4.463057994842529,
"rewards/margins": 4.45237922668457,
"rewards/rejected": -8.915437698364258,
"step": 1140
},
{
"epoch": 0.46,
"grad_norm": 20.100774899172244,
"learning_rate": 3.272542485937368e-07,
"logits/chosen": 38.019832611083984,
"logits/rejected": 39.65892791748047,
"logps/chosen": -99.71014404296875,
"logps/rejected": -152.48388671875,
"loss": 0.2683,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -4.522518157958984,
"rewards/margins": 4.272935390472412,
"rewards/rejected": -8.795454025268555,
"step": 1150
},
{
"epoch": 0.464,
"grad_norm": 17.271502510574486,
"learning_rate": 3.2392701251101167e-07,
"logits/chosen": 38.93779754638672,
"logits/rejected": 43.86870574951172,
"logps/chosen": -106.25166320800781,
"logps/rejected": -169.70816040039062,
"loss": 0.1857,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -4.724178314208984,
"rewards/margins": 4.432940483093262,
"rewards/rejected": -9.157119750976562,
"step": 1160
},
{
"epoch": 0.468,
"grad_norm": 14.85667949892512,
"learning_rate": 3.2058536421071914e-07,
"logits/chosen": 41.40235900878906,
"logits/rejected": 40.91047286987305,
"logps/chosen": -121.07177734375,
"logps/rejected": -163.6028289794922,
"loss": 0.2094,
"rewards/accuracies": 0.875,
"rewards/chosen": -5.4994378089904785,
"rewards/margins": 3.685161590576172,
"rewards/rejected": -9.184598922729492,
"step": 1170
},
{
"epoch": 0.472,
"grad_norm": 27.172740058479278,
"learning_rate": 3.172299551538164e-07,
"logits/chosen": 38.644805908203125,
"logits/rejected": 41.06559371948242,
"logps/chosen": -116.37313079833984,
"logps/rejected": -170.63470458984375,
"loss": 0.2396,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -5.4368085861206055,
"rewards/margins": 4.352160453796387,
"rewards/rejected": -9.788969039916992,
"step": 1180
},
{
"epoch": 0.476,
"grad_norm": 68.26652671242996,
"learning_rate": 3.1386143948394763e-07,
"logits/chosen": 39.66558074951172,
"logits/rejected": 39.35725784301758,
"logps/chosen": -110.7066421508789,
"logps/rejected": -148.9991455078125,
"loss": 0.2147,
"rewards/accuracies": 0.875,
"rewards/chosen": -5.205443382263184,
"rewards/margins": 3.4201292991638184,
"rewards/rejected": -8.62557315826416,
"step": 1190
},
{
"epoch": 0.48,
"grad_norm": 29.844573464953335,
"learning_rate": 3.104804738999169e-07,
"logits/chosen": 38.45096969604492,
"logits/rejected": 39.93912887573242,
"logps/chosen": -118.87516021728516,
"logps/rejected": -177.38160705566406,
"loss": 0.1226,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -5.623927116394043,
"rewards/margins": 4.835057735443115,
"rewards/rejected": -10.458983421325684,
"step": 1200
},
{
"epoch": 0.484,
"grad_norm": 33.168807520397415,
"learning_rate": 3.0708771752766395e-07,
"logits/chosen": 37.66096496582031,
"logits/rejected": 41.729278564453125,
"logps/chosen": -113.61344909667969,
"logps/rejected": -181.90975952148438,
"loss": 0.2327,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -5.367053985595703,
"rewards/margins": 5.373785495758057,
"rewards/rejected": -10.740839958190918,
"step": 1210
},
{
"epoch": 0.488,
"grad_norm": 35.949428687976074,
"learning_rate": 3.036838317917658e-07,
"logits/chosen": 39.2284049987793,
"logits/rejected": 39.56584167480469,
"logps/chosen": -116.41062927246094,
"logps/rejected": -161.64004516601562,
"loss": 0.2307,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -5.617162704467773,
"rewards/margins": 3.87888765335083,
"rewards/rejected": -9.496048927307129,
"step": 1220
},
{
"epoch": 0.492,
"grad_norm": 21.071566027005225,
"learning_rate": 3.002694802864912e-07,
"logits/chosen": 38.54627227783203,
"logits/rejected": 41.567474365234375,
"logps/chosen": -126.9264144897461,
"logps/rejected": -171.91610717773438,
"loss": 0.1884,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -6.11997127532959,
"rewards/margins": 3.630235195159912,
"rewards/rejected": -9.75020694732666,
"step": 1230
},
{
"epoch": 0.496,
"grad_norm": 19.472597346579764,
"learning_rate": 2.968453286464312e-07,
"logits/chosen": 38.088741302490234,
"logits/rejected": 40.73246383666992,
"logps/chosen": -110.57450103759766,
"logps/rejected": -164.86895751953125,
"loss": 0.1301,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -5.830416679382324,
"rewards/margins": 3.7488045692443848,
"rewards/rejected": -9.579221725463867,
"step": 1240
},
{
"epoch": 0.5,
"grad_norm": 9.297784140084044,
"learning_rate": 2.934120444167326e-07,
"logits/chosen": 39.20268249511719,
"logits/rejected": 39.98442077636719,
"logps/chosen": -110.739013671875,
"logps/rejected": -155.21205139160156,
"loss": 0.1205,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -5.348870277404785,
"rewards/margins": 4.167140007019043,
"rewards/rejected": -9.516010284423828,
"step": 1250
},
{
"epoch": 0.504,
"grad_norm": 15.15594138373471,
"learning_rate": 2.899702969229587e-07,
"logits/chosen": 39.848670959472656,
"logits/rejected": 38.895225524902344,
"logps/chosen": -113.55269622802734,
"logps/rejected": -164.986083984375,
"loss": 0.1671,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -5.258578777313232,
"rewards/margins": 4.728585243225098,
"rewards/rejected": -9.987164497375488,
"step": 1260
},
{
"epoch": 0.508,
"grad_norm": 58.68168492653791,
"learning_rate": 2.865207571406029e-07,
"logits/chosen": 39.46464920043945,
"logits/rejected": 37.720924377441406,
"logps/chosen": -124.67518615722656,
"logps/rejected": -169.3942413330078,
"loss": 0.2329,
"rewards/accuracies": 0.875,
"rewards/chosen": -5.639501571655273,
"rewards/margins": 4.678294658660889,
"rewards/rejected": -10.31779670715332,
"step": 1270
},
{
"epoch": 0.512,
"grad_norm": 26.11850298779012,
"learning_rate": 2.830640975642806e-07,
"logits/chosen": 36.690589904785156,
"logits/rejected": 37.93729782104492,
"logps/chosen": -115.35652160644531,
"logps/rejected": -169.6790008544922,
"loss": 0.1717,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -5.6271867752075195,
"rewards/margins": 4.886000633239746,
"rewards/rejected": -10.513187408447266,
"step": 1280
},
{
"epoch": 0.516,
"grad_norm": 18.8486793071701,
"learning_rate": 2.796009920766253e-07,
"logits/chosen": 34.7783203125,
"logits/rejected": 40.29841232299805,
"logps/chosen": -109.31401062011719,
"logps/rejected": -170.9484100341797,
"loss": 0.1634,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -5.725811958312988,
"rewards/margins": 4.456699848175049,
"rewards/rejected": -10.182512283325195,
"step": 1290
},
{
"epoch": 0.52,
"grad_norm": 47.45375394157672,
"learning_rate": 2.761321158169134e-07,
"logits/chosen": 37.883079528808594,
"logits/rejected": 40.114280700683594,
"logps/chosen": -124.6867904663086,
"logps/rejected": -172.8629150390625,
"loss": 0.2061,
"rewards/accuracies": 0.875,
"rewards/chosen": -6.434460639953613,
"rewards/margins": 3.7778351306915283,
"rewards/rejected": -10.212295532226562,
"step": 1300
},
{
"epoch": 0.524,
"grad_norm": 16.432297078281568,
"learning_rate": 2.726581450494451e-07,
"logits/chosen": 35.259639739990234,
"logits/rejected": 40.99456024169922,
"logps/chosen": -111.4013900756836,
"logps/rejected": -174.25094604492188,
"loss": 0.154,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -6.062236785888672,
"rewards/margins": 4.306510925292969,
