Jiazheng Li
init push
0ffb2d3
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.99562408835174,
"eval_steps": 200,
"global_step": 1797,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.02,
"grad_norm": 22.077117919921875,
"learning_rate": 9.999251052313705e-06,
"logits/chosen": 0.9130447506904602,
"logits/rejected": 0.9451152682304382,
"logps/chosen": -119.4351806640625,
"logps/rejected": -146.38978576660156,
"loss": 2.3628,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": 23.45577621459961,
"rewards/margins": 1.552248477935791,
"rewards/rejected": 21.903528213500977,
"step": 10
},
{
"epoch": 0.03,
"grad_norm": 22.769922256469727,
"learning_rate": 9.996974102027961e-06,
"logits/chosen": 0.5303283929824829,
"logits/rejected": 0.5954318642616272,
"logps/chosen": -131.7035369873047,
"logps/rejected": -162.93435668945312,
"loss": 2.0403,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 23.408517837524414,
"rewards/margins": 3.0596365928649902,
"rewards/rejected": 20.3488826751709,
"step": 20
},
{
"epoch": 0.05,
"grad_norm": 34.49003601074219,
"learning_rate": 9.993618904787861e-06,
"logits/chosen": 0.23024284839630127,
"logits/rejected": 0.2893335521221161,
"logps/chosen": -135.048828125,
"logps/rejected": -175.6779022216797,
"loss": 1.5462,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 22.8612003326416,
"rewards/margins": 3.707317352294922,
"rewards/rejected": 19.153881072998047,
"step": 30
},
{
"epoch": 0.07,
"grad_norm": 26.99772071838379,
"learning_rate": 9.98844090765316e-06,
"logits/chosen": -0.12471504509449005,
"logits/rejected": -0.05195974186062813,
"logps/chosen": -137.30029296875,
"logps/rejected": -179.8270721435547,
"loss": 1.6569,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": 21.461271286010742,
"rewards/margins": 3.4389548301696777,
"rewards/rejected": 18.022315979003906,
"step": 40
},
{
"epoch": 0.08,
"grad_norm": 39.197853088378906,
"learning_rate": 9.98173813574765e-06,
"logits/chosen": -0.5969198346138,
"logits/rejected": -0.5180742144584656,
"logps/chosen": -145.24537658691406,
"logps/rejected": -207.34591674804688,
"loss": 1.5324,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": 21.160579681396484,
"rewards/margins": 5.194663047790527,
"rewards/rejected": 15.965914726257324,
"step": 50
},
{
"epoch": 0.1,
"grad_norm": 28.039043426513672,
"learning_rate": 9.973512637851239e-06,
"logits/chosen": -0.972032904624939,
"logits/rejected": -0.9215275049209595,
"logps/chosen": -159.76718139648438,
"logps/rejected": -216.4589080810547,
"loss": 1.515,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": 19.871841430664062,
"rewards/margins": 5.058460235595703,
"rewards/rejected": 14.813380241394043,
"step": 60
},
{
"epoch": 0.12,
"grad_norm": 11.507994651794434,
"learning_rate": 9.963766928182676e-06,
"logits/chosen": -1.2355167865753174,
"logits/rejected": -1.1757726669311523,
"logps/chosen": -153.20492553710938,
"logps/rejected": -238.68008422851562,
"loss": 1.1533,
"rewards/accuracies": 0.862500011920929,
"rewards/chosen": 18.97603988647461,
"rewards/margins": 6.053138732910156,
"rewards/rejected": 12.922900199890137,
"step": 70
},
{
"epoch": 0.13,
"grad_norm": 8.204204559326172,
"learning_rate": 9.952503985631063e-06,
"logits/chosen": -1.262416958808899,
"logits/rejected": -1.1802202463150024,
"logps/chosen": -134.965576171875,
"logps/rejected": -221.21835327148438,
"loss": 1.0608,
"rewards/accuracies": 0.887499988079071,
"rewards/chosen": 20.67170524597168,
"rewards/margins": 6.517201900482178,
"rewards/rejected": 14.154504776000977,
"step": 80
},
{
"epoch": 0.15,
"grad_norm": 27.41866111755371,
"learning_rate": 9.939727252845304e-06,
"logits/chosen": -0.9036836624145508,
"logits/rejected": -0.8652833104133606,
"logps/chosen": -141.32794189453125,
"logps/rejected": -192.92916870117188,
"loss": 1.2558,
"rewards/accuracies": 0.887499988079071,
"rewards/chosen": 21.43319320678711,
"rewards/margins": 5.654860496520996,
"rewards/rejected": 15.77833080291748,
"step": 90
},
{
"epoch": 0.17,
"grad_norm": 31.713287353515625,
"learning_rate": 9.925440635181834e-06,
"logits/chosen": -0.7909407615661621,
"logits/rejected": -0.7640722393989563,
"logps/chosen": -128.05874633789062,
"logps/rejected": -197.05027770996094,
"loss": 1.3031,
"rewards/accuracies": 0.8125,
"rewards/chosen": 21.722820281982422,
"rewards/margins": 5.5480637550354,
"rewards/rejected": 16.174755096435547,
"step": 100
},
{
"epoch": 0.18,
"grad_norm": 5.262263298034668,
"learning_rate": 9.909648499510903e-06,
"logits/chosen": -0.7350383400917053,
"logits/rejected": -0.7220867872238159,
"logps/chosen": -141.88296508789062,
"logps/rejected": -192.28868103027344,
"loss": 1.2729,
"rewards/accuracies": 0.875,
"rewards/chosen": 20.970355987548828,
"rewards/margins": 6.049419403076172,
"rewards/rejected": 14.920933723449707,
"step": 110
},
{
"epoch": 0.2,
"grad_norm": 20.070709228515625,
"learning_rate": 9.892355672881781e-06,
"logits/chosen": -0.8407080769538879,
"logits/rejected": -0.8053030967712402,
"logps/chosen": -129.39187622070312,
"logps/rejected": -197.65792846679688,
"loss": 1.4401,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": 22.391189575195312,
"rewards/margins": 6.017077445983887,
"rewards/rejected": 16.374113082885742,
"step": 120
},
{
"epoch": 0.22,
"grad_norm": 15.006941795349121,
"learning_rate": 9.873567441047321e-06,
"logits/chosen": -0.965476393699646,
"logits/rejected": -0.8994636535644531,
"logps/chosen": -115.93741607666016,
"logps/rejected": -211.8863983154297,
"loss": 0.9361,
"rewards/accuracies": 0.875,
"rewards/chosen": 23.738582611083984,
"rewards/margins": 7.023127555847168,
"rewards/rejected": 16.715452194213867,
"step": 130
},
{
"epoch": 0.23,
"grad_norm": 21.54697608947754,
"learning_rate": 9.853289546848304e-06,
"logits/chosen": -1.1242530345916748,
"logits/rejected": -1.0690996646881104,
"logps/chosen": -136.72805786132812,
"logps/rejected": -215.2025146484375,
"loss": 0.994,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": 21.464555740356445,
"rewards/margins": 6.428616523742676,
"rewards/rejected": 15.035940170288086,
"step": 140
},
{
"epoch": 0.25,
"grad_norm": 36.22259521484375,
"learning_rate": 9.83152818845808e-06,
"logits/chosen": -1.238879680633545,
"logits/rejected": -1.212425947189331,
"logps/chosen": -135.1555938720703,
"logps/rejected": -223.7063751220703,
"loss": 1.5207,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": 21.200214385986328,
"rewards/margins": 6.609138488769531,
"rewards/rejected": 14.59107780456543,
"step": 150
},
{
"epoch": 0.27,
"grad_norm": 5.652963161468506,
"learning_rate": 9.808290017488018e-06,
"logits/chosen": -1.2760050296783447,
"logits/rejected": -1.229891061782837,
"logps/chosen": -124.10953521728516,
"logps/rejected": -205.0720977783203,
"loss": 1.1957,
"rewards/accuracies": 0.887499988079071,
"rewards/chosen": 21.773719787597656,
"rewards/margins": 6.910721778869629,
"rewards/rejected": 14.862997055053711,
"step": 160
},
{
"epoch": 0.28,
"grad_norm": 7.20241117477417,
"learning_rate": 9.783582136954363e-06,
"logits/chosen": -1.022456169128418,
"logits/rejected": -1.0098379850387573,
"logps/chosen": -138.91571044921875,
"logps/rejected": -193.36270141601562,
"loss": 1.1055,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": 21.772497177124023,
"rewards/margins": 7.529877662658691,
"rewards/rejected": 14.242624282836914,
"step": 170
},
{
"epoch": 0.3,
"grad_norm": 23.037525177001953,
"learning_rate": 9.757412099107122e-06,
"logits/chosen": -1.0896828174591064,
"logits/rejected": -1.049005150794983,
"logps/chosen": -126.07826232910156,
"logps/rejected": -211.6298370361328,
"loss": 1.1861,
"rewards/accuracies": 0.875,
"rewards/chosen": 22.460693359375,
"rewards/margins": 7.169064998626709,
"rewards/rejected": 15.29162883758545,
"step": 180
},
{
"epoch": 0.32,
"grad_norm": 29.62506866455078,
"learning_rate": 9.72978790312163e-06,
"logits/chosen": -0.8976786732673645,
"logits/rejected": -0.8687394857406616,
"logps/chosen": -119.01469421386719,
"logps/rejected": -198.20928955078125,
"loss": 0.9781,
"rewards/accuracies": 0.862500011920929,
"rewards/chosen": 23.478139877319336,
"rewards/margins": 6.4206695556640625,
"rewards/rejected": 17.057470321655273,
"step": 190
},
{
"epoch": 0.33,
"grad_norm": 36.63336181640625,
"learning_rate": 9.700717992653505e-06,
"logits/chosen": -1.0064117908477783,
"logits/rejected": -0.9567538499832153,
"logps/chosen": -119.5957260131836,
"logps/rejected": -198.21707153320312,
"loss": 1.1726,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": 22.104581832885742,
"rewards/margins": 6.333141803741455,
"rewards/rejected": 15.771438598632812,
"step": 200
},
{
"epoch": 0.33,
"eval_logits/chosen": -0.7826857566833496,
"eval_logits/rejected": -0.7607142925262451,
"eval_logps/chosen": -162.48268127441406,
"eval_logps/rejected": -199.3195037841797,
"eval_loss": 2.607863664627075,
"eval_rewards/accuracies": 0.5929077863693237,
"eval_rewards/chosen": 14.876009941101074,
"eval_rewards/margins": 1.181178092956543,
"eval_rewards/rejected": 13.694831848144531,
"eval_runtime": 941.4569,
"eval_samples_per_second": 0.749,
"eval_steps_per_second": 0.749,
"step": 200
},
{
"epoch": 0.35,
"grad_norm": 5.637684345245361,
"learning_rate": 9.670211253257753e-06,
"logits/chosen": -1.359417200088501,
"logits/rejected": -1.3465789556503296,
"logps/chosen": -123.56754302978516,
"logps/rejected": -221.25131225585938,
"loss": 0.8259,
"rewards/accuracies": 0.9375,
"rewards/chosen": 22.563716888427734,
"rewards/margins": 8.354880332946777,
"rewards/rejected": 14.208834648132324,
"step": 210
},
{
"epoch": 0.37,
"grad_norm": 18.743879318237305,
"learning_rate": 9.638277009672787e-06,
"logits/chosen": -1.294345736503601,
"logits/rejected": -1.2438859939575195,
"logps/chosen": -120.88092041015625,
"logps/rejected": -219.56845092773438,
"loss": 0.9568,
"rewards/accuracies": 0.887499988079071,
"rewards/chosen": 22.052473068237305,
"rewards/margins": 7.690165042877197,
"rewards/rejected": 14.36230754852295,
"step": 220
},
{
