phi-2-kto-0-20-i1 / trainer_state.json
BraylonDash's picture
Model save
39c3e54 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 625,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 7.936507936507937e-08,
"logits/chosen": 0.3974232077598572,
"logits/rejected": 0.3553540110588074,
"logps/chosen": -777.8718872070312,
"logps/rejected": -1263.3857421875,
"loss": 0.5,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.02,
"learning_rate": 7.936507936507937e-07,
"logits/chosen": 0.2189813256263733,
"logits/rejected": 0.11798671633005142,
"logps/chosen": -1287.5775146484375,
"logps/rejected": -2137.94189453125,
"loss": 0.4998,
"rewards/accuracies": 0.4236111044883728,
"rewards/chosen": 0.001081566559150815,
"rewards/margins": 0.002832952421158552,
"rewards/rejected": -0.0017513858620077372,
"step": 10
},
{
"epoch": 0.03,
"learning_rate": 1.5873015873015873e-06,
"logits/chosen": 0.254694402217865,
"logits/rejected": 0.17226830124855042,
"logps/chosen": -1060.751708984375,
"logps/rejected": -1972.5423583984375,
"loss": 0.4987,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.01862182281911373,
"rewards/margins": 0.011560038663446903,
"rewards/rejected": -0.03018186055123806,
"step": 20
},
{
"epoch": 0.05,
"learning_rate": 2.380952380952381e-06,
"logits/chosen": 0.24693536758422852,
"logits/rejected": 0.14691275358200073,
"logps/chosen": -1164.8314208984375,
"logps/rejected": -2265.688232421875,
"loss": 0.4942,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": -0.06745004653930664,
"rewards/margins": 0.06839548051357269,
"rewards/rejected": -0.13584552705287933,
"step": 30
},
{
"epoch": 0.06,
"learning_rate": 3.1746031746031746e-06,
"logits/chosen": 0.2351571023464203,
"logits/rejected": 0.08826713263988495,
"logps/chosen": -1165.0933837890625,
"logps/rejected": -2538.54443359375,
"loss": 0.4882,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": -0.18789814412593842,
"rewards/margins": 0.15809166431427002,
"rewards/rejected": -0.34598982334136963,
"step": 40
},
{
"epoch": 0.08,
"learning_rate": 3.968253968253968e-06,
"logits/chosen": 0.21212966740131378,
"logits/rejected": 0.055394046008586884,
"logps/chosen": -1518.728271484375,
"logps/rejected": -2687.27783203125,
"loss": 0.4803,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.5128198862075806,
"rewards/margins": 0.28584352135658264,
"rewards/rejected": -0.7986633777618408,
"step": 50
},
{
"epoch": 0.1,
"learning_rate": 4.761904761904762e-06,
"logits/chosen": 0.14459244906902313,
"logits/rejected": 0.03840586543083191,
"logps/chosen": -2544.2294921875,
"logps/rejected": -3382.06640625,
"loss": 0.4881,
"rewards/accuracies": 0.5,
"rewards/chosen": -1.1230318546295166,
"rewards/margins": 0.08744711428880692,
"rewards/rejected": -1.2104789018630981,
"step": 60
},
{
"epoch": 0.11,
"learning_rate": 4.998086282661188e-06,
"logits/chosen": 0.03217538818717003,
"logits/rejected": -0.029030317440629005,
"logps/chosen": -2578.112060546875,
"logps/rejected": -3333.111328125,
"loss": 0.4889,
"rewards/accuracies": 0.48124998807907104,
"rewards/chosen": -1.2647404670715332,
"rewards/margins": 0.06161295250058174,
"rewards/rejected": -1.3263534307479858,
"step": 70
},
{
"epoch": 0.13,
"learning_rate": 4.988720025682995e-06,
"logits/chosen": 0.1540413349866867,
"logits/rejected": 0.00599607964977622,
"logps/chosen": -2171.90966796875,
"logps/rejected": -3310.91259765625,
"loss": 0.4818,
"rewards/accuracies": 0.53125,
"rewards/chosen": -1.0253573656082153,
"rewards/margins": 0.32645487785339355,
"rewards/rejected": -1.3518123626708984,
"step": 80
},
{
"epoch": 0.14,
