zephyr-7b-UC-0-B0.5 / trainer_state.json
weijie210's picture
Model save
7b0bc3a 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": -2.9818074703216553,
"logits/rejected": -2.9741735458374023,
"logps/chosen": -197.26390075683594,
"logps/rejected": -149.23976135253906,
"loss": 0.6794,
"rewards/accuracies": 0.5277777910232544,
"rewards/chosen": 0.020052114501595497,
"rewards/margins": 0.0660661831498146,
"rewards/rejected": -0.04601407051086426,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 1.2345679012345677e-07,
"logits/chosen": -2.9329466819763184,
"logits/rejected": -2.8775200843811035,
"logps/chosen": -172.8914337158203,
"logps/rejected": -133.22866821289062,
"loss": 0.5158,
"rewards/accuracies": 0.875,
"rewards/chosen": 0.48435258865356445,
"rewards/margins": 0.6018184423446655,
"rewards/rejected": -0.11746580898761749,
"step": 20
},
{
"epoch": 0.04,
"learning_rate": 1.8518518518518516e-07,
"logits/chosen": -3.0486252307891846,
"logits/rejected": -3.003458023071289,
"logps/chosen": -178.92950439453125,
"logps/rejected": -143.7101287841797,
"loss": 0.3374,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": 0.6527904272079468,
"rewards/margins": 1.396834135055542,
"rewards/rejected": -0.7440436482429504,
"step": 30
},
{
"epoch": 0.05,
"learning_rate": 2.4691358024691354e-07,
"logits/chosen": -2.970651388168335,
"logits/rejected": -2.966358184814453,
"logps/chosen": -174.35009765625,
"logps/rejected": -139.04750061035156,
"loss": 0.2663,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": 0.44792431592941284,
"rewards/margins": 2.088245391845703,
"rewards/rejected": -1.6403213739395142,
"step": 40
},
{
"epoch": 0.06,
"learning_rate": 3.086419753086419e-07,
"logits/chosen": -2.9904656410217285,
"logits/rejected": -2.920414686203003,
"logps/chosen": -173.70179748535156,
"logps/rejected": -138.68955993652344,
"loss": 0.1817,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": 0.6215466856956482,
"rewards/margins": 2.8885910511016846,
"rewards/rejected": -2.2670440673828125,
"step": 50
},
{
"epoch": 0.07,
"learning_rate": 3.703703703703703e-07,
"logits/chosen": -3.000448703765869,
"logits/rejected": -2.8979151248931885,
"logps/chosen": -145.14614868164062,
"logps/rejected": -112.61439514160156,
"loss": 0.1912,
"rewards/accuracies": 0.875,
"rewards/chosen": 0.05986971780657768,
"rewards/margins": 3.269381046295166,
"rewards/rejected": -3.2095108032226562,
"step": 60
},
{
"epoch": 0.09,
"learning_rate": 4.320987654320987e-07,
"logits/chosen": -2.9225759506225586,
"logits/rejected": -2.8518130779266357,
"logps/chosen": -157.19439697265625,
"logps/rejected": -119.10911560058594,
"loss": 0.1713,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": 0.39277735352516174,
"rewards/margins": 4.21900749206543,
"rewards/rejected": -3.82623028755188,
"step": 70
},
{
"epoch": 0.1,
"learning_rate": 4.938271604938271e-07,
"logits/chosen": -2.842589855194092,
"logits/rejected": -2.805922031402588,
"logps/chosen": -168.32815551757812,
"logps/rejected": -141.83216857910156,
"loss": 0.1762,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -0.5003746151924133,
"rewards/margins": 5.050185203552246,
"rewards/rejected": -5.550559043884277,
"step": 80
},
{
"epoch": 0.11,
"learning_rate": 4.937845303867402e-07,
"logits/chosen": -2.9922776222229004,
"logits/rejected": -2.970557928085327,
"logps/chosen": -172.38787841796875,
"logps/rejected": -144.65328979492188,
"loss": 0.144,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -2.053427219390869,
"rewards/margins": 4.2831034660339355,
"rewards/rejected": -6.336531639099121,
"step": 90
},
{
"epoch": 0.12,
"learning_rate": 4.868784530386741e-07,
"logits/chosen": -2.897087574005127,
"logits/rejected": -2.9255499839782715,
"logps/chosen": -183.1656951904297,
"logps/rejected": -155.22946166992188,
"loss": 0.1584,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": 0.06980714946985245,
"rewards/margins": 5.842454433441162,
"rewards/rejected": -5.772646903991699,
"step": 100
},
{
"epoch": 0.14,
