HBART / trainer_state.json
Amin
First Model version
698009f
raw
history blame
No virus
49.5 kB
{
"best_metric": 2.0502805709838867,
"best_model_checkpoint": "results/models/bart-large-cnn-NewsRoomSmall/checkpoint-48701",
"epoch": 4.0,
"global_step": 194804,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 1.9949077021005728e-05,
"loss": 2.0989,
"step": 500
},
{
"epoch": 0.02,
"learning_rate": 1.9897743372826023e-05,
"loss": 2.0245,
"step": 1000
},
{
"epoch": 0.03,
"learning_rate": 1.9846409724646314e-05,
"loss": 1.9898,
"step": 1500
},
{
"epoch": 0.04,
"learning_rate": 1.9795076076466605e-05,
"loss": 1.9769,
"step": 2000
},
{
"epoch": 0.05,
"learning_rate": 1.9743742428286896e-05,
"loss": 1.9832,
"step": 2500
},
{
"epoch": 0.06,
"learning_rate": 1.9692408780107187e-05,
"loss": 1.9371,
"step": 3000
},
{
"epoch": 0.07,
"learning_rate": 1.9641177799223837e-05,
"loss": 1.9403,
"step": 3500
},
{
"epoch": 0.08,
"learning_rate": 1.958984415104413e-05,
"loss": 1.9412,
"step": 4000
},
{
"epoch": 0.09,
"learning_rate": 1.953851050286442e-05,
"loss": 1.9081,
"step": 4500
},
{
"epoch": 0.1,
"learning_rate": 1.948717685468471e-05,
"loss": 1.904,
"step": 5000
},
{
"epoch": 0.11,
"learning_rate": 1.9435843206505e-05,
"loss": 1.8917,
"step": 5500
},
{
"epoch": 0.12,
"learning_rate": 1.9384612225621652e-05,
"loss": 1.8901,
"step": 6000
},
{
"epoch": 0.13,
"learning_rate": 1.9333278577441943e-05,
"loss": 1.9037,
"step": 6500
},
{
"epoch": 0.14,
"learning_rate": 1.9281944929262234e-05,
"loss": 1.9027,
"step": 7000
},
{
"epoch": 0.15,
"learning_rate": 1.9230611281082525e-05,
"loss": 1.8839,
"step": 7500
},
{
"epoch": 0.16,
"learning_rate": 1.9179277632902817e-05,
"loss": 1.8959,
"step": 8000
},
{
"epoch": 0.17,
"learning_rate": 1.9128046652019467e-05,
"loss": 1.9039,
"step": 8500
},
{
"epoch": 0.18,
"learning_rate": 1.9076713003839758e-05,
"loss": 1.874,
"step": 9000
},
{
"epoch": 0.2,
"learning_rate": 1.902537935566005e-05,
"loss": 1.8835,
"step": 9500
},
{
"epoch": 0.21,
"learning_rate": 1.897404570748034e-05,
"loss": 1.8617,
"step": 10000
},
{
"epoch": 0.22,
"learning_rate": 1.892271205930063e-05,
"loss": 1.8594,
"step": 10500
},
{
"epoch": 0.23,
"learning_rate": 1.8871378411120926e-05,
"loss": 1.8483,
"step": 11000
},
{
"epoch": 0.24,
"learning_rate": 1.8820044762941213e-05,
"loss": 1.873,
"step": 11500
},
{
"epoch": 0.25,
"learning_rate": 1.8768813782057864e-05,
"loss": 1.8425,
"step": 12000
},
{
"epoch": 0.26,
"learning_rate": 1.8717480133878155e-05,
"loss": 1.834,
"step": 12500
},
{
"epoch": 0.27,
"learning_rate": 1.8666146485698446e-05,
"loss": 1.8399,
"step": 13000
},
{
"epoch": 0.28,
"learning_rate": 1.8614812837518737e-05,
"loss": 1.8352,
"step": 13500
},
{
"epoch": 0.29,
"learning_rate": 1.856347918933903e-05,
"loss": 1.8401,
"step": 14000
},
{
"epoch": 0.3,
"learning_rate": 1.851214554115932e-05,
"loss": 1.8416,
"step": 14500
},
{
"epoch": 0.31,
"learning_rate": 1.846081189297961e-05,
"loss": 1.8486,
"step": 15000
},
{
"epoch": 0.32,
"learning_rate": 1.8409478244799905e-05,
"loss": 1.8383,
"step": 15500
},
{
"epoch": 0.33,
"learning_rate": 1.8358247263916552e-05,
"loss": 1.8253,
"step": 16000
},
{
"epoch": 0.34,
"learning_rate": 1.8306913615736847e-05,
"loss": 1.8373,
"step": 16500
},
{
"epoch": 0.35,
"learning_rate": 1.8255579967557134e-05,
"loss": 1.8161,
"step": 17000
},
{
"epoch": 0.36,
"learning_rate": 1.820424631937743e-05,
"loss": 1.8242,
"step": 17500
},
{
"epoch": 0.37,
"learning_rate": 1.8152912671197716e-05,
"loss": 1.8319,
"step": 18000
},
{
"epoch": 0.38,
"learning_rate": 1.810157902301801e-05,
"loss": 1.8144,
"step": 18500
},
{
"epoch": 0.39,
"learning_rate": 1.80502453748383e-05,
"loss": 1.8316,
"step": 19000
},
{
"epoch": 0.4,
"learning_rate": 1.7998911726658593e-05,
"loss": 1.8223,
"step": 19500
},
{
"epoch": 0.41,
"learning_rate": 1.7947680745775244e-05,
"loss": 1.8062,
"step": 20000
},
{
"epoch": 0.42,
"learning_rate": 1.789644976489189e-05,
"loss": 1.8184,
"step": 20500
},
{
"epoch": 0.43,
"learning_rate": 1.7845116116712185e-05,
"loss": 1.8149,
"step": 21000
},
{
"epoch": 0.44,
"learning_rate": 1.7793782468532473e-05,
"loss": 1.8169,
"step": 21500
},
{
"epoch": 0.45,
"learning_rate": 1.7742448820352767e-05,
"loss": 1.8133,
"step": 22000
},
{
"epoch": 0.46,
"learning_rate": 1.7691115172173055e-05,
"loss": 1.818,
"step": 22500
},
{
"epoch": 0.47,
"learning_rate": 1.763988419128971e-05,
"loss": 1.8213,
"step": 23000
},
{
"epoch": 0.48,
"learning_rate": 1.758855054311e-05,
"loss": 1.8048,
"step": 23500
},
{
"epoch": 0.49,
"learning_rate": 1.753721689493029e-05,
"loss": 1.7975,
"step": 24000
},
{