"learning_rate": 4.9715789537359126e-06,
"logits/chosen": 0.18023057281970978,
"logits/rejected": -0.0267815999686718,
"logps/chosen": -2072.573974609375,
"logps/rejected": -3439.62744140625,
"loss": 0.477,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.79749995470047,
"rewards/margins": 0.3549983501434326,
"rewards/rejected": -1.152498483657837,
"step": 90
},
{
"epoch": 0.16,
"learning_rate": 4.946716615897932e-06,
"logits/chosen": 0.27958863973617554,
"logits/rejected": 0.08261282742023468,
"logps/chosen": -1780.5172119140625,
"logps/rejected": -2961.440185546875,
"loss": 0.4749,
"rewards/accuracies": 0.518750011920929,
"rewards/chosen": -0.6027330160140991,
"rewards/margins": 0.38158130645751953,
"rewards/rejected": -0.9843141436576843,
"step": 100
},
{
"epoch": 0.18,
"learning_rate": 4.9142106826480114e-06,
"logits/chosen": 0.139088436961174,
"logits/rejected": 0.03682307153940201,
"logps/chosen": -2555.55810546875,
"logps/rejected": -3190.896240234375,
"loss": 0.4867,
"rewards/accuracies": 0.44999998807907104,
"rewards/chosen": -0.9442774653434753,
"rewards/margins": 0.06666886806488037,
"rewards/rejected": -1.0109463930130005,
"step": 110
},
{
"epoch": 0.19,
"learning_rate": 4.874162703221823e-06,
"logits/chosen": 0.1611190289258957,
"logits/rejected": -0.012361553497612476,
"logps/chosen": -2050.61865234375,
"logps/rejected": -3838.018798828125,
"loss": 0.4732,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.9395005106925964,
"rewards/margins": 0.6280331015586853,
"rewards/rejected": -1.5675336122512817,
"step": 120
},
{
"epoch": 0.21,
"learning_rate": 4.826697788369752e-06,
"logits/chosen": 0.09378266334533691,
"logits/rejected": -0.05030001327395439,
"logps/chosen": -2598.26953125,
"logps/rejected": -3382.542236328125,
"loss": 0.479,
"rewards/accuracies": 0.512499988079071,
"rewards/chosen": -1.1976029872894287,
"rewards/margins": 0.14105060696601868,
"rewards/rejected": -1.338653564453125,
"step": 130
},
{
"epoch": 0.22,
"learning_rate": 4.7719642195082224e-06,
"logits/chosen": 0.14611801505088806,
"logits/rejected": 0.03608284890651703,
"logps/chosen": -1731.415771484375,
"logps/rejected": -2871.3349609375,
"loss": 0.4783,
"rewards/accuracies": 0.5062500238418579,
"rewards/chosen": -0.7977763414382935,
"rewards/margins": 0.4263841211795807,
"rewards/rejected": -1.2241604328155518,
"step": 140
},
{
"epoch": 0.24,
"learning_rate": 4.710132985485355e-06,
"logits/chosen": 0.1052960604429245,
"logits/rejected": -0.002506089163944125,
"logps/chosen": -1893.218505859375,
"logps/rejected": -2854.30859375,
"loss": 0.4777,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.6661534905433655,
"rewards/margins": 0.35293903946876526,
"rewards/rejected": -1.0190925598144531,
"step": 150
},
{
"epoch": 0.26,
"learning_rate": 4.641397248408122e-06,
"logits/chosen": 0.017198827117681503,
"logits/rejected": -0.11535916477441788,
"logps/chosen": -2269.91650390625,
"logps/rejected": -3919.4375,
"loss": 0.4746,
"rewards/accuracies": 0.5062500238418579,
"rewards/chosen": -0.9236265420913696,
"rewards/margins": 0.3831055164337158,
"rewards/rejected": -1.306731939315796,
"step": 160
},
{
"epoch": 0.27,
"learning_rate": 4.5659717401997655e-06,
"logits/chosen": 0.01966998353600502,
"logits/rejected": -0.08529923856258392,
"logps/chosen": -2126.437255859375,
"logps/rejected": -3403.6328125,
"loss": 0.4795,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -1.0373507738113403,
"rewards/margins": 0.41469430923461914,
"rewards/rejected": -1.452045202255249,
"step": 170
},
{
"epoch": 0.29,
"learning_rate": 4.4840920917726425e-06,