"learning_rate": 4.799723756906077e-07,
"logits/chosen": -3.016488790512085,
"logits/rejected": -2.9143893718719482,
"logps/chosen": -182.10484313964844,
"logps/rejected": -151.207275390625,
"loss": 0.185,
"rewards/accuracies": 0.875,
"rewards/chosen": -1.8662151098251343,
"rewards/margins": 4.957070350646973,
"rewards/rejected": -6.823284149169922,
"step": 110
},
{
"epoch": 0.15,
"learning_rate": 4.730662983425414e-07,
"logits/chosen": -2.906358242034912,
"logits/rejected": -2.8886070251464844,
"logps/chosen": -162.4820556640625,
"logps/rejected": -144.28274536132812,
"loss": 0.2087,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -0.8441675901412964,
"rewards/margins": 5.8418402671813965,
"rewards/rejected": -6.686008453369141,
"step": 120
},
{
"epoch": 0.16,
"learning_rate": 4.661602209944751e-07,
"logits/chosen": -3.044167995452881,
"logits/rejected": -2.989940881729126,
"logps/chosen": -192.1724090576172,
"logps/rejected": -160.28482055664062,
"loss": 0.1281,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -1.2753368616104126,
"rewards/margins": 5.963367462158203,
"rewards/rejected": -7.238703727722168,
"step": 130
},
{
"epoch": 0.17,
"learning_rate": 4.592541436464088e-07,
"logits/chosen": -2.9158730506896973,
"logits/rejected": -2.8665575981140137,
"logps/chosen": -176.7301788330078,
"logps/rejected": -157.2458038330078,
"loss": 0.121,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": 0.29454001784324646,
"rewards/margins": 7.677084445953369,
"rewards/rejected": -7.38254451751709,
"step": 140
},
{
"epoch": 0.19,
"learning_rate": 4.5234806629834255e-07,
"logits/chosen": -2.906966209411621,
"logits/rejected": -2.878993034362793,
"logps/chosen": -194.67013549804688,
"logps/rejected": -177.0902557373047,
"loss": 0.1565,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -2.4568727016448975,
"rewards/margins": 5.8625688552856445,
"rewards/rejected": -8.319440841674805,
"step": 150
},
{
"epoch": 0.2,
"learning_rate": 4.454419889502762e-07,
"logits/chosen": -3.035737991333008,
"logits/rejected": -2.9880356788635254,
"logps/chosen": -185.5491943359375,
"logps/rejected": -158.82350158691406,
"loss": 0.1474,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -1.9501594305038452,
"rewards/margins": 6.810189247131348,
"rewards/rejected": -8.76034927368164,
"step": 160
},
{
"epoch": 0.21,
"learning_rate": 4.3853591160220993e-07,
"logits/chosen": -3.051605463027954,
"logits/rejected": -2.9627249240875244,
"logps/chosen": -188.94488525390625,
"logps/rejected": -159.45440673828125,
"loss": 0.1317,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -3.886648178100586,
"rewards/margins": 6.363480567932129,
"rewards/rejected": -10.250128746032715,
"step": 170
},
{
"epoch": 0.22,
"learning_rate": 4.3162983425414365e-07,
"logits/chosen": -2.9437613487243652,
"logits/rejected": -2.923678398132324,
"logps/chosen": -197.8041229248047,
"logps/rejected": -179.19410705566406,
"loss": 0.1784,
"rewards/accuracies": 0.75,
"rewards/chosen": -2.9531164169311523,
"rewards/margins": 5.519814968109131,
"rewards/rejected": -8.472932815551758,
"step": 180
},
{
"epoch": 0.24,
"learning_rate": 4.247237569060773e-07,
"logits/chosen": -2.9907093048095703,
"logits/rejected": -2.91261625289917,
"logps/chosen": -183.89828491210938,
"logps/rejected": -151.09255981445312,
"loss": 0.1128,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -0.34959203004837036,
"rewards/margins": 6.793131351470947,
"rewards/rejected": -7.142723083496094,
"step": 190
},
{
"epoch": 0.25,
"learning_rate": 4.1781767955801103e-07,
"logits/chosen": -3.0003714561462402,
"logits/rejected": -2.9367072582244873,
"logps/chosen": -161.82725524902344,
"logps/rejected": -141.76551818847656,
"loss": 0.1327,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -1.736950159072876,
"rewards/margins": 6.6674089431762695,
"rewards/rejected": -8.404359817504883,
"step": 200
},
{
"epoch": 0.26,
"learning_rate": 4.1091160220994475e-07,
"logits/chosen": -2.9085640907287598,
"logits/rejected": -2.866804361343384,
