zephyr-7b-UC-0 / trainer_state.json
weijie210's picture
Model save
e405c5d verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 805,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 6.172839506172839e-09,
"logits/chosen": -2.8421168327331543,
"logits/rejected": -2.6747336387634277,
"logps/chosen": -92.33953094482422,
"logps/rejected": -44.262760162353516,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 6.172839506172839e-08,
"logits/chosen": -2.9818191528320312,
"logits/rejected": -2.9740567207336426,
"logps/chosen": -197.3586883544922,
"logps/rejected": -149.28749084472656,
"loss": 0.6901,
"rewards/accuracies": 0.3611111044883728,
"rewards/chosen": -0.005469343159347773,
"rewards/margins": 0.00850688572973013,
"rewards/rejected": -0.01397622935473919,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 1.2345679012345677e-07,
"logits/chosen": -2.9317967891693115,
"logits/rejected": -2.8763492107391357,
"logps/chosen": -172.7858428955078,
"logps/rejected": -133.58245849609375,
"loss": 0.6353,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": 0.1074294000864029,
"rewards/margins": 0.16630074381828308,
"rewards/rejected": -0.05887135863304138,
"step": 20
},
{
"epoch": 0.04,
"learning_rate": 1.8518518518518516e-07,
"logits/chosen": -3.0448741912841797,
"logits/rejected": -2.999143362045288,
"logps/chosen": -179.38113403320312,
"logps/rejected": -146.7749786376953,
"loss": 0.5047,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": 0.08539465069770813,
"rewards/margins": 0.540686845779419,
"rewards/rejected": -0.4552922248840332,
"step": 30
},
{
"epoch": 0.05,
"learning_rate": 2.4691358024691354e-07,
"logits/chosen": -2.9631247520446777,
"logits/rejected": -2.9552507400512695,
"logps/chosen": -174.05288696289062,
"logps/rejected": -143.798583984375,
"loss": 0.3928,
"rewards/accuracies": 0.875,
"rewards/chosen": 0.11930576711893082,
"rewards/margins": 0.9224799871444702,
"rewards/rejected": -0.8031741976737976,
"step": 40
},
{
"epoch": 0.06,
"learning_rate": 3.086419753086419e-07,
"logits/chosen": -2.955972194671631,
"logits/rejected": -2.880552053451538,
"logps/chosen": -183.1387481689453,
"logps/rejected": -157.20669555664062,
"loss": 0.2956,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -0.8193867802619934,
"rewards/margins": 1.485733985900879,
"rewards/rejected": -2.3051209449768066,
"step": 50
},
{
"epoch": 0.07,
"learning_rate": 3.703703703703703e-07,
"logits/chosen": -2.960036516189575,
"logits/rejected": -2.8513035774230957,
"logps/chosen": -154.39926147460938,
"logps/rejected": -133.40078735351562,
"loss": 0.2919,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -0.9133367538452148,
"rewards/margins": 1.8072048425674438,
"rewards/rejected": -2.720541477203369,
"step": 60
},
{
"epoch": 0.09,
"learning_rate": 4.320987654320987e-07,
"logits/chosen": -2.8771309852600098,
"logits/rejected": -2.797616481781006,
"logps/chosen": -165.77328491210938,
"logps/rejected": -142.69815063476562,
"loss": 0.2689,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.7793342471122742,
"rewards/margins": 2.3448145389556885,
"rewards/rejected": -3.1241488456726074,
"step": 70
},
{
"epoch": 0.1,
"learning_rate": 4.938271604938271e-07,
"logits/chosen": -2.7670953273773193,
"logits/rejected": -2.723829984664917,
"logps/chosen": -185.12596130371094,
"logps/rejected": -174.61465454101562,
"loss": 0.2633,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -1.7798545360565186,
"rewards/margins": 2.6085076332092285,
"rewards/rejected": -4.388362884521484,
"step": 80
},
{
"epoch": 0.11,
"learning_rate": 4.937845303867402e-07,
"logits/chosen": -2.933378219604492,
"logits/rejected": -2.9017395973205566,
"logps/chosen": -183.1439208984375,
"logps/rejected": -171.66964721679688,
"loss": 0.2236,
"rewards/accuracies": 0.875,
"rewards/chosen": -1.4862909317016602,
"rewards/margins": 2.482652187347412,
"rewards/rejected": -3.968942642211914,
"step": 90
},
{
"epoch": 0.12,
"learning_rate": 4.868784530386741e-07,
"logits/chosen": -2.7947914600372314,
"logits/rejected": -2.823068618774414,
