phi-2-gpo-v4-i1 / trainer_state.json
lole25's picture
Model save
eb61e81 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9996190476190476,
"eval_steps": 500,
"global_step": 656,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 7.575757575757576e-08,
"logits/chosen": 0.040165986865758896,
"logits/rejected": 0.1715753823518753,
"logps/chosen": -294.844482421875,
"logps/rejected": -361.2099914550781,
"loss": 0.3581,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.02,
"learning_rate": 7.575757575757576e-07,
"logits/chosen": 0.08047256618738174,
"logits/rejected": 0.3207971751689911,
"logps/chosen": -393.2153625488281,
"logps/rejected": -318.74615478515625,
"loss": 0.3404,
"rewards/accuracies": 0.4652777910232544,
"rewards/chosen": -0.00022995664039626718,
"rewards/margins": 2.2277235984802246e-05,
"rewards/rejected": -0.000252234167419374,
"step": 10
},
{
"epoch": 0.03,
"learning_rate": 1.5151515151515152e-06,
"logits/chosen": 0.07466734945774078,
"logits/rejected": 0.23236870765686035,
"logps/chosen": -332.4886169433594,
"logps/rejected": -281.853515625,
"loss": 0.3468,
"rewards/accuracies": 0.41874998807907104,
"rewards/chosen": 0.00011478399392217398,
"rewards/margins": -0.0012641319772228599,
"rewards/rejected": 0.0013789159711450338,
"step": 20
},
{
"epoch": 0.05,
"learning_rate": 2.2727272727272728e-06,
"logits/chosen": 0.07185273617506027,
"logits/rejected": 0.2604687213897705,
"logps/chosen": -358.33782958984375,
"logps/rejected": -292.1524963378906,
"loss": 0.3502,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": 0.0017305829096585512,
"rewards/margins": 0.001660021604038775,
"rewards/rejected": 7.056114554870874e-05,
"step": 30
},
{
"epoch": 0.06,
"learning_rate": 3.0303030303030305e-06,
"logits/chosen": 0.1199118122458458,
"logits/rejected": 0.2392597496509552,
"logps/chosen": -347.05926513671875,
"logps/rejected": -286.26080322265625,
"loss": 0.3353,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.005740107037127018,
"rewards/margins": 0.006530737970024347,
"rewards/rejected": -0.0007906301179900765,
"step": 40
},
{
"epoch": 0.08,
"learning_rate": 3.7878787878787882e-06,
"logits/chosen": 0.0741933211684227,
"logits/rejected": 0.3117237091064453,
"logps/chosen": -343.31170654296875,
"logps/rejected": -287.1304931640625,
"loss": 0.3286,
"rewards/accuracies": 0.6875,
"rewards/chosen": 0.016505183652043343,
"rewards/margins": 0.016020886600017548,
"rewards/rejected": 0.0004842969647143036,
"step": 50
},
{
"epoch": 0.09,
"learning_rate": 4.5454545454545455e-06,
"logits/chosen": 0.0762203261256218,
"logits/rejected": 0.2750852704048157,
"logps/chosen": -370.8611755371094,
"logps/rejected": -302.7222595214844,
"loss": 0.2924,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": 0.032092947512865067,
"rewards/margins": 0.04518315941095352,
"rewards/rejected": -0.013090210035443306,
"step": 60
},
{
"epoch": 0.11,
"learning_rate": 4.999432965739786e-06,
"logits/chosen": 0.07090188562870026,
"logits/rejected": 0.25236833095550537,
"logps/chosen": -321.5715026855469,
"logps/rejected": -301.0986022949219,
"loss": 0.2803,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": 0.05498770996928215,
"rewards/margins": 0.0691528171300888,
"rewards/rejected": -0.014165110886096954,
"step": 70
},
{
"epoch": 0.12,
"learning_rate": 4.9930567839810125e-06,
"logits/chosen": 0.10318852961063385,
"logits/rejected": 0.2712559401988983,
"logps/chosen": -358.31500244140625,
"logps/rejected": -302.6133117675781,
"loss": 0.2346,
"rewards/accuracies": 0.8062499761581421,
"rewards/chosen": 0.08596866577863693,
"rewards/margins": 0.1306326687335968,
"rewards/rejected": -0.044663988053798676,
"step": 80
},
{
"epoch": 0.14,
"learning_rate": 4.979613761906212e-06,
"logits/chosen": 0.1258704960346222,
"logits/rejected": 0.2653108239173889,