"rewards/rejected": -10.368746757507324,
"step": 1310
},
{
"epoch": 0.528,
"grad_norm": 30.21091242436278,
"learning_rate": 2.6917975703170465e-07,
"logits/chosen": 35.50202178955078,
"logits/rejected": 39.13561248779297,
"logps/chosen": -113.72837829589844,
"logps/rejected": -176.2806396484375,
"loss": 0.176,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -5.105679035186768,
"rewards/margins": 4.945074081420898,
"rewards/rejected": -10.050752639770508,
"step": 1320
},
{
"epoch": 0.532,
"grad_norm": 31.809974002259562,
"learning_rate": 2.6569762988232837e-07,
"logits/chosen": 35.528358459472656,
"logits/rejected": 39.22612380981445,
"logps/chosen": -108.7910385131836,
"logps/rejected": -159.57582092285156,
"loss": 0.2619,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -6.001819133758545,
"rewards/margins": 3.755993366241455,
"rewards/rejected": -9.757813453674316,
"step": 1330
},
{
"epoch": 0.536,
"grad_norm": 18.493232174978285,
"learning_rate": 2.6221244244890336e-07,
"logits/chosen": 34.30480194091797,
"logits/rejected": 38.207496643066406,
"logps/chosen": -104.4682388305664,
"logps/rejected": -172.13088989257812,
"loss": 0.1043,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -5.1313066482543945,
"rewards/margins": 5.272080421447754,
"rewards/rejected": -10.403387069702148,
"step": 1340
},
{
"epoch": 0.54,
"grad_norm": 24.170314835898346,
"learning_rate": 2.5872487417562527e-07,
"logits/chosen": 36.6849479675293,
"logits/rejected": 37.138465881347656,
"logps/chosen": -123.2945785522461,
"logps/rejected": -178.46588134765625,
"loss": 0.1769,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -6.805965423583984,
"rewards/margins": 4.122119426727295,
"rewards/rejected": -10.928085327148438,
"step": 1350
},
{
"epoch": 0.544,
"grad_norm": 23.047663073832965,
"learning_rate": 2.5523560497083924e-07,
"logits/chosen": 35.815589904785156,
"logits/rejected": 37.165496826171875,
"logps/chosen": -118.9620361328125,
"logps/rejected": -173.7703094482422,
"loss": 0.1205,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -6.1362409591674805,
"rewards/margins": 4.513824462890625,
"rewards/rejected": -10.650065422058105,
"step": 1360
},
{
"epoch": 0.548,
"grad_norm": 48.732887198712326,
"learning_rate": 2.5174531507449037e-07,
"logits/chosen": 36.993038177490234,
"logits/rejected": 36.96677780151367,
"logps/chosen": -132.6764678955078,
"logps/rejected": -172.9881591796875,
"loss": 0.2913,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -6.736947536468506,
"rewards/margins": 4.004305362701416,
"rewards/rejected": -10.741253852844238,
"step": 1370
},
{
"epoch": 0.552,
"grad_norm": 18.181102544014543,
"learning_rate": 2.482546849255096e-07,
"logits/chosen": 34.073097229003906,
"logits/rejected": 36.943172454833984,
"logps/chosen": -117.58740234375,
"logps/rejected": -197.7882537841797,
"loss": 0.1062,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -5.832341194152832,
"rewards/margins": 6.429389953613281,
"rewards/rejected": -12.26173210144043,
"step": 1380
},
{
"epoch": 0.556,
"grad_norm": 15.544623320366716,
"learning_rate": 2.447643950291608e-07,
"logits/chosen": 33.73606491088867,
"logits/rejected": 36.898773193359375,
"logps/chosen": -119.47406005859375,
"logps/rejected": -167.00711059570312,
"loss": 0.2418,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -6.214328289031982,
"rewards/margins": 4.625095844268799,
"rewards/rejected": -10.839425086975098,
"step": 1390
},
{
"epoch": 0.56,
"grad_norm": 77.37565224093566,
"learning_rate": 2.412751258243748e-07,
"logits/chosen": 33.67988204956055,
"logits/rejected": 34.09700393676758,
"logps/chosen": -128.5489501953125,
"logps/rejected": -176.24871826171875,
"loss": 0.1963,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -7.089766502380371,
"rewards/margins": 3.989410400390625,
"rewards/rejected": -11.07917594909668,
"step": 1400
},
{
"epoch": 0.564,
"grad_norm": 30.43460528340991,
"learning_rate": 2.3778755755109667e-07,
"logits/chosen": 34.34267044067383,
"logits/rejected": 37.77847671508789,
"logps/chosen": -105.62583923339844,
"logps/rejected": -190.62722778320312,
"loss": 0.1466,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -5.149474620819092,
"rewards/margins": 6.530264854431152,
"rewards/rejected": -11.679739952087402,
"step": 1410
},
{
"epoch": 0.568,
"grad_norm": 7.007072586126817,
"learning_rate": 2.3430237011767164e-07,
"logits/chosen": 33.648799896240234,
"logits/rejected": 36.46876525878906,
"logps/chosen": -113.22758483886719,
"logps/rejected": -178.4688262939453,
"loss": 0.1731,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -6.234652042388916,
"rewards/margins": 4.812324047088623,
"rewards/rejected": -11.046976089477539,
"step": 1420
},
{
"epoch": 0.572,
"grad_norm": 19.130152675478577,
"learning_rate": 2.3082024296829532e-07,
"logits/chosen": 35.6358757019043,
"logits/rejected": 36.253257751464844,
"logps/chosen": -113.02217864990234,
"logps/rejected": -185.16107177734375,
"loss": 0.1124,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -5.546517372131348,
"rewards/margins": 5.918461799621582,
"rewards/rejected": -11.464980125427246,
"step": 1430
},
{
"epoch": 0.576,
"grad_norm": 18.125883555971487,
"learning_rate": 2.2734185495055498e-07,
"logits/chosen": 35.95376968383789,
"logits/rejected": 37.0533332824707,
"logps/chosen": -121.62260437011719,
"logps/rejected": -175.02882385253906,
"loss": 0.1505,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -6.334244251251221,
"rewards/margins": 4.729000568389893,
"rewards/rejected": -11.06324577331543,
"step": 1440
},
{
"epoch": 0.58,
"grad_norm": 56.564917397952655,
"learning_rate": 2.2386788418308665e-07,
"logits/chosen": 33.461631774902344,
"logits/rejected": 37.12461853027344,
"logps/chosen": -112.4884262084961,
"logps/rejected": -202.30970764160156,
"loss": 0.0664,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -5.148140907287598,
"rewards/margins": 7.338738441467285,
"rewards/rejected": -12.486879348754883,
"step": 1450
},
{
"epoch": 0.584,
"grad_norm": 43.54182209696087,
"learning_rate": 2.2039900792337474e-07,
"logits/chosen": 32.17646026611328,
"logits/rejected": 35.69232940673828,
"logps/chosen": -121.44427490234375,
"logps/rejected": -192.20877075195312,
"loss": 0.1323,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -6.396025657653809,
"rewards/margins": 5.983513832092285,
"rewards/rejected": -12.379539489746094,
"step": 1460
},
{
"epoch": 0.588,
"grad_norm": 75.32127336294906,
"learning_rate": 2.1693590243571935e-07,
"logits/chosen": 34.178611755371094,
"logits/rejected": 36.59508514404297,
"logps/chosen": -123.4473648071289,
"logps/rejected": -192.17855834960938,
"loss": 0.2353,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -6.135283470153809,
"rewards/margins": 5.6739821434021,
"rewards/rejected": -11.80926513671875,
"step": 1470
},
{
"epoch": 0.592,
"grad_norm": 45.58891579183569,
"learning_rate": 2.1347924285939712e-07,
"logits/chosen": 34.34710693359375,
"logits/rejected": 34.652095794677734,
"logps/chosen": -125.2497329711914,
"logps/rejected": -172.30828857421875,
"loss": 0.2612,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -7.1709747314453125,
"rewards/margins": 4.1582560539245605,
"rewards/rejected": -11.329230308532715,