"logps/chosen": -186.4488983154297,
"logps/rejected": -154.49505615234375,
"loss": 0.0995,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -2.524721622467041,
"rewards/margins": 6.8670477867126465,
"rewards/rejected": -9.391769409179688,
"step": 210
},
{
"epoch": 0.27,
"learning_rate": 4.0400552486187846e-07,
"logits/chosen": -2.9839043617248535,
"logits/rejected": -2.905033588409424,
"logps/chosen": -196.44332885742188,
"logps/rejected": -162.53839111328125,
"loss": 0.1511,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -1.4525755643844604,
"rewards/margins": 6.9121527671813965,
"rewards/rejected": -8.364728927612305,
"step": 220
},
{
"epoch": 0.29,
"learning_rate": 3.970994475138121e-07,
"logits/chosen": -2.9412007331848145,
"logits/rejected": -2.897495746612549,
"logps/chosen": -169.09982299804688,
"logps/rejected": -140.54649353027344,
"loss": 0.1487,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -0.6759728193283081,
"rewards/margins": 5.637655258178711,
"rewards/rejected": -6.313628196716309,
"step": 230
},
{
"epoch": 0.3,
"learning_rate": 3.9019337016574584e-07,
"logits/chosen": -2.9135284423828125,
"logits/rejected": -2.9134836196899414,
"logps/chosen": -182.82675170898438,
"logps/rejected": -172.9473419189453,
"loss": 0.1548,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -2.683964490890503,
"rewards/margins": 7.280943393707275,
"rewards/rejected": -9.964908599853516,
"step": 240
},
{
"epoch": 0.31,
"learning_rate": 3.832872928176795e-07,
"logits/chosen": -2.9248108863830566,
"logits/rejected": -2.929591655731201,
"logps/chosen": -170.30747985839844,
"logps/rejected": -158.3248291015625,
"loss": 0.0943,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -0.5063244104385376,
"rewards/margins": 8.612382888793945,
"rewards/rejected": -9.118706703186035,
"step": 250
},
{
"epoch": 0.32,
"learning_rate": 3.763812154696133e-07,
"logits/chosen": -2.9988207817077637,
"logits/rejected": -2.945227861404419,
"logps/chosen": -176.06637573242188,
"logps/rejected": -152.9449005126953,
"loss": 0.1231,
"rewards/accuracies": 0.875,
"rewards/chosen": 0.20524081587791443,
"rewards/margins": 7.8652777671813965,
"rewards/rejected": -7.660037994384766,
"step": 260
},
{
"epoch": 0.34,
"learning_rate": 3.6947513812154694e-07,
"logits/chosen": -3.014315366744995,
"logits/rejected": -2.928999185562134,
"logps/chosen": -171.6766357421875,
"logps/rejected": -147.11154174804688,
"loss": 0.1415,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": 1.9650657176971436,
"rewards/margins": 6.9074506759643555,
"rewards/rejected": -4.942384243011475,
"step": 270
},
{
"epoch": 0.35,
"learning_rate": 3.6256906077348066e-07,
"logits/chosen": -2.907224178314209,
"logits/rejected": -2.906038761138916,
"logps/chosen": -182.75013732910156,
"logps/rejected": -167.35330200195312,
"loss": 0.1677,
"rewards/accuracies": 0.875,
"rewards/chosen": 0.3685176968574524,
"rewards/margins": 7.3049635887146,
"rewards/rejected": -6.936445713043213,
"step": 280
},
{
"epoch": 0.36,
"learning_rate": 3.556629834254143e-07,
"logits/chosen": -2.899120330810547,
"logits/rejected": -2.84541654586792,
"logps/chosen": -148.26084899902344,
"logps/rejected": -133.62619018554688,
"loss": 0.1569,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 0.6295042037963867,
"rewards/margins": 6.114689350128174,
"rewards/rejected": -5.485184669494629,
"step": 290
},
{
"epoch": 0.37,
"learning_rate": 3.4875690607734804e-07,
"logits/chosen": -2.965310573577881,
"logits/rejected": -2.8868935108184814,
"logps/chosen": -181.91659545898438,
"logps/rejected": -153.7183380126953,
"loss": 0.1737,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": 0.35049858689308167,
"rewards/margins": 7.459444999694824,
"rewards/rejected": -7.108946323394775,
"step": 300
},
{
"epoch": 0.39,
"learning_rate": 3.418508287292817e-07,
"logits/chosen": -2.978998899459839,
"logits/rejected": -2.9149255752563477,
"logps/chosen": -195.01406860351562,
"logps/rejected": -172.47183227539062,
