zephyr-7b-UFB-0 / trainer_state.json
weijie210's picture
Model save
9b4066e verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 443,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.111111111111111e-08,
"logits/chosen": -2.737483501434326,
"logits/rejected": -2.581545829772949,
"logps/chosen": -157.5369873046875,
"logps/rejected": -100.35362243652344,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.02,
"learning_rate": 1.111111111111111e-07,
"logits/chosen": -2.61592960357666,
"logits/rejected": -2.572232484817505,
"logps/chosen": -124.513671875,
"logps/rejected": -91.37435913085938,
"loss": 0.6778,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.04052966088056564,
"rewards/margins": 0.01970052346587181,
"rewards/rejected": 0.020829135552048683,
"step": 10
},
{
"epoch": 0.05,
"learning_rate": 2.222222222222222e-07,
"logits/chosen": -2.565521001815796,
"logits/rejected": -2.4634904861450195,
"logps/chosen": -114.85684967041016,
"logps/rejected": -80.09593963623047,
"loss": 0.5129,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": 0.4715304970741272,
"rewards/margins": 0.5839148759841919,
"rewards/rejected": -0.11238440126180649,
"step": 20
},
{
"epoch": 0.07,
"learning_rate": 3.333333333333333e-07,
"logits/chosen": -2.5056052207946777,
"logits/rejected": -2.4653897285461426,
"logps/chosen": -123.69754791259766,
"logps/rejected": -103.3398208618164,
"loss": 0.4006,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": 0.01100767869502306,
"rewards/margins": 1.2376883029937744,
"rewards/rejected": -1.2266806364059448,
"step": 30
},
{
"epoch": 0.09,
"learning_rate": 4.444444444444444e-07,
"logits/chosen": -2.4814658164978027,
"logits/rejected": -2.4201178550720215,
"logps/chosen": -136.6073760986328,
"logps/rejected": -117.58763122558594,
"loss": 0.3056,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -0.353680282831192,
"rewards/margins": 1.6959145069122314,
"rewards/rejected": -2.0495948791503906,
"step": 40
},
{
"epoch": 0.11,
"learning_rate": 4.937185929648241e-07,
"logits/chosen": -2.458440065383911,
"logits/rejected": -2.3304443359375,
"logps/chosen": -124.83992004394531,
"logps/rejected": -106.70866394042969,
"loss": 0.2679,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -0.31784066557884216,
"rewards/margins": 2.447019100189209,
"rewards/rejected": -2.764859914779663,
"step": 50
},
{
"epoch": 0.14,
"learning_rate": 4.811557788944723e-07,
"logits/chosen": -2.396515369415283,
"logits/rejected": -2.293870687484741,
"logps/chosen": -135.10882568359375,
"logps/rejected": -122.99256896972656,
"loss": 0.2447,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.10166273266077042,
"rewards/margins": 3.037165641784668,
"rewards/rejected": -3.1388285160064697,
"step": 60
},
{
"epoch": 0.16,
"learning_rate": 4.685929648241206e-07,
"logits/chosen": -2.355457305908203,
"logits/rejected": -2.2316598892211914,
"logps/chosen": -122.27568054199219,
"logps/rejected": -113.7423095703125,
"loss": 0.2718,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.5933733582496643,
"rewards/margins": 2.744062662124634,
"rewards/rejected": -3.3374359607696533,
"step": 70
},
{
"epoch": 0.18,
"learning_rate": 4.5603015075376885e-07,
"logits/chosen": -2.361450672149658,
"logits/rejected": -2.286520481109619,
"logps/chosen": -127.52728271484375,
"logps/rejected": -126.6650161743164,
"loss": 0.2313,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": -0.7555137872695923,
"rewards/margins": 3.271893262863159,
"rewards/rejected": -4.027407169342041,
"step": 80
},
{
"epoch": 0.2,
"learning_rate": 4.4346733668341706e-07,