"step": 1480
},
{
"epoch": 0.596,
"grad_norm": 15.251049246391148,
"learning_rate": 2.100297030770413e-07,
"logits/chosen": 34.02775573730469,
"logits/rejected": 35.586952209472656,
"logps/chosen": -118.30632019042969,
"logps/rejected": -179.9751434326172,
"loss": 0.0947,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -6.386832237243652,
"rewards/margins": 5.212705135345459,
"rewards/rejected": -11.599536895751953,
"step": 1490
},
{
"epoch": 0.6,
"grad_norm": 25.472331773437826,
"learning_rate": 2.065879555832674e-07,
"logits/chosen": 31.809621810913086,
"logits/rejected": 35.93482208251953,
"logps/chosen": -122.5445785522461,
"logps/rejected": -203.33628845214844,
"loss": 0.1295,
"rewards/accuracies": 1.0,
"rewards/chosen": -6.351222991943359,
"rewards/margins": 6.3509392738342285,
"rewards/rejected": -12.702162742614746,
"step": 1500
},
{
"epoch": 0.604,
"grad_norm": 36.99215746879388,
"learning_rate": 2.0315467135356878e-07,
"logits/chosen": 34.279014587402344,
"logits/rejected": 36.6539421081543,
"logps/chosen": -118.03910827636719,
"logps/rejected": -185.2469482421875,
"loss": 0.1324,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.5454583168029785,
"rewards/margins": 5.202276229858398,
"rewards/rejected": -11.747734069824219,
"step": 1510
},
{
"epoch": 0.608,
"grad_norm": 5.72817076275817,
"learning_rate": 1.9973051971350888e-07,
"logits/chosen": 32.36498260498047,
"logits/rejected": 36.872802734375,
"logps/chosen": -119.40788269042969,
"logps/rejected": -182.93051147460938,
"loss": 0.1548,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.524139404296875,
"rewards/margins": 4.851975440979004,
"rewards/rejected": -11.376115798950195,
"step": 1520
},
{
"epoch": 0.612,
"grad_norm": 62.7485215727146,
"learning_rate": 1.9631616820823418e-07,
"logits/chosen": 33.554473876953125,
"logits/rejected": 34.441917419433594,
"logps/chosen": -128.63380432128906,
"logps/rejected": -187.55038452148438,
"loss": 0.2279,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -6.242335319519043,
"rewards/margins": 5.341699600219727,
"rewards/rejected": -11.58403491973877,
"step": 1530
},
{
"epoch": 0.616,
"grad_norm": 50.55388655519452,
"learning_rate": 1.9291228247233603e-07,
"logits/chosen": 34.57413101196289,
"logits/rejected": 38.980873107910156,
"logps/chosen": -124.8611068725586,
"logps/rejected": -188.2270050048828,
"loss": 0.1368,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -6.550134181976318,
"rewards/margins": 4.793581008911133,
"rewards/rejected": -11.34371566772461,
"step": 1540
},
{
"epoch": 0.62,
"grad_norm": 20.98136994366918,
"learning_rate": 1.895195261000831e-07,
"logits/chosen": 32.666053771972656,
"logits/rejected": 36.28156661987305,
"logps/chosen": -123.55598449707031,
"logps/rejected": -190.2432098388672,
"loss": 0.2085,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -6.651708126068115,
"rewards/margins": 5.466277122497559,
"rewards/rejected": -12.117984771728516,
"step": 1550
},
{
"epoch": 0.624,
"grad_norm": 3.695214530883218,
"learning_rate": 1.861385605160524e-07,
"logits/chosen": 34.80469512939453,
"logits/rejected": 37.0577507019043,
"logps/chosen": -127.52436828613281,
"logps/rejected": -200.47572326660156,
"loss": 0.1962,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.4474639892578125,
"rewards/margins": 6.185749530792236,
"rewards/rejected": -12.633213996887207,
"step": 1560
},
{
"epoch": 0.628,
"grad_norm": 6.204487863528272,
"learning_rate": 1.8277004484618357e-07,
"logits/chosen": 35.11884689331055,
"logits/rejected": 34.14295959472656,
"logps/chosen": -123.8480224609375,
"logps/rejected": -175.71351623535156,
"loss": 0.1239,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -5.91684103012085,
"rewards/margins": 5.305278778076172,
"rewards/rejected": -11.22212028503418,
"step": 1570
},
{
"epoch": 0.632,
"grad_norm": 27.777656870634978,
"learning_rate": 1.7941463578928083e-07,
"logits/chosen": 31.193029403686523,
"logits/rejected": 34.23871612548828,
"logps/chosen": -110.49909973144531,
"logps/rejected": -183.0133514404297,
"loss": 0.1618,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -5.92338752746582,
"rewards/margins": 5.685771942138672,
"rewards/rejected": -11.609160423278809,
"step": 1580
},
{
"epoch": 0.636,
"grad_norm": 25.59557039318974,
"learning_rate": 1.760729874889884e-07,
"logits/chosen": 34.273590087890625,
"logits/rejected": 36.67070388793945,
"logps/chosen": -123.4420394897461,
"logps/rejected": -188.76451110839844,
"loss": 0.1413,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -5.854828834533691,
"rewards/margins": 5.536397457122803,
"rewards/rejected": -11.391225814819336,
"step": 1590
},
{
"epoch": 0.64,
"grad_norm": 54.20969538897634,
"learning_rate": 1.7274575140626315e-07,
"logits/chosen": 35.63384246826172,
"logits/rejected": 36.39768600463867,
"logps/chosen": -142.24166870117188,
"logps/rejected": -204.695556640625,
"loss": 0.2189,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -7.065789699554443,
"rewards/margins": 5.59266471862793,
"rewards/rejected": -12.658454895019531,
"step": 1600
},
{
"epoch": 0.644,
"grad_norm": 23.952776600344606,
"learning_rate": 1.6943357619237225e-07,
"logits/chosen": 30.349929809570312,
"logits/rejected": 35.475894927978516,
"logps/chosen": -136.0992431640625,
"logps/rejected": -200.9610595703125,
"loss": 0.1836,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -7.412652492523193,
"rewards/margins": 4.916618347167969,
"rewards/rejected": -12.32927131652832,
"step": 1610
},
{
"epoch": 0.648,
"grad_norm": 20.457675381739918,
"learning_rate": 1.6613710756243627e-07,
"logits/chosen": 34.76255416870117,
"logits/rejected": 34.33202362060547,
"logps/chosen": -123.5507583618164,
"logps/rejected": -174.77584838867188,
"loss": 0.1588,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -7.010063171386719,
"rewards/margins": 4.99750280380249,
"rewards/rejected": -12.007566452026367,
"step": 1620
},
{
"epoch": 0.652,
"grad_norm": 8.740451188853967,
"learning_rate": 1.6285698816954624e-07,
"logits/chosen": 32.92230987548828,
"logits/rejected": 37.1563720703125,
"logps/chosen": -125.0332260131836,
"logps/rejected": -198.4005126953125,
"loss": 0.146,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.896471977233887,
"rewards/margins": 5.607396602630615,
"rewards/rejected": -12.503868103027344,
"step": 1630
},
{
"epoch": 0.656,
"grad_norm": 12.82247945171976,
"learning_rate": 1.5959385747947695e-07,
"logits/chosen": 32.56360626220703,
"logits/rejected": 36.79113006591797,
"logps/chosen": -115.17347717285156,
"logps/rejected": -194.03790283203125,
"loss": 0.1129,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -5.800166130065918,
"rewards/margins": 5.930178165435791,
"rewards/rejected": -11.73034381866455,
"step": 1640
},
{
"epoch": 0.66,
"grad_norm": 54.05169324828869,
"learning_rate": 1.5634835164602196e-07,
"logits/chosen": 30.58074378967285,
"logits/rejected": 34.36854934692383,
"logps/chosen": -113.1893539428711,
"logps/rejected": -182.39627075195312,
"loss": 0.1431,
"rewards/accuracies": 0.875,
"rewards/chosen": -6.311237812042236,
"rewards/margins": 5.61757755279541,
"rewards/rejected": -11.928815841674805,
"step": 1650
},
{
"epoch": 0.664,