"logps/chosen": -316.4566345214844,
"logps/rejected": -285.59442138671875,
"loss": 0.2181,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": 0.10472643375396729,
"rewards/margins": 0.16386187076568604,
"rewards/rejected": -0.05913544446229935,
"step": 90
},
{
"epoch": 0.15,
"learning_rate": 4.959142005221991e-06,
"logits/chosen": 0.14865969121456146,
"logits/rejected": 0.2514886260032654,
"logps/chosen": -316.22650146484375,
"logps/rejected": -298.5392150878906,
"loss": 0.2379,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.09984966367483139,
"rewards/margins": 0.20773954689502716,
"rewards/rejected": -0.10788986831903458,
"step": 100
},
{
"epoch": 0.17,
"learning_rate": 4.931699543346854e-06,
"logits/chosen": 0.1114228144288063,
"logits/rejected": 0.2948494553565979,
"logps/chosen": -338.43450927734375,
"logps/rejected": -285.25494384765625,
"loss": 0.2098,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": 0.1445818692445755,
"rewards/margins": 0.19616642594337463,
"rewards/rejected": -0.05158457159996033,
"step": 110
},
{
"epoch": 0.18,
"learning_rate": 4.897364164920515e-06,
"logits/chosen": 0.10053505003452301,
"logits/rejected": 0.27514562010765076,
"logps/chosen": -338.99151611328125,
"logps/rejected": -297.2850646972656,
"loss": 0.2161,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": 0.14088504016399384,
"rewards/margins": 0.18905261158943176,
"rewards/rejected": -0.048167549073696136,
"step": 120
},
{
"epoch": 0.2,
"learning_rate": 4.8562331973035396e-06,
"logits/chosen": 0.08438269048929214,
"logits/rejected": 0.23364977538585663,
"logps/chosen": -322.02117919921875,
"logps/rejected": -303.7391357421875,
"loss": 0.2438,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": 0.146646648645401,
"rewards/margins": 0.1478952169418335,
"rewards/rejected": -0.0012485686456784606,
"step": 130
},
{
"epoch": 0.21,
"learning_rate": 4.808423230692374e-06,
"logits/chosen": 0.09936638176441193,
"logits/rejected": 0.24736297130584717,
"logps/chosen": -310.3603515625,
"logps/rejected": -278.47320556640625,
"loss": 0.222,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": 0.18732748925685883,
"rewards/margins": 0.1890837401151657,
"rewards/rejected": -0.0017562557477504015,
"step": 140
},
{
"epoch": 0.23,
"learning_rate": 4.754069787631761e-06,
"logits/chosen": 0.13391128182411194,
"logits/rejected": 0.2741778492927551,
"logps/chosen": -366.7373046875,
"logps/rejected": -295.2054138183594,
"loss": 0.2071,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": 0.23995013535022736,
"rewards/margins": 0.24667362868785858,
"rewards/rejected": -0.006723466329276562,
"step": 150
},
{
"epoch": 0.24,
"learning_rate": 4.693326938861367e-06,
"logits/chosen": 0.08330532908439636,
"logits/rejected": 0.23301962018013,
"logps/chosen": -295.6157531738281,
"logps/rejected": -277.6016540527344,
"loss": 0.2514,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": 0.19748535752296448,
"rewards/margins": 0.15948796272277832,
"rewards/rejected": 0.03799740970134735,
"step": 160
},
{
"epoch": 0.26,
"learning_rate": 4.626366866585528e-06,
"logits/chosen": 0.17037127912044525,
"logits/rejected": 0.3279545307159424,
"logps/chosen": -371.6747131347656,
"logps/rejected": -300.25250244140625,
"loss": 0.226,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": 0.2174214869737625,
"rewards/margins": 0.1945430338382721,
"rewards/rejected": 0.022878441959619522,
"step": 170
},
{
"epoch": 0.27,
"learning_rate": 4.553379376404085e-06,
"logits/chosen": 0.12852030992507935,
"logits/rejected": 0.23063895106315613,
"logps/chosen": -339.1563720703125,
"logps/rejected": -277.93841552734375,
"loss": 0.1992,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": 0.20981720089912415,
"rewards/margins": 0.22972619533538818,
"rewards/rejected": -0.019909001886844635,
