mbart_1209 / trainer_state.json
eunyounglee's picture
Upload 12 files
5fd36d9
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.9918182928318475,
"global_step": 171500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.985636946193791e-05,
"loss": 4.5112,
"step": 500
},
{
"epoch": 0.02,
"learning_rate": 4.9710994423413524e-05,
"loss": 2.8906,
"step": 1000
},
{
"epoch": 0.03,
"learning_rate": 4.956561938488914e-05,
"loss": 2.5728,
"step": 1500
},
{
"epoch": 0.03,
"learning_rate": 4.9420244346364754e-05,
"loss": 2.3159,
"step": 2000
},
{
"epoch": 0.04,
"learning_rate": 4.927516005791742e-05,
"loss": 2.0901,
"step": 2500
},
{
"epoch": 0.05,
"learning_rate": 4.9129785019393034e-05,
"loss": 1.9299,
"step": 3000
},
{
"epoch": 0.06,
"learning_rate": 4.898440998086865e-05,
"loss": 1.6719,
"step": 3500
},
{
"epoch": 0.07,
"learning_rate": 4.883903494234426e-05,
"loss": 1.5558,
"step": 4000
},
{
"epoch": 0.08,
"learning_rate": 4.869365990381988e-05,
"loss": 1.4674,
"step": 4500
},
{
"epoch": 0.09,
"learning_rate": 4.854828486529549e-05,
"loss": 1.3837,
"step": 5000
},
{
"epoch": 0.1,
"learning_rate": 4.840290982677111e-05,
"loss": 1.2387,
"step": 5500
},
{
"epoch": 0.1,
"learning_rate": 4.825753478824672e-05,
"loss": 1.2115,
"step": 6000
},
{
"epoch": 0.11,
"learning_rate": 4.811215974972234e-05,
"loss": 1.1154,
"step": 6500
},
{
"epoch": 0.12,
"learning_rate": 4.7966784711197945e-05,
"loss": 1.0496,
"step": 7000
},
{
"epoch": 0.13,
"learning_rate": 4.782140967267357e-05,
"loss": 0.9716,
"step": 7500
},
{
"epoch": 0.14,
"learning_rate": 4.767603463414918e-05,
"loss": 0.907,
"step": 8000
},
{
"epoch": 0.15,
"learning_rate": 4.753095034570184e-05,
"loss": 0.9429,
"step": 8500
},
{
"epoch": 0.16,
"learning_rate": 4.738557530717746e-05,
"loss": 0.9419,
"step": 9000
},
{
"epoch": 0.17,
"learning_rate": 4.7240200268653076e-05,
"loss": 0.8054,
"step": 9500
},
{
"epoch": 0.17,
"learning_rate": 4.7094825230128684e-05,
"loss": 0.7875,
"step": 10000
},
{
"epoch": 0.18,
"learning_rate": 4.6949450191604306e-05,
"loss": 0.7689,
"step": 10500
},
{
"epoch": 0.19,
"learning_rate": 4.680407515307992e-05,
"loss": 0.7609,
"step": 11000
},
{
"epoch": 0.2,
"learning_rate": 4.665870011455553e-05,
"loss": 0.7068,
"step": 11500
},
{
"epoch": 0.21,
"learning_rate": 4.651390657618524e-05,
"loss": 0.7209,
"step": 12000
},
{
"epoch": 0.22,
"learning_rate": 4.636853153766086e-05,
"loss": 0.64,
"step": 12500
},
{
"epoch": 0.23,
"learning_rate": 4.622315649913647e-05,
"loss": 0.6096,
"step": 13000
},
{
"epoch": 0.24,
"learning_rate": 4.607778146061209e-05,
"loss": 0.6555,
"step": 13500
},
{
"epoch": 0.24,
"learning_rate": 4.59324064220877e-05,
"loss": 0.6192,
"step": 14000
},
{
"epoch": 0.25,
"learning_rate": 4.578703138356332e-05,
"loss": 0.6229,
"step": 14500
},
{
"epoch": 0.26,
"learning_rate": 4.564165634503894e-05,
"loss": 0.5818,
"step": 15000
},
{
"epoch": 0.27,
"learning_rate": 4.549628130651455e-05,
"loss": 0.5736,
"step": 15500
},
{
"epoch": 0.28,
"learning_rate": 4.535090626799016e-05,
"loss": 0.5501,
"step": 16000
},
{
"epoch": 0.29,
"learning_rate": 4.520553122946578e-05,
"loss": 0.5234,
"step": 16500
},
{
"epoch": 0.3,
"learning_rate": 4.506044694101844e-05,
"loss": 0.5175,
"step": 17000
},
{
"epoch": 0.31,
"learning_rate": 4.4915071902494056e-05,
"loss": 0.5241,
"step": 17500
},
{
"epoch": 0.31,
"learning_rate": 4.476969686396967e-05,
"loss": 0.4967,
"step": 18000
},
{
"epoch": 0.32,
"learning_rate": 4.4624321825445286e-05,
"loss": 0.4896,
"step": 18500
},
{
"epoch": 0.33,
"learning_rate": 4.44789467869209e-05,
"loss": 0.5281,
"step": 19000
},
{
"epoch": 0.34,
"learning_rate": 4.4333571748396515e-05,
"loss": 0.4756,
"step": 19500
},
{
"epoch": 0.35,
"learning_rate": 4.418819670987213e-05,
"loss": 0.4733,
"step": 20000
},
{
"epoch": 0.36,
"learning_rate": 4.4042821671347745e-05,
"loss": 0.4575,
"step": 20500
},
{
"epoch": 0.37,
"learning_rate": 4.389744663282336e-05,
"loss": 0.4333,
"step": 21000
},
{
"epoch": 0.38,
