zephyr_0.2_2ep / trainer_state.json
chujiezheng's picture
Upload folder using huggingface_hub
28e3cae verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.9843342036553526,
"eval_steps": 500,
"global_step": 190,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"grad_norm": 6.530893553837012,
"learning_rate": 2.6315789473684208e-08,
"logits/chosen": -2.851747512817383,
"logits/rejected": -2.833996534347534,
"logps/chosen": -165.70089721679688,
"logps/rejected": -198.857666015625,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.05,
"grad_norm": 5.940017766790383,
"learning_rate": 1.3157894736842104e-07,
"logits/chosen": -2.7694077491760254,
"logits/rejected": -2.772179365158081,
"logps/chosen": -171.38229370117188,
"logps/rejected": -172.59909057617188,
"loss": 0.6932,
"rewards/accuracies": 0.3125,
"rewards/chosen": -0.00019297577091492712,
"rewards/margins": -0.0003426831099204719,
"rewards/rejected": 0.00014970726624596864,
"step": 5
},
{
"epoch": 0.1,
"grad_norm": 7.22860361762137,
"learning_rate": 2.631578947368421e-07,
"logits/chosen": -2.784412384033203,
"logits/rejected": -2.7932095527648926,
"logps/chosen": -190.19732666015625,
"logps/rejected": -194.6861572265625,
"loss": 0.6929,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": 0.0017464166739955544,
"rewards/margins": 0.0005834165494889021,
"rewards/rejected": 0.0011630002409219742,
"step": 10
},
{
"epoch": 0.16,
"grad_norm": 7.203100204300102,
"learning_rate": 3.9473684210526315e-07,
"logits/chosen": -2.8401434421539307,
"logits/rejected": -2.8505945205688477,
"logps/chosen": -198.759521484375,
"logps/rejected": -188.010986328125,
"loss": 0.6919,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": 0.012036855332553387,
"rewards/margins": 0.0030940533615648746,
"rewards/rejected": 0.0089428024366498,
"step": 15
},
{
"epoch": 0.21,
"grad_norm": 6.864101673229198,
"learning_rate": 4.999578104083306e-07,
"logits/chosen": -2.8547134399414062,
"logits/rejected": -2.864978075027466,
"logps/chosen": -165.3108673095703,
"logps/rejected": -177.08412170410156,
"loss": 0.6882,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": 0.034319791942834854,
"rewards/margins": 0.008450874127447605,
"rewards/rejected": 0.025868916884064674,
"step": 20
},
{
"epoch": 0.26,
"grad_norm": 6.3583895035816855,
"learning_rate": 4.984826693294873e-07,
"logits/chosen": -2.8253941535949707,
"logits/rejected": -2.8230338096618652,
"logps/chosen": -125.37552642822266,
"logps/rejected": -145.67837524414062,
"loss": 0.6822,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": 0.06096485257148743,
"rewards/margins": 0.02324753999710083,
"rewards/rejected": 0.0377173088490963,
"step": 25
},
{
"epoch": 0.31,
"grad_norm": 6.95430506418983,
"learning_rate": 4.949122667718934e-07,
"logits/chosen": -2.797011613845825,
"logits/rejected": -2.7968087196350098,
"logps/chosen": -156.46078491210938,
"logps/rejected": -166.90145874023438,
"loss": 0.6741,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": 0.030187183991074562,
"rewards/margins": 0.04637282341718674,
"rewards/rejected": -0.016185639426112175,
"step": 30
},
{
"epoch": 0.37,
"grad_norm": 7.0189636592719085,
"learning_rate": 4.892767091689785e-07,
"logits/chosen": -2.849172592163086,
"logits/rejected": -2.8440544605255127,
"logps/chosen": -193.01596069335938,
"logps/rejected": -207.4866180419922,
"loss": 0.664,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.01744670793414116,