"loss": 0.1305,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -8.093232154846191,
"rewards/margins": 7.233319282531738,
"rewards/rejected": -15.326550483703613,
"step": 310
},
{
"epoch": 0.4,
"learning_rate": 3.3494475138121547e-07,
"logits/chosen": -2.976937770843506,
"logits/rejected": -2.8859190940856934,
"logps/chosen": -199.7006072998047,
"logps/rejected": -168.48184204101562,
"loss": 0.094,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -6.615779876708984,
"rewards/margins": 7.833823204040527,
"rewards/rejected": -14.449602127075195,
"step": 320
},
{
"epoch": 0.41,
"learning_rate": 3.280386740331492e-07,
"logits/chosen": -2.887190103530884,
"logits/rejected": -2.877593517303467,
"logps/chosen": -167.93466186523438,
"logps/rejected": -152.63137817382812,
"loss": 0.1647,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -7.1772050857543945,
"rewards/margins": 6.209255218505859,
"rewards/rejected": -13.386459350585938,
"step": 330
},
{
"epoch": 0.42,
"learning_rate": 3.2113259668508285e-07,
"logits/chosen": -3.028519868850708,
"logits/rejected": -2.9872825145721436,
"logps/chosen": -209.1421661376953,
"logps/rejected": -189.716796875,
"loss": 0.1479,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -8.37966251373291,
"rewards/margins": 6.489211082458496,
"rewards/rejected": -14.868873596191406,
"step": 340
},
{
"epoch": 0.43,
"learning_rate": 3.1422651933701657e-07,
"logits/chosen": -2.91877818107605,
"logits/rejected": -2.86418080329895,
"logps/chosen": -188.1709747314453,
"logps/rejected": -169.12794494628906,
"loss": 0.12,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -8.10500431060791,
"rewards/margins": 7.924190521240234,
"rewards/rejected": -16.029193878173828,
"step": 350
},
{
"epoch": 0.45,
"learning_rate": 3.0732044198895023e-07,
"logits/chosen": -2.9679653644561768,
"logits/rejected": -2.9573211669921875,
"logps/chosen": -191.16941833496094,
"logps/rejected": -175.7834014892578,
"loss": 0.1287,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -6.326230525970459,
"rewards/margins": 7.559299468994141,
"rewards/rejected": -13.885531425476074,
"step": 360
},
{
"epoch": 0.46,
"learning_rate": 3.00414364640884e-07,
"logits/chosen": -2.9169762134552,
"logits/rejected": -2.834723949432373,
"logps/chosen": -178.90940856933594,
"logps/rejected": -150.42605590820312,
"loss": 0.0809,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -4.185718059539795,
"rewards/margins": 8.697784423828125,
"rewards/rejected": -12.883502006530762,
"step": 370
},
{
"epoch": 0.47,
"learning_rate": 2.9350828729281767e-07,
"logits/chosen": -3.0018362998962402,
"logits/rejected": -2.9604392051696777,
"logps/chosen": -170.78225708007812,
"logps/rejected": -154.3968505859375,
"loss": 0.1239,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -2.6663613319396973,
"rewards/margins": 8.932355880737305,
"rewards/rejected": -11.598716735839844,
"step": 380
},
{
"epoch": 0.48,
"learning_rate": 2.866022099447514e-07,
"logits/chosen": -2.994075298309326,
"logits/rejected": -2.9975550174713135,
"logps/chosen": -215.20736694335938,
"logps/rejected": -194.09291076660156,
"loss": 0.1248,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -3.196017265319824,
"rewards/margins": 10.618438720703125,
"rewards/rejected": -13.81445598602295,
"step": 390
},
{
"epoch": 0.5,
"learning_rate": 2.7969613259668505e-07,
"logits/chosen": -3.0385982990264893,
"logits/rejected": -2.966886043548584,
"logps/chosen": -178.92239379882812,
"logps/rejected": -152.96282958984375,
"loss": 0.1334,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -1.9255645275115967,
"rewards/margins": 9.632636070251465,
"rewards/rejected": -11.55820083618164,
"step": 400
},
{
"epoch": 0.51,
"learning_rate": 2.7279005524861877e-07,
"logits/chosen": -3.092007637023926,
"logits/rejected": -2.971647024154663,
"logps/chosen": -212.3997039794922,
"logps/rejected": -171.35873413085938,
"loss": 0.1333,
"rewards/accuracies": 0.875,
"rewards/chosen": -2.5890955924987793,
"rewards/margins": 8.974533081054688,
