notux-8x7b-v1 / trainer_state.json
dvilasuero's picture
dvilasuero HF staff
Model save
86c89d1
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 200,
"global_step": 921,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 5.3763440860215056e-09,
"logits/chosen": -1.3876744508743286,
"logits/rejected": -1.355853796005249,
"logps/chosen": -672.8331298828125,
"logps/rejected": -549.942138671875,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 5.3763440860215054e-08,
"logits/chosen": -1.1944549083709717,
"logits/rejected": -1.1097763776779175,
"logps/chosen": -368.380859375,
"logps/rejected": -353.98101806640625,
"loss": 0.709,
"rewards/accuracies": 0.4305555522441864,
"rewards/chosen": 0.015228834003210068,
"rewards/margins": 0.0040851375088095665,
"rewards/rejected": 0.011143695563077927,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 1.0752688172043011e-07,
"logits/chosen": -1.4370033740997314,
"logits/rejected": -1.4178457260131836,
"logps/chosen": -391.20050048828125,
"logps/rejected": -409.3836364746094,
"loss": 0.7198,
"rewards/accuracies": 0.512499988079071,
"rewards/chosen": -0.015771254897117615,
"rewards/margins": 0.035423360764980316,
"rewards/rejected": -0.05119461938738823,
"step": 20
},
{
"epoch": 0.03,
"learning_rate": 1.6129032258064515e-07,
"logits/chosen": -1.416802167892456,
"logits/rejected": -1.4156275987625122,
"logps/chosen": -411.7056579589844,
"logps/rejected": -425.6934509277344,
"loss": 0.6959,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.15962687134742737,
"rewards/margins": 0.045185185968875885,
"rewards/rejected": -0.20481204986572266,
"step": 30
},
{
"epoch": 0.04,
"learning_rate": 2.1505376344086022e-07,
"logits/chosen": -1.2398654222488403,
"logits/rejected": -1.2564724683761597,
"logps/chosen": -385.45782470703125,
"logps/rejected": -452.55535888671875,
"loss": 0.6365,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.3178340792655945,
"rewards/margins": 0.27168726921081543,
"rewards/rejected": -0.5895214080810547,
"step": 40
},
{
"epoch": 0.05,
"learning_rate": 2.6881720430107523e-07,
"logits/chosen": -1.2720705270767212,
"logits/rejected": -0.9971787333488464,
"logps/chosen": -339.6200866699219,
"logps/rejected": -361.6040954589844,
"loss": 0.5605,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.790237307548523,
"rewards/margins": 0.61395663022995,
"rewards/rejected": -1.4041939973831177,
"step": 50
},
{
"epoch": 0.07,
"learning_rate": 3.225806451612903e-07,
"logits/chosen": -1.350678563117981,
"logits/rejected": -1.3656706809997559,
"logps/chosen": -417.0431213378906,
"logps/rejected": -416.76165771484375,
"loss": 0.5485,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.473722219467163,
"rewards/margins": 1.0274063348770142,
"rewards/rejected": -2.5011284351348877,
"step": 60
},
{
"epoch": 0.08,
"learning_rate": 3.7634408602150537e-07,
"logits/chosen": -1.2629830837249756,
"logits/rejected": -1.415208339691162,
"logps/chosen": -377.01275634765625,
"logps/rejected": -445.7503967285156,
"loss": 0.5203,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -1.3981420993804932,
"rewards/margins": 0.6980546116828918,
"rewards/rejected": -2.0961966514587402,
"step": 70
},
{
"epoch": 0.09,
"learning_rate": 4.3010752688172043e-07,
"logits/chosen": -1.3798706531524658,
"logits/rejected": -1.3563942909240723,
"logps/chosen": -411.2007751464844,
"logps/rejected": -434.94476318359375,
"loss": 0.5067,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.8918856382369995,
"rewards/margins": 1.0622823238372803,
"rewards/rejected": -1.9541680812835693,
"step": 80
},
{
"epoch": 0.1,
"learning_rate": 4.838709677419355e-07,
"logits/chosen": -1.6255207061767578,
"logits/rejected": -1.610365629196167,
"logps/chosen": -355.00128173828125,
"logps/rejected": -383.69866943359375,
"loss": 0.511,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.8219491839408875,
"rewards/margins": 0.6381314992904663,
"rewards/rejected": -1.460080623626709,
"step": 90
},
{
"epoch": 0.11,
"learning_rate": 4.957729468599034e-07,
"logits/chosen": -1.3692975044250488,
"logits/rejected": -1.3177874088287354,
"logps/chosen": -378.91265869140625,
"logps/rejected": -361.2559509277344,
"loss": 0.4586,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.2890338897705078,
"rewards/margins": 1.1477439403533936,
"rewards/rejected": -1.4367780685424805,
"step": 100
},
{
"epoch": 0.12,
"learning_rate": 4.897342995169082e-07,
"logits/chosen": -1.3526300191879272,
"logits/rejected": -1.2792448997497559,
"logps/chosen": -378.30126953125,
"logps/rejected": -415.57464599609375,
"loss": 0.4725,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.5904244184494019,
"rewards/margins": 1.3282153606414795,
"rewards/rejected": -1.9186397790908813,
"step": 110
},
{
"epoch": 0.13,
"learning_rate": 4.83695652173913e-07,
"logits/chosen": -1.4887703657150269,