"learning_rate": 4.3752071594298975e-05,
"loss": 0.4247,
"step": 21500
},
{
"epoch": 0.38,
"learning_rate": 4.360669655577459e-05,
"loss": 0.4739,
"step": 22000
},
{
"epoch": 0.39,
"learning_rate": 4.3461321517250204e-05,
"loss": 0.4469,
"step": 22500
},
{
"epoch": 0.4,
"learning_rate": 4.331623722880287e-05,
"loss": 0.3953,
"step": 23000
},
{
"epoch": 0.41,
"learning_rate": 4.3170862190278484e-05,
"loss": 0.4534,
"step": 23500
},
{
"epoch": 0.42,
"learning_rate": 4.30254871517541e-05,
"loss": 0.4207,
"step": 24000
},
{
"epoch": 0.43,
"learning_rate": 4.288040286330676e-05,
"loss": 0.3691,
"step": 24500
},
{
"epoch": 0.44,
"learning_rate": 4.273502782478238e-05,
"loss": 0.3827,
"step": 25000
},
{
"epoch": 0.44,
"learning_rate": 4.2589652786257986e-05,
"loss": 0.4101,
"step": 25500
},
{
"epoch": 0.45,
"learning_rate": 4.244427774773361e-05,
"loss": 0.379,
"step": 26000
},
{
"epoch": 0.46,
"learning_rate": 4.229890270920922e-05,
"loss": 0.3616,
"step": 26500
},
{
"epoch": 0.47,
"learning_rate": 4.215381842076188e-05,
"loss": 0.4308,
"step": 27000
},
{
"epoch": 0.48,
"learning_rate": 4.20084433822375e-05,
"loss": 0.354,
"step": 27500
},
{
"epoch": 0.49,
"learning_rate": 4.186306834371311e-05,
"loss": 0.3633,
"step": 28000
},
{
"epoch": 0.5,
"learning_rate": 4.1717693305188725e-05,
"loss": 0.4011,
"step": 28500
},
{
"epoch": 0.51,
"learning_rate": 4.157231826666435e-05,
"loss": 0.3489,
"step": 29000
},
{
"epoch": 0.51,
"learning_rate": 4.142694322813996e-05,
"loss": 0.3721,
"step": 29500
},
{
"epoch": 0.52,
"learning_rate": 4.128185893969262e-05,
"loss": 0.3636,
"step": 30000
},
{
"epoch": 0.53,
"learning_rate": 4.1137065401322334e-05,
"loss": 0.3521,
"step": 30500
},
{
"epoch": 0.54,
"learning_rate": 4.099169036279795e-05,
"loss": 0.3484,
"step": 31000
},
{
"epoch": 0.55,
"learning_rate": 4.0846315324273564e-05,
"loss": 0.3303,
"step": 31500
},
{
"epoch": 0.56,
"learning_rate": 4.070094028574918e-05,
"loss": 0.3119,
"step": 32000
},
{
"epoch": 0.57,
"learning_rate": 4.055556524722479e-05,
"loss": 0.3495,
"step": 32500
},
{
"epoch": 0.58,
"learning_rate": 4.041019020870041e-05,
"loss": 0.3297,
"step": 33000
},
{
"epoch": 0.58,
"learning_rate": 4.026481517017602e-05,
"loss": 0.318,
"step": 33500
},
{
"epoch": 0.59,
"learning_rate": 4.011944013165164e-05,
"loss": 0.3036,
"step": 34000
},
{
"epoch": 0.6,
"learning_rate": 3.997406509312725e-05,
"loss": 0.3131,
"step": 34500
},
{
"epoch": 0.61,
"learning_rate": 3.982869005460287e-05,
"loss": 0.2966,
"step": 35000
},
{
"epoch": 0.62,
"learning_rate": 3.968331501607848e-05,
"loss": 0.3322,
"step": 35500
},
{
"epoch": 0.63,
"learning_rate": 3.953823072763114e-05,
"loss": 0.3059,
"step": 36000
},
{
"epoch": 0.64,
"learning_rate": 3.939285568910676e-05,
"loss": 0.3008,
"step": 36500
},
{
"epoch": 0.65,
"learning_rate": 3.9247480650582376e-05,
"loss": 0.2686,
"step": 37000
},
{
"epoch": 0.65,
"learning_rate": 3.9102105612057984e-05,
"loss": 0.3025,
"step": 37500
},
{
"epoch": 0.66,
"learning_rate": 3.8956730573533606e-05,
"loss": 0.295,
"step": 38000
},
{
"epoch": 0.67,
"learning_rate": 3.881135553500922e-05,
"loss": 0.3083,
"step": 38500
},
{
"epoch": 0.68,
"learning_rate": 3.8665980496484836e-05,
"loss": 0.3016,
"step": 39000
},
{
"epoch": 0.69,
"learning_rate": 3.8520605457960444e-05,
"loss": 0.3099,
"step": 39500
},
{
"epoch": 0.7,
"learning_rate": 3.8375230419436065e-05,
"loss": 0.2942,
"step": 40000
},
{
"epoch": 0.71,
"learning_rate": 3.822985538091168e-05,
"loss": 0.2642,
"step": 40500
},
{
"epoch": 0.72,
"learning_rate": 3.808448034238729e-05,
"loss": 0.268,
"step": 41000
},
{
"epoch": 0.72,
"learning_rate": 3.793910530386291e-05,
"loss": 0.2829,
"step": 41500
},
{
"epoch": 0.73,
"learning_rate": 3.779431176549262e-05,
"loss": 0.2487,
"step": 42000
},
{
"epoch": 0.74,
"learning_rate": 3.764893672696824e-05,
"loss": 0.2845,
"step": 42500
},
{
"epoch": 0.75,
"learning_rate": 3.750356168844385e-05,
"loss": 0.2894,
"step": 43000
},
{
"epoch": 0.76,