"logits/chosen": 0.05112285539507866,
"logits/rejected": -0.052227288484573364,
"logps/chosen": -2470.00537109375,
"logps/rejected": -3359.12109375,
"loss": 0.4825,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -1.1501070261001587,
"rewards/margins": 0.24429550766944885,
"rewards/rejected": -1.3944026231765747,
"step": 180
},
{
"epoch": 0.3,
"learning_rate": 4.396014096912182e-06,
"logits/chosen": 0.04223916679620743,
"logits/rejected": -0.13196751475334167,
"logps/chosen": -2119.87841796875,
"logps/rejected": -3257.087158203125,
"loss": 0.4842,
"rewards/accuracies": 0.4937500059604645,
"rewards/chosen": -1.0418171882629395,
"rewards/margins": 0.2414085417985916,
"rewards/rejected": -1.2832257747650146,
"step": 190
},
{
"epoch": 0.32,
"learning_rate": 4.302012913171584e-06,
"logits/chosen": 0.01791330613195896,
"logits/rejected": -0.14596834778785706,
"logps/chosen": -2362.22314453125,
"logps/rejected": -3574.25927734375,
"loss": 0.4675,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -1.061951994895935,
"rewards/margins": 0.25661829113960266,
"rewards/rejected": -1.3185702562332153,
"step": 200
},
{
"epoch": 0.34,
"learning_rate": 4.202382202273702e-06,
"logits/chosen": 0.012257062830030918,
"logits/rejected": -0.15737880766391754,
"logps/chosen": -2264.546630859375,
"logps/rejected": -4172.1708984375,
"loss": 0.4655,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -1.0106382369995117,
"rewards/margins": 0.853549599647522,
"rewards/rejected": -1.8641879558563232,
"step": 210
},
{
"epoch": 0.35,
"learning_rate": 4.097433212705492e-06,
"logits/chosen": -0.0625365823507309,
"logits/rejected": -0.27418404817581177,
"logps/chosen": -2211.255615234375,
"logps/rejected": -4339.79638671875,
"loss": 0.4675,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -1.0741349458694458,
"rewards/margins": 0.7618507742881775,
"rewards/rejected": -1.835985779762268,
"step": 220
},
{
"epoch": 0.37,
"learning_rate": 3.987493807371033e-06,
"logits/chosen": 0.07172416150569916,
"logits/rejected": 0.0005642950418405235,
"logps/chosen": -2197.752197265625,
"logps/rejected": -3114.567626953125,
"loss": 0.471,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.9756608009338379,
"rewards/margins": 0.24378642439842224,
"rewards/rejected": -1.2194470167160034,
"step": 230
},
{
"epoch": 0.38,
"learning_rate": 3.872907439340758e-06,
"logits/chosen": 0.01632564514875412,
"logits/rejected": -0.18748557567596436,
"logps/chosen": -2377.099853515625,
"logps/rejected": -4215.26953125,
"loss": 0.4747,
"rewards/accuracies": 0.5625,
"rewards/chosen": -1.054962396621704,
"rewards/margins": 0.6837267279624939,
"rewards/rejected": -1.7386891841888428,
"step": 240
},
{
"epoch": 0.4,
"learning_rate": 3.75403207889666e-06,
"logits/chosen": -0.012413917109370232,
"logits/rejected": -0.13336405158042908,
"logps/chosen": -2371.0888671875,
"logps/rejected": -3082.55322265625,
"loss": 0.4731,
"rewards/accuracies": 0.518750011920929,
"rewards/chosen": -0.9990957379341125,
"rewards/margins": 0.2219332903623581,
"rewards/rejected": -1.2210289239883423,
"step": 250
},
{
"epoch": 0.42,
"learning_rate": 3.631239095225417e-06,
"logits/chosen": 0.007460703607648611,
"logits/rejected": -0.1302846223115921,
"logps/chosen": -2123.05810546875,
"logps/rejected": -3470.561279296875,
"loss": 0.4727,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -1.0149245262145996,
"rewards/margins": 0.44816678762435913,
"rewards/rejected": -1.463091254234314,
"step": 260
},
{
"epoch": 0.43,
"learning_rate": 3.5049120962530608e-06,
"logits/chosen": -0.0549989752471447,
