gpt2-large-wechsel-ukrainian / trainer_state.json
benjamin's picture
initial commit
425f21f
raw history blame
No virus
61.1 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 18.383685286886188,
"global_step": 250000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.04,
"learning_rate": 1e-05,
"loss": 7.5478,
"step": 500
},
{
"epoch": 0.07,
"learning_rate": 2e-05,
"loss": 5.8238,
"step": 1000
},
{
"epoch": 0.11,
"learning_rate": 3e-05,
"loss": 5.0177,
"step": 1500
},
{
"epoch": 0.15,
"learning_rate": 4e-05,
"loss": 4.481,
"step": 2000
},
{
"epoch": 0.18,
"learning_rate": 5e-05,
"loss": 4.1159,
"step": 2500
},
{
"epoch": 0.22,
"learning_rate": 6e-05,
"loss": 3.8654,
"step": 3000
},
{
"epoch": 0.26,
"learning_rate": 7.000000000000001e-05,
"loss": 3.6867,
"step": 3500
},
{
"epoch": 0.29,
"learning_rate": 8e-05,
"loss": 3.5543,
"step": 4000
},
{
"epoch": 0.33,
"learning_rate": 8.999999999999999e-05,
"loss": 3.4529,
"step": 4500
},
{
"epoch": 0.37,
"learning_rate": 0.0001,
"loss": 3.3706,
"step": 5000
},
{
"epoch": 0.4,
"learning_rate": 0.00011,
"loss": 3.3042,
"step": 5500
},
{
"epoch": 0.44,
"learning_rate": 0.00012,
"loss": 3.2473,
"step": 6000
},
{
"epoch": 0.48,
"learning_rate": 0.00013000000000000002,
"loss": 3.2044,
"step": 6500
},
{
"epoch": 0.51,
"learning_rate": 0.00014000000000000001,
"loss": 3.1616,
"step": 7000
},
{
"epoch": 0.55,
"learning_rate": 0.00015,
"loss": 3.1317,
"step": 7500
},
{
"epoch": 0.59,
"learning_rate": 0.00016,
"loss": 3.1003,
"step": 8000
},
{
"epoch": 0.63,
"learning_rate": 0.00017,
"loss": 3.0747,
"step": 8500
},
{
"epoch": 0.66,
"learning_rate": 0.00017999999999999998,
"loss": 3.0469,
"step": 9000
},
{
"epoch": 0.7,
"learning_rate": 0.00019,
"loss": 3.0317,
"step": 9500
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 3.0075,
"step": 10000
},
{
"epoch": 0.77,
"learning_rate": 0.00021,
"loss": 2.9949,
"step": 10500
},
{
"epoch": 0.81,
"learning_rate": 0.00022,
"loss": 2.9786,
"step": 11000
},
{
"epoch": 0.85,
"learning_rate": 0.00023,
"loss": 2.9666,
"step": 11500
},
{
"epoch": 0.88,
"learning_rate": 0.00024,
"loss": 2.9489,
"step": 12000
},
{
"epoch": 0.92,
"learning_rate": 0.00025,
"loss": 2.9409,
"step": 12500
},
{
"epoch": 0.96,
"learning_rate": 0.00026000000000000003,
"loss": 2.9302,
"step": 13000
},
{
"epoch": 0.99,
"learning_rate": 0.00027,
"loss": 2.9231,
"step": 13500
},
{
"epoch": 1.03,
"learning_rate": 0.00028000000000000003,
"loss": 2.8963,
"step": 14000
},
{
"epoch": 1.07,
"learning_rate": 0.00029,
"loss": 2.8837,
"step": 14500
},
{
"epoch": 1.1,
"learning_rate": 0.0003,
"loss": 2.8798,
"step": 15000
},
{
"epoch": 1.14,
"learning_rate": 0.00031,
"loss": 2.8784,
"step": 15500
},
{
"epoch": 1.18,
"learning_rate": 0.00032,
"loss": 2.8685,
"step": 16000
},
{
"epoch": 1.21,
"learning_rate": 0.00033,
"loss": 2.8641,