"learning_rate": 3.735818664991946e-05,
"loss": 0.2812,
"step": 43500
},
{
"epoch": 0.77,
"learning_rate": 3.721281161139508e-05,
"loss": 0.2448,
"step": 44000
},
{
"epoch": 0.78,
"learning_rate": 3.70674365728707e-05,
"loss": 0.246,
"step": 44500
},
{
"epoch": 0.79,
"learning_rate": 3.6922061534346307e-05,
"loss": 0.2405,
"step": 45000
},
{
"epoch": 0.79,
"learning_rate": 3.677668649582192e-05,
"loss": 0.259,
"step": 45500
},
{
"epoch": 0.8,
"learning_rate": 3.663131145729754e-05,
"loss": 0.2679,
"step": 46000
},
{
"epoch": 0.81,
"learning_rate": 3.648593641877315e-05,
"loss": 0.2557,
"step": 46500
},
{
"epoch": 0.82,
"learning_rate": 3.6340561380248766e-05,
"loss": 0.2326,
"step": 47000
},
{
"epoch": 0.83,
"learning_rate": 3.619576784187848e-05,
"loss": 0.2394,
"step": 47500
},
{
"epoch": 0.84,
"learning_rate": 3.6050392803354095e-05,
"loss": 0.2612,
"step": 48000
},
{
"epoch": 0.85,
"learning_rate": 3.590501776482971e-05,
"loss": 0.2153,
"step": 48500
},
{
"epoch": 0.85,
"learning_rate": 3.5759642726305325e-05,
"loss": 0.2298,
"step": 49000
},
{
"epoch": 0.86,
"learning_rate": 3.561426768778094e-05,
"loss": 0.2299,
"step": 49500
},
{
"epoch": 0.87,
"learning_rate": 3.5468892649256555e-05,
"loss": 0.2372,
"step": 50000
},
{
"epoch": 0.88,
"learning_rate": 3.532351761073217e-05,
"loss": 0.2233,
"step": 50500
},
{
"epoch": 0.89,
"learning_rate": 3.5178142572207784e-05,
"loss": 0.2256,
"step": 51000
},
{
"epoch": 0.9,
"learning_rate": 3.50327675336834e-05,
"loss": 0.2409,
"step": 51500
},
{
"epoch": 0.91,
"learning_rate": 3.488739249515901e-05,
"loss": 0.2315,
"step": 52000
},
{
"epoch": 0.92,
"learning_rate": 3.474230820671168e-05,
"loss": 0.2129,
"step": 52500
},
{
"epoch": 0.92,
"learning_rate": 3.4596933168187287e-05,
"loss": 0.2215,
"step": 53000
},
{
"epoch": 0.93,
"learning_rate": 3.445155812966291e-05,
"loss": 0.22,
"step": 53500
},
{
"epoch": 0.94,
"learning_rate": 3.430618309113852e-05,
"loss": 0.2431,
"step": 54000
},
{
"epoch": 0.95,
"learning_rate": 3.416080805261414e-05,
"loss": 0.2271,
"step": 54500
},
{
"epoch": 0.96,
"learning_rate": 3.4015433014089746e-05,
"loss": 0.2015,
"step": 55000
},
{
"epoch": 0.97,
"learning_rate": 3.387034872564242e-05,
"loss": 0.2285,
"step": 55500
},
{
"epoch": 0.98,
"learning_rate": 3.3724973687118025e-05,
"loss": 0.227,
"step": 56000
},
{
"epoch": 0.99,
"learning_rate": 3.357959864859364e-05,
"loss": 0.2139,
"step": 56500
},
{
"epoch": 0.99,
"learning_rate": 3.343422361006926e-05,
"loss": 0.2038,
"step": 57000
},
{
"epoch": 1.0,
"eval_bleu": 0.0,
"eval_gen_len": 1.0,
"eval_loss": 0.13874725997447968,
"eval_runtime": 2091.8758,
"eval_samples_per_second": 12.179,
"eval_steps_per_second": 3.045,
"step": 57323
},
{
"epoch": 1.0,
"learning_rate": 3.328913932162192e-05,
"loss": 0.2153,
"step": 57500
},
{
"epoch": 1.01,
"learning_rate": 3.314376428309754e-05,
"loss": 0.1445,
"step": 58000
},
{
"epoch": 1.02,
"learning_rate": 3.299838924457315e-05,
"loss": 0.1382,
"step": 58500
},
{
"epoch": 1.03,
"learning_rate": 3.2853014206048764e-05,
"loss": 0.1251,
"step": 59000
},
{
"epoch": 1.04,
"learning_rate": 3.270792991760143e-05,
"loss": 0.1249,
"step": 59500
},
{
"epoch": 1.05,
"learning_rate": 3.2562554879077044e-05,
"loss": 0.135,
"step": 60000
},
{
"epoch": 1.06,
"learning_rate": 3.241747059062971e-05,
"loss": 0.1367,
"step": 60500
},
{
"epoch": 1.06,
"learning_rate": 3.227209555210532e-05,
"loss": 0.1398,
"step": 61000
},
{
"epoch": 1.07,
"learning_rate": 3.212672051358094e-05,
"loss": 0.1445,
"step": 61500
},
{
"epoch": 1.08,
"learning_rate": 3.198134547505655e-05,
"loss": 0.1367,
"step": 62000
},
{
"epoch": 1.09,
"learning_rate": 3.183597043653217e-05,
"loss": 0.1416,
"step": 62500
},
{
"epoch": 1.1,
"learning_rate": 3.169059539800778e-05,
"loss": 0.1378,
"step": 63000
},
{
"epoch": 1.11,
"learning_rate": 3.15452203594834e-05,
"loss": 0.1439,
"step": 63500
},
{
"epoch": 1.12,
"learning_rate": 3.139984532095901e-05,
"loss": 0.1413,
"step": 64000
},
{
"epoch": 1.13,