"logits/rejected": -1.3336336612701416,
"logps/chosen": -374.025390625,
"logps/rejected": -446.5819396972656,
"loss": 0.4726,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.5127348899841309,
"rewards/margins": 1.7226130962371826,
"rewards/rejected": -2.2353477478027344,
"step": 120
},
{
"epoch": 0.14,
"learning_rate": 4.776570048309178e-07,
"logits/chosen": -1.2795524597167969,
"logits/rejected": -1.3249471187591553,
"logps/chosen": -375.96710205078125,
"logps/rejected": -457.1031188964844,
"loss": 0.4781,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.473715215921402,
"rewards/margins": 1.3638942241668701,
"rewards/rejected": -1.8376095294952393,
"step": 130
},
{
"epoch": 0.15,
"learning_rate": 4.716183574879227e-07,
"logits/chosen": -0.8936523199081421,
"logits/rejected": -0.731735348701477,
"logps/chosen": -377.02410888671875,
"logps/rejected": -374.2434387207031,
"loss": 0.4771,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.4292902946472168,
"rewards/margins": 1.5224840641021729,
"rewards/rejected": -1.9517742395401,
"step": 140
},
{
"epoch": 0.16,
"learning_rate": 4.655797101449275e-07,
"logits/chosen": -1.3006255626678467,
"logits/rejected": -1.1428701877593994,
"logps/chosen": -422.60986328125,
"logps/rejected": -394.9052429199219,
"loss": 0.4141,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.40644779801368713,
"rewards/margins": 1.319599986076355,
"rewards/rejected": -1.7260475158691406,
"step": 150
},
{
"epoch": 0.17,
"learning_rate": 4.5954106280193235e-07,
"logits/chosen": -1.3789948225021362,
"logits/rejected": -1.2863258123397827,
"logps/chosen": -305.4602355957031,
"logps/rejected": -374.1790466308594,
"loss": 0.4965,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.9782305955886841,
"rewards/margins": 1.459176778793335,
"rewards/rejected": -2.4374070167541504,
"step": 160
},
{
"epoch": 0.18,
"learning_rate": 4.5350241545893717e-07,
"logits/chosen": -1.4751253128051758,
"logits/rejected": -1.1761484146118164,
"logps/chosen": -395.0107727050781,
"logps/rejected": -390.03076171875,
"loss": 0.441,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.5482892394065857,
"rewards/margins": 1.585138201713562,
"rewards/rejected": -2.133427143096924,
"step": 170
},
{
"epoch": 0.2,
"learning_rate": 4.47463768115942e-07,
"logits/chosen": -1.6447718143463135,
"logits/rejected": -1.6261088848114014,
"logps/chosen": -430.76129150390625,
"logps/rejected": -425.74969482421875,
"loss": 0.4392,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.5816106796264648,
"rewards/margins": 1.8813564777374268,
"rewards/rejected": -2.4629669189453125,
"step": 180
},
{
"epoch": 0.21,
"learning_rate": 4.414251207729469e-07,
"logits/chosen": -1.2417869567871094,
"logits/rejected": -1.3334053754806519,
"logps/chosen": -337.415771484375,
"logps/rejected": -356.4654235839844,
"loss": 0.4574,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.755621075630188,
"rewards/margins": 1.5105997323989868,
"rewards/rejected": -2.266220808029175,
"step": 190
},
{
"epoch": 0.22,
"learning_rate": 4.3538647342995165e-07,
"logits/chosen": -1.5782802104949951,
"logits/rejected": -1.6571115255355835,
"logps/chosen": -431.74151611328125,
"logps/rejected": -431.4637145996094,
"loss": 0.4384,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.07804969698190689,
"rewards/margins": 1.6836214065551758,
"rewards/rejected": -1.7616710662841797,
"step": 200
},
{
"epoch": 0.22,
"eval_logits/chosen": -1.4511131048202515,
"eval_logits/rejected": -1.3157293796539307,
"eval_logps/chosen": -397.8616638183594,
"eval_logps/rejected": -405.7994079589844,
"eval_loss": 0.4556037187576294,
"eval_rewards/accuracies": 0.7936508059501648,
"eval_rewards/chosen": -0.3274631202220917,
"eval_rewards/margins": 1.6173655986785889,
"eval_rewards/rejected": -1.944828748703003,
"eval_runtime": 405.1145,
"eval_samples_per_second": 4.937,
"eval_steps_per_second": 0.156,
"step": 200
},
{
"epoch": 0.23,
"learning_rate": 4.2934782608695653e-07,
"logits/chosen": -0.9520143270492554,
"logits/rejected": -1.01302969455719,
"logps/chosen": -444.7759704589844,
"logps/rejected": -401.30389404296875,
"loss": 0.4141,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.38320282101631165,
"rewards/margins": 1.236628770828247,
"rewards/rejected": -1.6198316812515259,
"step": 210
},
{
"epoch": 0.24,
"learning_rate": 4.2330917874396135e-07,
"logits/chosen": -1.601231336593628,
"logits/rejected": -1.4172089099884033,
"logps/chosen": -431.98272705078125,
"logps/rejected": -412.7859802246094,
"loss": 0.4817,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -0.31900420784950256,
"rewards/margins": 1.7263078689575195,
"rewards/rejected": -2.0453121662139893,
"step": 220
},
{
"epoch": 0.25,
"learning_rate": 4.172705314009662e-07,
"logits/chosen": -1.1527206897735596,
"logits/rejected": -0.8433942794799805,