"logits/rejected": -0.2491796910762787,
"logps/chosen": -2456.557861328125,
"logps/rejected": -4269.2412109375,
"loss": 0.4724,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -1.0748237371444702,
"rewards/margins": 0.7018817067146301,
"rewards/rejected": -1.7767053842544556,
"step": 270
},
{
"epoch": 0.45,
"learning_rate": 3.3754457302455464e-06,
"logits/chosen": 0.028259318321943283,
"logits/rejected": -0.15549519658088684,
"logps/chosen": -2343.92724609375,
"logps/rejected": -3815.20751953125,
"loss": 0.4775,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.9266031384468079,
"rewards/margins": 0.4884551167488098,
"rewards/rejected": -1.4150583744049072,
"step": 280
},
{
"epoch": 0.46,
"learning_rate": 3.2432444529190714e-06,
"logits/chosen": 0.09060511738061905,
"logits/rejected": -0.13999487459659576,
"logps/chosen": -1729.5335693359375,
"logps/rejected": -3556.65771484375,
"loss": 0.4629,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.5673225522041321,
"rewards/margins": 0.5148328542709351,
"rewards/rejected": -1.082155466079712,
"step": 290
},
{
"epoch": 0.48,
"learning_rate": 3.1087212639117057e-06,
"logits/chosen": 0.10731784254312515,
"logits/rejected": -0.12288031727075577,
"logps/chosen": -1606.456787109375,
"logps/rejected": -3169.525146484375,
"loss": 0.4802,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.47551971673965454,
"rewards/margins": 0.5530696511268616,
"rewards/rejected": -1.0285893678665161,
"step": 300
},
{
"epoch": 0.5,
"learning_rate": 2.9722964165636263e-06,
"logits/chosen": -0.040035147219896317,
"logits/rejected": -0.13161209225654602,
"logps/chosen": -2169.19677734375,
"logps/rejected": -3303.66064453125,
"loss": 0.4767,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.8213248252868652,
"rewards/margins": 0.4187691807746887,
"rewards/rejected": -1.2400939464569092,
"step": 310
},
{
"epoch": 0.51,
"learning_rate": 2.8343961050366275e-06,
"logits/chosen": -0.0054204524494707584,
"logits/rejected": -0.2861900329589844,
"logps/chosen": -1592.34375,
"logps/rejected": -3716.65869140625,
"loss": 0.4721,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.7492900490760803,
"rewards/margins": 0.8597043752670288,
"rewards/rejected": -1.608994483947754,
"step": 320
},
{
"epoch": 0.53,
"learning_rate": 2.695451132874385e-06,
"logits/chosen": -0.1356322020292282,
"logits/rejected": -0.292384535074234,
"logps/chosen": -1807.0335693359375,
"logps/rejected": -3166.426513671875,
"loss": 0.48,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.7727370858192444,
"rewards/margins": 0.4709666669368744,
"rewards/rejected": -1.2437037229537964,
"step": 330
},
{
"epoch": 0.54,
"learning_rate": 2.5558955671628964e-06,
"logits/chosen": -0.07628178596496582,
"logits/rejected": -0.20791587233543396,
"logps/chosen": -1737.909423828125,
"logps/rejected": -3203.9970703125,
"loss": 0.4668,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": -0.7647561430931091,
"rewards/margins": 0.49256449937820435,
"rewards/rejected": -1.2573206424713135,
"step": 340
},
{
"epoch": 0.56,
"learning_rate": 2.4161653824955654e-06,
"logits/chosen": -0.05267338082194328,
"logits/rejected": -0.25987547636032104,
"logps/chosen": -2673.15380859375,
"logps/rejected": -3792.9609375,
"loss": 0.4653,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": -1.1372772455215454,
"rewards/margins": 0.37971851229667664,
"rewards/rejected": -1.5169956684112549,
"step": 350
},
{
"epoch": 0.58,
"learning_rate": 2.2766970989791697e-06,
"logits/chosen": -0.15440881252288818,
"logits/rejected": -0.3189676105976105,
"logps/chosen": -2074.7294921875,