"rewards/rejected": -11.563628196716309,
"step": 410
},
{
"epoch": 0.52,
"learning_rate": 2.6588397790055243e-07,
"logits/chosen": -2.9924330711364746,
"logits/rejected": -2.9680898189544678,
"logps/chosen": -187.8248291015625,
"logps/rejected": -179.09518432617188,
"loss": 0.1164,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -2.1534762382507324,
"rewards/margins": 8.748167991638184,
"rewards/rejected": -10.901644706726074,
"step": 420
},
{
"epoch": 0.53,
"learning_rate": 2.589779005524862e-07,
"logits/chosen": -3.084371566772461,
"logits/rejected": -3.0335347652435303,
"logps/chosen": -200.3054962158203,
"logps/rejected": -174.8027801513672,
"loss": 0.1358,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -0.3474791646003723,
"rewards/margins": 8.830670356750488,
"rewards/rejected": -9.17814826965332,
"step": 430
},
{
"epoch": 0.55,
"learning_rate": 2.5207182320441986e-07,
"logits/chosen": -3.0466129779815674,
"logits/rejected": -2.9966206550598145,
"logps/chosen": -204.1211395263672,
"logps/rejected": -184.7081298828125,
"loss": 0.106,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -0.979719340801239,
"rewards/margins": 8.62094783782959,
"rewards/rejected": -9.600667953491211,
"step": 440
},
{
"epoch": 0.56,
"learning_rate": 2.451657458563536e-07,
"logits/chosen": -3.0778112411499023,
"logits/rejected": -3.046935558319092,
"logps/chosen": -193.56973266601562,
"logps/rejected": -168.67868041992188,
"loss": 0.0973,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -1.372416615486145,
"rewards/margins": 8.402765274047852,
"rewards/rejected": -9.775181770324707,
"step": 450
},
{
"epoch": 0.57,
"learning_rate": 2.3825966850828727e-07,
"logits/chosen": -2.967545986175537,
"logits/rejected": -2.895456314086914,
"logps/chosen": -157.90182495117188,
"logps/rejected": -134.09254455566406,
"loss": 0.1293,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 0.49174195528030396,
"rewards/margins": 9.59979248046875,
"rewards/rejected": -9.108049392700195,
"step": 460
},
{
"epoch": 0.58,
"learning_rate": 2.31353591160221e-07,
"logits/chosen": -3.0117809772491455,
"logits/rejected": -2.9402668476104736,
"logps/chosen": -190.03530883789062,
"logps/rejected": -168.36068725585938,
"loss": 0.0976,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -0.2254009246826172,
"rewards/margins": 9.530010223388672,
"rewards/rejected": -9.755411148071289,
"step": 470
},
{
"epoch": 0.6,
"learning_rate": 2.2444751381215468e-07,
"logits/chosen": -3.088365077972412,
"logits/rejected": -2.996295928955078,
"logps/chosen": -197.9468994140625,
"logps/rejected": -172.51962280273438,
"loss": 0.0996,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -3.3391284942626953,
"rewards/margins": 8.23055648803711,
"rewards/rejected": -11.569684982299805,
"step": 480
},
{
"epoch": 0.61,
"learning_rate": 2.175414364640884e-07,
"logits/chosen": -2.968456983566284,
"logits/rejected": -2.961946964263916,
"logps/chosen": -186.1309051513672,
"logps/rejected": -169.68812561035156,
"loss": 0.1217,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -0.4508451521396637,
"rewards/margins": 11.474364280700684,
"rewards/rejected": -11.925207138061523,
"step": 490
},
{
"epoch": 0.62,
"learning_rate": 2.1063535911602208e-07,
"logits/chosen": -2.978132724761963,
"logits/rejected": -2.919520616531372,
"logps/chosen": -196.43170166015625,
"logps/rejected": -174.66552734375,
"loss": 0.1262,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -5.012697696685791,
"rewards/margins": 8.436281204223633,
"rewards/rejected": -13.448979377746582,
"step": 500
},
{
"epoch": 0.62,
"eval_logits/chosen": -2.9003899097442627,
"eval_logits/rejected": -2.85654616355896,
"eval_logps/chosen": -188.17869567871094,
"eval_logps/rejected": -171.17919921875,
"eval_loss": 0.118459552526474,
"eval_rewards/accuracies": 0.8670212626457214,
"eval_rewards/chosen": -3.0113587379455566,
"eval_rewards/margins": 9.170669555664062,
"eval_rewards/rejected": -12.182028770446777,
"eval_runtime": 242.1612,
"eval_samples_per_second": 6.174,