"learning_rate": 3.125447028243463e-05,
"loss": 0.1345,
"step": 64500
},
{
"epoch": 1.13,
"learning_rate": 3.110938599398729e-05,
"loss": 0.1261,
"step": 65000
},
{
"epoch": 1.14,
"learning_rate": 3.09640109554629e-05,
"loss": 0.1403,
"step": 65500
},
{
"epoch": 1.15,
"learning_rate": 3.081863591693852e-05,
"loss": 0.1351,
"step": 66000
},
{
"epoch": 1.16,
"learning_rate": 3.0673260878414136e-05,
"loss": 0.1414,
"step": 66500
},
{
"epoch": 1.17,
"learning_rate": 3.0527885839889744e-05,
"loss": 0.1428,
"step": 67000
},
{
"epoch": 1.18,
"learning_rate": 3.0382510801365366e-05,
"loss": 0.1404,
"step": 67500
},
{
"epoch": 1.19,
"learning_rate": 3.023713576284098e-05,
"loss": 0.1611,
"step": 68000
},
{
"epoch": 1.19,
"learning_rate": 3.0091760724316592e-05,
"loss": 0.1342,
"step": 68500
},
{
"epoch": 1.2,
"learning_rate": 2.994638568579221e-05,
"loss": 0.1428,
"step": 69000
},
{
"epoch": 1.21,
"learning_rate": 2.9801010647267825e-05,
"loss": 0.1441,
"step": 69500
},
{
"epoch": 1.22,
"learning_rate": 2.9655635608743436e-05,
"loss": 0.1204,
"step": 70000
},
{
"epoch": 1.23,
"learning_rate": 2.9510551320296105e-05,
"loss": 0.132,
"step": 70500
},
{
"epoch": 1.24,
"learning_rate": 2.9365176281771716e-05,
"loss": 0.1363,
"step": 71000
},
{
"epoch": 1.25,
"learning_rate": 2.921980124324733e-05,
"loss": 0.1293,
"step": 71500
},
{
"epoch": 1.26,
"learning_rate": 2.9074426204722942e-05,
"loss": 0.1276,
"step": 72000
},
{
"epoch": 1.26,
"learning_rate": 2.892934191627561e-05,
"loss": 0.1413,
"step": 72500
},
{
"epoch": 1.27,
"learning_rate": 2.8783966877751222e-05,
"loss": 0.1259,
"step": 73000
},
{
"epoch": 1.28,
"learning_rate": 2.863859183922684e-05,
"loss": 0.1265,
"step": 73500
},
{
"epoch": 1.29,
"learning_rate": 2.8493216800702455e-05,
"loss": 0.1382,
"step": 74000
},
{
"epoch": 1.3,
"learning_rate": 2.8348132512255116e-05,
"loss": 0.1313,
"step": 74500
},
{
"epoch": 1.31,
"learning_rate": 2.8202757473730734e-05,
"loss": 0.1366,
"step": 75000
},
{
"epoch": 1.32,
"learning_rate": 2.805738243520635e-05,
"loss": 0.1283,
"step": 75500
},
{
"epoch": 1.33,
"learning_rate": 2.7912298146759007e-05,
"loss": 0.128,
"step": 76000
},
{
"epoch": 1.33,
"learning_rate": 2.776692310823463e-05,
"loss": 0.124,
"step": 76500
},
{
"epoch": 1.34,
"learning_rate": 2.762154806971024e-05,
"loss": 0.1176,
"step": 77000
},
{
"epoch": 1.35,
"learning_rate": 2.7476173031185855e-05,
"loss": 0.1314,
"step": 77500
},
{
"epoch": 1.36,
"learning_rate": 2.7330797992661473e-05,
"loss": 0.1146,
"step": 78000
},
{
"epoch": 1.37,
"learning_rate": 2.7185422954137084e-05,
"loss": 0.1198,
"step": 78500
},
{
"epoch": 1.38,
"learning_rate": 2.70400479156127e-05,
"loss": 0.1187,
"step": 79000
},
{
"epoch": 1.39,
"learning_rate": 2.689467287708831e-05,
"loss": 0.1201,
"step": 79500
},
{
"epoch": 1.4,
"learning_rate": 2.674929783856393e-05,
"loss": 0.1169,
"step": 80000
},
{
"epoch": 1.4,
"learning_rate": 2.6603922800039544e-05,
"loss": 0.1097,
"step": 80500
},
{
"epoch": 1.41,
"learning_rate": 2.6458547761515155e-05,
"loss": 0.1263,
"step": 81000
},
{
"epoch": 1.42,
"learning_rate": 2.6313172722990777e-05,
"loss": 0.1141,
"step": 81500
},
{
"epoch": 1.43,
"learning_rate": 2.6168088434543435e-05,
"loss": 0.1406,
"step": 82000
},
{
"epoch": 1.44,
"learning_rate": 2.602271339601905e-05,
"loss": 0.1233,
"step": 82500
},
{
"epoch": 1.45,
"learning_rate": 2.5877338357494668e-05,
"loss": 0.108,
"step": 83000
},
{
"epoch": 1.46,
"learning_rate": 2.5731963318970283e-05,
"loss": 0.1129,
"step": 83500
},
{
"epoch": 1.47,
"learning_rate": 2.5586588280445894e-05,
"loss": 0.1181,
"step": 84000
},
{
"epoch": 1.47,
"learning_rate": 2.544121324192151e-05,
"loss": 0.1305,
"step": 84500
},
{
"epoch": 1.48,
"learning_rate": 2.5295838203397127e-05,
"loss": 0.1111,
"step": 85000
},
{
"epoch": 1.49,
"learning_rate": 2.515075391494979e-05,
"loss": 0.1064,
"step": 85500
},
{
"epoch": 1.5,
"learning_rate": 2.5005378876425407e-05,
"loss": 0.103,
"step": 86000
},
{
"epoch": 1.51,