"grad_norm": 24.825767693321374,
"learning_rate": 1.5312110338697427e-07,
"logits/chosen": 35.91250228881836,
"logits/rejected": 36.06043243408203,
"logps/chosen": -124.50358581542969,
"logps/rejected": -201.7527313232422,
"loss": 0.1154,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.220883369445801,
"rewards/margins": 6.432749271392822,
"rewards/rejected": -12.653631210327148,
"step": 1660
},
{
"epoch": 0.668,
"grad_norm": 21.917052429686564,
"learning_rate": 1.4991274186077628e-07,
"logits/chosen": 32.15296936035156,
"logits/rejected": 34.598384857177734,
"logps/chosen": -128.67645263671875,
"logps/rejected": -194.93014526367188,
"loss": 0.2175,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.7162299156188965,
"rewards/margins": 5.552061557769775,
"rewards/rejected": -12.268292427062988,
"step": 1670
},
{
"epoch": 0.672,
"grad_norm": 73.93391750261412,
"learning_rate": 1.4672389254386457e-07,
"logits/chosen": 33.413238525390625,
"logits/rejected": 34.18181610107422,
"logps/chosen": -124.91861724853516,
"logps/rejected": -205.4623260498047,
"loss": 0.2138,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.610821723937988,
"rewards/margins": 6.807159423828125,
"rewards/rejected": -13.41797924041748,
"step": 1680
},
{
"epoch": 0.676,
"grad_norm": 14.047455580384543,
"learning_rate": 1.4355517710873182e-07,
"logits/chosen": 34.12297821044922,
"logits/rejected": 35.99665451049805,
"logps/chosen": -121.33824157714844,
"logps/rejected": -176.9901123046875,
"loss": 0.1439,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.118622779846191,
"rewards/margins": 5.124892234802246,
"rewards/rejected": -11.243515014648438,
"step": 1690
},
{
"epoch": 0.68,
"grad_norm": 0.7246303761815563,
"learning_rate": 1.404072133027306e-07,
"logits/chosen": 33.210655212402344,
"logits/rejected": 36.71358108520508,
"logps/chosen": -116.57881927490234,
"logps/rejected": -182.92950439453125,
"loss": 0.0992,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.269998550415039,
"rewards/margins": 5.008700370788574,
"rewards/rejected": -11.27869987487793,
"step": 1700
},
{
"epoch": 0.684,
"grad_norm": 34.85736611285892,
"learning_rate": 1.3728061482764235e-07,
"logits/chosen": 32.43885040283203,
"logits/rejected": 34.203834533691406,
"logps/chosen": -125.60906982421875,
"logps/rejected": -184.39035034179688,
"loss": 0.1539,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -6.697512149810791,
"rewards/margins": 5.102963924407959,
"rewards/rejected": -11.800477027893066,
"step": 1710
},
{
"epoch": 0.688,
"grad_norm": 9.586401996642,
"learning_rate": 1.341759912200346e-07,
"logits/chosen": 32.35176467895508,
"logits/rejected": 34.566925048828125,
"logps/chosen": -126.4211196899414,
"logps/rejected": -186.43441772460938,
"loss": 0.1744,
"rewards/accuracies": 0.875,
"rewards/chosen": -6.952803611755371,
"rewards/margins": 4.971192836761475,
"rewards/rejected": -11.923995971679688,
"step": 1720
},
{
"epoch": 0.692,
"grad_norm": 51.225200538375525,
"learning_rate": 1.3109394773243115e-07,
"logits/chosen": 32.60570526123047,
"logits/rejected": 34.30889892578125,
"logps/chosen": -112.5150375366211,
"logps/rejected": -185.24020385742188,
"loss": 0.1232,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -5.684107780456543,
"rewards/margins": 5.926398277282715,
"rewards/rejected": -11.610506057739258,
"step": 1730
},
{
"epoch": 0.696,
"grad_norm": 7.892076438359368,
"learning_rate": 1.2803508521531677e-07,
"logits/chosen": 32.87091827392578,
"logits/rejected": 35.00414276123047,
"logps/chosen": -115.19721984863281,
"logps/rejected": -186.33670043945312,
"loss": 0.1689,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -6.1829729080200195,
"rewards/margins": 5.768080711364746,
"rewards/rejected": -11.951054573059082,
"step": 1740
},
{
"epoch": 0.7,
"grad_norm": 71.60841944470864,
"learning_rate": 1.2500000000000005e-07,
"logits/chosen": 31.9818115234375,
"logits/rejected": 34.28449630737305,
"logps/chosen": -131.71206665039062,
"logps/rejected": -179.4838104248047,
"loss": 0.2651,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -7.21808385848999,
"rewards/margins": 4.244929313659668,
"rewards/rejected": -11.463014602661133,
"step": 1750
},
{
"epoch": 0.704,
"grad_norm": 11.403139100682084,
"learning_rate": 1.2198928378235715e-07,
"logits/chosen": 35.4173583984375,
"logits/rejected": 32.19343185424805,
"logps/chosen": -141.1168212890625,
"logps/rejected": -190.25442504882812,
"loss": 0.1845,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -7.428712368011475,
"rewards/margins": 4.997444152832031,
"rewards/rejected": -12.426156997680664,
"step": 1760
},
{
"epoch": 0.708,
"grad_norm": 15.614574564320913,
"learning_rate": 1.1900352350748024e-07,
"logits/chosen": 32.91938018798828,
"logits/rejected": 37.30883026123047,
"logps/chosen": -128.07351684570312,
"logps/rejected": -202.3236083984375,
"loss": 0.1459,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.7811689376831055,
"rewards/margins": 5.805022239685059,
"rewards/rejected": -12.586191177368164,
"step": 1770
},
{
"epoch": 0.712,
"grad_norm": 36.93773962705704,
"learning_rate": 1.1604330125525078e-07,
"logits/chosen": 33.32939910888672,
"logits/rejected": 34.7056770324707,
"logps/chosen": -126.90995025634766,
"logps/rejected": -206.4558563232422,
"loss": 0.109,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.557587623596191,
"rewards/margins": 6.45965576171875,
"rewards/rejected": -13.017242431640625,
"step": 1780
},
{
"epoch": 0.716,
"grad_norm": 19.64090712772826,
"learning_rate": 1.1310919412686245e-07,
"logits/chosen": 31.58943748474121,
"logits/rejected": 32.97970962524414,
"logps/chosen": -120.78947448730469,
"logps/rejected": -185.22781372070312,
"loss": 0.1554,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -6.556436061859131,
"rewards/margins": 5.089656829833984,
"rewards/rejected": -11.646093368530273,
"step": 1790
},
{
"epoch": 0.72,
"grad_norm": 1.8073260877539248,
"learning_rate": 1.1020177413231332e-07,
"logits/chosen": 32.85557556152344,
"logits/rejected": 36.344078063964844,
"logps/chosen": -119.52928161621094,
"logps/rejected": -186.12684631347656,
"loss": 0.1001,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -6.19125509262085,
"rewards/margins": 5.620010852813721,
"rewards/rejected": -11.81126594543457,
"step": 1800
},
{
"epoch": 0.724,
"grad_norm": 1.233410177278683,
"learning_rate": 1.073216080788921e-07,
"logits/chosen": 31.43521499633789,
"logits/rejected": 34.373809814453125,
"logps/chosen": -117.0105972290039,
"logps/rejected": -202.0782928466797,
"loss": 0.0878,
"rewards/accuracies": 1.0,
"rewards/chosen": -6.203524112701416,
"rewards/margins": 6.944014072418213,
"rewards/rejected": -13.147537231445312,
"step": 1810
},
{
"epoch": 0.728,
"grad_norm": 76.66139035605511,
"learning_rate": 1.0446925746067766e-07,
"logits/chosen": 34.948951721191406,
"logits/rejected": 38.018917083740234,
"logps/chosen": -128.2957305908203,
"logps/rejected": -192.70611572265625,
"loss": 0.1567,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -7.262060642242432,
"rewards/margins": 5.220892906188965,
"rewards/rejected": -12.482953071594238,
"step": 1820
},
{
"epoch": 0.732,
"grad_norm": 3.2686353366997087,