"epoch": 0.38,
"grad_norm": 4.425251483917236,
"learning_rate": 9.604925022970226e-06,
"logits/chosen": -1.4860092401504517,
"logits/rejected": -1.4260871410369873,
"logps/chosen": -116.0375747680664,
"logps/rejected": -233.6222381591797,
"loss": 0.8266,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 22.423358917236328,
"rewards/margins": 7.869799613952637,
"rewards/rejected": 14.553558349609375,
"step": 230
},
{
"epoch": 0.4,
"grad_norm": 15.595026969909668,
"learning_rate": 9.570165487571295e-06,
"logits/chosen": -1.29686439037323,
"logits/rejected": -1.2336632013320923,
"logps/chosen": -114.66429138183594,
"logps/rejected": -219.73941040039062,
"loss": 0.9711,
"rewards/accuracies": 0.875,
"rewards/chosen": 23.410022735595703,
"rewards/margins": 8.979305267333984,
"rewards/rejected": 14.430717468261719,
"step": 240
},
{
"epoch": 0.42,
"grad_norm": 8.037131309509277,
"learning_rate": 9.534009028130791e-06,
"logits/chosen": -1.3593500852584839,
"logits/rejected": -1.3431365489959717,
"logps/chosen": -116.7984619140625,
"logps/rejected": -203.38714599609375,
"loss": 1.0554,
"rewards/accuracies": 0.875,
"rewards/chosen": 22.437034606933594,
"rewards/margins": 7.695396423339844,
"rewards/rejected": 14.74163818359375,
"step": 250
},
{
"epoch": 0.43,
"grad_norm": 24.600027084350586,
"learning_rate": 9.496466696289533e-06,
"logits/chosen": -1.296525478363037,
"logits/rejected": -1.2403943538665771,
"logps/chosen": -125.29974365234375,
"logps/rejected": -220.273193359375,
"loss": 1.0706,
"rewards/accuracies": 0.875,
"rewards/chosen": 22.99033546447754,
"rewards/margins": 7.625124454498291,
"rewards/rejected": 15.365211486816406,
"step": 260
},
{
"epoch": 0.45,
"grad_norm": 12.19877815246582,
"learning_rate": 9.4575499672963e-06,
"logits/chosen": -1.330751657485962,
"logits/rejected": -1.2963988780975342,
"logps/chosen": -121.9324951171875,
"logps/rejected": -204.8725128173828,
"loss": 1.0879,
"rewards/accuracies": 0.875,
"rewards/chosen": 22.846813201904297,
"rewards/margins": 6.9288506507873535,
"rewards/rejected": 15.917961120605469,
"step": 270
},
{
"epoch": 0.47,
"grad_norm": 61.58235168457031,
"learning_rate": 9.417270736500284e-06,
"logits/chosen": -1.2155901193618774,
"logits/rejected": -1.1836094856262207,
"logps/chosen": -120.52559661865234,
"logps/rejected": -213.6286163330078,
"loss": 1.044,
"rewards/accuracies": 0.875,
"rewards/chosen": 22.38599967956543,
"rewards/margins": 7.965991973876953,
"rewards/rejected": 14.420007705688477,
"step": 280
},
{
"epoch": 0.48,
"grad_norm": 27.1291446685791,
"learning_rate": 9.375641315715147e-06,
"logits/chosen": -1.4014637470245361,
"logits/rejected": -1.3697010278701782,
"logps/chosen": -120.2726821899414,
"logps/rejected": -248.0132293701172,
"loss": 1.1305,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": 23.45905113220215,
"rewards/margins": 9.868653297424316,
"rewards/rejected": 13.5903959274292,
"step": 290
},
{
"epoch": 0.5,
"grad_norm": 4.111963748931885,
"learning_rate": 9.332674429455762e-06,
"logits/chosen": -1.3501853942871094,
"logits/rejected": -1.3238388299942017,
"logps/chosen": -120.29520416259766,
"logps/rejected": -219.8441619873047,
"loss": 0.9532,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": 23.571392059326172,
"rewards/margins": 10.177915573120117,
"rewards/rejected": 13.393475532531738,
"step": 300
},
{
"epoch": 0.52,
"grad_norm": 19.904033660888672,
"learning_rate": 9.288383211048827e-06,
"logits/chosen": -1.3211156129837036,
"logits/rejected": -1.2947032451629639,
"logps/chosen": -108.23663330078125,
"logps/rejected": -220.5325164794922,
"loss": 0.8943,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": 23.694904327392578,
"rewards/margins": 8.900300025939941,
"rewards/rejected": 14.794604301452637,
"step": 310
},
{
"epoch": 0.53,
"grad_norm": 5.557188987731934,
"learning_rate": 9.242781198618508e-06,
"logits/chosen": -1.4145915508270264,
"logits/rejected": -1.4117708206176758,
"logps/chosen": -116.71977233886719,
"logps/rejected": -205.55245971679688,
"loss": 0.8282,
"rewards/accuracies": 0.9375,
"rewards/chosen": 24.738483428955078,
"rewards/margins": 9.999399185180664,
"rewards/rejected": 14.739084243774414,
"step": 320
},
{
"epoch": 0.55,
"grad_norm": 38.25541687011719,
"learning_rate": 9.195882330948351e-06,
"logits/chosen": -1.4355194568634033,
"logits/rejected": -1.430084228515625,
"logps/chosen": -116.34031677246094,
"logps/rejected": -221.4161376953125,
"loss": 0.7625,
"rewards/accuracies": 0.9375,
"rewards/chosen": 24.25949478149414,
"rewards/margins": 10.02813720703125,
"rewards/rejected": 14.231356620788574,
"step": 330
},
{
"epoch": 0.57,
"grad_norm": 40.876747131347656,
"learning_rate": 9.147700943220737e-06,
"logits/chosen": -1.4224138259887695,
"logits/rejected": -1.3659865856170654,
"logps/chosen": -110.4036865234375,
"logps/rejected": -227.19296264648438,
"loss": 0.792,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": 24.380083084106445,
"rewards/margins": 9.536338806152344,
"rewards/rejected": 14.843744277954102,
"step": 340
},
{
"epoch": 0.58,
"grad_norm": 31.962615966796875,
"learning_rate": 9.098251762635162e-06,
"logits/chosen": -1.1947047710418701,
"logits/rejected": -1.1642345190048218,
"logps/chosen": -112.7500991821289,
"logps/rejected": -222.45657348632812,
"loss": 1.1409,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": 23.021337509155273,
"rewards/margins": 8.003363609313965,
"rewards/rejected": 15.017971992492676,
"step": 350
},
{
"epoch": 0.6,
"grad_norm": 22.922439575195312,
"learning_rate": 9.047549903906704e-06,
"logits/chosen": -1.4528108835220337,
"logits/rejected": -1.4263083934783936,
"logps/chosen": -112.0550537109375,
"logps/rejected": -224.366943359375,
"loss": 0.7604,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 24.369909286499023,
"rewards/margins": 9.932126998901367,
"rewards/rejected": 14.437784194946289,
"step": 360
},
{
"epoch": 0.62,
"grad_norm": 15.393943786621094,
"learning_rate": 8.99561086464603e-06,
"logits/chosen": -1.4283573627471924,
"logits/rejected": -1.4220378398895264,
"logps/chosen": -102.64058685302734,
"logps/rejected": -220.1312713623047,
"loss": 0.6298,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 24.416704177856445,
"rewards/margins": 9.940857887268066,
"rewards/rejected": 14.475845336914062,
"step": 370
},
{
"epoch": 0.63,
"grad_norm": 31.526750564575195,
"learning_rate": 8.942450520622371e-06,
"logits/chosen": -1.4221882820129395,
"logits/rejected": -1.4053022861480713,
"logps/chosen": -96.04936218261719,
"logps/rejected": -219.9590301513672,
"loss": 0.8734,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": 25.633686065673828,
"rewards/margins": 10.437232971191406,
"rewards/rejected": 15.196454048156738,
"step": 380
},
{
"epoch": 0.65,
"grad_norm": 21.012407302856445,
"learning_rate": 8.888085120910917e-06,
"logits/chosen": -1.5149024724960327,
"logits/rejected": -1.5197780132293701,
"logps/chosen": -121.2115249633789,
"logps/rejected": -217.0883026123047,
"loss": 0.9286,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": 23.104969024658203,
"rewards/margins": 8.534513473510742,
"rewards/rejected": 14.570454597473145,
"step": 390
},
{
"epoch": 0.67,
"grad_norm": 31.061025619506836,
"learning_rate": 8.83253128292609e-06,
"logits/chosen": -1.3974319696426392,
"logits/rejected": -1.365886926651001,
"logps/chosen": -106.92268371582031,
"logps/rejected": -213.5463409423828,
"loss": 1.0028,
"rewards/accuracies": 0.862500011920929,
"rewards/chosen": 23.568645477294922,
"rewards/margins": 9.186344146728516,
"rewards/rejected": 14.382299423217773,
"step": 400
},
{
"epoch": 0.67,
"eval_logits/chosen": -0.8742575645446777,
"eval_logits/rejected": -0.8668540120124817,
"eval_logps/chosen": -161.5126495361328,
"eval_logps/rejected": -197.9254608154297,
"eval_loss": 2.674311876296997,
"eval_rewards/accuracies": 0.5843971371650696,
"eval_rewards/chosen": 14.97301197052002,
"eval_rewards/margins": 1.1387754678726196,
"eval_rewards/rejected": 13.834238052368164,
"eval_runtime": 926.938,
"eval_samples_per_second": 0.761,
"eval_steps_per_second": 0.761,
"step": 400
},
{
"epoch": 0.68,
"grad_norm": 2.5695064067840576,
"learning_rate": 8.77580598734224e-06,
"logits/chosen": -1.3901078701019287,
"logits/rejected": -1.361106514930725,
"logps/chosen": -100.05810546875,
"logps/rejected": -217.40853881835938,
"loss": 0.8265,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": 24.2711181640625,
"rewards/margins": 9.111645698547363,
"rewards/rejected": 15.159472465515137,
"step": 410
},
{
"epoch": 0.7,
"grad_norm": 37.29205322265625,
"learning_rate": 8.717926572903315e-06,
"logits/chosen": -1.4033088684082031,
"logits/rejected": -1.3668584823608398,
"logps/chosen": -95.80254364013672,
"logps/rejected": -216.8607635498047,
"loss": 0.8961,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": 25.137277603149414,
"rewards/margins": 9.997251510620117,
"rewards/rejected": 15.140027046203613,
"step": 420
},
{
"epoch": 0.72,
"grad_norm": 13.271622657775879,
"learning_rate": 8.658910731123056e-06,
"logits/chosen": -1.2801754474639893,
"logits/rejected": -1.2473911046981812,
"logps/chosen": -97.76548767089844,
"logps/rejected": -216.43551635742188,
"loss": 0.8534,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": 24.32241439819336,
"rewards/margins": 9.534911155700684,
"rewards/rejected": 14.787504196166992,
"step": 430
},
{
"epoch": 0.73,
"grad_norm": 38.644073486328125,
"learning_rate": 8.598776500877398e-06,
"logits/chosen": -1.557314157485962,
"logits/rejected": -1.5338244438171387,
"logps/chosen": -101.72956085205078,
"logps/rejected": -228.32504272460938,
"loss": 0.7487,
"rewards/accuracies": 0.9375,
"rewards/chosen": 24.7454891204834,
"rewards/margins": 11.237323760986328,
"rewards/rejected": 13.50816535949707,
"step": 440
},
{
"epoch": 0.75,
"grad_norm": 3.0210349559783936,
"learning_rate": 8.537542262890664e-06,
"logits/chosen": -1.6236152648925781,
"logits/rejected": -1.5971102714538574,
"logps/chosen": -95.90837097167969,