"learning_rate": 2.4860003837901018e-05,
"loss": 0.1156,
"step": 86500
},
{
"epoch": 1.52,
"learning_rate": 2.4714628799376633e-05,
"loss": 0.1359,
"step": 87000
},
{
"epoch": 1.53,
"learning_rate": 2.4569253760852248e-05,
"loss": 0.1015,
"step": 87500
},
{
"epoch": 1.54,
"learning_rate": 2.4423878722327862e-05,
"loss": 0.105,
"step": 88000
},
{
"epoch": 1.54,
"learning_rate": 2.4278794433880524e-05,
"loss": 0.1157,
"step": 88500
},
{
"epoch": 1.55,
"learning_rate": 2.4133419395356142e-05,
"loss": 0.0975,
"step": 89000
},
{
"epoch": 1.56,
"learning_rate": 2.3988044356831757e-05,
"loss": 0.1053,
"step": 89500
},
{
"epoch": 1.57,
"learning_rate": 2.384296006838442e-05,
"loss": 0.0981,
"step": 90000
},
{
"epoch": 1.58,
"learning_rate": 2.3697585029860033e-05,
"loss": 0.1031,
"step": 90500
},
{
"epoch": 1.59,
"learning_rate": 2.355220999133565e-05,
"loss": 0.1124,
"step": 91000
},
{
"epoch": 1.6,
"learning_rate": 2.3406834952811263e-05,
"loss": 0.108,
"step": 91500
},
{
"epoch": 1.6,
"learning_rate": 2.3261459914286877e-05,
"loss": 0.0937,
"step": 92000
},
{
"epoch": 1.61,
"learning_rate": 2.3116084875762492e-05,
"loss": 0.1043,
"step": 92500
},
{
"epoch": 1.62,
"learning_rate": 2.2970709837238107e-05,
"loss": 0.0982,
"step": 93000
},
{
"epoch": 1.63,
"learning_rate": 2.2825625548790772e-05,
"loss": 0.1112,
"step": 93500
},
{
"epoch": 1.64,
"learning_rate": 2.2680250510266387e-05,
"loss": 0.1197,
"step": 94000
},
{
"epoch": 1.65,
"learning_rate": 2.2534875471742e-05,
"loss": 0.1042,
"step": 94500
},
{
"epoch": 1.66,
"learning_rate": 2.2389500433217616e-05,
"loss": 0.0888,
"step": 95000
},
{
"epoch": 1.67,
"learning_rate": 2.224412539469323e-05,
"loss": 0.1044,
"step": 95500
},
{
"epoch": 1.67,
"learning_rate": 2.2098750356168846e-05,
"loss": 0.0927,
"step": 96000
},
{
"epoch": 1.68,
"learning_rate": 2.195337531764446e-05,
"loss": 0.0971,
"step": 96500
},
{
"epoch": 1.69,
"learning_rate": 2.1808291029197122e-05,
"loss": 0.1112,
"step": 97000
},
{
"epoch": 1.7,
"learning_rate": 2.166291599067274e-05,
"loss": 0.1028,
"step": 97500
},
{
"epoch": 1.71,
"learning_rate": 2.1517540952148355e-05,
"loss": 0.0985,
"step": 98000
},
{
"epoch": 1.72,
"learning_rate": 2.1372165913623966e-05,
"loss": 0.1159,
"step": 98500
},
{
"epoch": 1.73,
"learning_rate": 2.1226790875099585e-05,
"loss": 0.0989,
"step": 99000
},
{
"epoch": 1.74,
"learning_rate": 2.1081415836575196e-05,
"loss": 0.1067,
"step": 99500
},
{
"epoch": 1.74,
"learning_rate": 2.0936040798050814e-05,
"loss": 0.1184,
"step": 100000
},
{
"epoch": 1.75,
"learning_rate": 2.0790665759526426e-05,
"loss": 0.081,
"step": 100500
},
{
"epoch": 1.76,
"learning_rate": 2.064529072100204e-05,
"loss": 0.0942,
"step": 101000
},
{
"epoch": 1.77,
"learning_rate": 2.049991568247766e-05,
"loss": 0.1033,
"step": 101500
},
{
"epoch": 1.78,
"learning_rate": 2.035454064395327e-05,
"loss": 0.0883,
"step": 102000
},
{
"epoch": 1.79,
"learning_rate": 2.020916560542889e-05,
"loss": 0.1017,
"step": 102500
},
{
"epoch": 1.8,
"learning_rate": 2.00643720670586e-05,
"loss": 0.0915,
"step": 103000
},
{
"epoch": 1.81,
"learning_rate": 1.9918997028534214e-05,
"loss": 0.1009,
"step": 103500
},
{
"epoch": 1.81,
"learning_rate": 1.9773621990009826e-05,
"loss": 0.0905,
"step": 104000
},
{
"epoch": 1.82,
"learning_rate": 1.962853770156249e-05,
"loss": 0.0989,
"step": 104500
},
{
"epoch": 1.83,
"learning_rate": 1.9483162663038105e-05,
"loss": 0.1035,
"step": 105000
},
{
"epoch": 1.84,
"learning_rate": 1.933778762451372e-05,
"loss": 0.0879,
"step": 105500
},
{
"epoch": 1.85,
"learning_rate": 1.9192412585989335e-05,
"loss": 0.1027,
"step": 106000
},
{
"epoch": 1.86,
"learning_rate": 1.9047037547464953e-05,
"loss": 0.0813,
"step": 106500
},
{
"epoch": 1.87,
"learning_rate": 1.8901662508940565e-05,
"loss": 0.0839,
"step": 107000
},
{
"epoch": 1.88,
"learning_rate": 1.8756287470416183e-05,
"loss": 0.0878,
"step": 107500
},
{
"epoch": 1.88,
"learning_rate": 1.8610912431891794e-05,
"loss": 0.0826,
"step": 108000
},
{