"learning_rate": 1.0164527834907466e-07,
"logits/chosen": 33.62763977050781,
"logits/rejected": 32.20967483520508,
"logps/chosen": -115.2950668334961,
"logps/rejected": -181.83990478515625,
"loss": 0.1879,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -6.048405170440674,
"rewards/margins": 6.285736083984375,
"rewards/rejected": -12.334141731262207,
"step": 1830
},
{
"epoch": 0.736,
"grad_norm": 19.72682359281921,
"learning_rate": 9.885022128440629e-08,
"logits/chosen": 30.57301902770996,
"logits/rejected": 33.91904067993164,
"logps/chosen": -123.07301330566406,
"logps/rejected": -196.01779174804688,
"loss": 0.1789,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -6.5850043296813965,
"rewards/margins": 5.699263572692871,
"rewards/rejected": -12.28426742553711,
"step": 1840
},
{
"epoch": 0.74,
"grad_norm": 27.782784282950555,
"learning_rate": 9.608463116858542e-08,
"logits/chosen": 32.8736457824707,
"logits/rejected": 33.712486267089844,
"logps/chosen": -124.52708435058594,
"logps/rejected": -185.42349243164062,
"loss": 0.2559,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -6.897887229919434,
"rewards/margins": 4.95641565322876,
"rewards/rejected": -11.854303359985352,
"step": 1850
},
{
"epoch": 0.744,
"grad_norm": 11.434937847700061,
"learning_rate": 9.334904715888494e-08,
"logits/chosen": 32.95331954956055,
"logits/rejected": 34.37748336791992,
"logps/chosen": -135.6911163330078,
"logps/rejected": -199.61776733398438,
"loss": 0.1256,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -7.312950134277344,
"rewards/margins": 5.76673698425293,
"rewards/rejected": -13.079686164855957,
"step": 1860
},
{
"epoch": 0.748,
"grad_norm": 21.711629131923438,
"learning_rate": 9.064400256282755e-08,
"logits/chosen": 32.11374282836914,
"logits/rejected": 32.81106185913086,
"logps/chosen": -126.60643005371094,
"logps/rejected": -190.9490966796875,
"loss": 0.1124,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -6.639864444732666,
"rewards/margins": 5.9442009925842285,
"rewards/rejected": -12.584066390991211,
"step": 1870
},
{
"epoch": 0.752,
"grad_norm": 4.882787255482076,
"learning_rate": 8.797002473421727e-08,
"logits/chosen": 32.480838775634766,
"logits/rejected": 34.38404846191406,
"logps/chosen": -120.6583023071289,
"logps/rejected": -187.3675079345703,
"loss": 0.0805,
"rewards/accuracies": 1.0,
"rewards/chosen": -6.4553680419921875,
"rewards/margins": 5.736647605895996,
"rewards/rejected": -12.192015647888184,
"step": 1880
},
{
"epoch": 0.756,
"grad_norm": 73.20268420406155,
"learning_rate": 8.532763497032986e-08,
"logits/chosen": 30.1946964263916,
"logits/rejected": 35.3504638671875,
"logps/chosen": -130.51853942871094,
"logps/rejected": -197.19638061523438,
"loss": 0.1867,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -7.48050594329834,
"rewards/margins": 5.310255527496338,
"rewards/rejected": -12.79076099395752,
"step": 1890
},
{
"epoch": 0.76,
"grad_norm": 62.43015390970909,
"learning_rate": 8.271734841028552e-08,
"logits/chosen": 30.940185546875,
"logits/rejected": 34.006492614746094,
"logps/chosen": -125.14324951171875,
"logps/rejected": -189.832763671875,
"loss": 0.2225,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -7.153977870941162,
"rewards/margins": 5.524373531341553,
"rewards/rejected": -12.678351402282715,
"step": 1900
},
{
"epoch": 0.764,
"grad_norm": 58.76354716466513,
"learning_rate": 8.013967393462093e-08,
"logits/chosen": 32.89315414428711,
"logits/rejected": 35.409217834472656,
"logps/chosen": -114.31298828125,
"logps/rejected": -197.35025024414062,
"loss": 0.089,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -5.788350582122803,
"rewards/margins": 6.997166633605957,
"rewards/rejected": -12.785517692565918,
"step": 1910
},
{
"epoch": 0.768,
"grad_norm": 40.55885429267448,
"learning_rate": 7.759511406608255e-08,
"logits/chosen": 30.79556655883789,
"logits/rejected": 33.01952362060547,
"logps/chosen": -136.01585388183594,
"logps/rejected": -189.73330688476562,
"loss": 0.134,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -7.579942226409912,
"rewards/margins": 4.825619220733643,
"rewards/rejected": -12.405561447143555,
"step": 1920
},
{
"epoch": 0.772,
"grad_norm": 11.57424881967435,
"learning_rate": 7.508416487165862e-08,
"logits/chosen": 28.126171112060547,
"logits/rejected": 33.28355026245117,
"logps/chosen": -123.10121154785156,
"logps/rejected": -203.12960815429688,
"loss": 0.093,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -7.236498832702637,
"rewards/margins": 5.934015274047852,
"rewards/rejected": -13.170514106750488,
"step": 1930
},
{
"epoch": 0.776,
"grad_norm": 21.840133243019153,
"learning_rate": 7.260731586586982e-08,
"logits/chosen": 31.944263458251953,
"logits/rejected": 30.55642318725586,
"logps/chosen": -123.80488586425781,
"logps/rejected": -195.6708526611328,
"loss": 0.162,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.678342342376709,
"rewards/margins": 6.705534934997559,
"rewards/rejected": -13.383877754211426,
"step": 1940
},
{
"epoch": 0.78,
"grad_norm": 13.922142563059072,
"learning_rate": 7.016504991533726e-08,
"logits/chosen": 32.323158264160156,
"logits/rejected": 34.14168167114258,
"logps/chosen": -122.30216979980469,
"logps/rejected": -184.92527770996094,
"loss": 0.178,
"rewards/accuracies": 0.875,
"rewards/chosen": -6.948062896728516,
"rewards/margins": 4.983668327331543,
"rewards/rejected": -11.931732177734375,
"step": 1950
},
{
"epoch": 0.784,
"grad_norm": 43.52548154571825,
"learning_rate": 6.775784314464716e-08,
"logits/chosen": 31.536640167236328,
"logits/rejected": 33.37910461425781,
"logps/chosen": -122.21321868896484,
"logps/rejected": -192.3970184326172,
"loss": 0.055,
"rewards/accuracies": 1.0,
"rewards/chosen": -6.536123752593994,
"rewards/margins": 6.116715908050537,
"rewards/rejected": -12.652839660644531,
"step": 1960
},
{
"epoch": 0.788,
"grad_norm": 14.960056288014227,
"learning_rate": 6.538616484352902e-08,
"logits/chosen": 30.65342140197754,
"logits/rejected": 31.002685546875,
"logps/chosen": -119.79139709472656,
"logps/rejected": -194.21560668945312,
"loss": 0.1468,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -6.551787376403809,
"rewards/margins": 6.742495536804199,
"rewards/rejected": -13.294283866882324,
"step": 1970
},
{
"epoch": 0.792,
"grad_norm": 63.96276766967657,
"learning_rate": 6.305047737536707e-08,
"logits/chosen": 32.422218322753906,
"logits/rejected": 34.42628479003906,
"logps/chosen": -141.8745574951172,
"logps/rejected": -196.88424682617188,
"loss": 0.2212,
"rewards/accuracies": 0.875,
"rewards/chosen": -8.096589088439941,
"rewards/margins": 4.696638584136963,
"rewards/rejected": -12.79322624206543,
"step": 1980
},
{
"epoch": 0.796,
"grad_norm": 6.994169229009863,
"learning_rate": 6.075123608706093e-08,
"logits/chosen": 31.055633544921875,
"logits/rejected": 33.48008346557617,
"logps/chosen": -122.30110931396484,
"logps/rejected": -194.860595703125,
"loss": 0.1579,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.244012355804443,
"rewards/margins": 6.198871612548828,
"rewards/rejected": -12.44288444519043,
"step": 1990
},
{
"epoch": 0.8,
"grad_norm": 21.842683348401124,
"learning_rate": 5.848888922025552e-08,
"logits/chosen": 31.886560440063477,