"logps/rejected": -224.8739471435547,
"loss": 0.6576,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 26.812732696533203,
"rewards/margins": 12.709887504577637,
"rewards/rejected": 14.102846145629883,
"step": 450
},
{
"epoch": 0.77,
"grad_norm": 31.46013641357422,
"learning_rate": 8.475226734117293e-06,
"logits/chosen": -1.275268793106079,
"logits/rejected": -1.2527453899383545,
"logps/chosen": -90.21646881103516,
"logps/rejected": -206.49935913085938,
"loss": 0.6837,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": 25.937389373779297,
"rewards/margins": 10.999462127685547,
"rewards/rejected": 14.93792724609375,
"step": 460
},
{
"epoch": 0.78,
"grad_norm": 8.342381477355957,
"learning_rate": 8.411848962020786e-06,
"logits/chosen": -1.283499002456665,
"logits/rejected": -1.2215421199798584,
"logps/chosen": -92.12187194824219,
"logps/rejected": -205.30447387695312,
"loss": 0.7592,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": 25.408042907714844,
"rewards/margins": 9.172635078430176,
"rewards/rejected": 16.235408782958984,
"step": 470
},
{
"epoch": 0.8,
"grad_norm": 21.772287368774414,
"learning_rate": 8.347428318751623e-06,
"logits/chosen": -1.44618821144104,
"logits/rejected": -1.3733718395233154,
"logps/chosen": -84.4385986328125,
"logps/rejected": -211.10861206054688,
"loss": 0.6625,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 26.974166870117188,
"rewards/margins": 11.247968673706055,
"rewards/rejected": 15.726198196411133,
"step": 480
},
{
"epoch": 0.82,
"grad_norm": 2.757871389389038,
"learning_rate": 8.281984495225938e-06,
"logits/chosen": -1.810275673866272,
"logits/rejected": -1.757764220237732,
"logps/chosen": -85.5127182006836,
"logps/rejected": -223.26803588867188,
"loss": 0.8267,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": 26.434656143188477,
"rewards/margins": 11.342155456542969,
"rewards/rejected": 15.092500686645508,
"step": 490
},
{
"epoch": 0.83,
"grad_norm": 23.952486038208008,
"learning_rate": 8.215537495106781e-06,
"logits/chosen": -2.0263965129852295,
"logits/rejected": -1.9776378870010376,
"logps/chosen": -78.3379898071289,
"logps/rejected": -242.8697967529297,
"loss": 0.655,
"rewards/accuracies": 0.9375,
"rewards/chosen": 28.698970794677734,
"rewards/margins": 14.983367919921875,
"rewards/rejected": 13.715606689453125,
"step": 500
},
{
"epoch": 0.85,
"grad_norm": 27.391632080078125,
"learning_rate": 8.148107628689736e-06,
"logits/chosen": -1.8597463369369507,
"logits/rejected": -1.8375170230865479,
"logps/chosen": -97.95671844482422,
"logps/rejected": -241.65005493164062,
"loss": 0.8206,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": 25.743389129638672,
"rewards/margins": 13.008593559265137,
"rewards/rejected": 12.734795570373535,
"step": 510
},
{
"epoch": 0.87,
"grad_norm": 32.186092376708984,
"learning_rate": 8.07971550669487e-06,
"logits/chosen": -1.7367337942123413,
"logits/rejected": -1.6837384700775146,
"logps/chosen": -97.98736572265625,
"logps/rejected": -236.2861785888672,
"loss": 0.7388,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 25.235164642333984,
"rewards/margins": 12.560583114624023,
"rewards/rejected": 12.674580574035645,
"step": 520
},
{
"epoch": 0.88,
"grad_norm": 2.0279977321624756,
"learning_rate": 8.01038203396682e-06,
"logits/chosen": -1.7110230922698975,
"logits/rejected": -1.6617202758789062,
"logps/chosen": -86.60490417480469,
"logps/rejected": -218.7451171875,
"loss": 0.7764,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": 27.18789291381836,
"rewards/margins": 12.605551719665527,
"rewards/rejected": 14.582344055175781,
"step": 530
},
{
"epoch": 0.9,
"grad_norm": 4.693509101867676,
"learning_rate": 7.940128403084979e-06,
"logits/chosen": -1.803138017654419,
"logits/rejected": -1.7637073993682861,
"logps/chosen": -82.79511260986328,
"logps/rejected": -221.371826171875,
"loss": 0.569,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 28.16043472290039,
"rewards/margins": 13.802328109741211,
"rewards/rejected": 14.358105659484863,
"step": 540
},
{
"epoch": 0.92,
"grad_norm": 27.00885772705078,
"learning_rate": 7.868976087885741e-06,
"logits/chosen": -1.769118309020996,
"logits/rejected": -1.7129123210906982,
"logps/chosen": -83.30874633789062,
"logps/rejected": -247.39993286132812,
"loss": 0.8273,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": 25.751117706298828,
"rewards/margins": 12.938650131225586,
"rewards/rejected": 12.812464714050293,
"step": 550
},
{
"epoch": 0.93,
"grad_norm": 36.580535888671875,
"learning_rate": 7.796946836898781e-06,
"logits/chosen": -1.8945682048797607,
"logits/rejected": -1.8643999099731445,
"logps/chosen": -83.80934143066406,
"logps/rejected": -228.9951629638672,
"loss": 0.9133,
"rewards/accuracies": 0.887499988079071,
"rewards/chosen": 25.17868995666504,
"rewards/margins": 11.825202941894531,
"rewards/rejected": 13.353485107421875,
"step": 560
},
{
"epoch": 0.95,
"grad_norm": 53.96921920776367,
"learning_rate": 7.724062666699359e-06,
"logits/chosen": -1.8547554016113281,
"logits/rejected": -1.8166589736938477,
"logps/chosen": -96.59284973144531,
"logps/rejected": -250.556640625,
"loss": 0.6751,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 25.961719512939453,
"rewards/margins": 14.7571382522583,
"rewards/rejected": 11.204582214355469,
"step": 570
},
{
"epoch": 0.97,
"grad_norm": 21.306289672851562,
"learning_rate": 7.650345855178695e-06,
"logits/chosen": -1.9091202020645142,
"logits/rejected": -1.8824758529663086,
"logps/chosen": -85.81441497802734,
"logps/rejected": -241.8865509033203,
"loss": 0.6111,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 25.982864379882812,
"rewards/margins": 13.588445663452148,
"rewards/rejected": 12.394417762756348,
"step": 580
},
{
"epoch": 0.98,
"grad_norm": 8.490675926208496,
"learning_rate": 7.575818934734482e-06,
"logits/chosen": -1.7473344802856445,
"logits/rejected": -1.7193949222564697,
"logps/chosen": -83.74988555908203,
"logps/rejected": -207.8963165283203,
"loss": 0.5498,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 26.92534828186035,
"rewards/margins": 12.920440673828125,
"rewards/rejected": 14.004905700683594,
"step": 590
},
{
"epoch": 1.0,
"grad_norm": 3.250255584716797,
"learning_rate": 7.500504685383589e-06,
"logits/chosen": -1.8129802942276,
"logits/rejected": -1.748167634010315,
"logps/chosen": -73.27245330810547,
"logps/rejected": -218.0284423828125,
"loss": 0.5127,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 27.97617530822754,
"rewards/margins": 13.350275993347168,
"rewards/rejected": 14.625898361206055,
"step": 600
},
{
"epoch": 1.0,
"eval_logits/chosen": -0.8561357855796814,
"eval_logits/rejected": -0.8501406908035278,
"eval_logps/chosen": -157.18006896972656,
"eval_logps/rejected": -196.33729553222656,
"eval_loss": 2.5238630771636963,
"eval_rewards/accuracies": 0.6000000238418579,
"eval_rewards/chosen": 15.406268119812012,
"eval_rewards/margins": 1.4132167100906372,
"eval_rewards/rejected": 13.993051528930664,
"eval_runtime": 959.2657,
"eval_samples_per_second": 0.735,
"eval_steps_per_second": 0.735,
"step": 600
},
{
"epoch": 1.02,
"grad_norm": 2.0936248302459717,
"learning_rate": 7.4244261277990935e-06,
"logits/chosen": -1.645000696182251,
"logits/rejected": -1.5917456150054932,
"logps/chosen": -73.57734680175781,
"logps/rejected": -235.9907684326172,
"loss": 0.4526,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 28.39139747619629,
"rewards/margins": 15.543986320495605,
"rewards/rejected": 12.847410202026367,
"step": 610
},
{
"epoch": 1.03,
"grad_norm": 3.0842278003692627,
"learning_rate": 7.347606516273741e-06,
"logits/chosen": -1.7528049945831299,
"logits/rejected": -1.7014877796173096,
"logps/chosen": -78.36769104003906,
"logps/rejected": -245.0220947265625,
"loss": 0.463,
"rewards/accuracies": 1.0,
"rewards/chosen": 28.718791961669922,
"rewards/margins": 15.73725700378418,
"rewards/rejected": 12.981534004211426,
"step": 620
},
{
"epoch": 1.05,
"grad_norm": 3.525000810623169,
"learning_rate": 7.270069331612e-06,
"logits/chosen": -1.6429353952407837,
"logits/rejected": -1.5838866233825684,
"logps/chosen": -79.90181732177734,
"logps/rejected": -221.4831085205078,
"loss": 0.4782,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 26.645065307617188,
"rewards/margins": 13.249574661254883,
"rewards/rejected": 13.395492553710938,
"step": 630
},
{
"epoch": 1.07,
"grad_norm": 24.071266174316406,
"learning_rate": 7.1918382739528804e-06,
"logits/chosen": -1.838573694229126,
"logits/rejected": -1.77816903591156,
"logps/chosen": -75.74369812011719,
"logps/rejected": -250.9163055419922,
"loss": 0.4283,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 27.663928985595703,
"rewards/margins": 14.643702507019043,
"rewards/rejected": 13.020225524902344,
"step": 640
},
{
"epoch": 1.08,
"grad_norm": 2.601551055908203,
"learning_rate": 7.112937255525722e-06,
"logits/chosen": -1.651479721069336,
"logits/rejected": -1.5933506488800049,
"logps/chosen": -69.61723327636719,
"logps/rejected": -234.6320037841797,
"loss": 0.4144,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 28.57497215270996,
"rewards/margins": 15.81470775604248,
"rewards/rejected": 12.760263442993164,
"step": 650
},
{
"epoch": 1.1,
"grad_norm": 2.3393607139587402,
"learning_rate": 7.033390393341133e-06,
"logits/chosen": -1.9843261241912842,
"logits/rejected": -1.8986890316009521,
"logps/chosen": -56.452239990234375,
"logps/rejected": -254.9853057861328,
"loss": 0.338,
"rewards/accuracies": 1.0,
"rewards/chosen": 28.793537139892578,
"rewards/margins": 16.600866317749023,
"rewards/rejected": 12.192671775817871,
"step": 660
},
{
"epoch": 1.12,
"grad_norm": 23.764225006103516,
"learning_rate": 6.953222001819347e-06,
"logits/chosen": -1.808882474899292,
"logits/rejected": -1.7674134969711304,
"logps/chosen": -70.89783477783203,
"logps/rejected": -221.87173461914062,
"loss": 0.4405,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 29.50577163696289,
"rewards/margins": 16.218791961669922,
"rewards/rejected": 13.286974906921387,
"step": 670
},
{