"step": 180
},
{
"epoch": 0.29,
"learning_rate": 4.474571359287791e-06,
"logits/chosen": 0.08759725093841553,
"logits/rejected": 0.26223450899124146,
"logps/chosen": -315.0005187988281,
"logps/rejected": -266.51715087890625,
"loss": 0.2166,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.19832256436347961,
"rewards/margins": 0.2093769758939743,
"rewards/rejected": -0.011054400354623795,
"step": 190
},
{
"epoch": 0.3,
"learning_rate": 4.3901662051233755e-06,
"logits/chosen": 0.1616448611021042,
"logits/rejected": 0.30854135751724243,
"logps/chosen": -382.2667236328125,
"logps/rejected": -286.1771240234375,
"loss": 0.2198,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": 0.19347664713859558,
"rewards/margins": 0.20408186316490173,
"rewards/rejected": -0.010605214163661003,
"step": 200
},
{
"epoch": 0.32,
"learning_rate": 4.30040316949064e-06,
"logits/chosen": 0.14471155405044556,
"logits/rejected": 0.26751500368118286,
"logps/chosen": -300.742919921875,
"logps/rejected": -251.2882080078125,
"loss": 0.23,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": 0.1704670488834381,
"rewards/margins": 0.18817836046218872,
"rewards/rejected": -0.0177113339304924,
"step": 210
},
{
"epoch": 0.34,
"learning_rate": 4.205536695466524e-06,
"logits/chosen": 0.16468700766563416,
"logits/rejected": 0.24206213653087616,
"logps/chosen": -294.42767333984375,
"logps/rejected": -272.92010498046875,
"loss": 0.2529,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": 0.17376992106437683,
"rewards/margins": 0.16116927564144135,
"rewards/rejected": 0.012600669637322426,
"step": 220
},
{
"epoch": 0.35,
"learning_rate": 4.105835692378557e-06,
"logits/chosen": 0.10714814811944962,
"logits/rejected": 0.2699413597583771,
"logps/chosen": -304.6961364746094,
"logps/rejected": -285.6314392089844,
"loss": 0.1973,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": 0.1876644492149353,
"rewards/margins": 0.20867836475372314,
"rewards/rejected": -0.02101389318704605,
"step": 230
},
{
"epoch": 0.37,
"learning_rate": 4.001582773552153e-06,
"logits/chosen": 0.06082786247134209,
"logits/rejected": 0.29382848739624023,
"logps/chosen": -378.81732177734375,
"logps/rejected": -309.0159912109375,
"loss": 0.1942,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": 0.16840167343616486,
"rewards/margins": 0.24110493063926697,
"rewards/rejected": -0.0727032721042633,
"step": 240
},
{
"epoch": 0.38,
"learning_rate": 3.893073455212438e-06,
"logits/chosen": 0.1116786003112793,
"logits/rejected": 0.2517862915992737,
"logps/chosen": -320.56365966796875,
"logps/rejected": -278.5521545410156,
"loss": 0.2346,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": 0.17214366793632507,
"rewards/margins": 0.2073363959789276,
"rewards/rejected": -0.03519275039434433,
"step": 250
},
{
"epoch": 0.4,
"learning_rate": 3.7806153188114027e-06,
"logits/chosen": 0.11448470503091812,
"logits/rejected": 0.22813239693641663,
"logps/chosen": -279.91162109375,
"logps/rejected": -260.7430114746094,
"loss": 0.2402,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": 0.1783895492553711,
"rewards/margins": 0.17652130126953125,
"rewards/rejected": 0.001868226332589984,
"step": 260
},
{
"epoch": 0.41,
"learning_rate": 3.6645271391548542e-06,
"logits/chosen": 0.06586723029613495,
"logits/rejected": 0.21862812340259552,
"logps/chosen": -302.5594787597656,
"logps/rejected": -270.1592712402344,
"loss": 0.212,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": 0.1828652024269104,
"rewards/margins": 0.2052091658115387,
"rewards/rejected": -0.022343963384628296,
"step": 270
},
{
"epoch": 0.43,
"learning_rate": 3.5451379808006014e-06,
"logits/chosen": 0.1570790708065033,
"logits/rejected": 0.288757860660553,
"logps/chosen": -347.58013916015625,
"logps/rejected": -294.15716552734375,