"rewards/margins": 0.05499963089823723,
"rewards/rejected": -0.07244633138179779,
"step": 35
},
{
"epoch": 0.42,
"grad_norm": 8.51056311217898,
"learning_rate": 4.816235168037004e-07,
"logits/chosen": -2.7640957832336426,
"logits/rejected": -2.776198387145996,
"logps/chosen": -137.4984130859375,
"logps/rejected": -163.426513671875,
"loss": 0.6517,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.046394020318984985,
"rewards/margins": 0.10987555980682373,
"rewards/rejected": -0.15626958012580872,
"step": 40
},
{
"epoch": 0.47,
"grad_norm": 11.159926429466159,
"learning_rate": 4.720172231068844e-07,
"logits/chosen": -2.723865032196045,
"logits/rejected": -2.7389719486236572,
"logps/chosen": -207.92044067382812,
"logps/rejected": -213.3718719482422,
"loss": 0.6536,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.19272716343402863,
"rewards/margins": 0.12425712496042252,
"rewards/rejected": -0.31698426604270935,
"step": 45
},
{
"epoch": 0.52,
"grad_norm": 12.393423854122126,
"learning_rate": 4.605388304968914e-07,
"logits/chosen": -2.718982696533203,
"logits/rejected": -2.7207980155944824,
"logps/chosen": -178.42510986328125,
"logps/rejected": -211.27377319335938,
"loss": 0.6282,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.040986787527799606,
"rewards/margins": 0.17269372940063477,
"rewards/rejected": -0.21368053555488586,
"step": 50
},
{
"epoch": 0.57,
"grad_norm": 21.049387798378756,
"learning_rate": 4.472851273490984e-07,
"logits/chosen": -2.5929465293884277,
"logits/rejected": -2.6008853912353516,
"logps/chosen": -160.97874450683594,
"logps/rejected": -186.36863708496094,
"loss": 0.6258,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.12865906953811646,
"rewards/margins": 0.20724160969257355,
"rewards/rejected": -0.3359006643295288,
"step": 55
},
{
"epoch": 0.63,
"grad_norm": 12.520960096291708,
"learning_rate": 4.323678718546552e-07,
"logits/chosen": -2.648369312286377,
"logits/rejected": -2.645002603530884,
"logps/chosen": -190.52731323242188,
"logps/rejected": -241.7861328125,
"loss": 0.6096,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.18569248914718628,
"rewards/margins": 0.27379462122917175,
"rewards/rejected": -0.45948711037635803,
"step": 60
},
{
"epoch": 0.68,
"grad_norm": 18.22398171524467,
"learning_rate": 4.159128496504053e-07,
"logits/chosen": -2.635317325592041,
"logits/rejected": -2.6489968299865723,
"logps/chosen": -185.21597290039062,
"logps/rejected": -226.77963256835938,
"loss": 0.5957,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.23138824105262756,
"rewards/margins": 0.33824095129966736,
"rewards/rejected": -0.5696292519569397,
"step": 65
},
{
"epoch": 0.73,
"grad_norm": 12.486094279730978,
"learning_rate": 3.9805881316624503e-07,
"logits/chosen": -2.5691702365875244,
"logits/rejected": -2.5781631469726562,
"logps/chosen": -209.5417938232422,
"logps/rejected": -230.74874877929688,
"loss": 0.597,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.3240502178668976,
"rewards/margins": 0.24576255679130554,
"rewards/rejected": -0.5698127150535583,
"step": 70
},
{
"epoch": 0.78,
"grad_norm": 16.18752353365872,
"learning_rate": 3.78956311633581e-07,
"logits/chosen": -2.5225658416748047,
"logits/rejected": -2.519923448562622,
"logps/chosen": -189.95606994628906,
"logps/rejected": -244.0750732421875,
"loss": 0.583,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.18096503615379333,