"step": 16500
},
{
"epoch": 1.25,
"learning_rate": 0.00034,
"loss": 2.8616,
"step": 17000
},
{
"epoch": 1.29,
"learning_rate": 0.00035,
"loss": 2.8567,
"step": 17500
},
{
"epoch": 1.32,
"learning_rate": 0.00035999999999999997,
"loss": 2.8544,
"step": 18000
},
{
"epoch": 1.36,
"learning_rate": 0.00037,
"loss": 2.846,
"step": 18500
},
{
"epoch": 1.4,
"learning_rate": 0.00038,
"loss": 2.8457,
"step": 19000
},
{
"epoch": 1.43,
"learning_rate": 0.00039000000000000005,
"loss": 2.8446,
"step": 19500
},
{
"epoch": 1.47,
"learning_rate": 0.0004,
"loss": 2.8356,
"step": 20000
},
{
"epoch": 1.51,
"learning_rate": 0.00041,
"loss": 2.8338,
"step": 20500
},
{
"epoch": 1.54,
"learning_rate": 0.00042,
"loss": 2.8304,
"step": 21000
},
{
"epoch": 1.58,
"learning_rate": 0.00043,
"loss": 2.8269,
"step": 21500
},
{
"epoch": 1.62,
"learning_rate": 0.00044,
"loss": 2.8268,
"step": 22000
},
{
"epoch": 1.65,
"learning_rate": 0.00045000000000000004,
"loss": 2.8224,
"step": 22500
},
{
"epoch": 1.69,
"learning_rate": 0.00046,
"loss": 2.8261,
"step": 23000
},
{
"epoch": 1.73,
"learning_rate": 0.00047,
"loss": 2.8189,
"step": 23500
},
{
"epoch": 1.76,
"learning_rate": 0.00048,
"loss": 2.8113,
"step": 24000
},
{
"epoch": 1.8,
"learning_rate": 0.00049,
"loss": 2.807,
"step": 24500
},
{
"epoch": 1.84,
"learning_rate": 0.0005,
"loss": 2.8181,
"step": 25000
},
{
"epoch": 1.88,
"learning_rate": 0.0004988888888888889,
"loss": 2.7993,
"step": 25500
},
{
"epoch": 1.91,
"learning_rate": 0.0004977777777777778,
"loss": 2.794,
"step": 26000
},
{
"epoch": 1.95,
"learning_rate": 0.0004966666666666666,
"loss": 2.795,
"step": 26500
},
{
"epoch": 1.99,
"learning_rate": 0.0004955555555555556,
"loss": 2.7797,
"step": 27000
},
{
"epoch": 2.02,
"learning_rate": 0.0004944444444444445,
"loss": 2.7595,
"step": 27500
},
{
"epoch": 2.06,
"learning_rate": 0.0004933333333333334,
"loss": 2.7412,
"step": 28000
},
{
"epoch": 2.1,
"learning_rate": 0.0004922222222222222,
"loss": 2.7442,
"step": 28500
},
{
"epoch": 2.13,
"learning_rate": 0.0004911111111111111,
"loss": 2.7406,
"step": 29000
},
{
"epoch": 2.17,
"learning_rate": 0.00049,
"loss": 2.7369,
"step": 29500
},
{
"epoch": 2.21,
"learning_rate": 0.0004888888888888889,
"loss": 2.7356,
"step": 30000
},
{
"epoch": 2.24,
"learning_rate": 0.0004877777777777778,
"loss": 2.7308,
"step": 30500
},
{
"epoch": 2.28,
"learning_rate": 0.0004866666666666667,
"loss": 2.7344,
"step": 31000
},
{
"epoch": 2.32,
"learning_rate": 0.0004855555555555556,
"loss": 2.7292,
"step": 31500
},
{
"epoch": 2.35,
"learning_rate": 0.00048444444444444446,
"loss": 2.724,
"step": 32000
},
{
"epoch": 2.39,
"learning_rate": 0.00048333333333333334,
"loss": 2.7212,
"step": 32500
},
{
"epoch": 2.43,
"learning_rate": 0.0004822222222222222,
"loss": 2.7227,
"step": 33000
},
{
"epoch": 2.46,