"eval_steps_per_second": 0.194,
"step": 500
},
{
"epoch": 0.63,
"learning_rate": 2.0372928176795578e-07,
"logits/chosen": -2.944871425628662,
"logits/rejected": -2.9157793521881104,
"logps/chosen": -180.36338806152344,
"logps/rejected": -159.6240692138672,
"loss": 0.1055,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -2.646656036376953,
"rewards/margins": 9.775663375854492,
"rewards/rejected": -12.422318458557129,
"step": 510
},
{
"epoch": 0.65,
"learning_rate": 1.968232044198895e-07,
"logits/chosen": -3.010129690170288,
"logits/rejected": -2.9667623043060303,
"logps/chosen": -171.2137908935547,
"logps/rejected": -152.80848693847656,
"loss": 0.1331,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -3.5951690673828125,
"rewards/margins": 7.591134548187256,
"rewards/rejected": -11.186304092407227,
"step": 520
},
{
"epoch": 0.66,
"learning_rate": 1.899171270718232e-07,
"logits/chosen": -3.0541915893554688,
"logits/rejected": -2.9794061183929443,
"logps/chosen": -179.92555236816406,
"logps/rejected": -154.8466339111328,
"loss": 0.1356,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -2.0221731662750244,
"rewards/margins": 8.44990062713623,
"rewards/rejected": -10.472073554992676,
"step": 530
},
{
"epoch": 0.67,
"learning_rate": 1.830110497237569e-07,
"logits/chosen": -2.930157423019409,
"logits/rejected": -2.9100921154022217,
"logps/chosen": -208.7513885498047,
"logps/rejected": -196.99765014648438,
"loss": 0.1238,
"rewards/accuracies": 0.875,
"rewards/chosen": -4.048346519470215,
"rewards/margins": 8.839378356933594,
"rewards/rejected": -12.887723922729492,
"step": 540
},
{
"epoch": 0.68,
"learning_rate": 1.7610497237569062e-07,
"logits/chosen": -2.9872794151306152,
"logits/rejected": -2.9665157794952393,
"logps/chosen": -181.928955078125,
"logps/rejected": -173.3523712158203,
"loss": 0.1339,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -2.8346304893493652,
"rewards/margins": 8.800179481506348,
"rewards/rejected": -11.634809494018555,
"step": 550
},
{
"epoch": 0.7,
"learning_rate": 1.691988950276243e-07,
"logits/chosen": -3.0336689949035645,
"logits/rejected": -3.022378444671631,
"logps/chosen": -197.01950073242188,
"logps/rejected": -182.48306274414062,
"loss": 0.1319,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -2.3179187774658203,
"rewards/margins": 9.856887817382812,
"rewards/rejected": -12.174806594848633,
"step": 560
},
{
"epoch": 0.71,
"learning_rate": 1.62292817679558e-07,
"logits/chosen": -2.9352331161499023,
"logits/rejected": -2.8333239555358887,
"logps/chosen": -183.40594482421875,
"logps/rejected": -157.8167266845703,
"loss": 0.0838,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -3.595484495162964,
"rewards/margins": 9.1626615524292,
"rewards/rejected": -12.758146286010742,
"step": 570
},
{
"epoch": 0.72,
"learning_rate": 1.5538674033149171e-07,
"logits/chosen": -3.005654811859131,
"logits/rejected": -2.9100375175476074,
"logps/chosen": -184.67526245117188,
"logps/rejected": -154.762451171875,
"loss": 0.1174,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -1.7798635959625244,
"rewards/margins": 10.110002517700195,
"rewards/rejected": -11.889866828918457,
"step": 580
},
{
"epoch": 0.73,
"learning_rate": 1.484806629834254e-07,
"logits/chosen": -3.069862127304077,
"logits/rejected": -2.978449583053589,
"logps/chosen": -209.2169647216797,
"logps/rejected": -170.54354858398438,
"loss": 0.122,
"rewards/accuracies": 0.875,
"rewards/chosen": -2.5865066051483154,
"rewards/margins": 8.170557022094727,
"rewards/rejected": -10.757063865661621,
"step": 590
},
{
"epoch": 0.75,
"learning_rate": 1.4157458563535912e-07,
"logits/chosen": -3.0297770500183105,
"logits/rejected": -2.956226110458374,
"logps/chosen": -179.11427307128906,
"logps/rejected": -155.3955535888672,
"loss": 0.1675,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -2.389403820037842,
"rewards/margins": 7.711805820465088,
"rewards/rejected": -10.10120964050293,
"step": 600
},
{
"epoch": 0.76,
"learning_rate": 1.346685082872928e-07,