"logps/chosen": -189.07904052734375,
"logps/rejected": -181.06781005859375,
"loss": 0.2324,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.5773736238479614,
"rewards/margins": 3.1609904766082764,
"rewards/rejected": -3.738363742828369,
"step": 100
},
{
"epoch": 0.14,
"learning_rate": 4.799723756906077e-07,
"logits/chosen": -2.972888946533203,
"logits/rejected": -2.856055498123169,
"logps/chosen": -194.32847595214844,
"logps/rejected": -180.61575317382812,
"loss": 0.2595,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -1.5956058502197266,
"rewards/margins": 2.7098982334136963,
"rewards/rejected": -4.305504322052002,
"step": 110
},
{
"epoch": 0.15,
"learning_rate": 4.730662983425414e-07,
"logits/chosen": -2.8165183067321777,
"logits/rejected": -2.7924342155456543,
"logps/chosen": -171.48434448242188,
"logps/rejected": -174.36788940429688,
"loss": 0.2177,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -1.069062352180481,
"rewards/margins": 3.2766518592834473,
"rewards/rejected": -4.345714092254639,
"step": 120
},
{
"epoch": 0.16,
"learning_rate": 4.661602209944751e-07,
"logits/chosen": -2.941241502761841,
"logits/rejected": -2.8772006034851074,
"logps/chosen": -206.2366180419922,
"logps/rejected": -196.18931579589844,
"loss": 0.1857,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -1.6614869832992554,
"rewards/margins": 3.376704454421997,
"rewards/rejected": -5.038191318511963,
"step": 130
},
{
"epoch": 0.17,
"learning_rate": 4.592541436464088e-07,
"logits/chosen": -2.7895777225494385,
"logits/rejected": -2.7228431701660156,
"logps/chosen": -185.4031524658203,
"logps/rejected": -193.98040771484375,
"loss": 0.174,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -0.8083890080451965,
"rewards/margins": 4.341578960418701,
"rewards/rejected": -5.149968147277832,
"step": 140
},
{
"epoch": 0.19,
"learning_rate": 4.5234806629834255e-07,
"logits/chosen": -2.73811674118042,
"logits/rejected": -2.7120604515075684,
"logps/chosen": -214.88565063476562,
"logps/rejected": -221.6265411376953,
"loss": 0.185,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -2.5129263401031494,
"rewards/margins": 3.6045918464660645,
"rewards/rejected": -6.117517948150635,
"step": 150
},
{
"epoch": 0.2,
"learning_rate": 4.454419889502762e-07,
"logits/chosen": -2.9045376777648926,
"logits/rejected": -2.8451740741729736,
"logps/chosen": -202.41903686523438,
"logps/rejected": -205.6617889404297,
"loss": 0.208,
"rewards/accuracies": 0.75,
"rewards/chosen": -2.077017307281494,
"rewards/margins": 4.358880996704102,
"rewards/rejected": -6.4358978271484375,
"step": 160
},
{
"epoch": 0.21,
"learning_rate": 4.3853591160220993e-07,
"logits/chosen": -2.8927552700042725,
"logits/rejected": -2.7776336669921875,
"logps/chosen": -203.76724243164062,
"logps/rejected": -200.79864501953125,
"loss": 0.1703,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -2.2595643997192383,
"rewards/margins": 3.9248855113983154,
"rewards/rejected": -6.184449672698975,
"step": 170
},
{
"epoch": 0.22,
"learning_rate": 4.3162983425414365e-07,
"logits/chosen": -2.7550511360168457,
"logits/rejected": -2.7225847244262695,
"logps/chosen": -225.6936798095703,
"logps/rejected": -231.35400390625,
"loss": 0.1733,
"rewards/accuracies": 0.75,
"rewards/chosen": -3.379578113555908,
"rewards/margins": 3.530996322631836,
"rewards/rejected": -6.910574436187744,
"step": 180
},
{
"epoch": 0.24,
"learning_rate": 4.247237569060773e-07,
"logits/chosen": -2.8363826274871826,
"logits/rejected": -2.7442939281463623,
"logps/chosen": -212.888427734375,
"logps/rejected": -198.71542358398438,
"loss": 0.218,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -2.968935251235962,
"rewards/margins": 3.221897602081299,
"rewards/rejected": -6.19083309173584,
"step": 190
},
{
"epoch": 0.25,
"learning_rate": 4.1781767955801103e-07,
"logits/chosen": -2.8280460834503174,
"logits/rejected": -2.7499425411224365,
"logps/chosen": -182.047607421875,
"logps/rejected": -188.87193298339844,
"loss": 0.1708,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -2.369425058364868,