"rewards/margins": 0.4236725866794586,
"rewards/rejected": -0.604637622833252,
"step": 75
},
{
"epoch": 0.84,
"grad_norm": 16.73585640066424,
"learning_rate": 3.587664216205183e-07,
"logits/chosen": -2.4873414039611816,
"logits/rejected": -2.488546133041382,
"logps/chosen": -186.70196533203125,
"logps/rejected": -229.10800170898438,
"loss": 0.5896,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.1871272623538971,
"rewards/margins": 0.35064736008644104,
"rewards/rejected": -0.5377745628356934,
"step": 80
},
{
"epoch": 0.89,
"grad_norm": 14.48535258453445,
"learning_rate": 3.376593887981886e-07,
"logits/chosen": -2.5321877002716064,
"logits/rejected": -2.5364866256713867,
"logps/chosen": -201.63038635253906,
"logps/rejected": -250.297607421875,
"loss": 0.5943,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.4144672751426697,
"rewards/margins": 0.3995351493358612,
"rewards/rejected": -0.8140023946762085,
"step": 85
},
{
"epoch": 0.94,
"grad_norm": 24.006493530801364,
"learning_rate": 3.1581319239114976e-07,
"logits/chosen": -2.46376633644104,
"logits/rejected": -2.4762072563171387,
"logps/chosen": -235.89602661132812,
"logps/rejected": -286.06207275390625,
"loss": 0.5905,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.6695653200149536,
"rewards/margins": 0.43523311614990234,
"rewards/rejected": -1.1047985553741455,
"step": 90
},
{
"epoch": 0.99,
"grad_norm": 13.899105116058017,
"learning_rate": 2.934120444167326e-07,
"logits/chosen": -2.5002081394195557,
"logits/rejected": -2.515636920928955,
"logps/chosen": -240.7231903076172,
"logps/rejected": -282.92974853515625,
"loss": 0.579,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.5511455535888672,
"rewards/margins": 0.450370728969574,
"rewards/rejected": -1.001516342163086,
"step": 95
},
{
"epoch": 1.04,
"grad_norm": 12.195720253264156,
"learning_rate": 2.706448363680831e-07,
"logits/chosen": -2.492196559906006,
"logits/rejected": -2.4908487796783447,
"logps/chosen": -217.4730224609375,
"logps/rejected": -285.6746826171875,
"loss": 0.4963,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.3912954330444336,
"rewards/margins": 0.6321467161178589,
"rewards/rejected": -1.023442268371582,
"step": 100
},
{
"epoch": 1.1,
"grad_norm": 13.625308588645998,
"learning_rate": 2.477035464388184e-07,
"logits/chosen": -2.3943684101104736,
"logits/rejected": -2.3905301094055176,
"logps/chosen": -160.0459442138672,
"logps/rejected": -262.6797790527344,
"loss": 0.4563,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": -0.28104764223098755,
"rewards/margins": 0.8251093029975891,
"rewards/rejected": -1.1061569452285767,
"step": 105
},
{
"epoch": 1.15,
"grad_norm": 16.514733311854357,
"learning_rate": 2.2478162071993296e-07,
"logits/chosen": -2.367004156112671,
"logits/rejected": -2.3704023361206055,
"logps/chosen": -212.1750946044922,
"logps/rejected": -292.59295654296875,
"loss": 0.462,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.39816007018089294,
"rewards/margins": 0.7662478089332581,
"rewards/rejected": -1.1644079685211182,
"step": 110
},
{
"epoch": 1.2,
"grad_norm": 15.624289502053674,
"learning_rate": 2.0207234201906545e-07,
"logits/chosen": -2.3321692943573,
"logits/rejected": -2.3237671852111816,
"logps/chosen": -199.48365783691406,
"logps/rejected": -294.98834228515625,
"loss": 0.4647,
"rewards/accuracies": 0.8062499761581421,
"rewards/chosen": -0.5382081866264343,