"logits/chosen": -2.3498244285583496,
"logits/rejected": -2.257256507873535,
"logps/chosen": -122.7756118774414,
"logps/rejected": -121.1026382446289,
"loss": 0.2307,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.7029477953910828,
"rewards/margins": 3.352973222732544,
"rewards/rejected": -4.0559210777282715,
"step": 90
},
{
"epoch": 0.23,
"learning_rate": 4.309045226130653e-07,
"logits/chosen": -2.4514331817626953,
"logits/rejected": -2.378006935119629,
"logps/chosen": -129.557861328125,
"logps/rejected": -123.76164245605469,
"loss": 0.2078,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": 0.20816242694854736,
"rewards/margins": 4.0717453956604,
"rewards/rejected": -3.8635826110839844,
"step": 100
},
{
"epoch": 0.25,
"learning_rate": 4.1834170854271357e-07,
"logits/chosen": -2.358485698699951,
"logits/rejected": -2.279703378677368,
"logps/chosen": -133.0589141845703,
"logps/rejected": -129.48451232910156,
"loss": 0.2371,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -0.7045987248420715,
"rewards/margins": 3.171238422393799,
"rewards/rejected": -3.8758368492126465,
"step": 110
},
{
"epoch": 0.27,
"learning_rate": 4.057788944723618e-07,
"logits/chosen": -2.3714640140533447,
"logits/rejected": -2.3009307384490967,
"logps/chosen": -129.02052307128906,
"logps/rejected": -130.68905639648438,
"loss": 0.1905,
"rewards/accuracies": 0.862500011920929,
"rewards/chosen": -1.2331945896148682,
"rewards/margins": 3.5529701709747314,
"rewards/rejected": -4.7861647605896,
"step": 120
},
{
"epoch": 0.29,
"learning_rate": 3.9321608040201003e-07,
"logits/chosen": -2.385488510131836,
"logits/rejected": -2.304129123687744,
"logps/chosen": -128.46060180664062,
"logps/rejected": -135.4064483642578,
"loss": 0.1812,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -0.6662181615829468,
"rewards/margins": 4.080785274505615,
"rewards/rejected": -4.74700403213501,
"step": 130
},
{
"epoch": 0.32,
"learning_rate": 3.806532663316583e-07,
"logits/chosen": -2.331386089324951,
"logits/rejected": -2.2633156776428223,
"logps/chosen": -138.77073669433594,
"logps/rejected": -150.08639526367188,
"loss": 0.1728,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.8451410531997681,
"rewards/margins": 4.410196781158447,
"rewards/rejected": -5.255337715148926,
"step": 140
},
{
"epoch": 0.34,
"learning_rate": 3.6809045226130655e-07,
"logits/chosen": -2.261727809906006,
"logits/rejected": -2.253206729888916,
"logps/chosen": -128.15460205078125,
"logps/rejected": -137.04916381835938,
"loss": 0.2448,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": -1.2643935680389404,
"rewards/margins": 4.03481912612915,
"rewards/rejected": -5.299212455749512,
"step": 150
},
{
"epoch": 0.36,
"learning_rate": 3.5552763819095475e-07,
"logits/chosen": -2.3198587894439697,
"logits/rejected": -2.2583937644958496,
"logps/chosen": -140.9436798095703,
"logps/rejected": -147.12094116210938,
"loss": 0.1615,
"rewards/accuracies": 0.887499988079071,
"rewards/chosen": -1.0014054775238037,
"rewards/margins": 4.486364364624023,
"rewards/rejected": -5.487771034240723,
"step": 160
},
{
"epoch": 0.38,
"learning_rate": 3.42964824120603e-07,
"logits/chosen": -2.291534185409546,
"logits/rejected": -2.208200454711914,
"logps/chosen": -141.3668670654297,
"logps/rejected": -155.08602905273438,
"loss": 0.216,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -1.5920617580413818,
"rewards/margins": 4.606629371643066,
"rewards/rejected": -6.198691368103027,
"step": 170
},
{
"epoch": 0.41,
"learning_rate": 3.3040201005025127e-07,
"logits/chosen": -2.387359142303467,
"logits/rejected": -2.28928279876709,