"logits/rejected": 35.48186492919922,
"logps/chosen": -135.24679565429688,
"logps/rejected": -206.4519500732422,
"loss": 0.0904,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -7.111126899719238,
"rewards/margins": 5.913634300231934,
"rewards/rejected": -13.024760246276855,
"step": 2000
},
{
"epoch": 0.804,
"grad_norm": 10.8862310531102,
"learning_rate": 5.6263877823955115e-08,
"logits/chosen": 33.398433685302734,
"logits/rejected": 33.78891372680664,
"logps/chosen": -123.2912368774414,
"logps/rejected": -192.1443634033203,
"loss": 0.1133,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -7.0102667808532715,
"rewards/margins": 5.909992218017578,
"rewards/rejected": -12.920259475708008,
"step": 2010
},
{
"epoch": 0.808,
"grad_norm": 42.03601830195942,
"learning_rate": 5.4076635668540065e-08,
"logits/chosen": 29.69512939453125,
"logits/rejected": 33.41075134277344,
"logps/chosen": -120.9056625366211,
"logps/rejected": -192.8750457763672,
"loss": 0.1104,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -6.7075395584106445,
"rewards/margins": 5.716131687164307,
"rewards/rejected": -12.423670768737793,
"step": 2020
},
{
"epoch": 0.812,
"grad_norm": 32.933513594571174,
"learning_rate": 5.192758916120235e-08,
"logits/chosen": 32.2730712890625,
"logits/rejected": 35.09003448486328,
"logps/chosen": -126.17935943603516,
"logps/rejected": -197.20889282226562,
"loss": 0.1222,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.885321140289307,
"rewards/margins": 5.830145835876465,
"rewards/rejected": -12.71546745300293,
"step": 2030
},
{
"epoch": 0.816,
"grad_norm": 42.07730167231684,
"learning_rate": 4.981715726281666e-08,
"logits/chosen": 32.14569091796875,
"logits/rejected": 33.87199401855469,
"logps/chosen": -137.12149047851562,
"logps/rejected": -203.73658752441406,
"loss": 0.2057,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -7.4929962158203125,
"rewards/margins": 5.718442916870117,
"rewards/rejected": -13.211441040039062,
"step": 2040
},
{
"epoch": 0.82,
"grad_norm": 15.321840192563384,
"learning_rate": 4.774575140626316e-08,
"logits/chosen": 30.928787231445312,
"logits/rejected": 33.557373046875,
"logps/chosen": -122.8592529296875,
"logps/rejected": -194.76742553710938,
"loss": 0.1891,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.503848075866699,
"rewards/margins": 6.216989040374756,
"rewards/rejected": -12.720837593078613,
"step": 2050
},
{
"epoch": 0.824,
"grad_norm": 71.00238324419556,
"learning_rate": 4.5713775416217875e-08,
"logits/chosen": 29.931808471679688,
"logits/rejected": 32.696876525878906,
"logps/chosen": -133.57138061523438,
"logps/rejected": -205.98837280273438,
"loss": 0.1357,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -7.4561920166015625,
"rewards/margins": 5.7885637283325195,
"rewards/rejected": -13.244755744934082,
"step": 2060
},
{
"epoch": 0.828,
"grad_norm": 9.35367075528383,
"learning_rate": 4.372162543042623e-08,
"logits/chosen": 31.20676612854004,
"logits/rejected": 31.955196380615234,
"logps/chosen": -135.50155639648438,
"logps/rejected": -201.00784301757812,
"loss": 0.0994,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -7.0842413902282715,
"rewards/margins": 6.034939289093018,
"rewards/rejected": -13.119178771972656,
"step": 2070
},
{
"epoch": 0.832,
"grad_norm": 24.304073923638086,
"learning_rate": 4.176968982247514e-08,
"logits/chosen": 29.5570068359375,
"logits/rejected": 33.91077423095703,
"logps/chosen": -126.53129577636719,
"logps/rejected": -205.4754638671875,
"loss": 0.1221,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -7.051833152770996,
"rewards/margins": 6.18133020401001,
"rewards/rejected": -13.233163833618164,
"step": 2080
},
{
"epoch": 0.836,
"grad_norm": 21.40116937016571,
"learning_rate": 3.9858349126078936e-08,
"logits/chosen": 30.234338760375977,
"logits/rejected": 35.29171371459961,
"logps/chosen": -118.93162536621094,
"logps/rejected": -211.0615692138672,
"loss": 0.0803,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.521851539611816,
"rewards/margins": 6.673744201660156,
"rewards/rejected": -13.195594787597656,
"step": 2090
},
{
"epoch": 0.84,
"grad_norm": 63.74043787739687,
"learning_rate": 3.798797596089351e-08,
"logits/chosen": 30.151935577392578,
"logits/rejected": 33.42070770263672,
"logps/chosen": -126.11729431152344,
"logps/rejected": -205.4882354736328,
"loss": 0.1006,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -7.028811454772949,
"rewards/margins": 6.618474006652832,
"rewards/rejected": -13.647287368774414,
"step": 2100
},
{
"epoch": 0.844,
"grad_norm": 12.990695199065254,
"learning_rate": 3.615893495987335e-08,
"logits/chosen": 30.569061279296875,
"logits/rejected": 33.5135383605957,
"logps/chosen": -118.1374282836914,
"logps/rejected": -183.0607147216797,
"loss": 0.2439,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -6.712668418884277,
"rewards/margins": 5.418641090393066,
"rewards/rejected": -12.131309509277344,
"step": 2110
},
{
"epoch": 0.848,
"grad_norm": 19.041922220697664,
"learning_rate": 3.437158269818563e-08,
"logits/chosen": 30.906530380249023,
"logits/rejected": 33.76823425292969,
"logps/chosen": -128.64764404296875,
"logps/rejected": -212.09213256835938,
"loss": 0.0389,
"rewards/accuracies": 1.0,
"rewards/chosen": -6.712831020355225,
"rewards/margins": 7.552107810974121,
"rewards/rejected": -14.264938354492188,
"step": 2120
},
{
"epoch": 0.852,
"grad_norm": 73.85387160932243,
"learning_rate": 3.262626762369525e-08,
"logits/chosen": 30.180126190185547,
"logits/rejected": 34.350563049316406,
"logps/chosen": -122.08283996582031,
"logps/rejected": -191.43997192382812,
"loss": 0.1989,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -6.999587059020996,
"rewards/margins": 5.256472587585449,
"rewards/rejected": -12.256059646606445,
"step": 2130
},
{
"epoch": 0.856,
"grad_norm": 23.568790751855996,
"learning_rate": 3.092332998903416e-08,
"logits/chosen": 29.4277286529541,
"logits/rejected": 31.689163208007812,
"logps/chosen": -136.5303192138672,
"logps/rejected": -214.65731811523438,
"loss": 0.108,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -8.013432502746582,
"rewards/margins": 6.365548133850098,
"rewards/rejected": -14.378979682922363,
"step": 2140
},
{
"epoch": 0.86,
"grad_norm": 16.09088977053807,
"learning_rate": 2.9263101785268252e-08,
"logits/chosen": 29.914642333984375,
"logits/rejected": 30.405963897705078,
"logps/chosen": -134.8611297607422,
"logps/rejected": -205.1629180908203,
"loss": 0.0879,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -7.248504638671875,
"rewards/margins": 6.40295934677124,
"rewards/rejected": -13.651464462280273,
"step": 2150
},
{
"epoch": 0.864,
"grad_norm": 97.3126034768412,
"learning_rate": 2.764590667717562e-08,
"logits/chosen": 31.892868041992188,
"logits/rejected": 33.220489501953125,
"logps/chosen": -138.26364135742188,
"logps/rejected": -187.60067749023438,
"loss": 0.2452,
"rewards/accuracies": 0.875,
"rewards/chosen": -7.446439266204834,
"rewards/margins": 4.706901550292969,
"rewards/rejected": -12.153340339660645,
"step": 2160
},
{
"epoch": 0.868,
"grad_norm": 34.25127629439165,
"learning_rate": 2.6072059940146772e-08,
"logits/chosen": 30.961380004882812,
"logits/rejected": 32.999900817871094,