"epoch": 0.5,
"learning_rate": 1.7485883246750582e-05,
"loss": 1.8023,
"step": 24500
},
{
"epoch": 0.51,
"learning_rate": 1.743465226586723e-05,
"loss": 1.8034,
"step": 25000
},
{
"epoch": 0.52,
"learning_rate": 1.7383318617687524e-05,
"loss": 1.8102,
"step": 25500
},
{
"epoch": 0.53,
"learning_rate": 1.7331984969507815e-05,
"loss": 1.7886,
"step": 26000
},
{
"epoch": 0.54,
"learning_rate": 1.7280651321328106e-05,
"loss": 1.798,
"step": 26500
},
{
"epoch": 0.55,
"learning_rate": 1.7229420340444757e-05,
"loss": 1.8022,
"step": 27000
},
{
"epoch": 0.56,
"learning_rate": 1.7178086692265048e-05,
"loss": 1.7992,
"step": 27500
},
{
"epoch": 0.57,
"learning_rate": 1.712675304408534e-05,
"loss": 1.7915,
"step": 28000
},
{
"epoch": 0.59,
"learning_rate": 1.707541939590563e-05,
"loss": 1.7804,
"step": 28500
},
{
"epoch": 0.6,
"learning_rate": 1.702408574772592e-05,
"loss": 1.7699,
"step": 29000
},
{
"epoch": 0.61,
"learning_rate": 1.6972854766842572e-05,
"loss": 1.7826,
"step": 29500
},
{
"epoch": 0.62,
"learning_rate": 1.6921623785959223e-05,
"loss": 1.7691,
"step": 30000
},
{
"epoch": 0.63,
"learning_rate": 1.6870290137779514e-05,
"loss": 1.7945,
"step": 30500
},
{
"epoch": 0.64,
"learning_rate": 1.6818956489599805e-05,
"loss": 1.7861,
"step": 31000
},
{
"epoch": 0.65,
"learning_rate": 1.6767622841420096e-05,
"loss": 1.7607,
"step": 31500
},
{
"epoch": 0.66,
"learning_rate": 1.6716289193240387e-05,
"loss": 1.7674,
"step": 32000
},
{
"epoch": 0.67,
"learning_rate": 1.6664955545060678e-05,
"loss": 1.7817,
"step": 32500
},
{
"epoch": 0.68,
"learning_rate": 1.661362189688097e-05,
"loss": 1.7751,
"step": 33000
},
{
"epoch": 0.69,
"learning_rate": 1.656228824870126e-05,
"loss": 1.7648,
"step": 33500
},
{
"epoch": 0.7,
"learning_rate": 1.651095460052155e-05,
"loss": 1.7611,
"step": 34000
},
{
"epoch": 0.71,
"learning_rate": 1.6459620952341842e-05,
"loss": 1.7567,
"step": 34500
},
{
"epoch": 0.72,
"learning_rate": 1.6408389971458493e-05,
"loss": 1.7598,
"step": 35000
},
{
"epoch": 0.73,
"learning_rate": 1.6357056323278784e-05,
"loss": 1.7591,
"step": 35500
},
{
"epoch": 0.74,
"learning_rate": 1.6305722675099075e-05,
"loss": 1.7789,
"step": 36000
},
{
"epoch": 0.75,
"learning_rate": 1.6254491694215725e-05,
"loss": 1.7779,
"step": 36500
},
{
"epoch": 0.76,
"learning_rate": 1.6203158046036016e-05,
"loss": 1.7388,
"step": 37000
},
{
"epoch": 0.77,
"learning_rate": 1.6151824397856308e-05,
"loss": 1.7636,
"step": 37500
},
{
"epoch": 0.78,
"learning_rate": 1.61004907496766e-05,
"loss": 1.7692,
"step": 38000
},
{
"epoch": 0.79,
"learning_rate": 1.604915710149689e-05,
"loss": 1.7595,
"step": 38500
},
{
"epoch": 0.8,
"learning_rate": 1.599782345331718e-05,
"loss": 1.7705,
"step": 39000
},
{
"epoch": 0.81,
"learning_rate": 1.594648980513747e-05,
"loss": 1.7357,
"step": 39500
},
{
"epoch": 0.82,
"learning_rate": 1.5895156156957766e-05,
"loss": 1.7714,
"step": 40000
},
{
"epoch": 0.83,
"learning_rate": 1.5843822508778054e-05,
"loss": 1.7471,
"step": 40500
},
{
"epoch": 0.84,
"learning_rate": 1.5792591527894704e-05,
"loss": 1.7513,
"step": 41000
},
{
"epoch": 0.85,
"learning_rate": 1.5741257879715e-05,
"loss": 1.7651,
"step": 41500
},
{
"epoch": 0.86,
"learning_rate": 1.5689924231535287e-05,
"loss": 1.7782,
"step": 42000
},
{
"epoch": 0.87,
"learning_rate": 1.563859058335558e-05,
"loss": 1.7573,
"step": 42500
},
{
"epoch": 0.88,
"learning_rate": 1.558725693517587e-05,
"loss": 1.7511,
"step": 43000
},
{
"epoch": 0.89,
"learning_rate": 1.5535923286996163e-05,
"loss": 1.7535,
"step": 43500
},
{
"epoch": 0.9,
"learning_rate": 1.548458963881645e-05,
"loss": 1.7525,
"step": 44000
},
{
"epoch": 0.91,
"learning_rate": 1.5433255990636745e-05,
"loss": 1.7629,
"step": 44500
},
{
"epoch": 0.92,
"learning_rate": 1.5382025009753392e-05,
"loss": 1.7534,
"step": 45000
},
{
"epoch": 0.93,
"learning_rate": 1.5330691361573687e-05,
"loss": 1.7444,
"step": 45500
},
{
"epoch": 0.94,
"learning_rate": 1.5279357713393978e-05,
"loss": 1.7479,
"step": 46000
},
{
"epoch": 0.95,
"learning_rate": 1.5228024065214269e-05,
"loss": 1.7384,
"step": 46500
},
{
"epoch": 0.97,
"learning_rate": 1.5176690417034558e-05,
"loss": 1.7526,
"step": 47000
},
{
"epoch": 0.98,
"learning_rate": 1.5125459436151209e-05,
"loss": 1.7422,
"step": 47500
},
{
"epoch": 0.99,
"learning_rate": 1.50741257879715e-05,
"loss": 1.7431,
"step": 48000
},
{
"epoch": 1.0,
"learning_rate": 1.5022792139791791e-05,
"loss": 1.7497,
"step": 48500
},
{
"epoch": 1.0,
"eval_gen_len": 58.64,
"eval_loss": 2.0502805709838867,
"eval_rouge1": 22.0185,
"eval_rouge2": 9.9661,
"eval_rougeL": 19.2037,
"eval_rougeLsum": 19.7843,