"loss": 0.1946,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": 0.21899041533470154,
"rewards/margins": 0.2574598491191864,
"rewards/rejected": -0.03846944123506546,
"step": 280
},
{
"epoch": 0.44,
"learning_rate": 3.4227862652892106e-06,
"logits/chosen": 0.10818709433078766,
"logits/rejected": 0.23891910910606384,
"logps/chosen": -348.48370361328125,
"logps/rejected": -307.1477966308594,
"loss": 0.201,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 0.1834907829761505,
"rewards/margins": 0.22308149933815002,
"rewards/rejected": -0.03959069401025772,
"step": 290
},
{
"epoch": 0.46,
"learning_rate": 3.2978188118513814e-06,
"logits/chosen": 0.11096982657909393,
"logits/rejected": 0.24753287434577942,
"logps/chosen": -296.8543701171875,
"logps/rejected": -281.43310546875,
"loss": 0.2306,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": 0.19204005599021912,
"rewards/margins": 0.18743662536144257,
"rewards/rejected": 0.0046034445986151695,
"step": 300
},
{
"epoch": 0.47,
"learning_rate": 3.1705898543111576e-06,
"logits/chosen": 0.07924026995897293,
"logits/rejected": 0.2654581665992737,
"logps/chosen": -327.4444274902344,
"logps/rejected": -319.9097595214844,
"loss": 0.2183,
"rewards/accuracies": 0.8062499761581421,
"rewards/chosen": 0.1851603388786316,
"rewards/margins": 0.23894396424293518,
"rewards/rejected": -0.05378361791372299,
"step": 310
},
{
"epoch": 0.49,
"learning_rate": 3.041460036971664e-06,
"logits/chosen": 0.09540507942438126,
"logits/rejected": 0.22888918220996857,
"logps/chosen": -334.81689453125,
"logps/rejected": -270.1705017089844,
"loss": 0.2204,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": 0.15298177301883698,
"rewards/margins": 0.2034742832183838,
"rewards/rejected": -0.05049251392483711,
"step": 320
},
{
"epoch": 0.5,
"learning_rate": 2.910795392329649e-06,
"logits/chosen": 0.11527317762374878,
"logits/rejected": 0.21196472644805908,
"logps/chosen": -315.2773132324219,
"logps/rejected": -283.607421875,
"loss": 0.1907,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": 0.19066214561462402,
"rewards/margins": 0.22805961966514587,
"rewards/rejected": -0.037397462874650955,
"step": 330
},
{
"epoch": 0.52,
"learning_rate": 2.7789663035166035e-06,
"logits/chosen": 0.054509587585926056,
"logits/rejected": 0.21440072357654572,
"logps/chosen": -327.4574279785156,
"logps/rejected": -277.25958251953125,
"loss": 0.201,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": 0.16702668368816376,
"rewards/margins": 0.22345618903636932,
"rewards/rejected": -0.05642951279878616,
"step": 340
},
{
"epoch": 0.53,
"learning_rate": 2.6463464544075344e-06,
"logits/chosen": 0.0780135840177536,
"logits/rejected": 0.2536531686782837,
"logps/chosen": -305.2668151855469,
"logps/rejected": -291.6106872558594,
"loss": 0.1948,
"rewards/accuracies": 0.793749988079071,
"rewards/chosen": 0.21040305495262146,
"rewards/margins": 0.27535635232925415,
"rewards/rejected": -0.06495330482721329,
"step": 350
},
{
"epoch": 0.55,
"learning_rate": 2.513311770373421e-06,
"logits/chosen": 0.12684503197669983,
"logits/rejected": 0.23648087680339813,
"logps/chosen": -299.1307067871094,
"logps/rejected": -288.98614501953125,
"loss": 0.2112,
"rewards/accuracies": 0.8062499761581421,
"rewards/chosen": 0.18101270496845245,
"rewards/margins": 0.24970810115337372,
"rewards/rejected": -0.06869538873434067,
"step": 360
},
{
"epoch": 0.56,
"learning_rate": 2.380239352679908e-06,
"logits/chosen": 0.050291478633880615,
"logits/rejected": 0.2234155833721161,
"logps/chosen": -296.54595947265625,
"logps/rejected": -269.51702880859375,
"loss": 0.2095,
"rewards/accuracies": 0.6875,
"rewards/chosen": 0.17090751230716705,
"rewards/margins": 0.19086746871471405,
"rewards/rejected": -0.019959963858127594,
"step": 370
},
{
"epoch": 0.58,
"learning_rate": 2.247506409552795e-06,