"epoch": 1.13,
"grad_norm": 5.011995315551758,
"learning_rate": 6.87245658535825e-06,
"logits/chosen": -1.7610832452774048,
"logits/rejected": -1.7023169994354248,
"logps/chosen": -64.40223693847656,
"logps/rejected": -239.07498168945312,
"loss": 0.3924,
"rewards/accuracies": 1.0,
"rewards/chosen": 28.532089233398438,
"rewards/margins": 15.520894050598145,
"rewards/rejected": 13.011195182800293,
"step": 680
},
{
"epoch": 1.15,
"grad_norm": 3.537167549133301,
"learning_rate": 6.791118830843311e-06,
"logits/chosen": -1.6986154317855835,
"logits/rejected": -1.6274213790893555,
"logps/chosen": -73.15770721435547,
"logps/rejected": -220.58633422851562,
"loss": 0.5515,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 27.489849090576172,
"rewards/margins": 13.804224967956543,
"rewards/rejected": 13.68562126159668,
"step": 690
},
{
"epoch": 1.17,
"grad_norm": 7.608737945556641,
"learning_rate": 6.709233600101761e-06,
"logits/chosen": -1.7815144062042236,
"logits/rejected": -1.7079941034317017,
"logps/chosen": -64.2397689819336,
"logps/rejected": -222.42898559570312,
"loss": 0.374,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.80181884765625,
"rewards/margins": 15.93518352508545,
"rewards/rejected": 13.8666353225708,
"step": 700
},
{
"epoch": 1.18,
"grad_norm": 7.344120025634766,
"learning_rate": 6.626825922303287e-06,
"logits/chosen": -1.482596755027771,
"logits/rejected": -1.398471713066101,
"logps/chosen": -69.20109558105469,
"logps/rejected": -227.3030548095703,
"loss": 0.4768,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 26.281591415405273,
"rewards/margins": 13.225834846496582,
"rewards/rejected": 13.055760383605957,
"step": 710
},
{
"epoch": 1.2,
"grad_norm": 2.493931293487549,
"learning_rate": 6.5439209863095675e-06,
"logits/chosen": -1.86245596408844,
"logits/rejected": -1.805001974105835,
"logps/chosen": -58.1787223815918,
"logps/rejected": -235.18923950195312,
"loss": 0.3453,
"rewards/accuracies": 1.0,
"rewards/chosen": 28.87502098083496,
"rewards/margins": 16.159029006958008,
"rewards/rejected": 12.715993881225586,
"step": 720
},
{
"epoch": 1.22,
"grad_norm": 5.012514114379883,
"learning_rate": 6.460544132975014e-06,
"logits/chosen": -1.6647535562515259,
"logits/rejected": -1.6346346139907837,
"logps/chosen": -67.99676513671875,
"logps/rejected": -234.99331665039062,
"loss": 0.4353,
"rewards/accuracies": 1.0,
"rewards/chosen": 27.812246322631836,
"rewards/margins": 16.04288673400879,
"rewards/rejected": 11.76936149597168,
"step": 730
},
{
"epoch": 1.23,
"grad_norm": 25.745906829833984,
"learning_rate": 6.376720847401042e-06,
"logits/chosen": -1.816080093383789,
"logits/rejected": -1.754172921180725,
"logps/chosen": -61.254425048828125,
"logps/rejected": -244.97998046875,
"loss": 0.3916,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 28.01883888244629,
"rewards/margins": 16.0445499420166,
"rewards/rejected": 11.974291801452637,
"step": 740
},
{
"epoch": 1.25,
"grad_norm": 3.14039945602417,
"learning_rate": 6.292476751146255e-06,
"logits/chosen": -1.8254092931747437,
"logits/rejected": -1.7369928359985352,
"logps/chosen": -59.7559814453125,
"logps/rejected": -227.97384643554688,
"loss": 0.41,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 29.0753173828125,
"rewards/margins": 16.199630737304688,
"rewards/rejected": 12.875683784484863,
"step": 750
},
{
"epoch": 1.27,
"grad_norm": 4.181332588195801,
"learning_rate": 6.207837594394913e-06,
"logits/chosen": -1.819637656211853,
"logits/rejected": -1.7874256372451782,
"logps/chosen": -60.003814697265625,
"logps/rejected": -217.7976531982422,
"loss": 0.3721,
"rewards/accuracies": 1.0,
"rewards/chosen": 28.516149520874023,
"rewards/margins": 15.223426818847656,
"rewards/rejected": 13.292726516723633,
"step": 760
},
{
"epoch": 1.28,
"grad_norm": 11.886590003967285,
"learning_rate": 6.1228292480861e-06,
"logits/chosen": -1.7983691692352295,
"logits/rejected": -1.707035779953003,
"logps/chosen": -70.19294738769531,
"logps/rejected": -250.7736358642578,
"loss": 0.5082,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 28.486637115478516,
"rewards/margins": 16.149328231811523,
"rewards/rejected": 12.337307929992676,
"step": 770
},
{
"epoch": 1.3,
"grad_norm": 2.3090531826019287,
"learning_rate": 6.037477696005966e-06,
"logits/chosen": -1.7597004175186157,
"logits/rejected": -1.6807353496551514,
"logps/chosen": -61.509925842285156,
"logps/rejected": -227.34652709960938,
"loss": 0.4032,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.05527687072754,
"rewards/margins": 16.505048751831055,
"rewards/rejected": 12.550226211547852,
"step": 780
},
{
"epoch": 1.32,
"grad_norm": 2.6496028900146484,
"learning_rate": 5.95180902684548e-06,
"logits/chosen": -1.6354789733886719,
"logits/rejected": -1.545203447341919,
"logps/chosen": -61.32403564453125,
"logps/rejected": -239.3839874267578,
"loss": 0.396,
"rewards/accuracies": 1.0,
"rewards/chosen": 28.665252685546875,
"rewards/margins": 16.576025009155273,
"rewards/rejected": 12.08923053741455,
"step": 790
},
{
"epoch": 1.33,
"grad_norm": 2.990304946899414,
"learning_rate": 5.8658494262261215e-06,
"logits/chosen": -1.8462188243865967,
"logits/rejected": -1.7873852252960205,
"logps/chosen": -64.35298919677734,
"logps/rejected": -239.20401000976562,
"loss": 0.3787,
"rewards/accuracies": 1.0,
"rewards/chosen": 28.686588287353516,
"rewards/margins": 15.4246826171875,
"rewards/rejected": 13.2619047164917,
"step": 800
},
{
"epoch": 1.33,
"eval_logits/chosen": -0.8357726335525513,
"eval_logits/rejected": -0.8384636044502258,
"eval_logps/chosen": -158.54798889160156,
"eval_logps/rejected": -197.1554718017578,
"eval_loss": 2.5951449871063232,
"eval_rewards/accuracies": 0.6141843795776367,
"eval_rewards/chosen": 15.269478797912598,
"eval_rewards/margins": 1.3582426309585571,
"eval_rewards/rejected": 13.911234855651855,
"eval_runtime": 933.4217,
"eval_samples_per_second": 0.755,
"eval_steps_per_second": 0.755,
"step": 800
},
{
"epoch": 1.35,
"grad_norm": 29.393373489379883,
"learning_rate": 5.779625168695943e-06,
"logits/chosen": -1.8400392532348633,
"logits/rejected": -1.7939599752426147,
"logps/chosen": -53.9841423034668,
"logps/rejected": -231.19735717773438,
"loss": 0.3973,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 29.610637664794922,
"rewards/margins": 16.809568405151367,
"rewards/rejected": 12.801069259643555,
"step": 810
},
{
"epoch": 1.37,
"grad_norm": 3.947079658508301,
"learning_rate": 5.6931626096984475e-06,
"logits/chosen": -1.7477422952651978,
"logits/rejected": -1.700312614440918,
"logps/chosen": -60.37701416015625,
"logps/rejected": -254.6928253173828,
"loss": 0.37,
"rewards/accuracies": 1.0,
"rewards/chosen": 27.82967185974121,
"rewards/margins": 15.517053604125977,
"rewards/rejected": 12.31261920928955,
"step": 820
},
{
"epoch": 1.38,
"grad_norm": 19.618087768554688,
"learning_rate": 5.6064881775167445e-06,
"logits/chosen": -1.7533241510391235,
"logits/rejected": -1.6893295049667358,
"logps/chosen": -61.474517822265625,
"logps/rejected": -223.3421173095703,
"loss": 0.4478,
"rewards/accuracies": 0.9624999761581421,
"rewards/chosen": 27.814483642578125,
"rewards/margins": 14.598594665527344,
"rewards/rejected": 13.215890884399414,
"step": 830
},
{
"epoch": 1.4,
"grad_norm": 3.5263216495513916,
"learning_rate": 5.5196283651954375e-06,
"logits/chosen": -1.5104930400848389,
"logits/rejected": -1.413916826248169,
"logps/chosen": -70.25590515136719,
"logps/rejected": -236.75833129882812,
"loss": 0.4924,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 28.405406951904297,
"rewards/margins": 15.316106796264648,
"rewards/rejected": 13.089299201965332,
"step": 840
},
{
"epoch": 1.42,
"grad_norm": 3.5469884872436523,
"learning_rate": 5.432609722442715e-06,
"logits/chosen": -1.7516591548919678,
"logits/rejected": -1.6808538436889648,
"logps/chosen": -69.64612579345703,
"logps/rejected": -237.8245086669922,
"loss": 0.4107,
"rewards/accuracies": 1.0,
"rewards/chosen": 28.481128692626953,
"rewards/margins": 16.653011322021484,
"rewards/rejected": 11.828117370605469,
"step": 850
},
{
"epoch": 1.43,
"grad_norm": 4.012500286102295,
"learning_rate": 5.345458847515133e-06,
"logits/chosen": -1.3920238018035889,
"logits/rejected": -1.284397840499878,
"logps/chosen": -67.01615905761719,
"logps/rejected": -207.93798828125,
"loss": 0.4615,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 27.689502716064453,
"rewards/margins": 13.398959159851074,
"rewards/rejected": 14.290542602539062,
"step": 860
},
{
"epoch": 1.45,
"grad_norm": 2.9533350467681885,
"learning_rate": 5.258202379087537e-06,
"logits/chosen": -1.7657880783081055,
"logits/rejected": -1.693655014038086,
"logps/chosen": -64.3235092163086,
"logps/rejected": -251.1502227783203,
"loss": 0.4456,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 28.66314125061035,
"rewards/margins": 16.5991268157959,
"rewards/rejected": 12.06401252746582,
"step": 870
},
{
"epoch": 1.47,
"grad_norm": 3.6378014087677,
"learning_rate": 5.170866988110656e-06,
"logits/chosen": -1.7154031991958618,
"logits/rejected": -1.657488226890564,
"logps/chosen": -66.95178985595703,
"logps/rejected": -236.646240234375,
"loss": 0.4076,
"rewards/accuracies": 1.0,
"rewards/chosen": 28.9368839263916,
"rewards/margins": 16.125226974487305,
"rewards/rejected": 12.811657905578613,
"step": 880
},
{
"epoch": 1.48,
"grad_norm": 2.8945438861846924,
"learning_rate": 5.083479369658807e-06,
"logits/chosen": -1.696937918663025,
"logits/rejected": -1.6348483562469482,
"logps/chosen": -58.26952362060547,
"logps/rejected": -230.33377075195312,
"loss": 0.3785,
"rewards/accuracies": 1.0,
"rewards/chosen": 28.738727569580078,
"rewards/margins": 15.88463306427002,
"rewards/rejected": 12.854090690612793,
"step": 890
},
{
"epoch": 1.5,
"grad_norm": 2.305314302444458,
"learning_rate": 4.9960662347702405e-06,
"logits/chosen": -1.5210245847702026,
"logits/rejected": -1.4456034898757935,
"logps/chosen": -54.75811004638672,
"logps/rejected": -232.2668914794922,
"loss": 0.3659,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 29.211429595947266,