"logits/chosen": -2.928321599960327,
"logits/rejected": -2.89027738571167,
"logps/chosen": -214.89675903320312,
"logps/rejected": -190.57569885253906,
"loss": 0.0856,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -1.7383911609649658,
"rewards/margins": 10.676383018493652,
"rewards/rejected": -12.414773941040039,
"step": 610
},
{
"epoch": 0.77,
"learning_rate": 1.277624309392265e-07,
"logits/chosen": -3.0483570098876953,
"logits/rejected": -2.9306087493896484,
"logps/chosen": -180.202880859375,
"logps/rejected": -148.0863037109375,
"loss": 0.1345,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -3.625720977783203,
"rewards/margins": 7.224545955657959,
"rewards/rejected": -10.85026741027832,
"step": 620
},
{
"epoch": 0.78,
"learning_rate": 1.2085635359116022e-07,
"logits/chosen": -3.087991237640381,
"logits/rejected": -3.0041158199310303,
"logps/chosen": -198.82481384277344,
"logps/rejected": -169.78836059570312,
"loss": 0.1162,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -2.9254348278045654,
"rewards/margins": 8.505046844482422,
"rewards/rejected": -11.430481910705566,
"step": 630
},
{
"epoch": 0.8,
"learning_rate": 1.1395027624309392e-07,
"logits/chosen": -2.9629359245300293,
"logits/rejected": -2.980262517929077,
"logps/chosen": -201.67306518554688,
"logps/rejected": -190.01260375976562,
"loss": 0.1416,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -4.385073661804199,
"rewards/margins": 8.24610710144043,
"rewards/rejected": -12.631181716918945,
"step": 640
},
{
"epoch": 0.81,
"learning_rate": 1.0704419889502763e-07,
"logits/chosen": -3.059577465057373,
"logits/rejected": -2.9308218955993652,
"logps/chosen": -221.16049194335938,
"logps/rejected": -187.49386596679688,
"loss": 0.103,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -3.075377941131592,
"rewards/margins": 10.026552200317383,
"rewards/rejected": -13.101930618286133,
"step": 650
},
{
"epoch": 0.82,
"learning_rate": 1.0013812154696132e-07,
"logits/chosen": -3.097217082977295,
"logits/rejected": -2.9983904361724854,
"logps/chosen": -198.12832641601562,
"logps/rejected": -173.28384399414062,
"loss": 0.122,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -3.524822950363159,
"rewards/margins": 8.267110824584961,
"rewards/rejected": -11.791935920715332,
"step": 660
},
{
"epoch": 0.83,
"learning_rate": 9.323204419889502e-08,
"logits/chosen": -3.0255918502807617,
"logits/rejected": -2.9517340660095215,
"logps/chosen": -174.54376220703125,
"logps/rejected": -154.30923461914062,
"loss": 0.1028,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -0.3350832462310791,
"rewards/margins": 10.506619453430176,
"rewards/rejected": -10.841703414916992,
"step": 670
},
{
"epoch": 0.84,
"learning_rate": 8.632596685082872e-08,
"logits/chosen": -2.9522705078125,
"logits/rejected": -2.905046224594116,
"logps/chosen": -184.12936401367188,
"logps/rejected": -156.78311157226562,
"loss": 0.1092,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -1.2172771692276,
"rewards/margins": 10.139422416687012,
"rewards/rejected": -11.35669994354248,
"step": 680
},
{
"epoch": 0.86,
"learning_rate": 7.941988950276243e-08,
"logits/chosen": -2.9998116493225098,
"logits/rejected": -2.93977689743042,
"logps/chosen": -159.04739379882812,
"logps/rejected": -137.8501434326172,
"loss": 0.1154,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -2.821601390838623,
"rewards/margins": 8.011626243591309,
"rewards/rejected": -10.833227157592773,
"step": 690
},
{
"epoch": 0.87,
"learning_rate": 7.251381215469612e-08,
"logits/chosen": -3.064391613006592,
"logits/rejected": -3.0105395317077637,
"logps/chosen": -216.6563720703125,
"logps/rejected": -190.38059997558594,
"loss": 0.143,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -3.8847603797912598,
"rewards/margins": 9.102531433105469,
"rewards/rejected": -12.987292289733887,
"step": 700
},
{
"epoch": 0.88,
"learning_rate": 6.560773480662984e-08,
"logits/chosen": -2.8895156383514404,
"logits/rejected": -2.873819589614868,