"logps/chosen": -130.8974609375,
"logps/rejected": -142.79708862304688,
"loss": 0.2023,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": -2.409898519515991,
"rewards/margins": 3.318312168121338,
"rewards/rejected": -5.72821044921875,
"step": 180
},
{
"epoch": 0.43,
"learning_rate": 3.178391959798995e-07,
"logits/chosen": -2.378145456314087,
"logits/rejected": -2.3165206909179688,
"logps/chosen": -131.51109313964844,
"logps/rejected": -142.97760009765625,
"loss": 0.1793,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -1.6101114749908447,
"rewards/margins": 3.964582920074463,
"rewards/rejected": -5.574694633483887,
"step": 190
},
{
"epoch": 0.45,
"learning_rate": 3.0527638190954773e-07,
"logits/chosen": -2.330974817276001,
"logits/rejected": -2.276876449584961,
"logps/chosen": -137.87515258789062,
"logps/rejected": -152.93994140625,
"loss": 0.1638,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": -1.3574166297912598,
"rewards/margins": 4.367352485656738,
"rewards/rejected": -5.724769115447998,
"step": 200
},
{
"epoch": 0.47,
"learning_rate": 2.92713567839196e-07,
"logits/chosen": -2.4077656269073486,
"logits/rejected": -2.3121707439422607,
"logps/chosen": -144.79739379882812,
"logps/rejected": -147.3674774169922,
"loss": 0.1603,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -1.5326565504074097,
"rewards/margins": 3.852278470993042,
"rewards/rejected": -5.384934902191162,
"step": 210
},
{
"epoch": 0.5,
"learning_rate": 2.8015075376884424e-07,
"logits/chosen": -2.2944283485412598,
"logits/rejected": -2.2064151763916016,
"logps/chosen": -130.70140075683594,
"logps/rejected": -147.8914794921875,
"loss": 0.1641,
"rewards/accuracies": 0.887499988079071,
"rewards/chosen": -1.207008957862854,
"rewards/margins": 4.792324542999268,
"rewards/rejected": -5.999332427978516,
"step": 220
},
{
"epoch": 0.52,
"learning_rate": 2.6758793969849245e-07,
"logits/chosen": -2.348940849304199,
"logits/rejected": -2.164310932159424,
"logps/chosen": -128.39413452148438,
"logps/rejected": -140.557373046875,
"loss": 0.166,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": -1.3905706405639648,
"rewards/margins": 4.686975479125977,
"rewards/rejected": -6.077546119689941,
"step": 230
},
{
"epoch": 0.54,
"learning_rate": 2.550251256281407e-07,
"logits/chosen": -2.395198106765747,
"logits/rejected": -2.2796173095703125,
"logps/chosen": -149.087646484375,
"logps/rejected": -157.66110229492188,
"loss": 0.1402,
"rewards/accuracies": 0.9375,
"rewards/chosen": -1.3839337825775146,
"rewards/margins": 4.870881080627441,
"rewards/rejected": -6.254815578460693,
"step": 240
},
{
"epoch": 0.56,
"learning_rate": 2.424623115577889e-07,
"logits/chosen": -2.4120841026306152,
"logits/rejected": -2.2922260761260986,
"logps/chosen": -133.91209411621094,
"logps/rejected": -144.4380340576172,
"loss": 0.1872,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -1.2101858854293823,
"rewards/margins": 4.869531154632568,
"rewards/rejected": -6.079718112945557,
"step": 250
},
{
"epoch": 0.59,
"learning_rate": 2.2989949748743717e-07,
"logits/chosen": -2.331249237060547,
"logits/rejected": -2.2102646827697754,
"logps/chosen": -124.6241683959961,
"logps/rejected": -138.6643524169922,
"loss": 0.1476,
"rewards/accuracies": 0.9750000238418579,
"rewards/chosen": -0.8653093576431274,
"rewards/margins": 4.82875394821167,
"rewards/rejected": -5.694063186645508,
"step": 260
},
{
"epoch": 0.61,
"learning_rate": 2.1733668341708543e-07,
"logits/chosen": -2.2793853282928467,
"logits/rejected": -2.1484787464141846,
"logps/chosen": -141.07778930664062,
"logps/rejected": -158.54676818847656,
"loss": 0.1925,