"logps/chosen": -383.67559814453125,
"logps/rejected": -369.19342041015625,
"loss": 0.45,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.5148889422416687,
"rewards/margins": 1.6342861652374268,
"rewards/rejected": -2.1491751670837402,
"step": 230
},
{
"epoch": 0.26,
"learning_rate": 4.11231884057971e-07,
"logits/chosen": -1.3957723379135132,
"logits/rejected": -1.4669153690338135,
"logps/chosen": -366.89093017578125,
"logps/rejected": -381.7056579589844,
"loss": 0.4573,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.2394193708896637,
"rewards/margins": 1.0727304220199585,
"rewards/rejected": -1.3121497631072998,
"step": 240
},
{
"epoch": 0.27,
"learning_rate": 4.0519323671497583e-07,
"logits/chosen": -1.1060564517974854,
"logits/rejected": -1.020811676979065,
"logps/chosen": -431.10791015625,
"logps/rejected": -398.845458984375,
"loss": 0.4008,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": -0.08395320922136307,
"rewards/margins": 1.6107537746429443,
"rewards/rejected": -1.694706916809082,
"step": 250
},
{
"epoch": 0.28,
"learning_rate": 3.9915458937198065e-07,
"logits/chosen": -1.0142405033111572,
"logits/rejected": -1.0058037042617798,
"logps/chosen": -404.4456481933594,
"logps/rejected": -417.4208984375,
"loss": 0.4342,
"rewards/accuracies": 0.8125,
"rewards/chosen": 0.09296660125255585,
"rewards/margins": 1.8066604137420654,
"rewards/rejected": -1.713693618774414,
"step": 260
},
{
"epoch": 0.29,
"learning_rate": 3.9311594202898553e-07,
"logits/chosen": -0.816851794719696,
"logits/rejected": -0.7470394372940063,
"logps/chosen": -395.8139343261719,
"logps/rejected": -355.42071533203125,
"loss": 0.4899,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.38039931654930115,
"rewards/margins": 1.4586222171783447,
"rewards/rejected": -1.8390214443206787,
"step": 270
},
{
"epoch": 0.3,
"learning_rate": 3.870772946859903e-07,
"logits/chosen": -1.0235198736190796,
"logits/rejected": -1.009104609489441,
"logps/chosen": -419.8531188964844,
"logps/rejected": -417.8365783691406,
"loss": 0.4258,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.24367530643939972,
"rewards/margins": 1.267002820968628,
"rewards/rejected": -1.5106781721115112,
"step": 280
},
{
"epoch": 0.31,
"learning_rate": 3.810386473429952e-07,
"logits/chosen": -1.1125423908233643,
"logits/rejected": -1.093120813369751,
"logps/chosen": -347.3122863769531,
"logps/rejected": -387.7191467285156,
"loss": 0.435,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": 0.060557298362255096,
"rewards/margins": 1.8192393779754639,
"rewards/rejected": -1.7586820125579834,
"step": 290
},
{
"epoch": 0.33,
"learning_rate": 3.75e-07,
"logits/chosen": -1.1970382928848267,
"logits/rejected": -1.1414059400558472,
"logps/chosen": -380.08892822265625,
"logps/rejected": -393.11126708984375,
"loss": 0.4137,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.4049530625343323,
"rewards/margins": 1.5264403820037842,
"rewards/rejected": -1.9313932657241821,
"step": 300
},
{
"epoch": 0.34,
"learning_rate": 3.689613526570048e-07,
"logits/chosen": -1.4467629194259644,
"logits/rejected": -1.4420421123504639,
"logps/chosen": -412.6943359375,
"logps/rejected": -411.64031982421875,
"loss": 0.4545,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.44115525484085083,
"rewards/margins": 1.5014537572860718,
"rewards/rejected": -1.9426085948944092,
"step": 310
},
{
"epoch": 0.35,
"learning_rate": 3.6292270531400966e-07,
"logits/chosen": -1.576790690422058,
"logits/rejected": -1.605373740196228,
"logps/chosen": -442.7998046875,
"logps/rejected": -471.8130798339844,
"loss": 0.426,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.1621936857700348,
"rewards/margins": 1.6487839221954346,
"rewards/rejected": -1.810977578163147,
"step": 320
},
{
"epoch": 0.36,
"learning_rate": 3.5688405797101443e-07,
"logits/chosen": -1.3066380023956299,
"logits/rejected": -1.1275885105133057,
"logps/chosen": -369.5173034667969,
"logps/rejected": -393.6142883300781,
"loss": 0.4594,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.3911195993423462,
"rewards/margins": 1.561603307723999,
"rewards/rejected": -1.9527229070663452,
"step": 330
},
{
"epoch": 0.37,
"learning_rate": 3.508454106280193e-07,
"logits/chosen": -1.460533857345581,
"logits/rejected": -1.3913938999176025,
"logps/chosen": -461.4322814941406,
"logps/rejected": -443.2294006347656,
"loss": 0.4343,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": 0.05905343219637871,
"rewards/margins": 1.777876615524292,
"rewards/rejected": -1.7188230752944946,
"step": 340
},
{
"epoch": 0.38,
"learning_rate": 3.4480676328502414e-07,
"logits/chosen": -1.0480554103851318,
"logits/rejected": -0.939221203327179,
"logps/chosen": -389.5008544921875,
"logps/rejected": -391.26678466796875,
"loss": 0.4464,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.5172668099403381,
"rewards/margins": 1.5226786136627197,