"rewards/margins": 4.022088050842285,
"rewards/rejected": -6.391513824462891,
"step": 200
},
{
"epoch": 0.26,
"learning_rate": 4.1091160220994475e-07,
"logits/chosen": -2.7316782474517822,
"logits/rejected": -2.6802334785461426,
"logps/chosen": -212.96212768554688,
"logps/rejected": -203.3765106201172,
"loss": 0.1689,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -3.15626859664917,
"rewards/margins": 3.610233783721924,
"rewards/rejected": -6.766502380371094,
"step": 210
},
{
"epoch": 0.27,
"learning_rate": 4.0400552486187846e-07,
"logits/chosen": -2.8573451042175293,
"logits/rejected": -2.751213550567627,
"logps/chosen": -225.369140625,
"logps/rejected": -213.1463623046875,
"loss": 0.2091,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -3.183094024658203,
"rewards/margins": 3.5506489276885986,
"rewards/rejected": -6.733743190765381,
"step": 220
},
{
"epoch": 0.29,
"learning_rate": 3.970994475138121e-07,
"logits/chosen": -2.7918601036071777,
"logits/rejected": -2.731823444366455,
"logps/chosen": -194.12893676757812,
"logps/rejected": -183.81556701660156,
"loss": 0.1979,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -2.6381051540374756,
"rewards/margins": 2.9515299797058105,
"rewards/rejected": -5.589634895324707,
"step": 230
},
{
"epoch": 0.3,
"learning_rate": 3.9019337016574584e-07,
"logits/chosen": -2.699711561203003,
"logits/rejected": -2.6905601024627686,
"logps/chosen": -212.59793090820312,
"logps/rejected": -232.6545867919922,
"loss": 0.1849,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -3.5139076709747314,
"rewards/margins": 4.449800491333008,
"rewards/rejected": -7.96370792388916,
"step": 240
},
{
"epoch": 0.31,
"learning_rate": 3.832872928176795e-07,
"logits/chosen": -2.721381425857544,
"logits/rejected": -2.7046854496002197,
"logps/chosen": -200.67886352539062,
"logps/rejected": -220.4629669189453,
"loss": 0.1447,
"rewards/accuracies": 0.875,
"rewards/chosen": -3.1384024620056152,
"rewards/margins": 4.899154186248779,
"rewards/rejected": -8.037556648254395,
"step": 250
},
{
"epoch": 0.32,
"learning_rate": 3.763812154696133e-07,
"logits/chosen": -2.8270747661590576,
"logits/rejected": -2.741473913192749,
"logps/chosen": -208.8101043701172,
"logps/rejected": -218.16537475585938,
"loss": 0.1567,
"rewards/accuracies": 0.875,
"rewards/chosen": -3.2333245277404785,
"rewards/margins": 4.820733547210693,
"rewards/rejected": -8.054059028625488,
"step": 260
},
{
"epoch": 0.34,
"learning_rate": 3.6947513812154694e-07,
"logits/chosen": -2.8693325519561768,
"logits/rejected": -2.7638630867004395,
"logps/chosen": -211.0121612548828,
"logps/rejected": -217.35910034179688,
"loss": 0.1731,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -3.540541410446167,
"rewards/margins": 4.472687721252441,
"rewards/rejected": -8.013228416442871,
"step": 270
},
{
"epoch": 0.35,
"learning_rate": 3.6256906077348066e-07,
"logits/chosen": -2.7264175415039062,
"logits/rejected": -2.7106542587280273,
"logps/chosen": -221.5767822265625,
"logps/rejected": -234.4220733642578,
"loss": 0.2261,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -3.8089592456817627,
"rewards/margins": 4.285206317901611,
"rewards/rejected": -8.094165802001953,
"step": 280
},
{
"epoch": 0.36,
"learning_rate": 3.556629834254143e-07,
"logits/chosen": -2.709073305130005,
"logits/rejected": -2.650305986404419,
"logps/chosen": -183.47171020507812,
"logps/rejected": -189.16053771972656,
"loss": 0.1949,
"rewards/accuracies": 0.75,
"rewards/chosen": -3.39518666267395,
"rewards/margins": 3.255284070968628,
"rewards/rejected": -6.6504716873168945,
"step": 290
},
{
"epoch": 0.37,
"learning_rate": 3.4875690607734804e-07,
"logits/chosen": -2.742584228515625,
"logits/rejected": -2.634887456893921,
"logps/chosen": -212.4958953857422,
"logps/rejected": -214.77978515625,
"loss": 0.1577,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -2.987830400466919,
"rewards/margins": 4.540104389190674,
"rewards/rejected": -7.527935028076172,
"step": 300
},
{
"epoch": 0.39,
"learning_rate": 3.418508287292817e-07,