"rewards/margins": 16.932321548461914,
"rewards/rejected": 12.279109954833984,
"step": 900
},
{
"epoch": 1.52,
"grad_norm": 30.96356964111328,
"learning_rate": 4.908654302282602e-06,
"logits/chosen": -1.9021522998809814,
"logits/rejected": -1.8719444274902344,
"logps/chosen": -55.43334197998047,
"logps/rejected": -241.4973602294922,
"loss": 0.3925,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 29.54325294494629,
"rewards/margins": 18.098033905029297,
"rewards/rejected": 11.44521713256836,
"step": 910
},
{
"epoch": 1.53,
"grad_norm": 2.172999858856201,
"learning_rate": 4.821270290666007e-06,
"logits/chosen": -1.9638334512710571,
"logits/rejected": -1.933005928993225,
"logps/chosen": -50.753273010253906,
"logps/rejected": -264.568603515625,
"loss": 0.3296,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 29.600269317626953,
"rewards/margins": 18.301733016967773,
"rewards/rejected": 11.29853630065918,
"step": 920
},
{
"epoch": 1.55,
"grad_norm": 3.1367807388305664,
"learning_rate": 4.733940909856239e-06,
"logits/chosen": -1.8073303699493408,
"logits/rejected": -1.769351601600647,
"logps/chosen": -56.24708938598633,
"logps/rejected": -268.60906982421875,
"loss": 0.349,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.238876342773438,
"rewards/margins": 17.679536819458008,
"rewards/rejected": 11.559335708618164,
"step": 930
},
{
"epoch": 1.57,
"grad_norm": 2.4563372135162354,
"learning_rate": 4.646692853090539e-06,
"logits/chosen": -1.8005622625350952,
"logits/rejected": -1.783666968345642,
"logps/chosen": -58.6412467956543,
"logps/rejected": -242.1209716796875,
"loss": 0.3657,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.608911514282227,
"rewards/margins": 18.011024475097656,
"rewards/rejected": 11.597890853881836,
"step": 940
},
{
"epoch": 1.58,
"grad_norm": 2.5796990394592285,
"learning_rate": 4.559552788748507e-06,
"logits/chosen": -2.053856134414673,
"logits/rejected": -2.0160202980041504,
"logps/chosen": -49.09900665283203,
"logps/rejected": -269.3837890625,
"loss": 0.2841,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.68400001525879,
"rewards/margins": 19.868282318115234,
"rewards/rejected": 10.815717697143555,
"step": 950
},
{
"epoch": 1.6,
"grad_norm": 2.9965362548828125,
"learning_rate": 4.472547352200615e-06,
"logits/chosen": -1.8396308422088623,
"logits/rejected": -1.8210567235946655,
"logps/chosen": -52.045127868652344,
"logps/rejected": -257.56011962890625,
"loss": 0.3359,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 29.730632781982422,
"rewards/margins": 18.14250373840332,
"rewards/rejected": 11.58813190460205,
"step": 960
},
{
"epoch": 1.62,
"grad_norm": 3.3091185092926025,
"learning_rate": 4.385703137666784e-06,
"logits/chosen": -1.7598381042480469,
"logits/rejected": -1.7196592092514038,
"logps/chosen": -55.36179733276367,
"logps/rejected": -260.2227478027344,
"loss": 0.3407,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.468708038330078,
"rewards/margins": 19.425325393676758,
"rewards/rejected": 11.043380737304688,
"step": 970
},
{
"epoch": 1.63,
"grad_norm": 1.9838672876358032,
"learning_rate": 4.2990466900875625e-06,
"logits/chosen": -1.8335222005844116,
"logits/rejected": -1.776532769203186,
"logps/chosen": -51.03464889526367,
"logps/rejected": -237.20083618164062,
"loss": 0.3608,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 29.984943389892578,
"rewards/margins": 17.165090560913086,
"rewards/rejected": 12.819851875305176,
"step": 980
},
{
"epoch": 1.65,
"grad_norm": 3.0028090476989746,
"learning_rate": 4.212604497010346e-06,
"logits/chosen": -1.7095162868499756,
"logits/rejected": -1.6648433208465576,
"logps/chosen": -48.533897399902344,
"logps/rejected": -234.913330078125,
"loss": 0.3397,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 30.055179595947266,
"rewards/margins": 17.712875366210938,
"rewards/rejected": 12.342303276062012,
"step": 990
},
{
"epoch": 1.67,
"grad_norm": 11.127403259277344,
"learning_rate": 4.126402980493171e-06,
"logits/chosen": -1.7350540161132812,
"logits/rejected": -1.6723353862762451,
"logps/chosen": -60.74663543701172,
"logps/rejected": -234.6821746826172,
"loss": 0.381,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.135547637939453,
"rewards/margins": 16.059070587158203,
"rewards/rejected": 13.07647705078125,
"step": 1000
},
{
"epoch": 1.67,
"eval_logits/chosen": -0.7807645797729492,
"eval_logits/rejected": -0.7846214771270752,
"eval_logps/chosen": -161.0571746826172,
"eval_logps/rejected": -201.4547576904297,
"eval_loss": 2.581395387649536,
"eval_rewards/accuracies": 0.6212766170501709,
"eval_rewards/chosen": 15.018560409545898,
"eval_rewards/margins": 1.537255883216858,
"eval_rewards/rejected": 13.481305122375488,
"eval_runtime": 933.6733,
"eval_samples_per_second": 0.755,
"eval_steps_per_second": 0.755,
"step": 1000
},
{
"epoch": 1.68,
"grad_norm": 2.986391305923462,
"learning_rate": 4.0404684890284815e-06,
"logits/chosen": -1.6968259811401367,
"logits/rejected": -1.62332022190094,
"logps/chosen": -54.682281494140625,
"logps/rejected": -259.25335693359375,
"loss": 0.3598,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 30.244094848632812,
"rewards/margins": 18.490741729736328,
"rewards/rejected": 11.753351211547852,
"step": 1010
},
{
"epoch": 1.7,
"grad_norm": 3.1635079383850098,
"learning_rate": 3.954827289489429e-06,
"logits/chosen": -1.8833897113800049,
"logits/rejected": -1.7493565082550049,
"logps/chosen": -53.182655334472656,
"logps/rejected": -240.314697265625,
"loss": 0.3551,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 30.2075252532959,
"rewards/margins": 17.21380615234375,
"rewards/rejected": 12.993721008300781,
"step": 1020
},
{
"epoch": 1.72,
"grad_norm": 2.4462807178497314,
"learning_rate": 3.86950555910108e-06,
"logits/chosen": -1.6473881006240845,
"logits/rejected": -1.5638983249664307,
"logps/chosen": -54.882896423339844,
"logps/rejected": -230.82162475585938,
"loss": 0.3772,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 29.62346839904785,
"rewards/margins": 15.928651809692383,
"rewards/rejected": 13.694814682006836,
"step": 1030
},
{
"epoch": 1.73,
"grad_norm": 4.168172836303711,
"learning_rate": 3.784529377439067e-06,
"logits/chosen": -1.6475791931152344,
"logits/rejected": -1.5899393558502197,
"logps/chosen": -59.67030715942383,
"logps/rejected": -221.40066528320312,
"loss": 0.3826,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 29.023029327392578,
"rewards/margins": 16.1686954498291,
"rewards/rejected": 12.854331970214844,
"step": 1040
},
{
"epoch": 1.75,
"grad_norm": 17.74908447265625,
"learning_rate": 3.699924718458036e-06,
"logits/chosen": -1.7890431880950928,
"logits/rejected": -1.7369229793548584,
"logps/chosen": -48.85768127441406,
"logps/rejected": -251.3222198486328,
"loss": 0.3095,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 30.195240020751953,
"rewards/margins": 17.660083770751953,
"rewards/rejected": 12.53515338897705,
"step": 1050
},
{
"epoch": 1.77,
"grad_norm": 6.809030532836914,
"learning_rate": 3.615717442552429e-06,
"logits/chosen": -1.7820018529891968,
"logits/rejected": -1.7489522695541382,
"logps/chosen": -54.51831817626953,
"logps/rejected": -246.6309356689453,
"loss": 0.337,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.5898380279541,
"rewards/margins": 18.221317291259766,
"rewards/rejected": 11.368522644042969,
"step": 1060
},
{
"epoch": 1.78,
"grad_norm": 7.795753002166748,
"learning_rate": 3.5319332886519393e-06,
"logits/chosen": -1.6910899877548218,
"logits/rejected": -1.6332054138183594,
"logps/chosen": -46.748443603515625,
"logps/rejected": -252.2515106201172,
"loss": 0.3168,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.91961669921875,
"rewards/margins": 18.060075759887695,
"rewards/rejected": 11.85954475402832,
"step": 1070
},
{
"epoch": 1.8,
"grad_norm": 4.77955436706543,
"learning_rate": 3.4485978663541233e-06,
"logits/chosen": -1.4919878244400024,
"logits/rejected": -1.4713367223739624,
"logps/chosen": -66.22251892089844,
"logps/rejected": -239.391357421875,
"loss": 0.4289,
"rewards/accuracies": 1.0,
"rewards/chosen": 28.419330596923828,
"rewards/margins": 17.399511337280273,
"rewards/rejected": 11.019818305969238,
"step": 1080
},
{
"epoch": 1.82,
"grad_norm": 2.38405179977417,
"learning_rate": 3.3657366480965158e-06,
"logits/chosen": -1.8986093997955322,
"logits/rejected": -1.8109142780303955,
"logps/chosen": -47.243350982666016,
"logps/rejected": -263.37347412109375,
"loss": 0.2852,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.1839656829834,
"rewards/margins": 18.492839813232422,
"rewards/rejected": 11.691123962402344,
"step": 1090
},
{
"epoch": 1.83,
"grad_norm": 5.4925994873046875,
"learning_rate": 3.2833749613706988e-06,
"logits/chosen": -1.583234429359436,
"logits/rejected": -1.497194766998291,
"logps/chosen": -58.58795166015625,
"logps/rejected": -240.7587432861328,
"loss": 0.4343,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 27.720067977905273,
"rewards/margins": 14.912628173828125,
"rewards/rejected": 12.807439804077148,
"step": 1100
},
{
"epoch": 1.85,
"grad_norm": 3.2018399238586426,
"learning_rate": 3.201537980980646e-06,
"logits/chosen": -1.3224772214889526,
"logits/rejected": -1.2666256427764893,
"logps/chosen": -54.696876525878906,
"logps/rejected": -243.18539428710938,
"loss": 0.3813,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 28.097915649414062,
"rewards/margins": 16.58795738220215,
"rewards/rejected": 11.50995922088623,
"step": 1110
},
{
"epoch": 1.87,
"grad_norm": 4.00109338760376,
"learning_rate": 3.1202507213477658e-06,
"logits/chosen": -1.6143379211425781,
"logits/rejected": -1.5576748847961426,
"logps/chosen": -56.855003356933594,
"logps/rejected": -233.1626739501953,
"loss": 0.3651,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.580801010131836,
"rewards/margins": 18.555524826049805,
"rewards/rejected": 12.025275230407715,
"step": 1120
},
{
"epoch": 1.88,
"grad_norm": 5.0345635414123535,
"learning_rate": 3.039538028864939e-06,