"eval_runtime": 17.9636,
"eval_samples_per_second": 5.567,
"eval_steps_per_second": 0.39,
"step": 48701
},
{
"epoch": 1.01,
"learning_rate": 1.4971458491612084e-05,
"loss": 1.6028,
"step": 49000
},
{
"epoch": 1.02,
"learning_rate": 1.4920124843432375e-05,
"loss": 1.4874,
"step": 49500
},
{
"epoch": 1.03,
"learning_rate": 1.4868791195252666e-05,
"loss": 1.5065,
"step": 50000
},
{
"epoch": 1.04,
"learning_rate": 1.4817560214369315e-05,
"loss": 1.4904,
"step": 50500
},
{
"epoch": 1.05,
"learning_rate": 1.4766226566189608e-05,
"loss": 1.505,
"step": 51000
},
{
"epoch": 1.06,
"learning_rate": 1.4714892918009897e-05,
"loss": 1.5026,
"step": 51500
},
{
"epoch": 1.07,
"learning_rate": 1.466355926983019e-05,
"loss": 1.5026,
"step": 52000
},
{
"epoch": 1.08,
"learning_rate": 1.4612225621650479e-05,
"loss": 1.5111,
"step": 52500
},
{
"epoch": 1.09,
"learning_rate": 1.4560891973470772e-05,
"loss": 1.5014,
"step": 53000
},
{
"epoch": 1.1,
"learning_rate": 1.4509558325291063e-05,
"loss": 1.502,
"step": 53500
},
{
"epoch": 1.11,
"learning_rate": 1.4458224677111354e-05,
"loss": 1.4973,
"step": 54000
},
{
"epoch": 1.12,
"learning_rate": 1.4406993696228005e-05,
"loss": 1.5042,
"step": 54500
},
{
"epoch": 1.13,
"learning_rate": 1.4355660048048297e-05,
"loss": 1.4914,
"step": 55000
},
{
"epoch": 1.14,
"learning_rate": 1.4304326399868587e-05,
"loss": 1.5026,
"step": 55500
},
{
"epoch": 1.15,
"learning_rate": 1.425299275168888e-05,
"loss": 1.4999,
"step": 56000
},
{
"epoch": 1.16,
"learning_rate": 1.4201761770805528e-05,
"loss": 1.5077,
"step": 56500
},
{
"epoch": 1.17,
"learning_rate": 1.415042812262582e-05,
"loss": 1.4962,
"step": 57000
},
{
"epoch": 1.18,
"learning_rate": 1.409909447444611e-05,
"loss": 1.5074,
"step": 57500
},
{
"epoch": 1.19,
"learning_rate": 1.4047760826266402e-05,
"loss": 1.5008,
"step": 58000
},
{
"epoch": 1.2,
"learning_rate": 1.3996529845383054e-05,
"loss": 1.5076,
"step": 58500
},
{
"epoch": 1.21,
"learning_rate": 1.3945196197203343e-05,
"loss": 1.5088,
"step": 59000
},
{
"epoch": 1.22,
"learning_rate": 1.3893862549023636e-05,
"loss": 1.4962,
"step": 59500
},
{
"epoch": 1.23,
"learning_rate": 1.3842528900843925e-05,
"loss": 1.4917,
"step": 60000
},
{
"epoch": 1.24,
"learning_rate": 1.3791195252664218e-05,
"loss": 1.4992,
"step": 60500
},
{
"epoch": 1.25,
"learning_rate": 1.3739964271780867e-05,
"loss": 1.4963,
"step": 61000
},
{
"epoch": 1.26,
"learning_rate": 1.368863062360116e-05,
"loss": 1.5008,
"step": 61500
},
{
"epoch": 1.27,
"learning_rate": 1.3637296975421451e-05,
"loss": 1.5009,
"step": 62000
},
{
"epoch": 1.28,
"learning_rate": 1.3585963327241742e-05,
"loss": 1.5127,
"step": 62500
},
{
"epoch": 1.29,
"learning_rate": 1.3534629679062033e-05,
"loss": 1.5035,
"step": 63000
},
{
"epoch": 1.3,
"learning_rate": 1.3483398698178682e-05,
"loss": 1.5129,
"step": 63500
},
{
"epoch": 1.31,
"learning_rate": 1.3432065049998975e-05,
"loss": 1.5134,
"step": 64000
},
{
"epoch": 1.32,
"learning_rate": 1.3380731401819264e-05,
"loss": 1.5131,
"step": 64500
},
{
"epoch": 1.33,
"learning_rate": 1.3329397753639557e-05,
"loss": 1.5008,
"step": 65000
},
{
"epoch": 1.34,
"learning_rate": 1.3278064105459846e-05,
"loss": 1.5022,
"step": 65500
},
{
"epoch": 1.36,
"learning_rate": 1.3226935791872857e-05,
"loss": 1.5003,
"step": 66000
},
{
"epoch": 1.37,
"learning_rate": 1.317560214369315e-05,
"loss": 1.519,
"step": 66500
},
{
"epoch": 1.38,
"learning_rate": 1.3124268495513439e-05,
"loss": 1.5058,
"step": 67000
},
{
"epoch": 1.39,
"learning_rate": 1.3072934847333732e-05,
"loss": 1.5252,
"step": 67500
},
{
"epoch": 1.4,
"learning_rate": 1.3021601199154024e-05,
"loss": 1.5102,
"step": 68000
},
{
"epoch": 1.41,
"learning_rate": 1.2970267550974314e-05,
"loss": 1.5059,
"step": 68500
},
{
"epoch": 1.42,
"learning_rate": 1.2918933902794606e-05,
"loss": 1.508,
"step": 69000
},
{
"epoch": 1.43,
"learning_rate": 1.2867600254614896e-05,
"loss": 1.4942,
"step": 69500
},
{
"epoch": 1.44,
"learning_rate": 1.2816266606435188e-05,
"loss": 1.5167,
"step": 70000
},
{
"epoch": 1.45,
"learning_rate": 1.2765035625551837e-05,
"loss": 1.5181,
"step": 70500
},
{
"epoch": 1.46,
"learning_rate": 1.2713701977372128e-05,
"loss": 1.508,
"step": 71000
},
{
"epoch": 1.47,
"learning_rate": 1.2662368329192421e-05,
"loss": 1.5116,
"step": 71500
},
{
"epoch": 1.48,
"learning_rate": 1.261103468101271e-05,
"loss": 1.5212,
"step": 72000
},
{
"epoch": 1.49,
"learning_rate": 1.2559701032833003e-05,
"loss": 1.4879,
"step": 72500
},
{
"epoch": 1.5,
"learning_rate": 1.2508470051949652e-05,
"loss": 1.5121,
"step": 73000
},
{
"epoch": 1.51,
"learning_rate": 1.2457136403769945e-05,