"logits/chosen": -2.803213119506836,
"logits/rejected": -2.705699920654297,
"logps/chosen": -201.23739624023438,
"logps/rejected": -210.5272979736328,
"loss": 0.1819,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -2.2409796714782715,
"rewards/margins": 4.629876136779785,
"rewards/rejected": -6.870855808258057,
"step": 310
},
{
"epoch": 0.4,
"learning_rate": 3.3494475138121547e-07,
"logits/chosen": -2.821776866912842,
"logits/rejected": -2.6838371753692627,
"logps/chosen": -205.13998413085938,
"logps/rejected": -210.2687225341797,
"loss": 0.1284,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -1.8670909404754639,
"rewards/margins": 5.2015180587768555,
"rewards/rejected": -7.068609714508057,
"step": 320
},
{
"epoch": 0.41,
"learning_rate": 3.280386740331492e-07,
"logits/chosen": -2.717777729034424,
"logits/rejected": -2.6688733100891113,
"logps/chosen": -181.1573028564453,
"logps/rejected": -195.77450561523438,
"loss": 0.2068,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -2.757702589035034,
"rewards/margins": 4.233901023864746,
"rewards/rejected": -6.991603851318359,
"step": 330
},
{
"epoch": 0.42,
"learning_rate": 3.2113259668508285e-07,
"logits/chosen": -2.834904193878174,
"logits/rejected": -2.751018524169922,
"logps/chosen": -236.5355224609375,
"logps/rejected": -249.8627166748047,
"loss": 0.1873,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -4.415268898010254,
"rewards/margins": 4.573099136352539,
"rewards/rejected": -8.988368034362793,
"step": 340
},
{
"epoch": 0.43,
"learning_rate": 3.1422651933701657e-07,
"logits/chosen": -2.702388048171997,
"logits/rejected": -2.5783984661102295,
"logps/chosen": -205.3250274658203,
"logps/rejected": -226.51025390625,
"loss": 0.1513,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -3.3364059925079346,
"rewards/margins": 5.607662200927734,
"rewards/rejected": -8.94406795501709,
"step": 350
},
{
"epoch": 0.45,
"learning_rate": 3.0732044198895023e-07,
"logits/chosen": -2.719581127166748,
"logits/rejected": -2.6670548915863037,
"logps/chosen": -215.56069946289062,
"logps/rejected": -238.80691528320312,
"loss": 0.1319,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -3.7043700218200684,
"rewards/margins": 5.375087261199951,
"rewards/rejected": -9.07945728302002,
"step": 360
},
{
"epoch": 0.46,
"learning_rate": 3.00414364640884e-07,
"logits/chosen": -2.687743902206421,
"logits/rejected": -2.534635305404663,
"logps/chosen": -197.8555145263672,
"logps/rejected": -212.0418243408203,
"loss": 0.1153,
"rewards/accuracies": 0.875,
"rewards/chosen": -2.731755495071411,
"rewards/margins": 6.006522178649902,
"rewards/rejected": -8.738277435302734,
"step": 370
},
{
"epoch": 0.47,
"learning_rate": 2.9350828729281767e-07,
"logits/chosen": -2.8031535148620605,
"logits/rejected": -2.7039589881896973,
"logps/chosen": -196.32485961914062,
"logps/rejected": -220.0038604736328,
"loss": 0.1482,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -3.087531805038452,
"rewards/margins": 5.792913436889648,
"rewards/rejected": -8.88044548034668,
"step": 380
},
{
"epoch": 0.48,
"learning_rate": 2.866022099447514e-07,
"logits/chosen": -2.762528657913208,
"logits/rejected": -2.7139079570770264,
"logps/chosen": -245.5567169189453,
"logps/rejected": -268.78729248046875,
"loss": 0.1335,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -3.6741371154785156,
"rewards/margins": 6.558190822601318,
"rewards/rejected": -10.232328414916992,
"step": 390
},
{
"epoch": 0.5,
"learning_rate": 2.7969613259668505e-07,
"logits/chosen": -2.8232762813568115,
"logits/rejected": -2.6880855560302734,
"logps/chosen": -210.9452362060547,
"logps/rejected": -228.576416015625,
"loss": 0.1482,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -3.587395191192627,
"rewards/margins": 6.285602569580078,
"rewards/rejected": -9.87299919128418,
"step": 400
},
{
"epoch": 0.51,
"learning_rate": 2.7279005524861877e-07,
"logits/chosen": -2.968181610107422,
"logits/rejected": -2.783268690109253,
"logps/chosen": -250.2484893798828,
"logps/rejected": -248.7700653076172,