"epoch": 1.89,
"learning_rate": 1.846553739336741e-05,
"loss": 0.0934,
"step": 108500
},
{
"epoch": 1.9,
"learning_rate": 1.8320162354843027e-05,
"loss": 0.0807,
"step": 109000
},
{
"epoch": 1.91,
"learning_rate": 1.817478731631864e-05,
"loss": 0.0856,
"step": 109500
},
{
"epoch": 1.92,
"learning_rate": 1.8029412277794257e-05,
"loss": 0.0819,
"step": 110000
},
{
"epoch": 1.93,
"learning_rate": 1.788403723926987e-05,
"loss": 0.0979,
"step": 110500
},
{
"epoch": 1.94,
"learning_rate": 1.7738662200745483e-05,
"loss": 0.1042,
"step": 111000
},
{
"epoch": 1.95,
"learning_rate": 1.7593287162221098e-05,
"loss": 0.0744,
"step": 111500
},
{
"epoch": 1.95,
"learning_rate": 1.7447912123696713e-05,
"loss": 0.0879,
"step": 112000
},
{
"epoch": 1.96,
"learning_rate": 1.730253708517233e-05,
"loss": 0.0849,
"step": 112500
},
{
"epoch": 1.97,
"learning_rate": 1.7157452796724992e-05,
"loss": 0.0767,
"step": 113000
},
{
"epoch": 1.98,
"learning_rate": 1.7012077758200607e-05,
"loss": 0.0859,
"step": 113500
},
{
"epoch": 1.99,
"learning_rate": 1.6866702719676222e-05,
"loss": 0.0818,
"step": 114000
},
{
"epoch": 2.0,
"learning_rate": 1.6721618431228887e-05,
"loss": 0.0827,
"step": 114500
},
{
"epoch": 2.0,
"eval_bleu": 0.0,
"eval_gen_len": 1.0,
"eval_loss": 0.04327930510044098,
"eval_runtime": 2076.3927,
"eval_samples_per_second": 12.27,
"eval_steps_per_second": 3.068,
"step": 114646
},
{
"epoch": 2.01,
"learning_rate": 1.6576243392704498e-05,
"loss": 0.0638,
"step": 115000
},
{
"epoch": 2.01,
"learning_rate": 1.6430868354180116e-05,
"loss": 0.0321,
"step": 115500
},
{
"epoch": 2.02,
"learning_rate": 1.6285493315655728e-05,
"loss": 0.0423,
"step": 116000
},
{
"epoch": 2.03,
"learning_rate": 1.6140118277131343e-05,
"loss": 0.0364,
"step": 116500
},
{
"epoch": 2.04,
"learning_rate": 1.5995033988684007e-05,
"loss": 0.0399,
"step": 117000
},
{
"epoch": 2.05,
"learning_rate": 1.5849658950159622e-05,
"loss": 0.0411,
"step": 117500
},
{
"epoch": 2.06,
"learning_rate": 1.5704283911635237e-05,
"loss": 0.034,
"step": 118000
},
{
"epoch": 2.07,
"learning_rate": 1.5558908873110852e-05,
"loss": 0.0467,
"step": 118500
},
{
"epoch": 2.08,
"learning_rate": 1.5413533834586467e-05,
"loss": 0.0395,
"step": 119000
},
{
"epoch": 2.08,
"learning_rate": 1.526815879606208e-05,
"loss": 0.0421,
"step": 119500
},
{
"epoch": 2.09,
"learning_rate": 1.5122783757537695e-05,
"loss": 0.0425,
"step": 120000
},
{
"epoch": 2.1,
"learning_rate": 1.4977408719013311e-05,
"loss": 0.0426,
"step": 120500
},
{
"epoch": 2.11,
"learning_rate": 1.4832324430565974e-05,
"loss": 0.0506,
"step": 121000
},
{
"epoch": 2.12,
"learning_rate": 1.468694939204159e-05,
"loss": 0.0341,
"step": 121500
},
{
"epoch": 2.13,
"learning_rate": 1.4541574353517204e-05,
"loss": 0.0429,
"step": 122000
},
{
"epoch": 2.14,
"learning_rate": 1.439619931499282e-05,
"loss": 0.039,
"step": 122500
},
{
"epoch": 2.15,
"learning_rate": 1.4250824276468433e-05,
"loss": 0.0445,
"step": 123000
},
{
"epoch": 2.15,
"learning_rate": 1.4105449237944048e-05,
"loss": 0.0367,
"step": 123500
},
{
"epoch": 2.16,
"learning_rate": 1.3960074199419665e-05,
"loss": 0.0408,
"step": 124000
},
{
"epoch": 2.17,
"learning_rate": 1.3814699160895278e-05,
"loss": 0.0432,
"step": 124500
},
{
"epoch": 2.18,
"learning_rate": 1.3669614872447942e-05,
"loss": 0.0319,
"step": 125000
},
{
"epoch": 2.19,
"learning_rate": 1.3524239833923557e-05,
"loss": 0.0547,
"step": 125500
},
{
"epoch": 2.2,
"learning_rate": 1.337886479539917e-05,
"loss": 0.0347,
"step": 126000
},
{
"epoch": 2.21,
"learning_rate": 1.3233489756874787e-05,
"loss": 0.0329,
"step": 126500
},
{
"epoch": 2.22,
"learning_rate": 1.30881147183504e-05,
"loss": 0.0381,
"step": 127000
},
{
"epoch": 2.22,
"learning_rate": 1.2942739679826017e-05,
"loss": 0.0407,
"step": 127500
},
{
"epoch": 2.23,
"learning_rate": 1.279765539137868e-05,
"loss": 0.0441,
"step": 128000
},
{
"epoch": 2.24,
"learning_rate": 1.2652280352854296e-05,
"loss": 0.0412,
"step": 128500
},
{
"epoch": 2.25,