"rewards/rejected": -2.039945125579834,
"step": 350
},
{
"epoch": 0.39,
"learning_rate": 3.3876811594202896e-07,
"logits/chosen": -1.367060899734497,
"logits/rejected": -1.250497579574585,
"logps/chosen": -374.1922302246094,
"logps/rejected": -377.8554382324219,
"loss": 0.4265,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.34825998544692993,
"rewards/margins": 1.6436443328857422,
"rewards/rejected": -1.9919040203094482,
"step": 360
},
{
"epoch": 0.4,
"learning_rate": 3.327294685990338e-07,
"logits/chosen": -1.4619824886322021,
"logits/rejected": -1.224457025527954,
"logps/chosen": -394.55694580078125,
"logps/rejected": -406.80389404296875,
"loss": 0.4343,
"rewards/accuracies": 0.8125,
"rewards/chosen": 0.11512075364589691,
"rewards/margins": 1.9784886837005615,
"rewards/rejected": -1.863368034362793,
"step": 370
},
{
"epoch": 0.41,
"learning_rate": 3.266908212560386e-07,
"logits/chosen": -1.5472246408462524,
"logits/rejected": -1.3760802745819092,
"logps/chosen": -370.1455993652344,
"logps/rejected": -360.59735107421875,
"loss": 0.4365,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.24435913562774658,
"rewards/margins": 1.2893067598342896,
"rewards/rejected": -1.5336658954620361,
"step": 380
},
{
"epoch": 0.42,
"learning_rate": 3.2065217391304344e-07,
"logits/chosen": -1.3668596744537354,
"logits/rejected": -1.2702538967132568,
"logps/chosen": -429.017822265625,
"logps/rejected": -415.56439208984375,
"loss": 0.4121,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.2810801863670349,
"rewards/margins": 1.7658510208129883,
"rewards/rejected": -2.046931028366089,
"step": 390
},
{
"epoch": 0.43,
"learning_rate": 3.146135265700483e-07,
"logits/chosen": -1.3646513223648071,
"logits/rejected": -1.072363257408142,
"logps/chosen": -390.0302429199219,
"logps/rejected": -395.0215148925781,
"loss": 0.4064,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.38891416788101196,
"rewards/margins": 1.990821123123169,
"rewards/rejected": -2.379735231399536,
"step": 400
},
{
"epoch": 0.43,
"eval_logits/chosen": -0.6539027094841003,
"eval_logits/rejected": -0.7660450339317322,
"eval_logps/chosen": -396.7496032714844,
"eval_logps/rejected": -408.44091796875,
"eval_loss": 0.4285692274570465,
"eval_rewards/accuracies": 0.8253968358039856,
"eval_rewards/chosen": -0.21625187993049622,
"eval_rewards/margins": 1.9927276372909546,
"eval_rewards/rejected": -2.208979606628418,
"eval_runtime": 409.2634,
"eval_samples_per_second": 4.887,
"eval_steps_per_second": 0.154,
"step": 400
},
{
"epoch": 0.45,
"learning_rate": 3.085748792270531e-07,
"logits/chosen": -1.2127052545547485,
"logits/rejected": -1.2303555011749268,
"logps/chosen": -399.98858642578125,
"logps/rejected": -388.2366638183594,
"loss": 0.4587,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.6597892642021179,
"rewards/margins": 1.7150468826293945,
"rewards/rejected": -2.3748364448547363,
"step": 410
},
{
"epoch": 0.46,
"learning_rate": 3.0253623188405797e-07,
"logits/chosen": -1.6032695770263672,
"logits/rejected": -1.3187893629074097,
"logps/chosen": -347.73553466796875,
"logps/rejected": -366.84814453125,
"loss": 0.4217,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.20497450232505798,
"rewards/margins": 1.4589314460754395,
"rewards/rejected": -1.6639057397842407,
"step": 420
},
{
"epoch": 0.47,
"learning_rate": 2.964975845410628e-07,
"logits/chosen": -1.4543535709381104,
"logits/rejected": -1.2742624282836914,
"logps/chosen": -423.95111083984375,
"logps/rejected": -437.7804260253906,
"loss": 0.4479,
"rewards/accuracies": 0.8125,
"rewards/chosen": 0.25907284021377563,
"rewards/margins": 2.2142040729522705,
"rewards/rejected": -1.9551312923431396,
"step": 430
},
{
"epoch": 0.48,
"learning_rate": 2.904589371980676e-07,
"logits/chosen": -1.9760487079620361,
"logits/rejected": -1.7164779901504517,
"logps/chosen": -386.17193603515625,
"logps/rejected": -450.7064514160156,
"loss": 0.4361,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.34018686413764954,
"rewards/margins": 1.6212705373764038,
"rewards/rejected": -1.9614572525024414,
"step": 440
},
{
"epoch": 0.49,
"learning_rate": 2.8442028985507245e-07,
"logits/chosen": -1.5040684938430786,
"logits/rejected": -1.4459855556488037,
"logps/chosen": -410.3436584472656,
"logps/rejected": -438.83837890625,
"loss": 0.4386,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -0.016602765768766403,
"rewards/margins": 1.8991400003433228,
"rewards/rejected": -1.9157428741455078,
"step": 450
},
{
"epoch": 0.5,
"learning_rate": 2.7838164251207727e-07,
"logits/chosen": -1.193297266960144,
"logits/rejected": -0.9360073804855347,
"logps/chosen": -415.88568115234375,
"logps/rejected": -389.2828674316406,
"loss": 0.4233,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": 0.08352740854024887,
"rewards/margins": 2.2130849361419678,