"loss": 0.16,
"rewards/accuracies": 0.875,
"rewards/chosen": -4.302699565887451,
"rewards/margins": 5.751161098480225,
"rewards/rejected": -10.053861618041992,
"step": 410
},
{
"epoch": 0.52,
"learning_rate": 2.6588397790055243e-07,
"logits/chosen": -2.843967914581299,
"logits/rejected": -2.7779390811920166,
"logps/chosen": -220.41262817382812,
"logps/rejected": -247.0951690673828,
"loss": 0.131,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -3.6894760131835938,
"rewards/margins": 5.2908501625061035,
"rewards/rejected": -8.980325698852539,
"step": 420
},
{
"epoch": 0.53,
"learning_rate": 2.589779005524862e-07,
"logits/chosen": -2.895799398422241,
"logits/rejected": -2.795949935913086,
"logps/chosen": -243.14785766601562,
"logps/rejected": -262.537109375,
"loss": 0.1558,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -4.353732109069824,
"rewards/margins": 6.2553300857543945,
"rewards/rejected": -10.609061241149902,
"step": 430
},
{
"epoch": 0.55,
"learning_rate": 2.5207182320441986e-07,
"logits/chosen": -2.8478636741638184,
"logits/rejected": -2.782703399658203,
"logps/chosen": -251.97689819335938,
"logps/rejected": -279.1159362792969,
"loss": 0.1489,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -4.981522083282471,
"rewards/margins": 6.379393100738525,
"rewards/rejected": -11.360913276672363,
"step": 440
},
{
"epoch": 0.56,
"learning_rate": 2.451657458563536e-07,
"logits/chosen": -2.9137637615203857,
"logits/rejected": -2.846818447113037,
"logps/chosen": -231.48171997070312,
"logps/rejected": -252.91494750976562,
"loss": 0.1211,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -4.0656819343566895,
"rewards/margins": 6.312979221343994,
"rewards/rejected": -10.378661155700684,
"step": 450
},
{
"epoch": 0.57,
"learning_rate": 2.3825966850828727e-07,
"logits/chosen": -2.820435047149658,
"logits/rejected": -2.7222158908843994,
"logps/chosen": -176.95230102539062,
"logps/rejected": -202.005859375,
"loss": 0.1487,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -1.8066993951797485,
"rewards/margins": 6.806242942810059,
"rewards/rejected": -8.612942695617676,
"step": 460
},
{
"epoch": 0.58,
"learning_rate": 2.31353591160221e-07,
"logits/chosen": -2.861485004425049,
"logits/rejected": -2.7676100730895996,
"logps/chosen": -223.93276977539062,
"logps/rejected": -244.85665893554688,
"loss": 0.1293,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -3.4348251819610596,
"rewards/margins": 6.165853023529053,
"rewards/rejected": -9.600679397583008,
"step": 470
},
{
"epoch": 0.6,
"learning_rate": 2.2444751381215468e-07,
"logits/chosen": -2.933964967727661,
"logits/rejected": -2.788942337036133,
"logps/chosen": -226.66796875,
"logps/rejected": -244.5904998779297,
"loss": 0.1224,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -3.5399329662323,
"rewards/margins": 5.981089115142822,
"rewards/rejected": -9.521021842956543,
"step": 480
},
{
"epoch": 0.61,
"learning_rate": 2.175414364640884e-07,
"logits/chosen": -2.798732280731201,
"logits/rejected": -2.7525930404663086,
"logps/chosen": -224.1324462890625,
"logps/rejected": -260.4334411621094,
"loss": 0.1676,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -3.89032244682312,
"rewards/margins": 7.569252967834473,
"rewards/rejected": -11.459574699401855,
"step": 490
},
{
"epoch": 0.62,
"learning_rate": 2.1063535911602208e-07,
"logits/chosen": -2.837965726852417,
"logits/rejected": -2.7505483627319336,
"logps/chosen": -236.68588256835938,
"logps/rejected": -258.0523681640625,
"loss": 0.1537,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -5.027956008911133,
"rewards/margins": 6.000524997711182,
"rewards/rejected": -11.028480529785156,
"step": 500
},
{
"epoch": 0.62,
"eval_logits/chosen": -2.780914545059204,
"eval_logits/rejected": -2.70768666267395,
"eval_logps/chosen": -219.7342071533203,
"eval_logps/rejected": -246.59507751464844,
"eval_loss": 0.1479674130678177,
"eval_rewards/accuracies": 0.8563829660415649,
"eval_rewards/chosen": -3.757824182510376,
"eval_rewards/margins": 6.220169544219971,