"learning_rate": 1.250690531432991e-05,
"loss": 0.0422,
"step": 129000
},
{
"epoch": 2.26,
"learning_rate": 1.2361530275805522e-05,
"loss": 0.0308,
"step": 129500
},
{
"epoch": 2.27,
"learning_rate": 1.2216155237281139e-05,
"loss": 0.0404,
"step": 130000
},
{
"epoch": 2.28,
"learning_rate": 1.2071070948833802e-05,
"loss": 0.041,
"step": 130500
},
{
"epoch": 2.29,
"learning_rate": 1.1925695910309417e-05,
"loss": 0.0331,
"step": 131000
},
{
"epoch": 2.29,
"learning_rate": 1.1780320871785032e-05,
"loss": 0.041,
"step": 131500
},
{
"epoch": 2.3,
"learning_rate": 1.1634945833260646e-05,
"loss": 0.0393,
"step": 132000
},
{
"epoch": 2.31,
"learning_rate": 1.1489570794736261e-05,
"loss": 0.036,
"step": 132500
},
{
"epoch": 2.32,
"learning_rate": 1.1344486506288924e-05,
"loss": 0.036,
"step": 133000
},
{
"epoch": 2.33,
"learning_rate": 1.1199111467764539e-05,
"loss": 0.0406,
"step": 133500
},
{
"epoch": 2.34,
"learning_rate": 1.1053736429240154e-05,
"loss": 0.0421,
"step": 134000
},
{
"epoch": 2.35,
"learning_rate": 1.090836139071577e-05,
"loss": 0.0345,
"step": 134500
},
{
"epoch": 2.36,
"learning_rate": 1.0762986352191383e-05,
"loss": 0.0489,
"step": 135000
},
{
"epoch": 2.36,
"learning_rate": 1.0617611313666998e-05,
"loss": 0.0298,
"step": 135500
},
{
"epoch": 2.37,
"learning_rate": 1.0472236275142613e-05,
"loss": 0.0339,
"step": 136000
},
{
"epoch": 2.38,
"learning_rate": 1.0327151986695278e-05,
"loss": 0.0362,
"step": 136500
},
{
"epoch": 2.39,
"learning_rate": 1.0181776948170891e-05,
"loss": 0.0298,
"step": 137000
},
{
"epoch": 2.4,
"learning_rate": 1.0036401909646506e-05,
"loss": 0.0433,
"step": 137500
},
{
"epoch": 2.41,
"learning_rate": 9.89102687112212e-06,
"loss": 0.0362,
"step": 138000
},
{
"epoch": 2.42,
"learning_rate": 9.745651832597737e-06,
"loss": 0.0365,
"step": 138500
},
{
"epoch": 2.42,
"learning_rate": 9.600276794073352e-06,
"loss": 0.0404,
"step": 139000
},
{
"epoch": 2.43,
"learning_rate": 9.455192505626015e-06,
"loss": 0.039,
"step": 139500
},
{
"epoch": 2.44,
"learning_rate": 9.30981746710163e-06,
"loss": 0.0361,
"step": 140000
},
{
"epoch": 2.45,
"learning_rate": 9.164442428577245e-06,
"loss": 0.0297,
"step": 140500
},
{
"epoch": 2.46,
"learning_rate": 9.01906739005286e-06,
"loss": 0.0405,
"step": 141000
},
{
"epoch": 2.47,
"learning_rate": 8.873692351528472e-06,
"loss": 0.0355,
"step": 141500
},
{
"epoch": 2.48,
"learning_rate": 8.728608063081137e-06,
"loss": 0.0361,
"step": 142000
},
{
"epoch": 2.49,
"learning_rate": 8.583233024556752e-06,
"loss": 0.0327,
"step": 142500
},
{
"epoch": 2.49,
"learning_rate": 8.437857986032367e-06,
"loss": 0.0285,
"step": 143000
},
{
"epoch": 2.5,
"learning_rate": 8.292482947507982e-06,
"loss": 0.0315,
"step": 143500
},
{
"epoch": 2.51,
"learning_rate": 8.147398659060646e-06,
"loss": 0.0401,
"step": 144000
},
{
"epoch": 2.52,
"learning_rate": 8.00202362053626e-06,
"loss": 0.0356,
"step": 144500
},
{
"epoch": 2.53,
"learning_rate": 7.856648582011874e-06,
"loss": 0.0257,
"step": 145000
},
{
"epoch": 2.54,
"learning_rate": 7.711273543487489e-06,
"loss": 0.0346,
"step": 145500
},
{
"epoch": 2.55,
"learning_rate": 7.566189255040153e-06,
"loss": 0.0326,
"step": 146000
},
{
"epoch": 2.56,
"learning_rate": 7.420814216515768e-06,
"loss": 0.0334,
"step": 146500
},
{
"epoch": 2.56,
"learning_rate": 7.27602067814548e-06,
"loss": 0.0309,
"step": 147000
},
{
"epoch": 2.57,
"learning_rate": 7.1306456396210946e-06,
"loss": 0.0439,
"step": 147500
},
{
"epoch": 2.58,
"learning_rate": 6.985270601096709e-06,
"loss": 0.036,
"step": 148000
},
{
"epoch": 2.59,
"learning_rate": 6.839895562572323e-06,
"loss": 0.0454,
"step": 148500
},
{
"epoch": 2.6,
"learning_rate": 6.69452052404794e-06,
"loss": 0.0307,
"step": 149000
},
{
"epoch": 2.61,
"learning_rate": 6.549145485523554e-06,
"loss": 0.031,
"step": 149500
},
{
"epoch": 2.62,
"learning_rate": 6.403770446999169e-06,
"loss": 0.0357,
"step": 150000
},
{
"epoch": 2.63,
"learning_rate": 6.2583954084747835e-06,
"loss": 0.0262,