"loss": 1.5094,
"step": 73500
},
{
"epoch": 1.52,
"learning_rate": 1.2405802755590234e-05,
"loss": 1.511,
"step": 74000
},
{
"epoch": 1.53,
"learning_rate": 1.2354469107410527e-05,
"loss": 1.512,
"step": 74500
},
{
"epoch": 1.54,
"learning_rate": 1.2303135459230816e-05,
"loss": 1.4996,
"step": 75000
},
{
"epoch": 1.55,
"learning_rate": 1.2251904478347467e-05,
"loss": 1.5122,
"step": 75500
},
{
"epoch": 1.56,
"learning_rate": 1.220057083016776e-05,
"loss": 1.5067,
"step": 76000
},
{
"epoch": 1.57,
"learning_rate": 1.2149339849284409e-05,
"loss": 1.5081,
"step": 76500
},
{
"epoch": 1.58,
"learning_rate": 1.2098006201104702e-05,
"loss": 1.4931,
"step": 77000
},
{
"epoch": 1.59,
"learning_rate": 1.2046672552924991e-05,
"loss": 1.5017,
"step": 77500
},
{
"epoch": 1.6,
"learning_rate": 1.1995338904745284e-05,
"loss": 1.4951,
"step": 78000
},
{
"epoch": 1.61,
"learning_rate": 1.1944005256565573e-05,
"loss": 1.5041,
"step": 78500
},
{
"epoch": 1.62,
"learning_rate": 1.1892671608385866e-05,
"loss": 1.4977,
"step": 79000
},
{
"epoch": 1.63,
"learning_rate": 1.1841337960206157e-05,
"loss": 1.517,
"step": 79500
},
{
"epoch": 1.64,
"learning_rate": 1.1790004312026448e-05,
"loss": 1.5056,
"step": 80000
},
{
"epoch": 1.65,
"learning_rate": 1.1738670663846739e-05,
"loss": 1.4893,
"step": 80500
},
{
"epoch": 1.66,
"learning_rate": 1.168733701566703e-05,
"loss": 1.5125,
"step": 81000
},
{
"epoch": 1.67,
"learning_rate": 1.163610603478368e-05,
"loss": 1.5074,
"step": 81500
},
{
"epoch": 1.68,
"learning_rate": 1.1584772386603973e-05,
"loss": 1.5206,
"step": 82000
},
{
"epoch": 1.69,
"learning_rate": 1.1533438738424263e-05,
"loss": 1.4934,
"step": 82500
},
{
"epoch": 1.7,
"learning_rate": 1.1482105090244556e-05,
"loss": 1.5052,
"step": 83000
},
{
"epoch": 1.71,
"learning_rate": 1.1430771442064845e-05,
"loss": 1.5036,
"step": 83500
},
{
"epoch": 1.72,
"learning_rate": 1.1379437793885138e-05,
"loss": 1.516,
"step": 84000
},
{
"epoch": 1.74,
"learning_rate": 1.1328206813001787e-05,
"loss": 1.4903,
"step": 84500
},
{
"epoch": 1.75,
"learning_rate": 1.1276873164822078e-05,
"loss": 1.4999,
"step": 85000
},
{
"epoch": 1.76,
"learning_rate": 1.122553951664237e-05,
"loss": 1.4822,
"step": 85500
},
{
"epoch": 1.77,
"learning_rate": 1.117420586846266e-05,
"loss": 1.5073,
"step": 86000
},
{
"epoch": 1.78,
"learning_rate": 1.1122974887579312e-05,
"loss": 1.5158,
"step": 86500
},
{
"epoch": 1.79,
"learning_rate": 1.1071641239399602e-05,
"loss": 1.4907,
"step": 87000
},
{
"epoch": 1.8,
"learning_rate": 1.1020410258516254e-05,
"loss": 1.5028,
"step": 87500
},
{
"epoch": 1.81,
"learning_rate": 1.0969076610336545e-05,
"loss": 1.4878,
"step": 88000
},
{
"epoch": 1.82,
"learning_rate": 1.0917742962156836e-05,
"loss": 1.5032,
"step": 88500
},
{
"epoch": 1.83,
"learning_rate": 1.0866409313977127e-05,
"loss": 1.5034,
"step": 89000
},
{
"epoch": 1.84,
"learning_rate": 1.0815075665797418e-05,
"loss": 1.5007,
"step": 89500
},
{
"epoch": 1.85,
"learning_rate": 1.076374201761771e-05,
"loss": 1.5083,
"step": 90000
},
{
"epoch": 1.86,
"learning_rate": 1.0712408369438e-05,
"loss": 1.5119,
"step": 90500
},
{
"epoch": 1.87,
"learning_rate": 1.0661074721258291e-05,
"loss": 1.5026,
"step": 91000
},
{
"epoch": 1.88,
"learning_rate": 1.060984374037494e-05,
"loss": 1.4866,
"step": 91500
},
{
"epoch": 1.89,
"learning_rate": 1.0558612759491593e-05,
"loss": 1.5006,
"step": 92000
},
{
"epoch": 1.9,
"learning_rate": 1.0507279111311884e-05,
"loss": 1.5035,
"step": 92500
},
{
"epoch": 1.91,
"learning_rate": 1.0455945463132175e-05,
"loss": 1.4945,
"step": 93000
},
{
"epoch": 1.92,
"learning_rate": 1.0404611814952466e-05,
"loss": 1.5115,
"step": 93500
},
{
"epoch": 1.93,
"learning_rate": 1.0353278166772759e-05,
"loss": 1.4946,
"step": 94000
},
{
"epoch": 1.94,
"learning_rate": 1.0301944518593048e-05,
"loss": 1.4847,
"step": 94500
},
{
"epoch": 1.95,
"learning_rate": 1.025061087041334e-05,
"loss": 1.4912,
"step": 95000
},
{
"epoch": 1.96,
"learning_rate": 1.019927722223363e-05,
"loss": 1.4912,
"step": 95500
},
{
"epoch": 1.97,
"learning_rate": 1.0147943574053923e-05,
"loss": 1.4949,
"step": 96000
},
{
"epoch": 1.98,
"learning_rate": 1.0096712593170572e-05,
"loss": 1.5093,
"step": 96500
},
{
"epoch": 1.99,
"learning_rate": 1.0045378944990864e-05,
"loss": 1.505,
"step": 97000
},
{
"epoch": 2.0,
"eval_gen_len": 58.69,
"eval_loss": 2.0827815532684326,
"eval_rouge1": 23.2685,
"eval_rouge2": 10.8781,
"eval_rougeL": 20.0935,
"eval_rougeLsum": 20.5489,
"eval_runtime": 18.0831,
"eval_samples_per_second": 5.53,
"eval_steps_per_second": 0.387,
"step": 97402
},
{
"epoch": 2.0,