"eval_rewards/rejected": -9.97799301147461,
"eval_runtime": 240.8343,
"eval_samples_per_second": 6.208,
"eval_steps_per_second": 0.195,
"step": 500
},
{
"epoch": 0.63,
"learning_rate": 2.0372928176795578e-07,
"logits/chosen": -2.8163836002349854,
"logits/rejected": -2.7235684394836426,
"logps/chosen": -206.2197265625,
"logps/rejected": -240.59561157226562,
"loss": 0.1149,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -3.1149652004241943,
"rewards/margins": 7.466650485992432,
"rewards/rejected": -10.58161449432373,
"step": 510
},
{
"epoch": 0.65,
"learning_rate": 1.968232044198895e-07,
"logits/chosen": -2.895784854888916,
"logits/rejected": -2.8176522254943848,
"logps/chosen": -211.858642578125,
"logps/rejected": -227.81741333007812,
"loss": 0.1775,
"rewards/accuracies": 0.875,
"rewards/chosen": -4.783520698547363,
"rewards/margins": 4.954631805419922,
"rewards/rejected": -9.738151550292969,
"step": 520
},
{
"epoch": 0.66,
"learning_rate": 1.899171270718232e-07,
"logits/chosen": -2.9654347896575928,
"logits/rejected": -2.8510003089904785,
"logps/chosen": -215.161376953125,
"logps/rejected": -225.93490600585938,
"loss": 0.1701,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -3.928018093109131,
"rewards/margins": 5.275222301483154,
"rewards/rejected": -9.203241348266602,
"step": 530
},
{
"epoch": 0.67,
"learning_rate": 1.830110497237569e-07,
"logits/chosen": -2.834559202194214,
"logits/rejected": -2.7768056392669678,
"logps/chosen": -248.0672149658203,
"logps/rejected": -277.980224609375,
"loss": 0.1456,
"rewards/accuracies": 0.875,
"rewards/chosen": -4.741249084472656,
"rewards/margins": 5.934549331665039,
"rewards/rejected": -10.675798416137695,
"step": 540
},
{
"epoch": 0.68,
"learning_rate": 1.7610497237569062e-07,
"logits/chosen": -2.878166913986206,
"logits/rejected": -2.8250679969787598,
"logps/chosen": -214.7487030029297,
"logps/rejected": -249.8466339111328,
"loss": 0.1641,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -3.848902940750122,
"rewards/margins": 6.1274847984313965,
"rewards/rejected": -9.976387023925781,
"step": 550
},
{
"epoch": 0.7,
"learning_rate": 1.691988950276243e-07,
"logits/chosen": -2.9045345783233643,
"logits/rejected": -2.85447096824646,
"logps/chosen": -237.44955444335938,
"logps/rejected": -270.6697998046875,
"loss": 0.1439,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -4.506590843200684,
"rewards/margins": 6.747040748596191,
"rewards/rejected": -11.253631591796875,
"step": 560
},
{
"epoch": 0.71,
"learning_rate": 1.62292817679558e-07,
"logits/chosen": -2.8267338275909424,
"logits/rejected": -2.673027515411377,
"logps/chosen": -219.37026977539062,
"logps/rejected": -235.9922332763672,
"loss": 0.1133,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -4.315529823303223,
"rewards/margins": 6.053648471832275,
"rewards/rejected": -10.369178771972656,
"step": 570
},
{
"epoch": 0.72,
"learning_rate": 1.5538674033149171e-07,
"logits/chosen": -2.883451223373413,
"logits/rejected": -2.7439939975738525,
"logps/chosen": -228.4202117919922,
"logps/rejected": -242.2844696044922,
"loss": 0.1627,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -4.730469226837158,
"rewards/margins": 6.399707317352295,
"rewards/rejected": -11.130178451538086,
"step": 580
},
{
"epoch": 0.73,
"learning_rate": 1.484806629834254e-07,
"logits/chosen": -2.9942538738250732,
"logits/rejected": -2.868765115737915,
"logps/chosen": -260.10406494140625,
"logps/rejected": -253.8542938232422,
"loss": 0.1594,
"rewards/accuracies": 0.875,
"rewards/chosen": -5.60601282119751,
"rewards/margins": 4.876471519470215,
"rewards/rejected": -10.482483863830566,
"step": 590
},
{
"epoch": 0.75,
"learning_rate": 1.4157458563535912e-07,
"logits/chosen": -2.9481163024902344,
"logits/rejected": -2.8333523273468018,
"logps/chosen": -226.2001953125,
"logps/rejected": -235.05014038085938,
"loss": 0.1784,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -5.186473846435547,
"rewards/margins": 4.799227714538574,
"rewards/rejected": -9.985700607299805,
"step": 600
},
{
"epoch": 0.76,
"learning_rate": 1.346685082872928e-07,