"logits/chosen": 0.10957841575145721,
"logits/rejected": 0.21787157654762268,
"logps/chosen": -305.3380432128906,
"logps/rejected": -283.4737854003906,
"loss": 0.2027,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 0.17482581734657288,
"rewards/margins": 0.22671441733837128,
"rewards/rejected": -0.0518886037170887,
"step": 380
},
{
"epoch": 0.59,
"learning_rate": 2.1154891869403436e-06,
"logits/chosen": 0.1384754180908203,
"logits/rejected": 0.25176170468330383,
"logps/chosen": -354.0063171386719,
"logps/rejected": -307.35455322265625,
"loss": 0.2073,
"rewards/accuracies": 0.793749988079071,
"rewards/chosen": 0.2005975991487503,
"rewards/margins": 0.27450570464134216,
"rewards/rejected": -0.07390810549259186,
"step": 390
},
{
"epoch": 0.61,
"learning_rate": 1.9845619020032552e-06,
"logits/chosen": 0.09188776463270187,
"logits/rejected": 0.2664358913898468,
"logps/chosen": -328.11083984375,
"logps/rejected": -286.30419921875,
"loss": 0.1895,
"rewards/accuracies": 0.8062499761581421,
"rewards/chosen": 0.1963554322719574,
"rewards/margins": 0.23746831715106964,
"rewards/rejected": -0.04111289232969284,
"step": 400
},
{
"epoch": 0.62,
"learning_rate": 1.8550956823554708e-06,
"logits/chosen": 0.055462319403886795,
"logits/rejected": 0.18411260843276978,
"logps/chosen": -326.14703369140625,
"logps/rejected": -275.77069091796875,
"loss": 0.235,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": 0.1863834410905838,
"rewards/margins": 0.22951778769493103,
"rewards/rejected": -0.043134383857250214,
"step": 410
},
{
"epoch": 0.64,
"learning_rate": 1.7274575140626318e-06,
"logits/chosen": 0.09742958843708038,
"logits/rejected": 0.23133966326713562,
"logps/chosen": -334.5967712402344,
"logps/rejected": -302.8537902832031,
"loss": 0.2048,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": 0.21493402123451233,
"rewards/margins": 0.26205217838287354,
"rewards/rejected": -0.047118157148361206,
"step": 420
},
{
"epoch": 0.66,
"learning_rate": 1.6020092013802002e-06,
"logits/chosen": 0.061180900782346725,
"logits/rejected": 0.22411946952342987,
"logps/chosen": -326.1264953613281,
"logps/rejected": -270.6374206542969,
"loss": 0.2211,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": 0.19443106651306152,
"rewards/margins": 0.2181394100189209,
"rewards/rejected": -0.023708324879407883,
"step": 430
},
{
"epoch": 0.67,
"learning_rate": 1.4791063411799938e-06,
"logits/chosen": 0.11679482460021973,
"logits/rejected": 0.20075193047523499,
"logps/chosen": -313.3033752441406,
"logps/rejected": -301.5423278808594,
"loss": 0.1941,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": 0.18806949257850647,
"rewards/margins": 0.22217002511024475,
"rewards/rejected": -0.03410057723522186,
"step": 440
},
{
"epoch": 0.69,
"learning_rate": 1.3590973149722103e-06,
"logits/chosen": 0.11450199782848358,
"logits/rejected": 0.2375846803188324,
"logps/chosen": -324.9442443847656,
"logps/rejected": -282.5760498046875,
"loss": 0.2164,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": 0.1744813621044159,
"rewards/margins": 0.21334946155548096,
"rewards/rejected": -0.03886810690164566,
"step": 450
},
{
"epoch": 0.7,
"learning_rate": 1.2423223013801946e-06,
"logits/chosen": 0.10633231699466705,
"logits/rejected": 0.2275623381137848,
"logps/chosen": -359.95428466796875,
"logps/rejected": -296.4113464355469,
"loss": 0.1969,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": 0.21253260970115662,
"rewards/margins": 0.24374982714653015,
"rewards/rejected": -0.03121720813214779,
"step": 460
},
{
"epoch": 0.72,
"learning_rate": 1.1291123118671665e-06,
"logits/chosen": 0.03388429060578346,
"logits/rejected": 0.17446021735668182,
"logps/chosen": -308.4808349609375,
"logps/rejected": -263.65283203125,
"loss": 0.2128,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": 0.21648752689361572,