"rewards/rejected": -2.1295576095581055,
"step": 460
},
{
"epoch": 0.51,
"learning_rate": 2.723429951690821e-07,
"logits/chosen": -1.5409491062164307,
"logits/rejected": -1.7222496271133423,
"logps/chosen": -426.3505859375,
"logps/rejected": -415.15753173828125,
"loss": 0.3939,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -0.04843442887067795,
"rewards/margins": 2.11486554145813,
"rewards/rejected": -2.163300037384033,
"step": 470
},
{
"epoch": 0.52,
"learning_rate": 2.66304347826087e-07,
"logits/chosen": -1.1968003511428833,
"logits/rejected": -1.0116466283798218,
"logps/chosen": -381.65277099609375,
"logps/rejected": -407.2720642089844,
"loss": 0.4468,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": 0.10518859326839447,
"rewards/margins": 2.6760292053222656,
"rewards/rejected": -2.570840358734131,
"step": 480
},
{
"epoch": 0.53,
"learning_rate": 2.6026570048309175e-07,
"logits/chosen": -1.5621907711029053,
"logits/rejected": -1.4111429452896118,
"logps/chosen": -331.765380859375,
"logps/rejected": -345.84417724609375,
"loss": 0.4083,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.2680264115333557,
"rewards/margins": 1.9610168933868408,
"rewards/rejected": -2.229043483734131,
"step": 490
},
{
"epoch": 0.54,
"learning_rate": 2.5422705314009663e-07,
"logits/chosen": -1.137390375137329,
"logits/rejected": -0.9998054504394531,
"logps/chosen": -393.7897644042969,
"logps/rejected": -426.21612548828125,
"loss": 0.4487,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.3346656262874603,
"rewards/margins": 1.5842936038970947,
"rewards/rejected": -1.9189590215682983,
"step": 500
},
{
"epoch": 0.55,
"learning_rate": 2.4818840579710145e-07,
"logits/chosen": -1.2915681600570679,
"logits/rejected": -1.1393409967422485,
"logps/chosen": -476.5746154785156,
"logps/rejected": -397.6057434082031,
"loss": 0.4763,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.33603593707084656,
"rewards/margins": 1.7989603281021118,
"rewards/rejected": -2.134996175765991,
"step": 510
},
{
"epoch": 0.56,
"learning_rate": 2.421497584541063e-07,
"logits/chosen": -1.4665111303329468,
"logits/rejected": -1.4302623271942139,
"logps/chosen": -410.9242248535156,
"logps/rejected": -435.70721435546875,
"loss": 0.4378,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.36040204763412476,
"rewards/margins": 1.7777122259140015,
"rewards/rejected": -2.1381144523620605,
"step": 520
},
{
"epoch": 0.58,
"learning_rate": 2.361111111111111e-07,
"logits/chosen": -1.3417316675186157,
"logits/rejected": -1.2472248077392578,
"logps/chosen": -396.3175354003906,
"logps/rejected": -406.9625244140625,
"loss": 0.3857,
"rewards/accuracies": 0.862500011920929,
"rewards/chosen": 0.186844140291214,
"rewards/margins": 2.318281650543213,
"rewards/rejected": -2.131437301635742,
"step": 530
},
{
"epoch": 0.59,
"learning_rate": 2.3007246376811593e-07,
"logits/chosen": -1.5599995851516724,
"logits/rejected": -1.4291542768478394,
"logps/chosen": -411.170654296875,
"logps/rejected": -423.13177490234375,
"loss": 0.4244,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.1071757823228836,
"rewards/margins": 1.971361517906189,
"rewards/rejected": -2.0785374641418457,
"step": 540
},
{
"epoch": 0.6,
"learning_rate": 2.2403381642512075e-07,
"logits/chosen": -1.2990379333496094,
"logits/rejected": -1.1973732709884644,
"logps/chosen": -418.79437255859375,
"logps/rejected": -472.189453125,
"loss": 0.4643,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": 0.147103950381279,
"rewards/margins": 2.323150157928467,
"rewards/rejected": -2.176046371459961,
"step": 550
},
{
"epoch": 0.61,
"learning_rate": 2.1799516908212558e-07,
"logits/chosen": -1.8142582178115845,
"logits/rejected": -1.725685715675354,
"logps/chosen": -374.6809387207031,
"logps/rejected": -410.0592346191406,
"loss": 0.4145,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -0.23340897262096405,
"rewards/margins": 2.3226966857910156,
"rewards/rejected": -2.556105852127075,
"step": 560
},
{
"epoch": 0.62,
"learning_rate": 2.1195652173913043e-07,
"logits/chosen": -1.4505486488342285,
"logits/rejected": -1.3042548894882202,
"logps/chosen": -409.4254150390625,
"logps/rejected": -374.3670654296875,
"loss": 0.4178,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": 0.1292637288570404,
"rewards/margins": 2.543600082397461,
"rewards/rejected": -2.414336681365967,
"step": 570
},
{
"epoch": 0.63,
"learning_rate": 2.0591787439613526e-07,
"logits/chosen": -0.5316571593284607,
"logits/rejected": -0.45167016983032227,
"logps/chosen": -386.24517822265625,
"logps/rejected": -434.59393310546875,
"loss": 0.3853,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.4753696322441101,
"rewards/margins": 1.9324595928192139,
"rewards/rejected": -2.4078292846679688,
"step": 580
},
{
"epoch": 0.64,
"learning_rate": 1.9987922705314008e-07,
"logits/chosen": -1.3159304857254028,