"logits/chosen": -2.820510149002075,
"logits/rejected": -2.734377384185791,
"logps/chosen": -261.7950134277344,
"logps/rejected": -286.0494689941406,
"loss": 0.104,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -5.037507057189941,
"rewards/margins": 6.992823600769043,
"rewards/rejected": -12.030329704284668,
"step": 610
},
{
"epoch": 0.77,
"learning_rate": 1.277624309392265e-07,
"logits/chosen": -2.909327507019043,
"logits/rejected": -2.750497817993164,
"logps/chosen": -224.2764129638672,
"logps/rejected": -231.7129364013672,
"loss": 0.1504,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -5.132498741149902,
"rewards/margins": 5.400217056274414,
"rewards/rejected": -10.532715797424316,
"step": 620
},
{
"epoch": 0.78,
"learning_rate": 1.2085635359116022e-07,
"logits/chosen": -2.9544167518615723,
"logits/rejected": -2.8220763206481934,
"logps/chosen": -245.1588134765625,
"logps/rejected": -261.6687927246094,
"loss": 0.1255,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -5.218487739562988,
"rewards/margins": 6.255653381347656,
"rewards/rejected": -11.474142074584961,
"step": 630
},
{
"epoch": 0.8,
"learning_rate": 1.1395027624309392e-07,
"logits/chosen": -2.8086042404174805,
"logits/rejected": -2.785409927368164,
"logps/chosen": -253.0203857421875,
"logps/rejected": -281.2375793457031,
"loss": 0.16,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -6.011744976043701,
"rewards/margins": 5.636991024017334,
"rewards/rejected": -11.648736000061035,
"step": 640
},
{
"epoch": 0.81,
"learning_rate": 1.0704419889502763e-07,
"logits/chosen": -2.9435603618621826,
"logits/rejected": -2.7380692958831787,
"logps/chosen": -261.493896484375,
"logps/rejected": -278.2936096191406,
"loss": 0.1174,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -4.648413181304932,
"rewards/margins": 7.05194616317749,
"rewards/rejected": -11.700358390808105,
"step": 650
},
{
"epoch": 0.82,
"learning_rate": 1.0013812154696132e-07,
"logits/chosen": -2.9581801891326904,
"logits/rejected": -2.8110265731811523,
"logps/chosen": -244.45413208007812,
"logps/rejected": -264.9123840332031,
"loss": 0.1429,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -5.337543487548828,
"rewards/margins": 6.183696746826172,
"rewards/rejected": -11.521239280700684,
"step": 660
},
{
"epoch": 0.83,
"learning_rate": 9.323204419889502e-08,
"logits/chosen": -2.905041217803955,
"logits/rejected": -2.7901289463043213,
"logps/chosen": -200.6727752685547,
"logps/rejected": -228.93032836914062,
"loss": 0.1408,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -2.679917812347412,
"rewards/margins": 6.9505295753479,
"rewards/rejected": -9.630447387695312,
"step": 670
},
{
"epoch": 0.84,
"learning_rate": 8.632596685082872e-08,
"logits/chosen": -2.856194257736206,
"logits/rejected": -2.761082410812378,
"logps/chosen": -218.3007049560547,
"logps/rejected": -234.4198455810547,
"loss": 0.1398,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -3.6605896949768066,
"rewards/margins": 6.374427795410156,
"rewards/rejected": -10.035017013549805,
"step": 680
},
{
"epoch": 0.86,
"learning_rate": 7.941988950276243e-08,
"logits/chosen": -2.896955966949463,
"logits/rejected": -2.786186933517456,
"logps/chosen": -197.8193817138672,
"logps/rejected": -214.23001098632812,
"loss": 0.1442,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -4.441520690917969,
"rewards/margins": 5.363109588623047,
"rewards/rejected": -9.804631233215332,
"step": 690
},
{
"epoch": 0.87,
"learning_rate": 7.251381215469612e-08,
"logits/chosen": -2.957908868789673,
"logits/rejected": -2.8377201557159424,
"logps/chosen": -270.6792907714844,
"logps/rejected": -286.62213134765625,
"loss": 0.1486,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -6.179241180419922,
"rewards/margins": 6.042364597320557,
"rewards/rejected": -12.22160530090332,
"step": 700
},
{
"epoch": 0.88,
"learning_rate": 6.560773480662984e-08,
"logits/chosen": -2.774834156036377,
"logits/rejected": -2.703329563140869,
"logps/chosen": -202.3549041748047,
"logps/rejected": -244.16653442382812,