"step": 150500
},
{
"epoch": 2.63,
"learning_rate": 6.113020369950398e-06,
"loss": 0.0296,
"step": 151000
},
{
"epoch": 2.64,
"learning_rate": 5.967645331426013e-06,
"loss": 0.0312,
"step": 151500
},
{
"epoch": 2.65,
"learning_rate": 5.822270292901628e-06,
"loss": 0.0313,
"step": 152000
},
{
"epoch": 2.66,
"learning_rate": 5.676895254377243e-06,
"loss": 0.0237,
"step": 152500
},
{
"epoch": 2.67,
"learning_rate": 5.531520215852858e-06,
"loss": 0.0208,
"step": 153000
},
{
"epoch": 2.68,
"learning_rate": 5.3861451773284725e-06,
"loss": 0.0312,
"step": 153500
},
{
"epoch": 2.69,
"learning_rate": 5.240770138804087e-06,
"loss": 0.0354,
"step": 154000
},
{
"epoch": 2.7,
"learning_rate": 5.095395100279702e-06,
"loss": 0.0294,
"step": 154500
},
{
"epoch": 2.7,
"learning_rate": 4.950020061755316e-06,
"loss": 0.0222,
"step": 155000
},
{
"epoch": 2.71,
"learning_rate": 4.804645023230931e-06,
"loss": 0.0279,
"step": 155500
},
{
"epoch": 2.72,
"learning_rate": 4.6592699847065466e-06,
"loss": 0.0437,
"step": 156000
},
{
"epoch": 2.73,
"learning_rate": 4.5141856962592096e-06,
"loss": 0.0317,
"step": 156500
},
{
"epoch": 2.74,
"learning_rate": 4.368810657734824e-06,
"loss": 0.0286,
"step": 157000
},
{
"epoch": 2.75,
"learning_rate": 4.223435619210439e-06,
"loss": 0.0334,
"step": 157500
},
{
"epoch": 2.76,
"learning_rate": 4.078060580686054e-06,
"loss": 0.0379,
"step": 158000
},
{
"epoch": 2.77,
"learning_rate": 3.932976292238718e-06,
"loss": 0.0279,
"step": 158500
},
{
"epoch": 2.77,
"learning_rate": 3.7876012537143327e-06,
"loss": 0.0385,
"step": 159000
},
{
"epoch": 2.78,
"learning_rate": 3.642226215189947e-06,
"loss": 0.0287,
"step": 159500
},
{
"epoch": 2.79,
"learning_rate": 3.4968511766655615e-06,
"loss": 0.0275,
"step": 160000
},
{
"epoch": 2.8,
"learning_rate": 3.3514761381411767e-06,
"loss": 0.0321,
"step": 160500
},
{
"epoch": 2.81,
"learning_rate": 3.2061010996167916e-06,
"loss": 0.0274,
"step": 161000
},
{
"epoch": 2.82,
"learning_rate": 3.0607260610924064e-06,
"loss": 0.0357,
"step": 161500
},
{
"epoch": 2.83,
"learning_rate": 2.9153510225680212e-06,
"loss": 0.0304,
"step": 162000
},
{
"epoch": 2.83,
"learning_rate": 2.7699759840436356e-06,
"loss": 0.0209,
"step": 162500
},
{
"epoch": 2.84,
"learning_rate": 2.624600945519251e-06,
"loss": 0.0216,
"step": 163000
},
{
"epoch": 2.85,
"learning_rate": 2.4792259069948657e-06,
"loss": 0.0288,
"step": 163500
},
{
"epoch": 2.86,
"learning_rate": 2.33385086847048e-06,
"loss": 0.0258,
"step": 164000
},
{
"epoch": 2.87,
"learning_rate": 2.1890573301001925e-06,
"loss": 0.0294,
"step": 164500
},
{
"epoch": 2.88,
"learning_rate": 2.0436822915758073e-06,
"loss": 0.0287,
"step": 165000
},
{
"epoch": 2.89,
"learning_rate": 1.898307253051422e-06,
"loss": 0.0262,
"step": 165500
},
{
"epoch": 2.9,
"learning_rate": 1.7532229646040858e-06,
"loss": 0.0359,
"step": 166000
},
{
"epoch": 2.9,
"learning_rate": 1.6078479260797006e-06,
"loss": 0.0255,
"step": 166500
},
{
"epoch": 2.91,
"learning_rate": 1.4624728875553154e-06,
"loss": 0.0241,
"step": 167000
},
{
"epoch": 2.92,
"learning_rate": 1.3173885991079788e-06,
"loss": 0.025,
"step": 167500
},
{
"epoch": 2.93,
"learning_rate": 1.1720135605835934e-06,
"loss": 0.0262,
"step": 168000
},
{
"epoch": 2.94,
"learning_rate": 1.0266385220592085e-06,
"loss": 0.0251,
"step": 168500
},
{
"epoch": 2.95,
"learning_rate": 8.812634835348232e-07,
"loss": 0.0264,
"step": 169000
},
{
"epoch": 2.96,
"learning_rate": 7.35888445010438e-07,
"loss": 0.0284,
"step": 169500
},
{
"epoch": 2.97,
"learning_rate": 5.905134064860527e-07,
"loss": 0.0246,
"step": 170000
},
{
"epoch": 2.97,
"learning_rate": 4.4513836796166756e-07,
"loss": 0.0244,
"step": 170500
},
{
"epoch": 2.98,
"learning_rate": 2.9976332943728233e-07,
"loss": 0.0306,
"step": 171000
},
{
"epoch": 2.99,
"learning_rate": 1.543882909128971e-07,
"loss": 0.0237,
"step": 171500
}
],
"max_steps": 171969,
"num_train_epochs": 3,
"total_flos": 3.801898817534362e+16,
"trial_name": null,
"trial_params": null
}