"logits/rejected": -1.087425947189331,
"logps/chosen": -377.9801025390625,
"logps/rejected": -377.0760803222656,
"loss": 0.4461,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -0.38399359583854675,
"rewards/margins": 2.1969008445739746,
"rewards/rejected": -2.580894708633423,
"step": 590
},
{
"epoch": 0.65,
"learning_rate": 1.938405797101449e-07,
"logits/chosen": -0.9003721475601196,
"logits/rejected": -0.6628047227859497,
"logps/chosen": -435.5120544433594,
"logps/rejected": -446.45550537109375,
"loss": 0.3952,
"rewards/accuracies": 0.875,
"rewards/chosen": 0.012645396403968334,
"rewards/margins": 1.9577815532684326,
"rewards/rejected": -1.9451364278793335,
"step": 600
},
{
"epoch": 0.65,
"eval_logits/chosen": -0.7205740809440613,
"eval_logits/rejected": -0.6783490777015686,
"eval_logps/chosen": -395.8981628417969,
"eval_logps/rejected": -407.95367431640625,
"eval_loss": 0.42750075459480286,
"eval_rewards/accuracies": 0.8015872836112976,
"eval_rewards/chosen": -0.13110622763633728,
"eval_rewards/margins": 2.0291495323181152,
"eval_rewards/rejected": -2.1602559089660645,
"eval_runtime": 404.1135,
"eval_samples_per_second": 4.949,
"eval_steps_per_second": 0.156,
"step": 600
},
{
"epoch": 0.66,
"learning_rate": 1.8780193236714976e-07,
"logits/chosen": -0.49158763885498047,
"logits/rejected": -0.30927062034606934,
"logps/chosen": -421.99005126953125,
"logps/rejected": -378.0501708984375,
"loss": 0.4263,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.14008215069770813,
"rewards/margins": 1.6353435516357422,
"rewards/rejected": -1.775425672531128,
"step": 610
},
{
"epoch": 0.67,
"learning_rate": 1.8176328502415459e-07,
"logits/chosen": -1.3565137386322021,
"logits/rejected": -1.1652584075927734,
"logps/chosen": -450.35540771484375,
"logps/rejected": -436.74761962890625,
"loss": 0.4252,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.03007013536989689,
"rewards/margins": 1.9853531122207642,
"rewards/rejected": -2.01542329788208,
"step": 620
},
{
"epoch": 0.68,
"learning_rate": 1.757246376811594e-07,
"logits/chosen": -0.015371406450867653,
"logits/rejected": -0.01580933667719364,
"logps/chosen": -382.2900390625,
"logps/rejected": -360.01446533203125,
"loss": 0.3848,
"rewards/accuracies": 0.949999988079071,
"rewards/chosen": 0.17353440821170807,
"rewards/margins": 2.286402940750122,
"rewards/rejected": -2.1128687858581543,
"step": 630
},
{
"epoch": 0.69,
"learning_rate": 1.6968599033816424e-07,
"logits/chosen": 0.3802258372306824,
"logits/rejected": 0.4669532775878906,
"logps/chosen": -355.05322265625,
"logps/rejected": -390.06976318359375,
"loss": 0.4137,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.6333318948745728,
"rewards/margins": 1.6801398992538452,
"rewards/rejected": -2.313471794128418,
"step": 640
},
{
"epoch": 0.71,
"learning_rate": 1.636473429951691e-07,
"logits/chosen": 0.4362594485282898,
"logits/rejected": 0.5603054761886597,
"logps/chosen": -411.2711486816406,
"logps/rejected": -398.11346435546875,
"loss": 0.4552,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.7625397443771362,
"rewards/margins": 1.8215090036392212,
"rewards/rejected": -2.5840485095977783,
"step": 650
},
{
"epoch": 0.72,
"learning_rate": 1.5760869565217392e-07,
"logits/chosen": 0.4388657212257385,
"logits/rejected": 0.8059428334236145,
"logps/chosen": -406.7152404785156,
"logps/rejected": -383.4807434082031,
"loss": 0.4061,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.5559307336807251,
"rewards/margins": 1.9154608249664307,
"rewards/rejected": -2.4713916778564453,
"step": 660
},
{
"epoch": 0.73,
"learning_rate": 1.5157004830917874e-07,
"logits/chosen": 0.707136332988739,
"logits/rejected": 0.8469961881637573,
"logps/chosen": -373.77032470703125,
"logps/rejected": -383.32525634765625,
"loss": 0.4709,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.159020334482193,
"rewards/margins": 1.6874653100967407,
"rewards/rejected": -1.8464854955673218,
"step": 670
},
{
"epoch": 0.74,
"learning_rate": 1.4553140096618357e-07,
"logits/chosen": -0.08916174620389938,
"logits/rejected": -0.07962872833013535,
"logps/chosen": -403.15765380859375,
"logps/rejected": -382.7056884765625,
"loss": 0.4123,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": 0.1752249300479889,
"rewards/margins": 2.1825637817382812,
"rewards/rejected": -2.007338762283325,
"step": 680
},
{
"epoch": 0.75,
"learning_rate": 1.3949275362318842e-07,
"logits/chosen": 0.2861272692680359,
"logits/rejected": 0.5865569114685059,
"logps/chosen": -438.97808837890625,
"logps/rejected": -424.977294921875,
"loss": 0.3888,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.10896565020084381,
"rewards/margins": 1.5186336040496826,
"rewards/rejected": -1.6275993585586548,
"step": 690
},
{
"epoch": 0.76,
"learning_rate": 1.3345410628019324e-07,
"logits/chosen": 0.7959792017936707,
"logits/rejected": 1.0797499418258667,