"rewards/accuracies": 0.9375,
"rewards/chosen": -0.8885013461112976,
"rewards/margins": 5.88812780380249,
"rewards/rejected": -6.776629447937012,
"step": 270
},
{
"epoch": 0.63,
"learning_rate": 2.0477386934673366e-07,
"logits/chosen": -2.3435816764831543,
"logits/rejected": -2.222494125366211,
"logps/chosen": -132.23934936523438,
"logps/rejected": -152.41224670410156,
"loss": 0.1719,
"rewards/accuracies": 0.862500011920929,
"rewards/chosen": -0.8898094892501831,
"rewards/margins": 5.38167667388916,
"rewards/rejected": -6.271485805511475,
"step": 280
},
{
"epoch": 0.65,
"learning_rate": 1.922110552763819e-07,
"logits/chosen": -2.337050437927246,
"logits/rejected": -2.2481775283813477,
"logps/chosen": -143.0239715576172,
"logps/rejected": -167.92628479003906,
"loss": 0.1408,
"rewards/accuracies": 0.887499988079071,
"rewards/chosen": -1.425594449043274,
"rewards/margins": 5.652303695678711,
"rewards/rejected": -7.0778985023498535,
"step": 290
},
{
"epoch": 0.68,
"learning_rate": 1.7964824120603015e-07,
"logits/chosen": -2.3036749362945557,
"logits/rejected": -2.1755778789520264,
"logps/chosen": -143.76683044433594,
"logps/rejected": -168.64463806152344,
"loss": 0.1393,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": -1.336169719696045,
"rewards/margins": 5.389135837554932,
"rewards/rejected": -6.725306034088135,
"step": 300
},
{
"epoch": 0.7,
"learning_rate": 1.6708542713567838e-07,
"logits/chosen": -2.2387304306030273,
"logits/rejected": -2.1477370262145996,
"logps/chosen": -134.2172393798828,
"logps/rejected": -152.63461303710938,
"loss": 0.1579,
"rewards/accuracies": 0.862500011920929,
"rewards/chosen": -1.4482789039611816,
"rewards/margins": 4.973227500915527,
"rewards/rejected": -6.421506404876709,
"step": 310
},
{
"epoch": 0.72,
"learning_rate": 1.5452261306532663e-07,
"logits/chosen": -2.2377147674560547,
"logits/rejected": -2.1322357654571533,
"logps/chosen": -135.234130859375,
"logps/rejected": -161.66934204101562,
"loss": 0.1088,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": -1.7265955209732056,
"rewards/margins": 5.747531890869141,
"rewards/rejected": -7.474127769470215,
"step": 320
},
{
"epoch": 0.74,
"learning_rate": 1.4195979899497487e-07,
"logits/chosen": -2.236829996109009,
"logits/rejected": -2.154937982559204,
"logps/chosen": -141.66830444335938,
"logps/rejected": -166.32823181152344,
"loss": 0.1488,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -1.782355546951294,
"rewards/margins": 5.470345497131348,
"rewards/rejected": -7.2527008056640625,
"step": 330
},
{
"epoch": 0.77,
"learning_rate": 1.2939698492462312e-07,
"logits/chosen": -2.3523616790771484,
"logits/rejected": -2.24237322807312,
"logps/chosen": -134.5939178466797,
"logps/rejected": -169.53317260742188,
"loss": 0.1579,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": -1.731003999710083,
"rewards/margins": 6.211657524108887,
"rewards/rejected": -7.942662239074707,
"step": 340
},
{
"epoch": 0.79,
"learning_rate": 1.1683417085427135e-07,
"logits/chosen": -2.2935945987701416,
"logits/rejected": -2.2166569232940674,
"logps/chosen": -137.88473510742188,
"logps/rejected": -160.505126953125,
"loss": 0.1508,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.9919565916061401,
"rewards/margins": 5.5159502029418945,
"rewards/rejected": -6.507906913757324,
"step": 350
},
{
"epoch": 0.81,
"learning_rate": 1.0427135678391958e-07,
"logits/chosen": -2.320068120956421,
"logits/rejected": -2.2384986877441406,
"logps/chosen": -134.75381469726562,
"logps/rejected": -160.6112518310547,
"loss": 0.1696,
"rewards/accuracies": 0.875,
"rewards/chosen": -2.4626049995422363,