"logps/rejected": -3670.177001953125,
"loss": 0.47,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.8276017904281616,
"rewards/margins": 0.6911835670471191,
"rewards/rejected": -1.5187852382659912,
"step": 360
},
{
"epoch": 0.59,
"learning_rate": 2.1379264185356545e-06,
"logits/chosen": -0.03989617899060249,
"logits/rejected": -0.29078492522239685,
"logps/chosen": -2240.99951171875,
"logps/rejected": -3693.21875,
"loss": 0.4675,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.982672393321991,
"rewards/margins": 0.5129915475845337,
"rewards/rejected": -1.4956640005111694,
"step": 370
},
{
"epoch": 0.61,
"learning_rate": 2.000286863759934e-06,
"logits/chosen": -0.18141961097717285,
"logits/rejected": -0.34517520666122437,
"logps/chosen": -2447.95458984375,
"logps/rejected": -4179.51220703125,
"loss": 0.4721,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -1.249446153640747,
"rewards/margins": 0.7962532639503479,
"rewards/rejected": -2.04569935798645,
"step": 380
},
{
"epoch": 0.62,
"learning_rate": 1.8642084235859764e-06,
"logits/chosen": -0.10085698217153549,
"logits/rejected": -0.4116978645324707,
"logps/chosen": -2039.9761962890625,
"logps/rejected": -4429.87451171875,
"loss": 0.4625,
"rewards/accuracies": 0.625,
"rewards/chosen": -1.0368822813034058,
"rewards/margins": 1.2251580953598022,
"rewards/rejected": -2.262040615081787,
"step": 390
},
{
"epoch": 0.64,
"learning_rate": 1.7301162099921013e-06,
"logits/chosen": -0.07682979851961136,
"logits/rejected": -0.35371267795562744,
"logps/chosen": -2304.14306640625,
"logps/rejected": -3896.870361328125,
"loss": 0.4642,
"rewards/accuracies": 0.5625,
"rewards/chosen": -1.131152868270874,
"rewards/margins": 0.7709552645683289,
"rewards/rejected": -1.9021081924438477,
"step": 400
},
{
"epoch": 0.66,
"learning_rate": 1.5984291299420117e-06,
"logits/chosen": -0.07511943578720093,
"logits/rejected": -0.28092044591903687,
"logps/chosen": -2233.59814453125,
"logps/rejected": -3915.47900390625,
"loss": 0.4691,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -1.0048208236694336,
"rewards/margins": 0.6850495934486389,
"rewards/rejected": -1.6898702383041382,
"step": 410
},
{
"epoch": 0.67,
"learning_rate": 1.4695585767104092e-06,
"logits/chosen": -0.005914182402193546,
"logits/rejected": -0.24251346290111542,
"logps/chosen": -1424.0343017578125,
"logps/rejected": -3496.74755859375,
"loss": 0.4581,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.6118324398994446,
"rewards/margins": 0.6595968008041382,
"rewards/rejected": -1.2714293003082275,
"step": 420
},
{
"epoch": 0.69,
"learning_rate": 1.3439071446815452e-06,
"logits/chosen": -0.05694418027997017,
"logits/rejected": -0.27310264110565186,
"logps/chosen": -1811.8056640625,
"logps/rejected": -3876.968017578125,
"loss": 0.4613,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.7894518971443176,
"rewards/margins": 0.9580751657485962,
"rewards/rejected": -1.7475271224975586,
"step": 430
},
{
"epoch": 0.7,
"learning_rate": 1.2218673716356919e-06,
"logits/chosen": -0.04861157387495041,
"logits/rejected": -0.22785380482673645,
"logps/chosen": -2162.256103515625,
"logps/rejected": -3135.13330078125,
"loss": 0.4752,
"rewards/accuracies": 0.4749999940395355,
"rewards/chosen": -0.9659271240234375,
"rewards/margins": 0.3405894339084625,
"rewards/rejected": -1.3065165281295776,
"step": 440
},
{
"epoch": 0.72,
"learning_rate": 1.103820512452661e-06,
"logits/chosen": -0.0886421948671341,
"logits/rejected": -0.24363021552562714,
"logps/chosen": -2278.705078125,
"logps/rejected": -3871.57861328125,