"logps/chosen": -123.99442291259766,
"logps/rejected": -199.34922790527344,
"loss": 0.1935,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -6.214547634124756,
"rewards/margins": 6.843503475189209,
"rewards/rejected": -13.058052062988281,
"step": 2170
},
{
"epoch": 0.872,
"grad_norm": 24.468484869974596,
"learning_rate": 2.4541868398721576e-08,
"logits/chosen": 31.01216697692871,
"logits/rejected": 32.20166778564453,
"logps/chosen": -128.03175354003906,
"logps/rejected": -194.68743896484375,
"loss": 0.0858,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -7.191014766693115,
"rewards/margins": 5.623617649078369,
"rewards/rejected": -12.8146333694458,
"step": 2180
},
{
"epoch": 0.876,
"grad_norm": 2.66007251335378,
"learning_rate": 2.3055630366772856e-08,
"logits/chosen": 29.776905059814453,
"logits/rejected": 32.563453674316406,
"logps/chosen": -130.4777069091797,
"logps/rejected": -196.01138305664062,
"loss": 0.1665,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -6.950491428375244,
"rewards/margins": 6.007107734680176,
"rewards/rejected": -12.957598686218262,
"step": 2190
},
{
"epoch": 0.88,
"grad_norm": 28.407188913509714,
"learning_rate": 2.1613635589349756e-08,
"logits/chosen": 31.004446029663086,
"logits/rejected": 31.923709869384766,
"logps/chosen": -129.14511108398438,
"logps/rejected": -189.40194702148438,
"loss": 0.152,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -6.862164497375488,
"rewards/margins": 5.448739051818848,
"rewards/rejected": -12.310903549194336,
"step": 2200
},
{
"epoch": 0.884,
"grad_norm": 27.05189743480488,
"learning_rate": 2.0216165186191404e-08,
"logits/chosen": 29.659343719482422,
"logits/rejected": 32.05767059326172,
"logps/chosen": -132.9698028564453,
"logps/rejected": -204.43704223632812,
"loss": 0.054,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -7.252483367919922,
"rewards/margins": 5.930403709411621,
"rewards/rejected": -13.182887077331543,
"step": 2210
},
{
"epoch": 0.888,
"grad_norm": 55.43405815780065,
"learning_rate": 1.8863491596921743e-08,
"logits/chosen": 32.782066345214844,
"logits/rejected": 32.522891998291016,
"logps/chosen": -136.2987060546875,
"logps/rejected": -198.53738403320312,
"loss": 0.2253,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -7.071662902832031,
"rewards/margins": 6.142470359802246,
"rewards/rejected": -13.214132308959961,
"step": 2220
},
{
"epoch": 0.892,
"grad_norm": 51.20992346555615,
"learning_rate": 1.7555878527937163e-08,
"logits/chosen": 31.446645736694336,
"logits/rejected": 35.24169921875,
"logps/chosen": -118.80924224853516,
"logps/rejected": -208.82534790039062,
"loss": 0.0807,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -6.682692527770996,
"rewards/margins": 6.966187477111816,
"rewards/rejected": -13.648880004882812,
"step": 2230
},
{
"epoch": 0.896,
"grad_norm": 2.6516558135690262,
"learning_rate": 1.629358090099639e-08,
"logits/chosen": 29.057031631469727,
"logits/rejected": 32.2264404296875,
"logps/chosen": -120.19584655761719,
"logps/rejected": -200.8718719482422,
"loss": 0.0942,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -6.863171577453613,
"rewards/margins": 6.121987342834473,
"rewards/rejected": -12.98515796661377,
"step": 2240
},
{
"epoch": 0.9,
"grad_norm": 19.604673890955695,
"learning_rate": 1.507684480352292e-08,
"logits/chosen": 31.421955108642578,
"logits/rejected": 33.76438522338867,
"logps/chosen": -122.68287658691406,
"logps/rejected": -201.4803009033203,
"loss": 0.1018,
"rewards/accuracies": 1.0,
"rewards/chosen": -6.61349630355835,
"rewards/margins": 7.108196258544922,
"rewards/rejected": -13.721692085266113,
"step": 2250
},
{
"epoch": 0.904,
"grad_norm": 18.823513111687696,
"learning_rate": 1.390590744062975e-08,
"logits/chosen": 29.491558074951172,
"logits/rejected": 31.476119995117188,
"logps/chosen": -130.34339904785156,
"logps/rejected": -202.07351684570312,
"loss": 0.1108,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -7.389586448669434,
"rewards/margins": 6.204103469848633,
"rewards/rejected": -13.593690872192383,
"step": 2260
},
{
"epoch": 0.908,
"grad_norm": 41.18098938197995,
"learning_rate": 1.2780997088875866e-08,
"logits/chosen": 32.09846878051758,
"logits/rejected": 34.34507751464844,
"logps/chosen": -122.9142074584961,
"logps/rejected": -186.44857788085938,
"loss": 0.1196,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -6.446104526519775,
"rewards/margins": 5.684818267822266,
"rewards/rejected": -12.130921363830566,
"step": 2270
},
{
"epoch": 0.912,
"grad_norm": 26.449033993731902,
"learning_rate": 1.1702333051763268e-08,
"logits/chosen": 30.299280166625977,
"logits/rejected": 33.13639831542969,
"logps/chosen": -125.92582702636719,
"logps/rejected": -201.4826202392578,
"loss": 0.0914,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -6.8485307693481445,
"rewards/margins": 6.146922588348389,
"rewards/rejected": -12.995452880859375,
"step": 2280
},
{
"epoch": 0.916,
"grad_norm": 3.868894147937223,
"learning_rate": 1.0670125616983189e-08,
"logits/chosen": 31.07822036743164,
"logits/rejected": 33.824302673339844,
"logps/chosen": -133.55230712890625,
"logps/rejected": -204.15899658203125,
"loss": 0.2123,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -7.182616233825684,
"rewards/margins": 5.8420729637146,
"rewards/rejected": -13.024690628051758,
"step": 2290
},
{
"epoch": 0.92,
"grad_norm": 3.3807387885526454,
"learning_rate": 9.684576015420275e-09,
"logits/chosen": 29.672496795654297,
"logits/rejected": 37.18809509277344,
"logps/chosen": -126.36643981933594,
"logps/rejected": -208.7083740234375,
"loss": 0.158,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -7.423619270324707,
"rewards/margins": 5.624678611755371,
"rewards/rejected": -13.048298835754395,
"step": 2300
},
{
"epoch": 0.924,
"grad_norm": 66.31275321980026,
"learning_rate": 8.745876381922146e-09,
"logits/chosen": 30.262380599975586,
"logits/rejected": 33.276100158691406,
"logps/chosen": -124.06925964355469,
"logps/rejected": -195.91195678710938,
"loss": 0.1534,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -6.720442295074463,
"rewards/margins": 6.354052543640137,
"rewards/rejected": -13.074495315551758,
"step": 2310
},
{
"epoch": 0.928,
"grad_norm": 7.064746922185591,
"learning_rate": 7.85420971784223e-09,
"logits/chosen": 31.4277400970459,
"logits/rejected": 30.434642791748047,
"logps/chosen": -133.32081604003906,
"logps/rejected": -188.7599639892578,
"loss": 0.2425,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -7.45855712890625,
"rewards/margins": 5.378756999969482,
"rewards/rejected": -12.837315559387207,
"step": 2320
},
{
"epoch": 0.932,
"grad_norm": 4.788662000330739,
"learning_rate": 7.009749855363456e-09,
"logits/chosen": 30.91952896118164,
"logits/rejected": 32.42715072631836,
"logps/chosen": -133.62342834472656,
"logps/rejected": -188.62930297851562,
"loss": 0.3079,
"rewards/accuracies": 0.875,
"rewards/chosen": -7.070608615875244,
"rewards/margins": 5.031331539154053,
"rewards/rejected": -12.101941108703613,
"step": 2330
},
{
"epoch": 0.936,
"grad_norm": 10.233821689329831,
"learning_rate": 6.2126614236091834e-09,
"logits/chosen": 28.668411254882812,
"logits/rejected": 32.377296447753906,
"logps/chosen": -117.2125244140625,