"rewards/margins": 0.22755059599876404,
"rewards/rejected": -0.011063081212341785,
"step": 470
},
{
"epoch": 0.73,
"learning_rate": 1.019788252448267e-06,
"logits/chosen": 0.12791678309440613,
"logits/rejected": 0.28256458044052124,
"logps/chosen": -379.8475341796875,
"logps/rejected": -303.40545654296875,
"loss": 0.182,
"rewards/accuracies": 0.8062499761581421,
"rewards/chosen": 0.22206160426139832,
"rewards/margins": 0.2580808699131012,
"rewards/rejected": -0.036019258201122284,
"step": 480
},
{
"epoch": 0.75,
"learning_rate": 9.146600140475945e-07,
"logits/chosen": 0.11331765353679657,
"logits/rejected": 0.2163075953722,
"logps/chosen": -366.92926025390625,
"logps/rejected": -308.08148193359375,
"loss": 0.2333,
"rewards/accuracies": 0.71875,
"rewards/chosen": 0.1822303682565689,
"rewards/margins": 0.232683464884758,
"rewards/rejected": -0.050453104078769684,
"step": 490
},
{
"epoch": 0.76,
"learning_rate": 8.140255940787059e-07,
"logits/chosen": 0.1049843281507492,
"logits/rejected": 0.20622961223125458,
"logps/chosen": -283.49822998046875,
"logps/rejected": -262.69561767578125,
"loss": 0.2244,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": 0.19038459658622742,
"rewards/margins": 0.19227764010429382,
"rewards/rejected": -0.001893045729957521,
"step": 500
},
{
"epoch": 0.78,
"learning_rate": 7.181702517385789e-07,
"logits/chosen": 0.06920811533927917,
"logits/rejected": 0.22420334815979004,
"logps/chosen": -313.43670654296875,
"logps/rejected": -280.72442626953125,
"loss": 0.2521,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.1772725135087967,
"rewards/margins": 0.1995859146118164,
"rewards/rejected": -0.022313417866826057,
"step": 510
},
{
"epoch": 0.79,
"learning_rate": 6.273656994094232e-07,
"logits/chosen": 0.0857834741473198,
"logits/rejected": 0.1759118139743805,
"logps/chosen": -308.9149475097656,
"logps/rejected": -283.34747314453125,
"loss": 0.1955,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": 0.2155129462480545,
"rewards/margins": 0.22527408599853516,
"rewards/rejected": -0.009761162102222443,
"step": 520
},
{
"epoch": 0.81,
"learning_rate": 5.418693324604082e-07,
"logits/chosen": 0.057937733829021454,
"logits/rejected": 0.24161191284656525,
"logps/chosen": -333.49603271484375,
"logps/rejected": -280.8927917480469,
"loss": 0.2266,
"rewards/accuracies": 0.793749988079071,
"rewards/chosen": 0.21429701149463654,
"rewards/margins": 0.22269944846630096,
"rewards/rejected": -0.008402440696954727,
"step": 530
},
{
"epoch": 0.82,
"learning_rate": 4.619234996325314e-07,
"logits/chosen": 0.07015545666217804,
"logits/rejected": 0.2190779447555542,
"logps/chosen": -344.3935852050781,
"logps/rejected": -303.2052917480469,
"loss": 0.2045,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": 0.21076282858848572,
"rewards/margins": 0.25355157256126404,
"rewards/rejected": -0.04278876259922981,
"step": 540
},
{
"epoch": 0.84,
"learning_rate": 3.877548160747768e-07,
"logits/chosen": 0.08561773598194122,
"logits/rejected": 0.2999951243400574,
"logps/chosen": -330.68475341796875,
"logps/rejected": -281.82275390625,
"loss": 0.2013,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": 0.20333750545978546,
"rewards/margins": 0.23657293617725372,
"rewards/rejected": -0.033235400915145874,
"step": 550
},
{
"epoch": 0.85,
"learning_rate": 3.195735209788528e-07,
"logits/chosen": 0.13458076119422913,
"logits/rejected": 0.27361050248146057,
"logps/chosen": -318.90203857421875,
"logps/rejected": -284.5428466796875,
"loss": 0.2178,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.19831877946853638,
"rewards/margins": 0.23214980959892273,
"rewards/rejected": -0.03383101895451546,
"step": 560
},
{
"epoch": 0.87,
"learning_rate": 2.5757288163336806e-07,
"logits/chosen": 0.11606297641992569,