"loss": 0.4697,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -1.0452690124511719,
"rewards/margins": 0.5815836191177368,
"rewards/rejected": -1.6268523931503296,
"step": 450
},
{
"epoch": 0.74,
"learning_rate": 9.901353480633468e-07,
"logits/chosen": -0.01507838536053896,
"logits/rejected": -0.27143269777297974,
"logps/chosen": -1932.046875,
"logps/rejected": -3638.012939453125,
"loss": 0.4705,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.8395353555679321,
"rewards/margins": 0.6771036982536316,
"rewards/rejected": -1.5166391134262085,
"step": 460
},
{
"epoch": 0.75,
"learning_rate": 8.811670333701544e-07,
"logits/chosen": -0.06362856924533844,
"logits/rejected": -0.2936163544654846,
"logps/chosen": -1988.303955078125,
"logps/rejected": -4044.038330078125,
"loss": 0.4641,
"rewards/accuracies": 0.5062500238418579,
"rewards/chosen": -0.8915748596191406,
"rewards/margins": 0.8164836168289185,
"rewards/rejected": -1.7080585956573486,
"step": 470
},
{
"epoch": 0.77,
"learning_rate": 7.772559877354341e-07,
"logits/chosen": -0.030277037993073463,
"logits/rejected": -0.21407613158226013,
"logps/chosen": -2318.84912109375,
"logps/rejected": -4174.81298828125,
"loss": 0.4745,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.9398199319839478,
"rewards/margins": 0.7878143787384033,
"rewards/rejected": -1.7276341915130615,
"step": 480
},
{
"epoch": 0.78,
"learning_rate": 6.787268315040604e-07,
"logits/chosen": -0.1001216396689415,
"logits/rejected": -0.24970397353172302,
"logps/chosen": -2327.34375,
"logps/rejected": -3589.403564453125,
"loss": 0.4775,
"rewards/accuracies": 0.5062500238418579,
"rewards/chosen": -0.9565193057060242,
"rewards/margins": 0.5348533987998962,
"rewards/rejected": -1.4913727045059204,
"step": 490
},
{
"epoch": 0.8,
"learning_rate": 5.858873718824829e-07,
"logits/chosen": -0.08827606588602066,
"logits/rejected": -0.2271912395954132,
"logps/chosen": -2329.033935546875,
"logps/rejected": -3762.891357421875,
"loss": 0.4701,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.9019953608512878,
"rewards/margins": 0.6341419219970703,
"rewards/rejected": -1.536137342453003,
"step": 500
},
{
"epoch": 0.82,
"learning_rate": 4.990276413423817e-07,
"logits/chosen": 0.006130737718194723,
"logits/rejected": -0.2663702070713043,
"logps/chosen": -1954.1243896484375,
"logps/rejected": -4118.31689453125,
"loss": 0.4608,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.7334184050559998,
"rewards/margins": 0.8962429761886597,
"rewards/rejected": -1.6296613216400146,
"step": 510
},
{
"epoch": 0.83,
"learning_rate": 4.184189915529796e-07,
"logits/chosen": -0.022313248366117477,
"logits/rejected": -0.2796049416065216,
"logps/chosen": -2069.07861328125,
"logps/rejected": -4346.296875,
"loss": 0.4631,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.8130847215652466,
"rewards/margins": 0.9093513488769531,
"rewards/rejected": -1.7224359512329102,
"step": 520
},
{
"epoch": 0.85,
"learning_rate": 3.4431324567258176e-07,
"logits/chosen": -0.006228646729141474,
"logits/rejected": -0.18093259632587433,
"logps/chosen": -1853.189453125,
"logps/rejected": -3369.416748046875,
"loss": 0.4685,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.7921710014343262,
"rewards/margins": 0.5718700885772705,
"rewards/rejected": -1.3640410900115967,
"step": 530
},
{
"epoch": 0.86,
"learning_rate": 2.769419116476052e-07,
"logits/chosen": -0.0418720506131649,
"logits/rejected": -0.30269795656204224,
"logps/chosen": -2341.14501953125,
"logps/rejected": -4083.86083984375,
"loss": 0.4729,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.9471173286437988,