"logps/chosen": -399.54302978515625,
"logps/rejected": -411.88824462890625,
"loss": 0.4496,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.02021363377571106,
"rewards/margins": 2.1036648750305176,
"rewards/rejected": -2.1238787174224854,
"step": 700
},
{
"epoch": 0.77,
"learning_rate": 1.2741545893719807e-07,
"logits/chosen": 0.061588358134031296,
"logits/rejected": 0.5792427659034729,
"logps/chosen": -389.0401611328125,
"logps/rejected": -442.1222229003906,
"loss": 0.461,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -0.3942088484764099,
"rewards/margins": 1.9804880619049072,
"rewards/rejected": -2.3746962547302246,
"step": 710
},
{
"epoch": 0.78,
"learning_rate": 1.213768115942029e-07,
"logits/chosen": -0.3789304494857788,
"logits/rejected": -0.5976569056510925,
"logps/chosen": -383.9620056152344,
"logps/rejected": -411.39923095703125,
"loss": 0.3725,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -0.1539289653301239,
"rewards/margins": 2.0121874809265137,
"rewards/rejected": -2.166116237640381,
"step": 720
},
{
"epoch": 0.79,
"learning_rate": 1.1533816425120772e-07,
"logits/chosen": -0.32424452900886536,
"logits/rejected": 0.054877616465091705,
"logps/chosen": -399.9856262207031,
"logps/rejected": -433.40008544921875,
"loss": 0.4335,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.20857331156730652,
"rewards/margins": 1.7776143550872803,
"rewards/rejected": -1.9861876964569092,
"step": 730
},
{
"epoch": 0.8,
"learning_rate": 1.0929951690821256e-07,
"logits/chosen": 0.7587161064147949,
"logits/rejected": 0.4672706127166748,
"logps/chosen": -394.1853942871094,
"logps/rejected": -397.8048095703125,
"loss": 0.4497,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -0.04534666985273361,
"rewards/margins": 2.2218165397644043,
"rewards/rejected": -2.267162799835205,
"step": 740
},
{
"epoch": 0.81,
"learning_rate": 1.0326086956521738e-07,
"logits/chosen": -0.007064342498779297,
"logits/rejected": -0.10647717863321304,
"logps/chosen": -451.4822692871094,
"logps/rejected": -446.7669372558594,
"loss": 0.4347,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.011669009923934937,
"rewards/margins": 1.8043409585952759,
"rewards/rejected": -1.8160098791122437,
"step": 750
},
{
"epoch": 0.83,
"learning_rate": 9.722222222222222e-08,
"logits/chosen": -0.4753515124320984,
"logits/rejected": -0.590586245059967,
"logps/chosen": -471.35137939453125,
"logps/rejected": -436.88531494140625,
"loss": 0.386,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": 0.01547972857952118,
"rewards/margins": 2.1215872764587402,
"rewards/rejected": -2.106107473373413,
"step": 760
},
{
"epoch": 0.84,
"learning_rate": 9.118357487922705e-08,
"logits/chosen": -0.4006249010562897,
"logits/rejected": -0.5747838616371155,
"logps/chosen": -375.1439514160156,
"logps/rejected": -383.0394287109375,
"loss": 0.3963,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.3230701684951782,
"rewards/margins": 2.1252005100250244,
"rewards/rejected": -2.448270797729492,
"step": 770
},
{
"epoch": 0.85,
"learning_rate": 8.514492753623189e-08,
"logits/chosen": 0.18706265091896057,
"logits/rejected": 0.14009952545166016,
"logps/chosen": -381.85443115234375,
"logps/rejected": -417.75146484375,
"loss": 0.3904,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -0.3124622702598572,
"rewards/margins": 1.5956885814666748,
"rewards/rejected": -1.9081509113311768,
"step": 780
},
{
"epoch": 0.86,
"learning_rate": 7.910628019323671e-08,
"logits/chosen": 0.359174907207489,
"logits/rejected": 0.3385276198387146,
"logps/chosen": -429.9090270996094,
"logps/rejected": -362.47857666015625,
"loss": 0.4715,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.23555055260658264,
"rewards/margins": 1.2514636516571045,
"rewards/rejected": -1.4870140552520752,
"step": 790
},
{
"epoch": 0.87,
"learning_rate": 7.306763285024155e-08,
"logits/chosen": -0.1668163537979126,
"logits/rejected": 0.2065034657716751,
"logps/chosen": -394.82257080078125,
"logps/rejected": -394.52142333984375,
"loss": 0.3909,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.4898454248905182,
"rewards/margins": 1.7736278772354126,
"rewards/rejected": -2.2634730339050293,
"step": 800
},
{
"epoch": 0.87,
"eval_logits/chosen": -0.773809552192688,
"eval_logits/rejected": -0.84580397605896,
"eval_logps/chosen": -396.8601989746094,
"eval_logps/rejected": -409.49676513671875,
"eval_loss": 0.41672879457473755,
"eval_rewards/accuracies": 0.8134920597076416,
"eval_rewards/chosen": -0.22731684148311615,
"eval_rewards/margins": 2.0872485637664795,
"eval_rewards/rejected": -2.3145651817321777,
"eval_runtime": 391.0713,
"eval_samples_per_second": 5.114,
"eval_steps_per_second": 0.161,
"step": 800
},
{
"epoch": 0.88,
"learning_rate": 6.702898550724638e-08,
"logits/chosen": -0.8055841326713562,
"logits/rejected": -0.8581466674804688,