"logits/chosen": -1.6249618530273438,
"logits/rejected": -1.6053358316421509,
"logps/chosen": -57.446815490722656,
"logps/rejected": -246.0026092529297,
"loss": 0.3495,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.101943969726562,
"rewards/margins": 17.34486198425293,
"rewards/rejected": 11.757081985473633,
"step": 1130
},
{
"epoch": 1.9,
"grad_norm": 2.065876007080078,
"learning_rate": 2.9594245743019477e-06,
"logits/chosen": -1.8201977014541626,
"logits/rejected": -1.8249000310897827,
"logps/chosen": -52.830360412597656,
"logps/rejected": -243.8861083984375,
"loss": 0.3389,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": 31.09743881225586,
"rewards/margins": 18.6008358001709,
"rewards/rejected": 12.496602058410645,
"step": 1140
},
{
"epoch": 1.92,
"grad_norm": 4.4573588371276855,
"learning_rate": 2.8799348452645515e-06,
"logits/chosen": -1.7230621576309204,
"logits/rejected": -1.6711606979370117,
"logps/chosen": -52.5847282409668,
"logps/rejected": -234.39968872070312,
"loss": 0.3286,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.327728271484375,
"rewards/margins": 18.147693634033203,
"rewards/rejected": 12.180032730102539,
"step": 1150
},
{
"epoch": 1.93,
"grad_norm": 14.367749214172363,
"learning_rate": 2.801093138709582e-06,
"logits/chosen": -1.7738536596298218,
"logits/rejected": -1.687819480895996,
"logps/chosen": -49.588436126708984,
"logps/rejected": -238.64501953125,
"loss": 0.3485,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 29.91376304626465,
"rewards/margins": 17.238935470581055,
"rewards/rejected": 12.674827575683594,
"step": 1160
},
{
"epoch": 1.95,
"grad_norm": 9.441333770751953,
"learning_rate": 2.722923553518285e-06,
"logits/chosen": -1.5205357074737549,
"logits/rejected": -1.480017900466919,
"logps/chosen": -57.287017822265625,
"logps/rejected": -243.42919921875,
"loss": 0.3522,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.02022933959961,
"rewards/margins": 17.072509765625,
"rewards/rejected": 11.947721481323242,
"step": 1170
},
{
"epoch": 1.97,
"grad_norm": 12.407715797424316,
"learning_rate": 2.6454499831302223e-06,
"logits/chosen": -1.6342418193817139,
"logits/rejected": -1.5738509893417358,
"logps/chosen": -60.34296798706055,
"logps/rejected": -226.05990600585938,
"loss": 0.5105,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 30.577117919921875,
"rewards/margins": 17.6352481842041,
"rewards/rejected": 12.941868782043457,
"step": 1180
},
{
"epoch": 1.98,
"grad_norm": 3.367372512817383,
"learning_rate": 2.5686961082399716e-06,
"logits/chosen": -1.8151267766952515,
"logits/rejected": -1.7858302593231201,
"logps/chosen": -57.720130920410156,
"logps/rejected": -260.25494384765625,
"loss": 0.3546,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.963693618774414,
"rewards/margins": 18.514841079711914,
"rewards/rejected": 11.44885540008545,
"step": 1190
},
{
"epoch": 2.0,
"grad_norm": 3.254979133605957,
"learning_rate": 2.4926853895588343e-06,
"logits/chosen": -1.7511732578277588,
"logits/rejected": -1.6836206912994385,
"logps/chosen": -41.321022033691406,
"logps/rejected": -248.452880859375,
"loss": 0.2993,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 31.503093719482422,
"rewards/margins": 19.927473068237305,
"rewards/rejected": 11.575621604919434,
"step": 1200
},
{
"epoch": 2.0,
"eval_logits/chosen": -0.7554205060005188,
"eval_logits/rejected": -0.758985161781311,
"eval_logps/chosen": -160.93551635742188,
"eval_logps/rejected": -202.3504638671875,
"eval_loss": 2.5816383361816406,
"eval_rewards/accuracies": 0.6382978558540344,
"eval_rewards/chosen": 15.030726432800293,
"eval_rewards/margins": 1.6389882564544678,
"eval_rewards/rejected": 13.39173698425293,
"eval_runtime": 930.5524,
"eval_samples_per_second": 0.758,
"eval_steps_per_second": 0.758,
"step": 1200
},
{
"epoch": 2.02,
"grad_norm": 2.5797388553619385,
"learning_rate": 2.417441060643809e-06,
"logits/chosen": -1.783384084701538,
"logits/rejected": -1.7092987298965454,
"logps/chosen": -39.536407470703125,
"logps/rejected": -250.8395538330078,
"loss": 0.2602,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.661413192749023,
"rewards/margins": 19.599422454833984,
"rewards/rejected": 12.061990737915039,
"step": 1210
},
{
"epoch": 2.03,
"grad_norm": 3.2678422927856445,
"learning_rate": 2.342986120795978e-06,
"logits/chosen": -1.6378501653671265,
"logits/rejected": -1.6011245250701904,
"logps/chosen": -50.96418762207031,
"logps/rejected": -246.3638153076172,
"loss": 0.3088,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.954126358032227,
"rewards/margins": 19.514755249023438,
"rewards/rejected": 11.439367294311523,
"step": 1220
},
{
"epoch": 2.05,
"grad_norm": 2.6984269618988037,
"learning_rate": 2.2693433280305127e-06,
"logits/chosen": -1.7482858896255493,
"logits/rejected": -1.7367308139801025,
"logps/chosen": -43.29402160644531,
"logps/rejected": -253.78939819335938,
"loss": 0.2572,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.882701873779297,
"rewards/margins": 21.47805404663086,
"rewards/rejected": 10.40464973449707,
"step": 1230
},
{
"epoch": 2.07,
"grad_norm": 3.53029727935791,
"learning_rate": 2.19653519212041e-06,
"logits/chosen": -1.7963911294937134,
"logits/rejected": -1.7613227367401123,
"logps/chosen": -44.7977294921875,
"logps/rejected": -243.6311798095703,
"loss": 0.2862,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.34331703186035,
"rewards/margins": 19.52144432067871,
"rewards/rejected": 10.821873664855957,
"step": 1240
},
{
"epoch": 2.08,
"grad_norm": 1.513169288635254,
"learning_rate": 2.124583967716136e-06,
"logits/chosen": -2.0151898860931396,
"logits/rejected": -1.9886600971221924,
"logps/chosen": -38.372108459472656,
"logps/rejected": -267.5242919921875,
"loss": 0.2252,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.756805419921875,
"rewards/margins": 20.560382843017578,
"rewards/rejected": 11.196426391601562,
"step": 1250
},
{
"epoch": 2.1,
"grad_norm": 2.537264347076416,
"learning_rate": 2.053511647543218e-06,
"logits/chosen": -1.4479506015777588,
"logits/rejected": -1.408527135848999,
"logps/chosen": -49.19340896606445,
"logps/rejected": -229.5871124267578,
"loss": 0.3208,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.4880428314209,
"rewards/margins": 17.939146041870117,
"rewards/rejected": 11.548896789550781,
"step": 1260
},
{
"epoch": 2.12,
"grad_norm": 2.1548752784729004,
"learning_rate": 1.98333995567992e-06,
"logits/chosen": -1.5170295238494873,
"logits/rejected": -1.4900450706481934,
"logps/chosen": -50.18815231323242,
"logps/rejected": -242.35986328125,
"loss": 0.3042,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.04024887084961,
"rewards/margins": 19.42563247680664,
"rewards/rejected": 11.614619255065918,
"step": 1270
},
{
"epoch": 2.13,
"grad_norm": 2.6000428199768066,
"learning_rate": 1.9140903409170276e-06,
"logits/chosen": -1.59740149974823,
"logits/rejected": -1.5533560514450073,
"logps/chosen": -49.30552291870117,
"logps/rejected": -255.7830810546875,
"loss": 0.3064,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.910289764404297,
"rewards/margins": 19.63786506652832,
"rewards/rejected": 11.272429466247559,
"step": 1280
},
{
"epoch": 2.15,
"grad_norm": 2.6881704330444336,
"learning_rate": 1.8457839702017783e-06,
"logits/chosen": -1.788030982017517,
"logits/rejected": -1.7674751281738281,
"logps/chosen": -38.47369384765625,
"logps/rejected": -241.8249053955078,
"loss": 0.2368,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.96920394897461,
"rewards/margins": 19.61397933959961,
"rewards/rejected": 11.355224609375,
"step": 1290
},
{
"epoch": 2.17,
"grad_norm": 1.9455982446670532,
"learning_rate": 1.7784417221679346e-06,
"logits/chosen": -1.7658576965332031,
"logits/rejected": -1.7436749935150146,
"logps/chosen": -43.32396697998047,
"logps/rejected": -260.0502624511719,
"loss": 0.2658,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.932819366455078,
"rewards/margins": 20.97707176208496,
"rewards/rejected": 9.955748558044434,
"step": 1300
},
{
"epoch": 2.18,
"grad_norm": 2.7157328128814697,
"learning_rate": 1.7120841807539867e-06,
"logits/chosen": -1.5605119466781616,
"logits/rejected": -1.533529281616211,
"logps/chosen": -45.2177848815918,
"logps/rejected": -252.9283905029297,
"loss": 0.2779,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.456167221069336,
"rewards/margins": 20.617300033569336,
"rewards/rejected": 10.8388671875,
"step": 1310
},
{
"epoch": 2.2,
"grad_norm": 4.16888952255249,
"learning_rate": 1.6467316289114365e-06,
"logits/chosen": -1.7607736587524414,
"logits/rejected": -1.72702157497406,
"logps/chosen": -52.18560028076172,
"logps/rejected": -237.4579620361328,
"loss": 0.3116,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.68082618713379,
"rewards/margins": 19.22063446044922,
"rewards/rejected": 11.460187911987305,
"step": 1320
},
{
"epoch": 2.22,
"grad_norm": 3.37455153465271,
"learning_rate": 1.5824040424050763e-06,
"logits/chosen": -1.8833210468292236,
"logits/rejected": -1.844347357749939,
"logps/chosen": -40.77017593383789,
"logps/rejected": -248.3719940185547,
"loss": 0.2508,
"rewards/accuracies": 1.0,
"rewards/chosen": 32.0168342590332,
"rewards/margins": 21.67666244506836,
"rewards/rejected": 10.34017562866211,
"step": 1330
},
{
"epoch": 2.23,
"grad_norm": 1.9589617252349854,
"learning_rate": 1.5191210837071695e-06,
"logits/chosen": -1.8483455181121826,
"logits/rejected": -1.8569103479385376,
"logps/chosen": -37.26815414428711,
"logps/rejected": -267.3866882324219,
"loss": 0.231,
"rewards/accuracies": 1.0,
"rewards/chosen": 32.11620330810547,
"rewards/margins": 21.5185604095459,
"rewards/rejected": 10.597644805908203,
"step": 1340
},
{
"epoch": 2.25,
"grad_norm": 3.257993221282959,
"learning_rate": 1.4569020959873809e-06,
"logits/chosen": -1.786929726600647,
"logits/rejected": -1.7171766757965088,
"logps/chosen": -38.82440948486328,
"logps/rejected": -267.8984375,
"loss": 0.2323,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.542959213256836,