"rewards/margins": 0.7745588421821594,
"rewards/rejected": -1.3127670288085938,
"step": 115
},
{
"epoch": 1.25,
"grad_norm": 16.693818527228895,
"learning_rate": 1.7976720005660767e-07,
"logits/chosen": -2.3294899463653564,
"logits/rejected": -2.3380885124206543,
"logps/chosen": -236.26138305664062,
"logps/rejected": -317.8197326660156,
"loss": 0.4734,
"rewards/accuracies": 0.793749988079071,
"rewards/chosen": -0.5919082164764404,
"rewards/margins": 0.8398138284683228,
"rewards/rejected": -1.4317219257354736,
"step": 120
},
{
"epoch": 1.31,
"grad_norm": 18.248981197195587,
"learning_rate": 1.5805427678152674e-07,
"logits/chosen": -2.3658409118652344,
"logits/rejected": -2.3801846504211426,
"logps/chosen": -252.34451293945312,
"logps/rejected": -347.273193359375,
"loss": 0.434,
"rewards/accuracies": 0.84375,
"rewards/chosen": -0.6862586736679077,
"rewards/margins": 0.8617037534713745,
"rewards/rejected": -1.5479624271392822,
"step": 125
},
{
"epoch": 1.36,
"grad_norm": 18.599023816787692,
"learning_rate": 1.371166604222777e-07,
"logits/chosen": -2.3468356132507324,
"logits/rejected": -2.3559298515319824,
"logps/chosen": -222.77267456054688,
"logps/rejected": -324.4059143066406,
"loss": 0.4462,
"rewards/accuracies": 0.831250011920929,
"rewards/chosen": -0.609306275844574,
"rewards/margins": 0.9141504168510437,
"rewards/rejected": -1.5234566926956177,
"step": 130
},
{
"epoch": 1.41,
"grad_norm": 18.279705611333885,
"learning_rate": 1.1713090164588606e-07,
"logits/chosen": -2.3474090099334717,
"logits/rejected": -2.3513846397399902,
"logps/chosen": -237.4060516357422,
"logps/rejected": -349.4905700683594,
"loss": 0.4436,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": -0.6407004594802856,
"rewards/margins": 0.92046058177948,
"rewards/rejected": -1.5611611604690552,
"step": 135
},
{
"epoch": 1.46,
"grad_norm": 21.064741442339784,
"learning_rate": 9.826552484321085e-08,
"logits/chosen": -2.2716422080993652,
"logits/rejected": -2.270905017852783,
"logps/chosen": -254.72885131835938,
"logps/rejected": -353.58477783203125,
"loss": 0.4319,
"rewards/accuracies": 0.8500000238418579,
"rewards/chosen": -0.7879461050033569,
"rewards/margins": 1.0126745700836182,
"rewards/rejected": -1.800620675086975,
"step": 140
},
{
"epoch": 1.51,
"grad_norm": 18.966914430346105,
"learning_rate": 8.067960709356478e-08,
"logits/chosen": -2.1971030235290527,
"logits/rejected": -2.2067959308624268,
"logps/chosen": -210.01760864257812,
"logps/rejected": -323.2452697753906,
"loss": 0.422,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.6746965646743774,
"rewards/margins": 1.0784823894500732,
"rewards/rejected": -1.7531789541244507,
"step": 145
},
{
"epoch": 1.57,
"grad_norm": 19.251567468071876,
"learning_rate": 6.452143679117964e-08,
"logits/chosen": -2.269256114959717,
"logits/rejected": -2.2733891010284424,
"logps/chosen": -269.5815124511719,
"logps/rejected": -360.28729248046875,
"loss": 0.4286,
"rewards/accuracies": 0.793749988079071,
"rewards/chosen": -0.8714927434921265,
"rewards/margins": 0.9906436204910278,
"rewards/rejected": -1.8621364831924438,
"step": 150
},
{
"epoch": 1.62,
"grad_norm": 20.881384471810865,
"learning_rate": 4.992726324427901e-08,
"logits/chosen": -2.2466838359832764,
"logits/rejected": -2.2517459392547607,
"logps/chosen": -229.53665161132812,
"logps/rejected": -345.5362243652344,
"loss": 0.4455,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.835593581199646,