"learning_rate": 9.994045296811154e-06,
"loss": 1.4514,
"step": 97500
},
{
"epoch": 2.01,
"learning_rate": 9.942711648631445e-06,
"loss": 1.2686,
"step": 98000
},
{
"epoch": 2.02,
"learning_rate": 9.891480667748097e-06,
"loss": 1.269,
"step": 98500
},
{
"epoch": 2.03,
"learning_rate": 9.840147019568388e-06,
"loss": 1.2746,
"step": 99000
},
{
"epoch": 2.04,
"learning_rate": 9.78881337138868e-06,
"loss": 1.2654,
"step": 99500
},
{
"epoch": 2.05,
"learning_rate": 9.73747972320897e-06,
"loss": 1.2771,
"step": 100000
},
{
"epoch": 2.06,
"learning_rate": 9.68624874232562e-06,
"loss": 1.2854,
"step": 100500
},
{
"epoch": 2.07,
"learning_rate": 9.63491509414591e-06,
"loss": 1.269,
"step": 101000
},
{
"epoch": 2.08,
"learning_rate": 9.583581445966203e-06,
"loss": 1.2924,
"step": 101500
},
{
"epoch": 2.09,
"learning_rate": 9.532350465082854e-06,
"loss": 1.2741,
"step": 102000
},
{
"epoch": 2.1,
"learning_rate": 9.481016816903145e-06,
"loss": 1.2818,
"step": 102500
},
{
"epoch": 2.11,
"learning_rate": 9.429683168723436e-06,
"loss": 1.2764,
"step": 103000
},
{
"epoch": 2.13,
"learning_rate": 9.378349520543727e-06,
"loss": 1.2852,
"step": 103500
},
{
"epoch": 2.14,
"learning_rate": 9.327015872364018e-06,
"loss": 1.282,
"step": 104000
},
{
"epoch": 2.15,
"learning_rate": 9.27568222418431e-06,
"loss": 1.2787,
"step": 104500
},
{
"epoch": 2.16,
"learning_rate": 9.22445124330096e-06,
"loss": 1.2763,
"step": 105000
},
{
"epoch": 2.17,
"learning_rate": 9.173117595121251e-06,
"loss": 1.2758,
"step": 105500
},
{
"epoch": 2.18,
"learning_rate": 9.121783946941542e-06,
"loss": 1.2788,
"step": 106000
},
{
"epoch": 2.19,
"learning_rate": 9.070450298761833e-06,
"loss": 1.2916,
"step": 106500
},
{
"epoch": 2.2,
"learning_rate": 9.019116650582124e-06,
"loss": 1.285,
"step": 107000
},
{
"epoch": 2.21,
"learning_rate": 8.967783002402417e-06,
"loss": 1.2996,
"step": 107500
},
{
"epoch": 2.22,
"learning_rate": 8.916552021519066e-06,
"loss": 1.2857,
"step": 108000
},
{
"epoch": 2.23,
"learning_rate": 8.865218373339357e-06,
"loss": 1.2822,
"step": 108500
},
{
"epoch": 2.24,
"learning_rate": 8.813884725159648e-06,
"loss": 1.2912,
"step": 109000
},
{
"epoch": 2.25,
"learning_rate": 8.762551076979939e-06,
"loss": 1.2792,
"step": 109500
},
{
"epoch": 2.26,
"learning_rate": 8.71121742880023e-06,
"loss": 1.2979,
"step": 110000
},
{
"epoch": 2.27,
"learning_rate": 8.659883780620521e-06,
"loss": 1.2776,
"step": 110500
},
{
"epoch": 2.28,
"learning_rate": 8.608550132440812e-06,
"loss": 1.2855,
"step": 111000
},
{
"epoch": 2.29,
"learning_rate": 8.557216484261103e-06,
"loss": 1.2906,
"step": 111500
},
{
"epoch": 2.3,
"learning_rate": 8.505985503377756e-06,
"loss": 1.2918,
"step": 112000
},
{
"epoch": 2.31,
"learning_rate": 8.454651855198047e-06,
"loss": 1.2911,
"step": 112500
},
{
"epoch": 2.32,
"learning_rate": 8.403318207018338e-06,
"loss": 1.2917,
"step": 113000
},
{
"epoch": 2.33,
"learning_rate": 8.351984558838629e-06,
"loss": 1.2884,
"step": 113500
},
{
"epoch": 2.34,
"learning_rate": 8.30065091065892e-06,
"loss": 1.2882,
"step": 114000
},
{
"epoch": 2.35,
"learning_rate": 8.24952259707193e-06,
"loss": 1.282,
"step": 114500
},
{
"epoch": 2.36,
"learning_rate": 8.198188948892221e-06,
"loss": 1.2964,
"step": 115000
},
{
"epoch": 2.37,
"learning_rate": 8.146855300712512e-06,
"loss": 1.2933,
"step": 115500
},
{
"epoch": 2.38,
"learning_rate": 8.095521652532803e-06,
"loss": 1.2897,
"step": 116000
},
{
"epoch": 2.39,
"learning_rate": 8.044290671649454e-06,
"loss": 1.2811,
"step": 116500
},
{
"epoch": 2.4,
"learning_rate": 7.992957023469745e-06,
"loss": 1.281,
"step": 117000
},
{
"epoch": 2.41,
"learning_rate": 7.941623375290036e-06,
"loss": 1.2865,
"step": 117500
},
{
"epoch": 2.42,
"learning_rate": 7.890289727110327e-06,
"loss": 1.2891,
"step": 118000
},
{
"epoch": 2.43,
"learning_rate": 7.838956078930618e-06,
"loss": 1.287,
"step": 118500
},
{
"epoch": 2.44,
"learning_rate": 7.787622430750909e-06,
"loss": 1.283,
"step": 119000
},
{
"epoch": 2.45,
"learning_rate": 7.7362887825712e-06,
"loss": 1.278,
"step": 119500
},
{
"epoch": 2.46,
"learning_rate": 7.684955134391491e-06,
"loss": 1.2964,
"step": 120000
},
{
"epoch": 2.47,
"learning_rate": 7.633621486211782e-06,
"loss": 1.2913,
"step": 120500
},
{
"epoch": 2.48,
"learning_rate": 7.582390505328433e-06,
"loss": 1.2948,
"step": 121000
},
{
"epoch": 2.49,
"learning_rate": 7.531056857148724e-06,
"loss": 1.2951,
"step": 121500
},
{
"epoch": 2.51,
"learning_rate": 7.479723208969015e-06,
"loss": 1.2967,
"step": 122000
},
{
"epoch": 2.52,
"learning_rate": 7.428389560789306e-06,
"loss": 1.2929,
"step": 122500
},
{