"logits/rejected": 0.22765600681304932,
"logps/chosen": -313.04278564453125,
"logps/rejected": -281.8597717285156,
"loss": 0.1959,
"rewards/accuracies": 0.78125,
"rewards/chosen": 0.21330790221691132,
"rewards/margins": 0.25364676117897034,
"rewards/rejected": -0.04033887758851051,
"step": 570
},
{
"epoch": 0.88,
"learning_rate": 2.019286455866981e-07,
"logits/chosen": 0.057523488998413086,
"logits/rejected": 0.19763953983783722,
"logps/chosen": -299.98626708984375,
"logps/rejected": -252.2355194091797,
"loss": 0.2448,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.2040693461894989,
"rewards/margins": 0.18340806663036346,
"rewards/rejected": 0.020661287009716034,
"step": 580
},
{
"epoch": 0.9,
"learning_rate": 1.5279854247146703e-07,
"logits/chosen": 0.10874730348587036,
"logits/rejected": 0.26722806692123413,
"logps/chosen": -326.7654724121094,
"logps/rejected": -274.41162109375,
"loss": 0.2041,
"rewards/accuracies": 0.7562500238418579,
"rewards/chosen": 0.2091568410396576,
"rewards/margins": 0.2396632879972458,
"rewards/rejected": -0.030506467446684837,
"step": 590
},
{
"epoch": 0.91,
"learning_rate": 1.1032183690276754e-07,
"logits/chosen": 0.12037558853626251,
"logits/rejected": 0.19973725080490112,
"logps/chosen": -314.14910888671875,
"logps/rejected": -270.01043701171875,
"loss": 0.2167,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 0.21347804367542267,
"rewards/margins": 0.22940710186958313,
"rewards/rejected": -0.015929043292999268,
"step": 600
},
{
"epoch": 0.93,
"learning_rate": 7.46189337174788e-08,
"logits/chosen": 0.08730605989694595,
"logits/rejected": 0.2364271879196167,
"logps/chosen": -289.814453125,
"logps/rejected": -251.5435791015625,
"loss": 0.2221,
"rewards/accuracies": 0.71875,
"rewards/chosen": 0.19059757888317108,
"rewards/margins": 0.19377049803733826,
"rewards/rejected": -0.0031729289330542088,
"step": 610
},
{
"epoch": 0.94,
"learning_rate": 4.579103667367385e-08,
"logits/chosen": 0.13279291987419128,
"logits/rejected": 0.2206091433763504,
"logps/chosen": -342.04498291015625,
"logps/rejected": -276.22613525390625,
"loss": 0.1984,
"rewards/accuracies": 0.71875,
"rewards/chosen": 0.19695451855659485,
"rewards/margins": 0.2266121655702591,
"rewards/rejected": -0.029657626524567604,
"step": 620
},
{
"epoch": 0.96,
"learning_rate": 2.3919861577572924e-08,
"logits/chosen": 0.0992569848895073,
"logits/rejected": 0.26812419295310974,
"logps/chosen": -336.68817138671875,
"logps/rejected": -261.6818542480469,
"loss": 0.2111,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": 0.21192236244678497,
"rewards/margins": 0.2388623207807541,
"rewards/rejected": -0.026939954608678818,
"step": 630
},
{
"epoch": 0.98,
"learning_rate": 9.067404651211808e-09,
"logits/chosen": 0.04493387043476105,
"logits/rejected": 0.1637609452009201,
"logps/chosen": -332.1258544921875,
"logps/rejected": -283.83709716796875,
"loss": 0.1939,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": 0.20134110748767853,
"rewards/margins": 0.23969101905822754,
"rewards/rejected": -0.03834990784525871,
"step": 640
},
{
"epoch": 0.99,
"learning_rate": 1.2757667974155896e-09,
"logits/chosen": 0.1145804151892662,
"logits/rejected": 0.23529252409934998,
"logps/chosen": -351.08197021484375,
"logps/rejected": -293.947998046875,
"loss": 0.2055,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": 0.2066545933485031,
"rewards/margins": 0.22757765650749207,
"rewards/rejected": -0.020923063158988953,
"step": 650
},
{
"epoch": 1.0,
"step": 656,
"total_flos": 0.0,
"train_loss": 0.22703545004492853,
"train_runtime": 7888.8763,
"train_samples_per_second": 2.662,
"train_steps_per_second": 0.083
}
],
"logging_steps": 10,
"max_steps": 656,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 100,
"total_flos": 0.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}