"logps/chosen": -165.34201049804688,
"logps/rejected": -158.01522827148438,
"loss": 0.1274,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -2.2765231132507324,
"rewards/margins": 10.176279067993164,
"rewards/rejected": -12.452800750732422,
"step": 710
},
{
"epoch": 0.89,
"learning_rate": 5.870165745856354e-08,
"logits/chosen": -2.992870569229126,
"logits/rejected": -2.9288392066955566,
"logps/chosen": -166.22451782226562,
"logps/rejected": -149.21884155273438,
"loss": 0.0849,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -0.1316981315612793,
"rewards/margins": 11.313945770263672,
"rewards/rejected": -11.44564437866211,
"step": 720
},
{
"epoch": 0.91,
"learning_rate": 5.1795580110497236e-08,
"logits/chosen": -3.0519726276397705,
"logits/rejected": -2.9871127605438232,
"logps/chosen": -198.3760528564453,
"logps/rejected": -173.3822021484375,
"loss": 0.1439,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -4.240297317504883,
"rewards/margins": 9.328596115112305,
"rewards/rejected": -13.568893432617188,
"step": 730
},
{
"epoch": 0.92,
"learning_rate": 4.488950276243094e-08,
"logits/chosen": -2.905094623565674,
"logits/rejected": -2.898705005645752,
"logps/chosen": -166.13040161132812,
"logps/rejected": -166.46963500976562,
"loss": 0.0909,
"rewards/accuracies": 0.875,
"rewards/chosen": -2.6403870582580566,
"rewards/margins": 10.037248611450195,
"rewards/rejected": -12.677635192871094,
"step": 740
},
{
"epoch": 0.93,
"learning_rate": 3.7983425414364637e-08,
"logits/chosen": -3.0315964221954346,
"logits/rejected": -2.945363998413086,
"logps/chosen": -186.29241943359375,
"logps/rejected": -159.93798828125,
"loss": 0.1194,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -3.548847198486328,
"rewards/margins": 9.53381061553955,
"rewards/rejected": -13.082656860351562,
"step": 750
},
{
"epoch": 0.94,
"learning_rate": 3.107734806629834e-08,
"logits/chosen": -2.901912212371826,
"logits/rejected": -2.8841569423675537,
"logps/chosen": -165.88009643554688,
"logps/rejected": -153.7960662841797,
"loss": 0.1276,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -3.673292875289917,
"rewards/margins": 10.183616638183594,
"rewards/rejected": -13.856908798217773,
"step": 760
},
{
"epoch": 0.96,
"learning_rate": 2.4171270718232044e-08,
"logits/chosen": -3.0154058933258057,
"logits/rejected": -2.9658362865448,
"logps/chosen": -207.61538696289062,
"logps/rejected": -184.4959259033203,
"loss": 0.1581,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -3.740403652191162,
"rewards/margins": 10.590685844421387,
"rewards/rejected": -14.331089973449707,
"step": 770
},
{
"epoch": 0.97,
"learning_rate": 1.7265193370165747e-08,
"logits/chosen": -3.099858045578003,
"logits/rejected": -2.996018171310425,
"logps/chosen": -209.8662109375,
"logps/rejected": -179.3186492919922,
"loss": 0.0908,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": -0.9476644396781921,
"rewards/margins": 12.100104331970215,
"rewards/rejected": -13.047770500183105,
"step": 780
},
{
"epoch": 0.98,
"learning_rate": 1.0359116022099446e-08,
"logits/chosen": -3.0576415061950684,
"logits/rejected": -2.9471096992492676,
"logps/chosen": -175.6331787109375,
"logps/rejected": -154.61416625976562,
"loss": 0.0915,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -3.3139986991882324,
"rewards/margins": 9.281652450561523,
"rewards/rejected": -12.595651626586914,
"step": 790
},
{
"epoch": 0.99,
"learning_rate": 3.453038674033149e-09,
"logits/chosen": -2.9590699672698975,
"logits/rejected": -2.9163191318511963,
"logps/chosen": -206.9409942626953,
"logps/rejected": -183.0679931640625,
"loss": 0.1379,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -2.519080638885498,
"rewards/margins": 11.627119064331055,
"rewards/rejected": -14.146200180053711,
"step": 800
},
{
"epoch": 1.0,
"step": 805,
"total_flos": 0.0,
"train_loss": 0.14663515942437308,
"train_runtime": 4275.5798,
"train_samples_per_second": 3.012,
"train_steps_per_second": 0.188
}
],
"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
}