"logps/rejected": -195.70059204101562,
"loss": 0.0607,
"rewards/accuracies": 1.0,
"rewards/chosen": -6.799081325531006,
"rewards/margins": 6.002202987670898,
"rewards/rejected": -12.801284790039062,
"step": 2340
},
{
"epoch": 0.94,
"grad_norm": 12.346515996086536,
"learning_rate": 5.463099816548577e-09,
"logits/chosen": 31.012216567993164,
"logits/rejected": 35.09211349487305,
"logps/chosen": -144.66915893554688,
"logps/rejected": -219.4803009033203,
"loss": 0.0968,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -7.998269557952881,
"rewards/margins": 5.874829292297363,
"rewards/rejected": -13.873098373413086,
"step": 2350
},
{
"epoch": 0.944,
"grad_norm": 33.1665678427252,
"learning_rate": 4.761211162702117e-09,
"logits/chosen": 29.088903427124023,
"logits/rejected": 34.32123947143555,
"logps/chosen": -127.30374908447266,
"logps/rejected": -211.43215942382812,
"loss": 0.1457,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -7.1538214683532715,
"rewards/margins": 6.435636043548584,
"rewards/rejected": -13.589457511901855,
"step": 2360
},
{
"epoch": 0.948,
"grad_norm": 23.425984241535982,
"learning_rate": 4.107132296653548e-09,
"logits/chosen": 28.269973754882812,
"logits/rejected": 31.716510772705078,
"logps/chosen": -125.16642761230469,
"logps/rejected": -197.10952758789062,
"loss": 0.1364,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -7.0634284019470215,
"rewards/margins": 5.970315933227539,
"rewards/rejected": -13.033744812011719,
"step": 2370
},
{
"epoch": 0.952,
"grad_norm": 10.233047916013119,
"learning_rate": 3.5009907323737818e-09,
"logits/chosen": 32.33917999267578,
"logits/rejected": 31.942947387695312,
"logps/chosen": -150.05201721191406,
"logps/rejected": -198.63037109375,
"loss": 0.1165,
"rewards/accuracies": 1.0,
"rewards/chosen": -7.604508876800537,
"rewards/margins": 5.146544933319092,
"rewards/rejected": -12.751053810119629,
"step": 2380
},
{
"epoch": 0.956,
"grad_norm": 43.68344404125229,
"learning_rate": 2.9429046383618038e-09,
"logits/chosen": 32.400047302246094,
"logits/rejected": 32.68498992919922,
"logps/chosen": -131.60537719726562,
"logps/rejected": -197.2787628173828,
"loss": 0.1184,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -7.099400520324707,
"rewards/margins": 6.27431058883667,
"rewards/rejected": -13.373710632324219,
"step": 2390
},
{
"epoch": 0.96,
"grad_norm": 59.46460968091823,
"learning_rate": 2.4329828146074096e-09,
"logits/chosen": 30.623287200927734,
"logits/rejected": 32.035682678222656,
"logps/chosen": -131.70294189453125,
"logps/rejected": -198.9145050048828,
"loss": 0.099,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -7.071308135986328,
"rewards/margins": 6.146972179412842,
"rewards/rejected": -13.218279838562012,
"step": 2400
},
{
"epoch": 0.964,
"grad_norm": 56.79097277412103,
"learning_rate": 1.9713246713805587e-09,
"logits/chosen": 31.465078353881836,
"logits/rejected": 34.087913513183594,
"logps/chosen": -137.09597778320312,
"logps/rejected": -203.9413604736328,
"loss": 0.1021,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -7.499648094177246,
"rewards/margins": 5.972261905670166,
"rewards/rejected": -13.47191047668457,
"step": 2410
},
{
"epoch": 0.968,
"grad_norm": 0.8585461059925868,
"learning_rate": 1.5580202098509076e-09,
"logits/chosen": 29.60024070739746,
"logits/rejected": 32.122642517089844,
"logps/chosen": -138.75808715820312,
"logps/rejected": -205.38681030273438,
"loss": 0.1414,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -7.920256614685059,
"rewards/margins": 5.451854228973389,
"rewards/rejected": -13.372111320495605,
"step": 2420
},
{
"epoch": 0.972,
"grad_norm": 6.596142925147252,
"learning_rate": 1.1931500045422038e-09,
"logits/chosen": 27.959314346313477,
"logits/rejected": 32.65019226074219,
"logps/chosen": -126.5806655883789,
"logps/rejected": -199.76040649414062,
"loss": 0.2261,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -6.968267917633057,
"rewards/margins": 5.849569797515869,
"rewards/rejected": -12.817837715148926,
"step": 2430
},
{
"epoch": 0.976,
"grad_norm": 35.62534557996734,
"learning_rate": 8.767851876239074e-10,
"logits/chosen": 32.697261810302734,
"logits/rejected": 35.91864013671875,
"logps/chosen": -123.58503723144531,
"logps/rejected": -206.4244842529297,
"loss": 0.1628,
"rewards/accuracies": 1.0,
"rewards/chosen": -6.5748610496521,
"rewards/margins": 6.952831268310547,
"rewards/rejected": -13.527691841125488,
"step": 2440
},
{
"epoch": 0.98,
"grad_norm": 27.658485623502656,
"learning_rate": 6.089874350439505e-10,
"logits/chosen": 31.490856170654297,
"logits/rejected": 32.93403625488281,
"logps/chosen": -138.20703125,
"logps/rejected": -193.5495147705078,
"loss": 0.216,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -7.787785530090332,
"rewards/margins": 4.913016319274902,
"rewards/rejected": -12.700800895690918,
"step": 2450
},
{
"epoch": 0.984,
"grad_norm": 0.08797343919690617,
"learning_rate": 3.898089545047445e-10,
"logits/chosen": 31.071331024169922,
"logits/rejected": 32.911354064941406,
"logps/chosen": -130.91525268554688,
"logps/rejected": -205.3039093017578,
"loss": 0.0549,
"rewards/accuracies": 1.0,
"rewards/chosen": -7.0166826248168945,
"rewards/margins": 6.3437819480896,
"rewards/rejected": -13.360466003417969,
"step": 2460
},
{
"epoch": 0.988,
"grad_norm": 1.2212789051560087,
"learning_rate": 2.1929247528540418e-10,
"logits/chosen": 28.965045928955078,
"logits/rejected": 32.625064849853516,
"logps/chosen": -127.5809326171875,
"logps/rejected": -214.167724609375,
"loss": 0.0704,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -6.630097389221191,
"rewards/margins": 7.368367671966553,
"rewards/rejected": -13.998464584350586,
"step": 2470
},
{
"epoch": 0.992,
"grad_norm": 91.06898861755354,
"learning_rate": 9.747123991141193e-11,
"logits/chosen": 30.026744842529297,
"logits/rejected": 32.432098388671875,
"logps/chosen": -131.2405548095703,
"logps/rejected": -194.84970092773438,
"loss": 0.2616,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -7.285294532775879,
"rewards/margins": 5.499347686767578,
"rewards/rejected": -12.78464126586914,
"step": 2480
},
{
"epoch": 0.996,
"grad_norm": 89.82748805014613,
"learning_rate": 2.4368997673940294e-11,
"logits/chosen": 29.70314598083496,
"logits/rejected": 32.16979217529297,
"logps/chosen": -126.3776626586914,
"logps/rejected": -199.11517333984375,
"loss": 0.174,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -7.552847385406494,
"rewards/margins": 6.022560119628906,
"rewards/rejected": -13.575407028198242,
"step": 2490
},
{
"epoch": 1.0,
"grad_norm": 18.11358569727584,
"learning_rate": 0.0,
"logits/chosen": 30.229150772094727,
"logits/rejected": 34.69512176513672,
"logps/chosen": -133.08348083496094,
"logps/rejected": -207.73001098632812,
"loss": 0.1378,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -7.222729682922363,
"rewards/margins": 5.665229320526123,
"rewards/rejected": -12.887959480285645,
"step": 2500
},
{
"epoch": 1.0,
"step": 2500,
"total_flos": 0.0,
"train_loss": 0.26106402876377105,
"train_runtime": 17697.4864,
"train_samples_per_second": 1.13,
"train_steps_per_second": 0.141
}
],
"logging_steps": 10,
"max_steps": 2500,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}