"rewards/margins": 20.44384002685547,
"rewards/rejected": 11.099119186401367,
"step": 1350
},
{
"epoch": 2.27,
"grad_norm": 2.4979026317596436,
"learning_rate": 1.3957660972003167e-06,
"logits/chosen": -1.6343905925750732,
"logits/rejected": -1.6573156118392944,
"logps/chosen": -45.56403732299805,
"logps/rejected": -253.07199096679688,
"loss": 0.2767,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.549020767211914,
"rewards/margins": 20.293630599975586,
"rewards/rejected": 10.255391120910645,
"step": 1360
},
{
"epoch": 2.28,
"grad_norm": 1.8096556663513184,
"learning_rate": 1.3357317742724658e-06,
"logits/chosen": -1.5966780185699463,
"logits/rejected": -1.570854902267456,
"logps/chosen": -48.08317184448242,
"logps/rejected": -251.5933380126953,
"loss": 0.2885,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.559154510498047,
"rewards/margins": 19.232830047607422,
"rewards/rejected": 11.326326370239258,
"step": 1370
},
{
"epoch": 2.3,
"grad_norm": 2.8283846378326416,
"learning_rate": 1.2768174773903263e-06,
"logits/chosen": -1.86501145362854,
"logits/rejected": -1.822446584701538,
"logps/chosen": -47.01121139526367,
"logps/rejected": -263.478515625,
"loss": 0.2764,
"rewards/accuracies": 1.0,
"rewards/chosen": 32.028045654296875,
"rewards/margins": 20.432361602783203,
"rewards/rejected": 11.595685958862305,
"step": 1380
},
{
"epoch": 2.32,
"grad_norm": 2.776090621948242,
"learning_rate": 1.2190412143914536e-06,
"logits/chosen": -1.513856291770935,
"logits/rejected": -1.46199631690979,
"logps/chosen": -42.10026931762695,
"logps/rejected": -247.57339477539062,
"loss": 0.2768,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.788198471069336,
"rewards/margins": 19.206623077392578,
"rewards/rejected": 10.58157730102539,
"step": 1390
},
{
"epoch": 2.33,
"grad_norm": 2.4359893798828125,
"learning_rate": 1.1624206452601623e-06,
"logits/chosen": -1.7050933837890625,
"logits/rejected": -1.6352756023406982,
"logps/chosen": -48.3925666809082,
"logps/rejected": -255.81640625,
"loss": 0.2917,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.2396240234375,
"rewards/margins": 19.301578521728516,
"rewards/rejected": 10.938047409057617,
"step": 1400
},
{
"epoch": 2.33,
"eval_logits/chosen": -0.8292198181152344,
"eval_logits/rejected": -0.8336656093597412,
"eval_logps/chosen": -165.87318420410156,
"eval_logps/rejected": -207.3828887939453,
"eval_loss": 2.626953363418579,
"eval_rewards/accuracies": 0.6425532102584839,
"eval_rewards/chosen": 14.536958694458008,
"eval_rewards/margins": 1.6484651565551758,
"eval_rewards/rejected": 12.888493537902832,
"eval_runtime": 931.7841,
"eval_samples_per_second": 0.757,
"eval_steps_per_second": 0.757,
"step": 1400
},
{
"epoch": 2.35,
"grad_norm": 2.7294886112213135,
"learning_rate": 1.1069730767295394e-06,
"logits/chosen": -1.7735283374786377,
"logits/rejected": -1.7543232440948486,
"logps/chosen": -40.81293869018555,
"logps/rejected": -259.8978576660156,
"loss": 0.2463,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.607662200927734,
"rewards/margins": 20.695466995239258,
"rewards/rejected": 10.912198066711426,
"step": 1410
},
{
"epoch": 2.37,
"grad_norm": 2.734895944595337,
"learning_rate": 1.0527154569914472e-06,
"logits/chosen": -1.5312741994857788,
"logits/rejected": -1.4923324584960938,
"logps/chosen": -42.63439178466797,
"logps/rejected": -244.82046508789062,
"loss": 0.2697,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.897241592407227,
"rewards/margins": 18.947063446044922,
"rewards/rejected": 10.950177192687988,
"step": 1420
},
{
"epoch": 2.38,
"grad_norm": 2.473545789718628,
"learning_rate": 9.996643705161125e-07,
"logits/chosen": -1.6525815725326538,
"logits/rejected": -1.6200284957885742,
"logps/chosen": -46.02938461303711,
"logps/rejected": -261.2445068359375,
"loss": 0.2916,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.908071517944336,
"rewards/margins": 19.2839298248291,
"rewards/rejected": 10.624141693115234,
"step": 1430
},
{
"epoch": 2.4,
"grad_norm": 2.676964044570923,
"learning_rate": 9.47836032982884e-07,
"logits/chosen": -1.5494722127914429,
"logits/rejected": -1.4936171770095825,
"logps/chosen": -47.933658599853516,
"logps/rejected": -252.27401733398438,
"loss": 0.3175,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.43538475036621,
"rewards/margins": 18.781505584716797,
"rewards/rejected": 10.653879165649414,
"step": 1440
},
{
"epoch": 2.42,
"grad_norm": 2.895836591720581,
"learning_rate": 8.972462863237341e-07,
"logits/chosen": -1.6493561267852783,
"logits/rejected": -1.5964713096618652,
"logps/chosen": -44.10541915893555,
"logps/rejected": -271.24859619140625,
"loss": 0.2734,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.411285400390625,
"rewards/margins": 19.59650993347168,
"rewards/rejected": 9.814775466918945,
"step": 1450
},
{
"epoch": 2.43,
"grad_norm": 1.9935178756713867,
"learning_rate": 8.479105938809701e-07,
"logits/chosen": -1.485527515411377,
"logits/rejected": -1.435868263244629,
"logps/chosen": -50.575355529785156,
"logps/rejected": -253.8799591064453,
"loss": 0.3184,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.476688385009766,
"rewards/margins": 18.307661056518555,
"rewards/rejected": 11.169027328491211,
"step": 1460
},
{
"epoch": 2.45,
"grad_norm": 2.0916085243225098,
"learning_rate": 7.998440356807075e-07,
"logits/chosen": -1.8527294397354126,
"logits/rejected": -1.8244997262954712,
"logps/chosen": -34.15217208862305,
"logps/rejected": -262.5540466308594,
"loss": 0.2042,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.451465606689453,
"rewards/margins": 20.44220542907715,
"rewards/rejected": 11.009258270263672,
"step": 1470
},
{
"epoch": 2.47,
"grad_norm": 3.5937747955322266,
"learning_rate": 7.530613038234646e-07,
"logits/chosen": -1.5336072444915771,
"logits/rejected": -1.4814928770065308,
"logps/chosen": -42.957305908203125,
"logps/rejected": -248.6365966796875,
"loss": 0.2776,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.6094970703125,
"rewards/margins": 20.037199020385742,
"rewards/rejected": 11.572298049926758,
"step": 1480
},
{
"epoch": 2.48,
"grad_norm": 2.563070774078369,
"learning_rate": 7.075766979933674e-07,
"logits/chosen": -1.6885433197021484,
"logits/rejected": -1.663726806640625,
"logps/chosen": -35.838836669921875,
"logps/rejected": -253.3392333984375,
"loss": 0.2286,
"rewards/accuracies": 1.0,
"rewards/chosen": 32.139644622802734,
"rewards/margins": 20.651351928710938,
"rewards/rejected": 11.48829174041748,
"step": 1490
},
{
"epoch": 2.5,
"grad_norm": 1.7424818277359009,
"learning_rate": 6.634041210872743e-07,
"logits/chosen": -1.8395694494247437,
"logits/rejected": -1.8219817876815796,
"logps/chosen": -43.62278366088867,
"logps/rejected": -261.8132629394531,
"loss": 0.259,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.96283531188965,
"rewards/margins": 20.133371353149414,
"rewards/rejected": 10.829463005065918,
"step": 1500
},
{
"epoch": 2.52,
"grad_norm": 3.5391411781311035,
"learning_rate": 6.205570749652002e-07,
"logits/chosen": -1.5959376096725464,
"logits/rejected": -1.5644586086273193,
"logps/chosen": -47.74095916748047,
"logps/rejected": -254.2753448486328,
"loss": 0.3061,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.81204605102539,
"rewards/margins": 19.811317443847656,
"rewards/rejected": 11.00072956085205,
"step": 1510
},
{
"epoch": 2.53,
"grad_norm": 2.486269235610962,
"learning_rate": 5.790486563233145e-07,
"logits/chosen": -1.8033177852630615,
"logits/rejected": -1.7998552322387695,
"logps/chosen": -35.27519226074219,
"logps/rejected": -256.3056640625,
"loss": 0.2171,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.810821533203125,
"rewards/margins": 20.776798248291016,
"rewards/rejected": 11.034022331237793,
"step": 1520
},
{
"epoch": 2.55,
"grad_norm": 2.615795612335205,
"learning_rate": 5.388915526907862e-07,
"logits/chosen": -1.7600593566894531,
"logits/rejected": -1.7274726629257202,
"logps/chosen": -38.91143035888672,
"logps/rejected": -256.5948791503906,
"loss": 0.2455,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.60187339782715,
"rewards/margins": 19.52530288696289,
"rewards/rejected": 11.076571464538574,
"step": 1530
},
{
"epoch": 2.57,
"grad_norm": 3.510915994644165,
"learning_rate": 5.000980385516935e-07,
"logits/chosen": -1.6004310846328735,
"logits/rejected": -1.5690761804580688,
"logps/chosen": -41.87531280517578,
"logps/rejected": -251.2759246826172,
"loss": 0.2711,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.125137329101562,
"rewards/margins": 18.800806045532227,
"rewards/rejected": 11.32433032989502,
"step": 1540
},
{
"epoch": 2.58,
"grad_norm": 3.1003832817077637,
"learning_rate": 4.626799715931812e-07,
"logits/chosen": -1.6454312801361084,
"logits/rejected": -1.6356630325317383,
"logps/chosen": -47.9179801940918,
"logps/rejected": -259.88311767578125,
"loss": 0.293,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.139379501342773,
"rewards/margins": 18.82354736328125,
"rewards/rejected": 11.315831184387207,
"step": 1550
},
{
"epoch": 2.6,
"grad_norm": 2.93013334274292,
"learning_rate": 4.2664878908102556e-07,
"logits/chosen": -1.88095223903656,
"logits/rejected": -1.8393617868423462,
"logps/chosen": -32.803287506103516,
"logps/rejected": -261.6094055175781,
"loss": 0.2059,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.75473976135254,
"rewards/margins": 20.777175903320312,
"rewards/rejected": 10.977563858032227,
"step": 1560
},
{
"epoch": 2.62,
"grad_norm": 2.920027017593384,
"learning_rate": 3.9201550436370026e-07,
"logits/chosen": -1.7541553974151611,
"logits/rejected": -1.7304025888442993,
"logps/chosen": -40.56893539428711,
"logps/rejected": -264.2108459472656,
"loss": 0.2521,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.008968353271484,
"rewards/margins": 20.572416305541992,
"rewards/rejected": 10.436552047729492,
"step": 1570
},
{
"epoch": 2.63,
"grad_norm": 2.7568492889404297,
"learning_rate": 3.587907035060195e-07,
"logits/chosen": -1.5946696996688843,
"logits/rejected": -1.577462911605835,