"epoch": 2.53,
"learning_rate": 7.377158579905958e-06,
"loss": 1.2803,
"step": 123000
},
{
"epoch": 2.54,
"learning_rate": 7.325824931726249e-06,
"loss": 1.2916,
"step": 123500
},
{
"epoch": 2.55,
"learning_rate": 7.27449128354654e-06,
"loss": 1.294,
"step": 124000
},
{
"epoch": 2.56,
"learning_rate": 7.223157635366831e-06,
"loss": 1.2977,
"step": 124500
},
{
"epoch": 2.57,
"learning_rate": 7.171823987187122e-06,
"loss": 1.2951,
"step": 125000
},
{
"epoch": 2.58,
"learning_rate": 7.120593006303772e-06,
"loss": 1.2824,
"step": 125500
},
{
"epoch": 2.59,
"learning_rate": 7.0692593581240645e-06,
"loss": 1.2776,
"step": 126000
},
{
"epoch": 2.6,
"learning_rate": 7.0179257099443555e-06,
"loss": 1.2846,
"step": 126500
},
{
"epoch": 2.61,
"learning_rate": 6.966694729061005e-06,
"loss": 1.2986,
"step": 127000
},
{
"epoch": 2.62,
"learning_rate": 6.9153610808812964e-06,
"loss": 1.2913,
"step": 127500
},
{
"epoch": 2.63,
"learning_rate": 6.8640274327015875e-06,
"loss": 1.2924,
"step": 128000
},
{
"epoch": 2.64,
"learning_rate": 6.8126937845218785e-06,
"loss": 1.2834,
"step": 128500
},
{
"epoch": 2.65,
"learning_rate": 6.7613601363421695e-06,
"loss": 1.2963,
"step": 129000
},
{
"epoch": 2.66,
"learning_rate": 6.710129155458821e-06,
"loss": 1.2848,
"step": 129500
},
{
"epoch": 2.67,
"learning_rate": 6.658795507279112e-06,
"loss": 1.2774,
"step": 130000
},
{
"epoch": 2.68,
"learning_rate": 6.607461859099403e-06,
"loss": 1.303,
"step": 130500
},
{
"epoch": 2.69,
"learning_rate": 6.556128210919694e-06,
"loss": 1.2995,
"step": 131000
},
{
"epoch": 2.7,
"learning_rate": 6.504794562739985e-06,
"loss": 1.2789,
"step": 131500
},
{
"epoch": 2.71,
"learning_rate": 6.453460914560276e-06,
"loss": 1.2873,
"step": 132000
},
{
"epoch": 2.72,
"learning_rate": 6.402127266380568e-06,
"loss": 1.2966,
"step": 132500
},
{
"epoch": 2.73,
"learning_rate": 6.350793618200859e-06,
"loss": 1.2904,
"step": 133000
},
{
"epoch": 2.74,
"learning_rate": 6.299562637317509e-06,
"loss": 1.2777,
"step": 133500
},
{
"epoch": 2.75,
"learning_rate": 6.2482289891378e-06,
"loss": 1.2784,
"step": 134000
},
{
"epoch": 2.76,
"learning_rate": 6.196895340958091e-06,
"loss": 1.2644,
"step": 134500
},
{
"epoch": 2.77,
"learning_rate": 6.145561692778382e-06,
"loss": 1.2932,
"step": 135000
},
{
"epoch": 2.78,
"learning_rate": 6.094330711895034e-06,
"loss": 1.2863,
"step": 135500
},
{
"epoch": 2.79,
"learning_rate": 6.042997063715325e-06,
"loss": 1.3051,
"step": 136000
},
{
"epoch": 2.8,
"learning_rate": 5.991766082831976e-06,
"loss": 1.2866,
"step": 136500
},
{
"epoch": 2.81,
"learning_rate": 5.940432434652267e-06,
"loss": 1.2989,
"step": 137000
},
{
"epoch": 2.82,
"learning_rate": 5.889098786472558e-06,
"loss": 1.2696,
"step": 137500
},
{
"epoch": 2.83,
"learning_rate": 5.837765138292849e-06,
"loss": 1.3041,
"step": 138000
},
{
"epoch": 2.84,
"learning_rate": 5.786431490113141e-06,
"loss": 1.2941,
"step": 138500
},
{
"epoch": 2.85,
"learning_rate": 5.735097841933432e-06,
"loss": 1.2806,
"step": 139000
},
{
"epoch": 2.86,
"learning_rate": 5.683764193753723e-06,
"loss": 1.298,
"step": 139500
},
{
"epoch": 2.87,
"learning_rate": 5.632430545574014e-06,
"loss": 1.2748,
"step": 140000
},
{
"epoch": 2.88,
"learning_rate": 5.581199564690664e-06,
"loss": 1.2838,
"step": 140500
},
{
"epoch": 2.9,
"learning_rate": 5.529968583807314e-06,
"loss": 1.298,
"step": 141000
},
{
"epoch": 2.91,
"learning_rate": 5.478634935627606e-06,
"loss": 1.2884,
"step": 141500
},
{
"epoch": 2.92,
"learning_rate": 5.427301287447897e-06,
"loss": 1.2849,
"step": 142000
},
{
"epoch": 2.93,
"learning_rate": 5.375967639268188e-06,
"loss": 1.2805,
"step": 142500
},
{
"epoch": 2.94,
"learning_rate": 5.324633991088479e-06,
"loss": 1.2881,
"step": 143000
},
{
"epoch": 2.95,
"learning_rate": 5.27330034290877e-06,
"loss": 1.2981,
"step": 143500
},
{
"epoch": 2.96,
"learning_rate": 5.221966694729061e-06,
"loss": 1.2751,
"step": 144000
},
{
"epoch": 2.97,
"learning_rate": 5.1706330465493524e-06,
"loss": 1.2822,
"step": 144500
},
{
"epoch": 2.98,
"learning_rate": 5.119402065666002e-06,
"loss": 1.3016,
"step": 145000
},
{
"epoch": 2.99,
"learning_rate": 5.068068417486295e-06,
"loss": 1.2856,
"step": 145500
},
{
"epoch": 3.0,
"learning_rate": 5.016734769306586e-06,
"loss": 1.2799,
"step": 146000
},
{
"epoch": 3.0,
"eval_gen_len": 58.57,
"eval_loss": 2.115015983581543,
"eval_rouge1": 23.0677,
"eval_rouge2": 10.5086,
"eval_rougeL": 19.9851,
"eval_rougeLsum": 20.6597,
"eval_runtime": 18.1865,
"eval_samples_per_second": 5.499,
"eval_steps_per_second": 0.385,
"step": 146103
},
{
"epoch": 3.01,
"learning_rate": 4.965401121126876e-06,