"rewards/margins": 5.313817024230957,
"rewards/rejected": -7.77642297744751,
"step": 360
},
{
"epoch": 0.84,
"learning_rate": 9.170854271356783e-08,
"logits/chosen": -2.2218480110168457,
"logits/rejected": -2.1257214546203613,
"logps/chosen": -147.00196838378906,
"logps/rejected": -171.19464111328125,
"loss": 0.1453,
"rewards/accuracies": 0.8999999761581421,
"rewards/chosen": -1.5424976348876953,
"rewards/margins": 5.8535027503967285,
"rewards/rejected": -7.395999908447266,
"step": 370
},
{
"epoch": 0.86,
"learning_rate": 7.914572864321607e-08,
"logits/chosen": -2.27662992477417,
"logits/rejected": -2.157721519470215,
"logps/chosen": -131.88534545898438,
"logps/rejected": -158.86459350585938,
"loss": 0.1955,
"rewards/accuracies": 0.875,
"rewards/chosen": -2.0121521949768066,
"rewards/margins": 4.82122802734375,
"rewards/rejected": -6.833380222320557,
"step": 380
},
{
"epoch": 0.88,
"learning_rate": 6.658291457286432e-08,
"logits/chosen": -2.4072272777557373,
"logits/rejected": -2.285789966583252,
"logps/chosen": -140.86204528808594,
"logps/rejected": -169.72288513183594,
"loss": 0.1312,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -0.6351500749588013,
"rewards/margins": 6.939029693603516,
"rewards/rejected": -7.574179649353027,
"step": 390
},
{
"epoch": 0.9,
"learning_rate": 5.4020100502512555e-08,
"logits/chosen": -2.2212846279144287,
"logits/rejected": -2.141895294189453,
"logps/chosen": -126.89100646972656,
"logps/rejected": -161.2249755859375,
"loss": 0.2109,
"rewards/accuracies": 0.925000011920929,
"rewards/chosen": -0.795007050037384,
"rewards/margins": 6.2724409103393555,
"rewards/rejected": -7.067447662353516,
"step": 400
},
{
"epoch": 0.93,
"learning_rate": 4.14572864321608e-08,
"logits/chosen": -2.370591878890991,
"logits/rejected": -2.2174134254455566,
"logps/chosen": -146.6661834716797,
"logps/rejected": -165.058837890625,
"loss": 0.1407,
"rewards/accuracies": 0.9125000238418579,
"rewards/chosen": -0.8686304092407227,
"rewards/margins": 6.0188398361206055,
"rewards/rejected": -6.8874711990356445,
"step": 410
},
{
"epoch": 0.95,
"learning_rate": 2.8894472361809044e-08,
"logits/chosen": -2.3981685638427734,
"logits/rejected": -2.2463793754577637,
"logps/chosen": -147.17486572265625,
"logps/rejected": -157.16622924804688,
"loss": 0.1245,
"rewards/accuracies": 0.875,
"rewards/chosen": -1.544703483581543,
"rewards/margins": 5.296902179718018,
"rewards/rejected": -6.841605186462402,
"step": 420
},
{
"epoch": 0.97,
"learning_rate": 1.6331658291457288e-08,
"logits/chosen": -2.3454318046569824,
"logits/rejected": -2.2091023921966553,
"logps/chosen": -153.2548370361328,
"logps/rejected": -168.34854125976562,
"loss": 0.1576,
"rewards/accuracies": 0.887499988079071,
"rewards/chosen": -1.147238850593567,
"rewards/margins": 5.976444721221924,
"rewards/rejected": -7.123683929443359,
"step": 430
},
{
"epoch": 0.99,
"learning_rate": 3.768844221105527e-09,
"logits/chosen": -2.2823286056518555,
"logits/rejected": -2.1865620613098145,
"logps/chosen": -123.73600769042969,
"logps/rejected": -146.88194274902344,
"loss": 0.1626,
"rewards/accuracies": 0.875,
"rewards/chosen": -1.5477970838546753,
"rewards/margins": 5.064129829406738,
"rewards/rejected": -6.611927032470703,
"step": 440
},
{
"epoch": 1.0,
"step": 443,
"total_flos": 0.0,
"train_loss": 0.2053286122078282,
"train_runtime": 6354.2323,
"train_samples_per_second": 2.231,
"train_steps_per_second": 0.07
}
],
"logging_steps": 10,
"max_steps": 443,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 500,
"total_flos": 0.0,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}