"loss": 0.1335,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -4.156593322753906,
"rewards/margins": 6.949100494384766,
"rewards/rejected": -11.105693817138672,
"step": 710
},
{
"epoch": 0.89,
"learning_rate": 5.870165745856354e-08,
"logits/chosen": -2.8945531845092773,
"logits/rejected": -2.7444348335266113,
"logps/chosen": -190.70558166503906,
"logps/rejected": -232.6694793701172,
"loss": 0.1084,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -2.4744460582733154,
"rewards/margins": 8.159748077392578,
"rewards/rejected": -10.634195327758789,
"step": 720
},
{
"epoch": 0.91,
"learning_rate": 5.1795580110497236e-08,
"logits/chosen": -2.9436967372894287,
"logits/rejected": -2.8356709480285645,
"logps/chosen": -236.69949340820312,
"logps/rejected": -253.46145629882812,
"loss": 0.1855,
"rewards/accuracies": 0.875,
"rewards/chosen": -4.680401802062988,
"rewards/margins": 6.0413007736206055,
"rewards/rejected": -10.72170352935791,
"step": 730
},
{
"epoch": 0.92,
"learning_rate": 4.488950276243094e-08,
"logits/chosen": -2.7878644466400146,
"logits/rejected": -2.745734930038452,
"logps/chosen": -197.5469970703125,
"logps/rejected": -246.12252807617188,
"loss": 0.1106,
"rewards/accuracies": 0.875,
"rewards/chosen": -3.669739246368408,
"rewards/margins": 6.831077575683594,
"rewards/rejected": -10.500818252563477,
"step": 740
},
{
"epoch": 0.93,
"learning_rate": 3.7983425414364637e-08,
"logits/chosen": -2.918865203857422,
"logits/rejected": -2.7714521884918213,
"logps/chosen": -224.0564727783203,
"logps/rejected": -241.874267578125,
"loss": 0.1357,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -4.486175537109375,
"rewards/margins": 6.323982238769531,
"rewards/rejected": -10.810157775878906,
"step": 750
},
{
"epoch": 0.94,
"learning_rate": 3.107734806629834e-08,
"logits/chosen": -2.789961338043213,
"logits/rejected": -2.702083110809326,
"logps/chosen": -197.3814697265625,
"logps/rejected": -234.6428680419922,
"loss": 0.1653,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -3.8847968578338623,
"rewards/margins": 6.971263885498047,
"rewards/rejected": -10.856060981750488,
"step": 760
},
{
"epoch": 0.96,
"learning_rate": 2.4171270718232044e-08,
"logits/chosen": -2.886162042617798,
"logits/rejected": -2.782167673110962,
"logps/chosen": -251.7931365966797,
"logps/rejected": -276.71856689453125,
"loss": 0.1811,
"rewards/accuracies": 0.75,
"rewards/chosen": -5.165857791900635,
"rewards/margins": 6.922626495361328,
"rewards/rejected": -12.088483810424805,
"step": 770
},
{
"epoch": 0.97,
"learning_rate": 1.7265193370165747e-08,
"logits/chosen": -2.998176336288452,
"logits/rejected": -2.831453800201416,
"logps/chosen": -238.40634155273438,
"logps/rejected": -264.1795349121094,
"loss": 0.1153,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -3.043548345565796,
"rewards/margins": 8.052096366882324,
"rewards/rejected": -11.0956449508667,
"step": 780
},
{
"epoch": 0.98,
"learning_rate": 1.0359116022099446e-08,
"logits/chosen": -2.931680202484131,
"logits/rejected": -2.758117198944092,
"logps/chosen": -215.9856719970703,
"logps/rejected": -238.5784912109375,
"loss": 0.133,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -4.6980485916137695,
"rewards/margins": 6.217514991760254,
"rewards/rejected": -10.915563583374023,
"step": 790
},
{
"epoch": 0.99,
"learning_rate": 3.453038674033149e-09,
"logits/chosen": -2.8354744911193848,
"logits/rejected": -2.7321650981903076,
"logps/chosen": -240.4857940673828,
"logps/rejected": -271.0428771972656,
"loss": 0.1495,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -3.8582940101623535,
"rewards/margins": 7.768431186676025,
"rewards/rejected": -11.626726150512695,
"step": 800
},
{
"epoch": 1.0,
"step": 805,
"total_flos": 0.0,
"train_loss": 0.18541636852003773,
"train_runtime": 4261.3779,
"train_samples_per_second": 3.022,
"train_steps_per_second": 0.189
}
],
"logging_steps": 10,
"max_steps": 805,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 500,
"total_flos": 0.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}