"logps/chosen": -43.67282485961914,
"logps/rejected": -259.1739807128906,
"loss": 0.2793,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.5866641998291,
"rewards/margins": 18.64238166809082,
"rewards/rejected": 10.944284439086914,
"step": 1580
},
{
"epoch": 2.65,
"grad_norm": 2.2731993198394775,
"learning_rate": 3.269845420533824e-07,
"logits/chosen": -1.7446489334106445,
"logits/rejected": -1.701433777809143,
"logps/chosen": -37.443397521972656,
"logps/rejected": -254.092529296875,
"loss": 0.279,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 31.358057022094727,
"rewards/margins": 19.601125717163086,
"rewards/rejected": 11.756929397583008,
"step": 1590
},
{
"epoch": 2.67,
"grad_norm": 3.0889647006988525,
"learning_rate": 2.9660674192761753e-07,
"logits/chosen": -1.5739749670028687,
"logits/rejected": -1.5466270446777344,
"logps/chosen": -46.05111312866211,
"logps/rejected": -245.4317626953125,
"loss": 0.2881,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.035449981689453,
"rewards/margins": 18.32761573791504,
"rewards/rejected": 10.707832336425781,
"step": 1600
},
{
"epoch": 2.67,
"eval_logits/chosen": -0.8468231558799744,
"eval_logits/rejected": -0.8503096103668213,
"eval_logps/chosen": -167.3941192626953,
"eval_logps/rejected": -209.29457092285156,
"eval_loss": 2.6357526779174805,
"eval_rewards/accuracies": 0.6468085050582886,
"eval_rewards/chosen": 14.38486385345459,
"eval_rewards/margins": 1.6875391006469727,
"eval_rewards/rejected": 12.697324752807617,
"eval_runtime": 934.2741,
"eval_samples_per_second": 0.755,
"eval_steps_per_second": 0.755,
"step": 1600
},
{
"epoch": 2.68,
"grad_norm": 3.4763405323028564,
"learning_rate": 2.676665884553559e-07,
"logits/chosen": -1.6974531412124634,
"logits/rejected": -1.6716630458831787,
"logps/chosen": -45.91952133178711,
"logps/rejected": -240.70156860351562,
"loss": 0.288,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.833568572998047,
"rewards/margins": 19.545482635498047,
"rewards/rejected": 11.288084030151367,
"step": 1610
},
{
"epoch": 2.7,
"grad_norm": 2.333751916885376,
"learning_rate": 2.401729275298753e-07,
"logits/chosen": -1.663021445274353,
"logits/rejected": -1.6606937646865845,
"logps/chosen": -41.728370666503906,
"logps/rejected": -257.09600830078125,
"loss": 0.2824,
"rewards/accuracies": 0.987500011920929,
"rewards/chosen": 31.06747817993164,
"rewards/margins": 21.039501190185547,
"rewards/rejected": 10.02797794342041,
"step": 1620
},
{
"epoch": 2.72,
"grad_norm": 3.9069831371307373,
"learning_rate": 2.1413416290723966e-07,
"logits/chosen": -1.7814449071884155,
"logits/rejected": -1.7398021221160889,
"logps/chosen": -39.289886474609375,
"logps/rejected": -252.5143585205078,
"loss": 0.2494,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.424646377563477,
"rewards/margins": 20.74696922302246,
"rewards/rejected": 10.677675247192383,
"step": 1630
},
{
"epoch": 2.73,
"grad_norm": 2.8962225914001465,
"learning_rate": 1.8955825363760172e-07,
"logits/chosen": -1.6453288793563843,
"logits/rejected": -1.6407811641693115,
"logps/chosen": -48.07708740234375,
"logps/rejected": -257.2157287597656,
"loss": 0.2834,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.098657608032227,
"rewards/margins": 19.977779388427734,
"rewards/rejected": 10.120881080627441,
"step": 1640
},
{
"epoch": 2.75,
"grad_norm": 3.3172779083251953,
"learning_rate": 1.6645271163242106e-07,
"logits/chosen": -1.6090644598007202,
"logits/rejected": -1.5861512422561646,
"logps/chosen": -46.873756408691406,
"logps/rejected": -253.38778686523438,
"loss": 0.2934,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.051366806030273,
"rewards/margins": 18.360876083374023,
"rewards/rejected": 10.6904935836792,
"step": 1650
},
{
"epoch": 2.77,
"grad_norm": 3.810804605484009,
"learning_rate": 1.448245993683639e-07,
"logits/chosen": -1.6912416219711304,
"logits/rejected": -1.6398060321807861,
"logps/chosen": -42.75129318237305,
"logps/rejected": -255.9859619140625,
"loss": 0.2546,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.15047836303711,
"rewards/margins": 19.870410919189453,
"rewards/rejected": 11.280068397521973,
"step": 1660
},
{
"epoch": 2.78,
"grad_norm": 2.5805416107177734,
"learning_rate": 1.2468052772857786e-07,
"logits/chosen": -1.6538400650024414,
"logits/rejected": -1.6332565546035767,
"logps/chosen": -39.50425338745117,
"logps/rejected": -243.5617218017578,
"loss": 0.2568,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.246089935302734,
"rewards/margins": 18.635494232177734,
"rewards/rejected": 10.610593795776367,
"step": 1670
},
{
"epoch": 2.8,
"grad_norm": 3.1274917125701904,
"learning_rate": 1.060266539819932e-07,
"logits/chosen": -1.642760992050171,
"logits/rejected": -1.6645612716674805,
"logps/chosen": -52.89581298828125,
"logps/rejected": -240.6681671142578,
"loss": 0.3194,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.198993682861328,
"rewards/margins": 18.33091926574707,
"rewards/rejected": 10.868074417114258,
"step": 1680
},
{
"epoch": 2.82,
"grad_norm": 4.011273384094238,
"learning_rate": 8.886867990128722e-08,
"logits/chosen": -1.5743844509124756,
"logits/rejected": -1.5511356592178345,
"logps/chosen": -42.92226028442383,
"logps/rejected": -239.4049072265625,
"loss": 0.2702,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.1669921875,
"rewards/margins": 20.256168365478516,
"rewards/rejected": 10.91082763671875,
"step": 1690
},
{
"epoch": 2.83,
"grad_norm": 3.313141345977783,
"learning_rate": 7.321185002006848e-08,
"logits/chosen": -1.7491827011108398,
"logits/rejected": -1.710752248764038,
"logps/chosen": -36.811859130859375,
"logps/rejected": -276.8099365234375,
"loss": 0.2412,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.183746337890625,
"rewards/margins": 20.148881912231445,
"rewards/rejected": 10.034868240356445,
"step": 1700
},
{
"epoch": 2.85,
"grad_norm": 2.1705501079559326,
"learning_rate": 5.906095002982615e-08,
"logits/chosen": -1.6982301473617554,
"logits/rejected": -1.6695678234100342,
"logps/chosen": -39.64902877807617,
"logps/rejected": -265.86370849609375,
"loss": 0.258,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.876617431640625,
"rewards/margins": 21.336835861206055,
"rewards/rejected": 10.539777755737305,
"step": 1710
},
{
"epoch": 2.87,
"grad_norm": 3.416759729385376,
"learning_rate": 4.642030531712582e-08,
"logits/chosen": -1.6781730651855469,
"logits/rejected": -1.6320860385894775,
"logps/chosen": -44.278289794921875,
"logps/rejected": -270.3473205566406,
"loss": 0.2723,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.628204345703125,
"rewards/margins": 20.757869720458984,
"rewards/rejected": 10.87033748626709,
"step": 1720
},
{
"epoch": 2.88,
"grad_norm": 2.6415388584136963,
"learning_rate": 3.5293779641508156e-08,
"logits/chosen": -1.7743009328842163,
"logits/rejected": -1.7475459575653076,
"logps/chosen": -41.21794891357422,
"logps/rejected": -267.9061584472656,
"loss": 0.2469,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.46852684020996,
"rewards/margins": 19.723846435546875,
"rewards/rejected": 11.74467945098877,
"step": 1730
},
{
"epoch": 2.9,
"grad_norm": 3.2879960536956787,
"learning_rate": 2.5684773954482433e-08,
"logits/chosen": -1.7669696807861328,
"logits/rejected": -1.705436110496521,
"logps/chosen": -38.20746612548828,
"logps/rejected": -279.90460205078125,
"loss": 0.2434,
"rewards/accuracies": 1.0,
"rewards/chosen": 31.641986846923828,
"rewards/margins": 21.152555465698242,
"rewards/rejected": 10.489428520202637,
"step": 1740
},
{
"epoch": 2.92,
"grad_norm": 2.513577461242676,
"learning_rate": 1.7596225359988728e-08,
"logits/chosen": -1.8220970630645752,
"logits/rejected": -1.7889318466186523,
"logps/chosen": -40.94774627685547,
"logps/rejected": -259.34320068359375,
"loss": 0.2511,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.275482177734375,
"rewards/margins": 19.547029495239258,
"rewards/rejected": 10.728452682495117,
"step": 1750
},
{
"epoch": 2.93,
"grad_norm": 3.861309766769409,
"learning_rate": 1.1030606216637097e-08,
"logits/chosen": -1.4674508571624756,
"logits/rejected": -1.425626277923584,
"logps/chosen": -48.60122299194336,
"logps/rejected": -243.1677703857422,
"loss": 0.3138,
"rewards/accuracies": 1.0,
"rewards/chosen": 28.8472957611084,
"rewards/margins": 19.466501235961914,
"rewards/rejected": 9.380796432495117,
"step": 1760
},
{
"epoch": 2.95,
"grad_norm": 3.0707430839538574,
"learning_rate": 5.989923382003216e-09,
"logits/chosen": -1.5614324808120728,
"logits/rejected": -1.5531564950942993,
"logps/chosen": -50.88349151611328,
"logps/rejected": -241.66943359375,
"loss": 0.3286,
"rewards/accuracies": 1.0,
"rewards/chosen": 29.8480167388916,
"rewards/margins": 17.89462661743164,
"rewards/rejected": 11.953387260437012,
"step": 1770
},
{
"epoch": 2.97,
"grad_norm": 1.1481070518493652,
"learning_rate": 2.4757175992079496e-09,
"logits/chosen": -1.654233694076538,
"logits/rejected": -1.6192829608917236,
"logps/chosen": -43.71112060546875,
"logps/rejected": -246.0106201171875,
"loss": 0.2606,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.80698013305664,
"rewards/margins": 19.817432403564453,
"rewards/rejected": 10.989545822143555,
"step": 1780
},
{
"epoch": 2.98,
"grad_norm": 1.825456976890564,
"learning_rate": 4.890630259724027e-10,
"logits/chosen": -1.703809380531311,
"logits/rejected": -1.6901493072509766,
"logps/chosen": -53.00714874267578,
"logps/rejected": -266.7369384765625,
"loss": 0.3145,
"rewards/accuracies": 1.0,
"rewards/chosen": 30.395549774169922,
"rewards/margins": 19.31685447692871,
"rewards/rejected": 11.078699111938477,
"step": 1790
},
{
"epoch": 3.0,
"step": 1797,
"total_flos": 5.920748380997222e+18,
"train_loss": 0.5555986505650652,
"train_runtime": 48641.0087,
"train_samples_per_second": 0.296,
"train_steps_per_second": 0.037
}
],
"logging_steps": 10,
"max_steps": 1797,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 200,
"total_flos": 5.920748380997222e+18,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}