"rewards/margins": 0.6616966128349304,
"rewards/rejected": -1.6088138818740845,
"step": 540
},
{
"epoch": 0.88,
"learning_rate": 2.1651545897676512e-07,
"logits/chosen": -0.08359251916408539,
"logits/rejected": -0.31541210412979126,
"logps/chosen": -1795.650390625,
"logps/rejected": -3922.41259765625,
"loss": 0.4713,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.6783340573310852,
"rewards/margins": 0.7878134846687317,
"rewards/rejected": -1.4661474227905273,
"step": 550
},
{
"epoch": 0.9,
"learning_rate": 1.6322266119983222e-07,
"logits/chosen": -0.08543523401021957,
"logits/rejected": -0.22058483958244324,
"logps/chosen": -2356.09619140625,
"logps/rejected": -2975.605712890625,
"loss": 0.4693,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.9343336224555969,
"rewards/margins": 0.29671743512153625,
"rewards/rejected": -1.231050968170166,
"step": 560
},
{
"epoch": 0.91,
"learning_rate": 1.1723000616502167e-07,
"logits/chosen": -0.07000622898340225,
"logits/rejected": -0.2409258633852005,
"logps/chosen": -2480.648681640625,
"logps/rejected": -3360.589111328125,
"loss": 0.4772,
"rewards/accuracies": 0.5062500238418579,
"rewards/chosen": -1.0692129135131836,
"rewards/margins": 0.2927553355693817,
"rewards/rejected": -1.3619682788848877,
"step": 570
},
{
"epoch": 0.93,
"learning_rate": 7.868117591737585e-08,
"logits/chosen": -0.02028440684080124,
"logits/rejected": -0.17536571621894836,
"logps/chosen": -2015.087646484375,
"logps/rejected": -3354.944580078125,
"loss": 0.4748,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.7683865427970886,
"rewards/margins": 0.602486252784729,
"rewards/rejected": -1.3708727359771729,
"step": 580
},
{
"epoch": 0.94,
"learning_rate": 4.769659783295383e-08,
"logits/chosen": -0.12467147409915924,
"logits/rejected": -0.2483750879764557,
"logps/chosen": -2129.997314453125,
"logps/rejected": -3120.9951171875,
"loss": 0.4782,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.9002014398574829,
"rewards/margins": 0.3127484619617462,
"rewards/rejected": -1.2129498720169067,
"step": 590
},
{
"epoch": 0.96,
"learning_rate": 2.4373068401120358e-08,
"logits/chosen": -0.02018025331199169,
"logits/rejected": -0.2830565869808197,
"logps/chosen": -1880.4671630859375,
"logps/rejected": -3650.389404296875,
"loss": 0.4589,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.7423778772354126,
"rewards/margins": 0.7750624418258667,
"rewards/rejected": -1.5174401998519897,
"step": 600
},
{
"epoch": 0.98,
"learning_rate": 8.78345083022425e-09,
"logits/chosen": -0.017909971997141838,
"logits/rejected": -0.2379368245601654,
"logps/chosen": -2240.98193359375,
"logps/rejected": -3461.29345703125,
"loss": 0.467,
"rewards/accuracies": 0.518750011920929,
"rewards/chosen": -1.001007080078125,
"rewards/margins": 0.4783032536506653,
"rewards/rejected": -1.479310154914856,
"step": 610
},
{
"epoch": 0.99,
"learning_rate": 9.764474213677654e-10,
"logits/chosen": -0.011101929470896721,
"logits/rejected": -0.18822148442268372,
"logps/chosen": -1725.9417724609375,
"logps/rejected": -3190.5283203125,
"loss": 0.4695,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.661989688873291,
"rewards/margins": 0.6126888394355774,
"rewards/rejected": -1.2746784687042236,
"step": 620
},
{
"epoch": 1.0,
"step": 625,
"total_flos": 0.0,
"train_loss": 0.4739649757385254,
"train_runtime": 8016.6369,
"train_samples_per_second": 2.495,
"train_steps_per_second": 0.078
}
],
"logging_steps": 10,
"max_steps": 625,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 20,
"total_flos": 0.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}