"rewards/margins": 0.9400846362113953,
"rewards/rejected": -1.775678277015686,
"step": 155
},
{
"epoch": 1.67,
"grad_norm": 21.71541477038859,
"learning_rate": 3.702014779041826e-08,
"logits/chosen": -2.186835765838623,
"logits/rejected": -2.1860439777374268,
"logps/chosen": -272.96112060546875,
"logps/rejected": -380.77685546875,
"loss": 0.4318,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -1.01149320602417,
"rewards/margins": 1.0301023721694946,
"rewards/rejected": -2.041595697402954,
"step": 160
},
{
"epoch": 1.72,
"grad_norm": 23.87035641942232,
"learning_rate": 2.5908926115744994e-08,
"logits/chosen": -2.3079283237457275,
"logits/rejected": -2.320064067840576,
"logps/chosen": -277.4764099121094,
"logps/rejected": -361.50311279296875,
"loss": 0.4383,
"rewards/accuracies": 0.7562500238418579,
"rewards/chosen": -1.0418604612350464,
"rewards/margins": 0.8757144808769226,
"rewards/rejected": -1.9175748825073242,
"step": 165
},
{
"epoch": 1.78,
"grad_norm": 21.532935828831377,
"learning_rate": 1.6687290528135722e-08,
"logits/chosen": -2.2234318256378174,
"logits/rejected": -2.235042095184326,
"logps/chosen": -243.2294158935547,
"logps/rejected": -327.7035217285156,
"loss": 0.433,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.8599830865859985,
"rewards/margins": 0.809087872505188,
"rewards/rejected": -1.669070839881897,
"step": 170
},
{
"epoch": 1.83,
"grad_norm": 22.322246598167897,
"learning_rate": 9.432999922687396e-09,
"logits/chosen": -2.2345292568206787,
"logits/rejected": -2.2446045875549316,
"logps/chosen": -282.7356872558594,
"logps/rejected": -373.1545104980469,
"loss": 0.4213,
"rewards/accuracies": 0.84375,
"rewards/chosen": -0.9868669509887695,
"rewards/margins": 1.0404006242752075,
"rewards/rejected": -2.0272676944732666,
"step": 175
},
{
"epoch": 1.88,
"grad_norm": 19.226686156526466,
"learning_rate": 4.207224101311246e-09,
"logits/chosen": -2.2762608528137207,
"logits/rejected": -2.290148973464966,
"logps/chosen": -289.68035888671875,
"logps/rejected": -380.23883056640625,
"loss": 0.4364,
"rewards/accuracies": 0.8374999761581421,
"rewards/chosen": -1.0276674032211304,
"rewards/margins": 0.9330458641052246,
"rewards/rejected": -1.9607131481170654,
"step": 180
},
{
"epoch": 1.93,
"grad_norm": 21.532509699275593,
"learning_rate": 1.0540279752731252e-09,
"logits/chosen": -2.232375383377075,
"logits/rejected": -2.242682456970215,
"logps/chosen": -278.67156982421875,
"logps/rejected": -388.5971374511719,
"loss": 0.4303,
"rewards/accuracies": 0.78125,
"rewards/chosen": -1.0453999042510986,
"rewards/margins": 0.9826061129570007,
"rewards/rejected": -2.028006076812744,
"step": 185
},
{
"epoch": 1.98,
"grad_norm": 20.268430701124647,
"learning_rate": 0.0,
"logits/chosen": -2.247197151184082,
"logits/rejected": -2.25414776802063,
"logps/chosen": -272.12359619140625,
"logps/rejected": -399.1447448730469,
"loss": 0.4268,
"rewards/accuracies": 0.8062499761581421,
"rewards/chosen": -1.060980200767517,
"rewards/margins": 1.1344108581542969,
"rewards/rejected": -2.1953911781311035,
"step": 190
},
{
"epoch": 1.98,
"step": 190,
"total_flos": 0.0,
"train_loss": 0.5396593997353002,
"train_runtime": 4845.5764,
"train_samples_per_second": 5.047,
"train_steps_per_second": 0.039
}
],
"logging_steps": 5,
"max_steps": 190,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 500,
"total_flos": 0.0,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}