"loss": 1.153,
"step": 146500
},
{
"epoch": 3.02,
"learning_rate": 4.914067472947167e-06,
"loss": 1.1045,
"step": 147000
},
{
"epoch": 3.03,
"learning_rate": 4.862836492063818e-06,
"loss": 1.1333,
"step": 147500
},
{
"epoch": 3.04,
"learning_rate": 4.811502843884109e-06,
"loss": 1.1321,
"step": 148000
},
{
"epoch": 3.05,
"learning_rate": 4.7601691957044e-06,
"loss": 1.1175,
"step": 148500
},
{
"epoch": 3.06,
"learning_rate": 4.708835547524692e-06,
"loss": 1.1196,
"step": 149000
},
{
"epoch": 3.07,
"learning_rate": 4.657501899344983e-06,
"loss": 1.1221,
"step": 149500
},
{
"epoch": 3.08,
"learning_rate": 4.606168251165274e-06,
"loss": 1.1262,
"step": 150000
},
{
"epoch": 3.09,
"learning_rate": 4.554834602985566e-06,
"loss": 1.1147,
"step": 150500
},
{
"epoch": 3.1,
"learning_rate": 4.503603622102216e-06,
"loss": 1.1275,
"step": 151000
},
{
"epoch": 3.11,
"learning_rate": 4.452269973922507e-06,
"loss": 1.1155,
"step": 151500
},
{
"epoch": 3.12,
"learning_rate": 4.400936325742799e-06,
"loss": 1.1162,
"step": 152000
},
{
"epoch": 3.13,
"learning_rate": 4.34960267756309e-06,
"loss": 1.1204,
"step": 152500
},
{
"epoch": 3.14,
"learning_rate": 4.298269029383381e-06,
"loss": 1.1135,
"step": 153000
},
{
"epoch": 3.15,
"learning_rate": 4.246935381203672e-06,
"loss": 1.1153,
"step": 153500
},
{
"epoch": 3.16,
"learning_rate": 4.195601733023963e-06,
"loss": 1.1198,
"step": 154000
},
{
"epoch": 3.17,
"learning_rate": 4.144268084844254e-06,
"loss": 1.1327,
"step": 154500
},
{
"epoch": 3.18,
"learning_rate": 4.092934436664545e-06,
"loss": 1.1097,
"step": 155000
},
{
"epoch": 3.19,
"learning_rate": 4.041703455781196e-06,
"loss": 1.1378,
"step": 155500
},
{
"epoch": 3.2,
"learning_rate": 3.990369807601487e-06,
"loss": 1.1282,
"step": 156000
},
{
"epoch": 3.21,
"learning_rate": 3.939036159421778e-06,
"loss": 1.1256,
"step": 156500
},
{
"epoch": 3.22,
"learning_rate": 3.8878051785384286e-06,
"loss": 1.1362,
"step": 157000
},
{
"epoch": 3.23,
"learning_rate": 3.83647153035872e-06,
"loss": 1.1225,
"step": 157500
},
{
"epoch": 3.24,
"learning_rate": 3.7851378821790106e-06,
"loss": 1.1247,
"step": 158000
},
{
"epoch": 3.25,
"learning_rate": 3.733804233999302e-06,
"loss": 1.1327,
"step": 158500
},
{
"epoch": 3.26,
"learning_rate": 3.6824705858195936e-06,
"loss": 1.1277,
"step": 159000
},
{
"epoch": 3.28,
"learning_rate": 3.6311369376398846e-06,
"loss": 1.1271,
"step": 159500
},
{
"epoch": 3.29,
"learning_rate": 3.579803289460176e-06,
"loss": 1.1356,
"step": 160000
},
{
"epoch": 3.3,
"learning_rate": 3.528469641280467e-06,
"loss": 1.1197,
"step": 160500
},
{
"epoch": 3.31,
"learning_rate": 3.477341327693477e-06,
"loss": 1.1391,
"step": 161000
},
{
"epoch": 3.32,
"learning_rate": 3.426007679513768e-06,
"loss": 1.1144,
"step": 161500
},
{
"epoch": 3.33,
"learning_rate": 3.374674031334059e-06,
"loss": 1.1291,
"step": 162000
},
{
"epoch": 3.34,
"learning_rate": 3.3233403831543502e-06,
"loss": 1.1129,
"step": 162500
},
{
"epoch": 3.35,
"learning_rate": 3.2720067349746417e-06,
"loss": 1.1279,
"step": 163000
},
{
"epoch": 3.36,
"learning_rate": 3.2206730867949327e-06,
"loss": 1.1118,
"step": 163500
},
{
"epoch": 3.37,
"learning_rate": 3.169442105911583e-06,
"loss": 1.1295,
"step": 164000
},
{
"epoch": 3.38,
"learning_rate": 3.1181084577318745e-06,
"loss": 1.1182,
"step": 164500
},
{
"epoch": 3.39,
"learning_rate": 3.0667748095521655e-06,
"loss": 1.1283,
"step": 165000
},
{
"epoch": 3.4,
"learning_rate": 3.0154411613724566e-06,
"loss": 1.1112,
"step": 165500
},
{
"epoch": 3.41,
"learning_rate": 2.9641075131927476e-06,
"loss": 1.1313,
"step": 166000
},
{
"epoch": 3.42,
"learning_rate": 2.9127738650130386e-06,
"loss": 1.1318,
"step": 166500
},
{
"epoch": 3.43,
"learning_rate": 2.86144021683333e-06,
"loss": 1.1199,
"step": 167000
},
{
"epoch": 3.44,
"learning_rate": 2.810106568653621e-06,
"loss": 1.1231,
"step": 167500
},
{
"epoch": 3.45,
"learning_rate": 2.758772920473912e-06,
"loss": 1.1332,
"step": 168000
},
{
"epoch": 3.46,
"learning_rate": 2.7076446068869222e-06,
"loss": 1.1247,
"step": 168500
},
{
"epoch": 3.47,
"learning_rate": 2.656310958707214e-06,
"loss": 1.1271,
"step": 169000
},
{
"epoch": 3.48,
"learning_rate": 2.604977310527505e-06,
"loss": 1.1131,
"step": 169500
},
{
"epoch": 3.49,
"learning_rate": 2.553643662347796e-06,
"loss": 1.1231,
"step": 170000
},
{
"epoch": 3.5,
"learning_rate": 2.502412681464447e-06,
"loss": 1.1218,
"step": 170500
},
{
"epoch": 3.51,
"learning_rate": 2.4510790332847375e-06,
"loss": 1.1151,
"step": 171000
},
{
"epoch": 3.52,
"learning_rate": 2.399745385105029e-06,
"loss": 1.1153,
"step": 171500
},
{