"logps/chosen": -434.89300537109375,
"logps/rejected": -407.73486328125,
"loss": 0.4191,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -0.5859915018081665,
"rewards/margins": 1.8876674175262451,
"rewards/rejected": -2.473658800125122,
"step": 810
},
{
"epoch": 0.89,
"learning_rate": 6.09903381642512e-08,
"logits/chosen": -0.6149953603744507,
"logits/rejected": -0.6610177755355835,
"logps/chosen": -403.4602355957031,
"logps/rejected": -412.859619140625,
"loss": 0.3632,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": 0.0069200992584228516,
"rewards/margins": 2.4245333671569824,
"rewards/rejected": -2.4176132678985596,
"step": 820
},
{
"epoch": 0.9,
"learning_rate": 5.4951690821256036e-08,
"logits/chosen": -0.7099069356918335,
"logits/rejected": -0.5322675704956055,
"logps/chosen": -477.89080810546875,
"logps/rejected": -509.3751525878906,
"loss": 0.4242,
"rewards/accuracies": 0.862500011920929,
"rewards/chosen": 0.04087377339601517,
"rewards/margins": 2.5105957984924316,
"rewards/rejected": -2.469722270965576,
"step": 830
},
{
"epoch": 0.91,
"learning_rate": 4.891304347826087e-08,
"logits/chosen": -0.8323551416397095,
"logits/rejected": -0.5835217833518982,
"logps/chosen": -344.49261474609375,
"logps/rejected": -416.62738037109375,
"loss": 0.4114,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.43376049399375916,
"rewards/margins": 1.836294412612915,
"rewards/rejected": -2.270055055618286,
"step": 840
},
{
"epoch": 0.92,
"learning_rate": 4.28743961352657e-08,
"logits/chosen": -0.8896607160568237,
"logits/rejected": -0.9991308450698853,
"logps/chosen": -429.5093688964844,
"logps/rejected": -462.163330078125,
"loss": 0.4213,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 0.14300577342510223,
"rewards/margins": 2.3328614234924316,
"rewards/rejected": -2.1898555755615234,
"step": 850
},
{
"epoch": 0.93,
"learning_rate": 3.6835748792270526e-08,
"logits/chosen": -1.5217692852020264,
"logits/rejected": -1.4032458066940308,
"logps/chosen": -445.5403747558594,
"logps/rejected": -484.6028747558594,
"loss": 0.4084,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.3494121730327606,
"rewards/margins": 1.832707166671753,
"rewards/rejected": -2.182119369506836,
"step": 860
},
{
"epoch": 0.94,
"learning_rate": 3.0797101449275364e-08,
"logits/chosen": -1.3731000423431396,
"logits/rejected": -1.262548804283142,
"logps/chosen": -421.68524169921875,
"logps/rejected": -415.681884765625,
"loss": 0.4108,
"rewards/accuracies": 0.8125,
"rewards/chosen": 0.045568324625492096,
"rewards/margins": 2.3908283710479736,
"rewards/rejected": -2.34525990486145,
"step": 870
},
{
"epoch": 0.96,
"learning_rate": 2.475845410628019e-08,
"logits/chosen": -1.434045433998108,
"logits/rejected": -1.2689034938812256,
"logps/chosen": -386.4149169921875,
"logps/rejected": -396.9994812011719,
"loss": 0.4121,
"rewards/accuracies": 0.824999988079071,
"rewards/chosen": -0.10745694488286972,
"rewards/margins": 2.171525478363037,
"rewards/rejected": -2.278982400894165,
"step": 880
},
{
"epoch": 0.97,
"learning_rate": 1.8719806763285022e-08,
"logits/chosen": -1.6423695087432861,
"logits/rejected": -1.5149376392364502,
"logps/chosen": -413.06158447265625,
"logps/rejected": -437.2244567871094,
"loss": 0.3935,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.15360334515571594,
"rewards/margins": 2.0728042125701904,
"rewards/rejected": -2.226407527923584,
"step": 890
},
{
"epoch": 0.98,
"learning_rate": 1.2681159420289856e-08,
"logits/chosen": -0.9922491312026978,
"logits/rejected": -1.0585081577301025,
"logps/chosen": -401.11627197265625,
"logps/rejected": -402.2158508300781,
"loss": 0.3954,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.1775648146867752,
"rewards/margins": 2.2243430614471436,
"rewards/rejected": -2.4019076824188232,
"step": 900
},
{
"epoch": 0.99,
"learning_rate": 6.642512077294686e-09,
"logits/chosen": 0.17671330273151398,
"logits/rejected": 0.08089754730463028,
"logps/chosen": -331.13409423828125,
"logps/rejected": -373.9659729003906,
"loss": 0.3714,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.24092864990234375,
"rewards/margins": 1.8302351236343384,
"rewards/rejected": -2.0711638927459717,
"step": 910
},
{
"epoch": 1.0,
"learning_rate": 6.038647342995168e-10,
"logits/chosen": -0.6910379528999329,
"logits/rejected": -0.7387306094169617,
"logps/chosen": -440.9806213378906,
"logps/rejected": -403.29083251953125,
"loss": 0.3947,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -0.311662882566452,
"rewards/margins": 2.3290438652038574,
"rewards/rejected": -2.640706777572632,
"step": 920
},
{
"epoch": 1.0,
"step": 921,
"total_flos": 0.0,
"train_loss": 0.4461688995361328,
"train_runtime": 44067.2139,
"train_samples_per_second": 1.337,
"train_steps_per_second": 0.021
}
],
"logging_steps": 10,
"max_steps": 921,
"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
}