"epoch": 3.53,
"learning_rate": 2.34841173692532e-06,
"loss": 1.1159,
"step": 172000
},
{
"epoch": 3.54,
"learning_rate": 2.297078088745611e-06,
"loss": 1.1214,
"step": 172500
},
{
"epoch": 3.55,
"learning_rate": 2.2457444405659025e-06,
"loss": 1.1259,
"step": 173000
},
{
"epoch": 3.56,
"learning_rate": 2.1944107923861935e-06,
"loss": 1.1348,
"step": 173500
},
{
"epoch": 3.57,
"learning_rate": 2.1430771442064846e-06,
"loss": 1.121,
"step": 174000
},
{
"epoch": 3.58,
"learning_rate": 2.0918461633231353e-06,
"loss": 1.1177,
"step": 174500
},
{
"epoch": 3.59,
"learning_rate": 2.0405125151434264e-06,
"loss": 1.1201,
"step": 175000
},
{
"epoch": 3.6,
"learning_rate": 1.9891788669637174e-06,
"loss": 1.1316,
"step": 175500
},
{
"epoch": 3.61,
"learning_rate": 1.937845218784009e-06,
"loss": 1.1277,
"step": 176000
},
{
"epoch": 3.62,
"learning_rate": 1.8865115706042999e-06,
"loss": 1.1211,
"step": 176500
},
{
"epoch": 3.63,
"learning_rate": 1.8351779224245911e-06,
"loss": 1.1206,
"step": 177000
},
{
"epoch": 3.64,
"learning_rate": 1.7839469415412417e-06,
"loss": 1.1177,
"step": 177500
},
{
"epoch": 3.65,
"learning_rate": 1.7326132933615327e-06,
"loss": 1.127,
"step": 178000
},
{
"epoch": 3.67,
"learning_rate": 1.681279645181824e-06,
"loss": 1.1204,
"step": 178500
},
{
"epoch": 3.68,
"learning_rate": 1.629945997002115e-06,
"loss": 1.126,
"step": 179000
},
{
"epoch": 3.69,
"learning_rate": 1.5787150161187655e-06,
"loss": 1.1168,
"step": 179500
},
{
"epoch": 3.7,
"learning_rate": 1.527484035235416e-06,
"loss": 1.1124,
"step": 180000
},
{
"epoch": 3.71,
"learning_rate": 1.4761503870557073e-06,
"loss": 1.1112,
"step": 180500
},
{
"epoch": 3.72,
"learning_rate": 1.4248167388759983e-06,
"loss": 1.1164,
"step": 181000
},
{
"epoch": 3.73,
"learning_rate": 1.3734830906962898e-06,
"loss": 1.1178,
"step": 181500
},
{
"epoch": 3.74,
"learning_rate": 1.322149442516581e-06,
"loss": 1.1106,
"step": 182000
},
{
"epoch": 3.75,
"learning_rate": 1.270815794336872e-06,
"loss": 1.127,
"step": 182500
},
{
"epoch": 3.76,
"learning_rate": 1.2194821461571631e-06,
"loss": 1.1225,
"step": 183000
},
{
"epoch": 3.77,
"learning_rate": 1.1681484979774544e-06,
"loss": 1.1211,
"step": 183500
},
{
"epoch": 3.78,
"learning_rate": 1.116917517094105e-06,
"loss": 1.1109,
"step": 184000
},
{
"epoch": 3.79,
"learning_rate": 1.0655838689143961e-06,
"loss": 1.1099,
"step": 184500
},
{
"epoch": 3.8,
"learning_rate": 1.0142502207346874e-06,
"loss": 1.1094,
"step": 185000
},
{
"epoch": 3.81,
"learning_rate": 9.629165725549784e-07,
"loss": 1.1355,
"step": 185500
},
{
"epoch": 3.82,
"learning_rate": 9.115829243752696e-07,
"loss": 1.112,
"step": 186000
},
{
"epoch": 3.83,
"learning_rate": 8.603519434919201e-07,
"loss": 1.1186,
"step": 186500
},
{
"epoch": 3.84,
"learning_rate": 8.090182953122114e-07,
"loss": 1.1284,
"step": 187000
},
{
"epoch": 3.85,
"learning_rate": 7.576846471325025e-07,
"loss": 1.1177,
"step": 187500
},
{
"epoch": 3.86,
"learning_rate": 7.063509989527936e-07,
"loss": 1.1208,
"step": 188000
},
{
"epoch": 3.87,
"learning_rate": 6.550173507730848e-07,
"loss": 1.1142,
"step": 188500
},
{
"epoch": 3.88,
"learning_rate": 6.037863698897354e-07,
"loss": 1.1091,
"step": 189000
},
{
"epoch": 3.89,
"learning_rate": 5.524527217100266e-07,
"loss": 1.1055,
"step": 189500
},
{
"epoch": 3.9,
"learning_rate": 5.011190735303177e-07,
"loss": 1.1321,
"step": 190000
},
{
"epoch": 3.91,
"learning_rate": 4.4978542535060883e-07,
"loss": 1.1177,
"step": 190500
},
{
"epoch": 3.92,
"learning_rate": 3.9855444446725943e-07,
"loss": 1.1174,
"step": 191000
},
{
"epoch": 3.93,
"learning_rate": 3.4732346358391e-07,
"loss": 1.1177,
"step": 191500
},
{
"epoch": 3.94,
"learning_rate": 2.9598981540420117e-07,
"loss": 1.1043,
"step": 192000
},
{
"epoch": 3.95,
"learning_rate": 2.4465616722449236e-07,
"loss": 1.1063,
"step": 192500
},
{
"epoch": 3.96,
"learning_rate": 1.933225190447835e-07,
"loss": 1.1233,
"step": 193000
},
{
"epoch": 3.97,
"learning_rate": 1.4198887086507466e-07,
"loss": 1.1186,
"step": 193500
},
{
"epoch": 3.98,
"learning_rate": 9.06552226853658e-08,
"loss": 1.1049,
"step": 194000
},
{
"epoch": 3.99,
"learning_rate": 3.9424241802016386e-08,
"loss": 1.1201,
"step": 194500
},
{
"epoch": 4.0,
"eval_gen_len": 58.68,
"eval_loss": 2.1514933109283447,
"eval_rouge1": 22.8624,
"eval_rouge2": 10.7296,
"eval_rougeL": 20.1567,
"eval_rougeLsum": 20.6088,
"eval_runtime": 18.3644,
"eval_samples_per_second": 5.445,
"eval_steps_per_second": 0.381,
"step": 194804
}
],
"max_steps": 194804,
"num_train_epochs": 4,
"total_flos": 6.740081427542286e+18,
"trial_name": null,
"trial_params": null
}