"learning_rate": 0.0004811111111111111,
"loss": 2.716,
"step": 33500
},
{
"epoch": 2.5,
"learning_rate": 0.00048,
"loss": 2.7139,
"step": 34000
},
{
"epoch": 2.54,
"learning_rate": 0.0004788888888888889,
"loss": 2.7083,
"step": 34500
},
{
"epoch": 2.57,
"learning_rate": 0.0004777777777777778,
"loss": 2.7062,
"step": 35000
},
{
"epoch": 2.61,
"learning_rate": 0.0004766666666666667,
"loss": 2.7066,
"step": 35500
},
{
"epoch": 2.65,
"learning_rate": 0.00047555555555555556,
"loss": 2.7028,
"step": 36000
},
{
"epoch": 2.68,
"learning_rate": 0.00047444444444444444,
"loss": 2.6972,
"step": 36500
},
{
"epoch": 2.72,
"learning_rate": 0.00047333333333333336,
"loss": 2.6951,
"step": 37000
},
{
"epoch": 2.76,
"learning_rate": 0.00047222222222222224,
"loss": 2.6941,
"step": 37500
},
{
"epoch": 2.79,
"learning_rate": 0.0004711111111111111,
"loss": 2.6943,
"step": 38000
},
{
"epoch": 2.83,
"learning_rate": 0.00047,
"loss": 2.6884,
"step": 38500
},
{
"epoch": 2.87,
"learning_rate": 0.0004688888888888889,
"loss": 2.6858,
"step": 39000
},
{
"epoch": 2.9,
"learning_rate": 0.0004677777777777778,
"loss": 2.685,
"step": 39500
},
{
"epoch": 2.94,
"learning_rate": 0.00046666666666666666,
"loss": 2.6823,
"step": 40000
},
{
"epoch": 2.98,
"learning_rate": 0.0004655555555555556,
"loss": 2.6765,
"step": 40500
},
{
"epoch": 3.01,
"learning_rate": 0.00046444444444444446,
"loss": 2.6626,
"step": 41000
},
{
"epoch": 3.05,
"learning_rate": 0.00046333333333333334,
"loss": 2.6395,
"step": 41500
},
{
"epoch": 3.09,
"learning_rate": 0.0004622222222222222,
"loss": 2.6378,
"step": 42000
},
{
"epoch": 3.13,
"learning_rate": 0.00046111111111111114,
"loss": 2.6419,
"step": 42500
},
{
"epoch": 3.16,
"learning_rate": 0.00046,
"loss": 2.6451,
"step": 43000
},
{
"epoch": 3.2,
"learning_rate": 0.0004588888888888889,
"loss": 2.6396,
"step": 43500
},
{
"epoch": 3.24,
"learning_rate": 0.0004577777777777778,
"loss": 2.6415,
"step": 44000
},
{
"epoch": 3.27,
"learning_rate": 0.0004566666666666667,
"loss": 2.6467,
"step": 44500
},
{
"epoch": 3.31,
"learning_rate": 0.00045555555555555556,
"loss": 2.64,
"step": 45000
},
{
"epoch": 3.35,
"learning_rate": 0.00045444444444444444,
"loss": 2.6391,
"step": 45500
},
{
"epoch": 3.38,
"learning_rate": 0.0004533333333333333,
"loss": 2.6377,
"step": 46000
},
{
"epoch": 3.42,
"learning_rate": 0.00045222222222222224,
"loss": 2.639,
"step": 46500
},
{
"epoch": 3.46,
"learning_rate": 0.0004511111111111111,
"loss": 2.6368,
"step": 47000
},
{
"epoch": 3.49,
"learning_rate": 0.00045000000000000004,
"loss": 2.6374,
"step": 47500
},
{
"epoch": 3.53,
"learning_rate": 0.0004488888888888889,
"loss": 2.6349,
"step": 48000
},
{
"epoch": 3.57,
"learning_rate": 0.0004477777777777778,
"loss": 2.6321,
"step": 48500
},
{
"epoch": 3.6,
"learning_rate": 0.00044666666666666666,
"loss": 2.6286,
"step": 49000
},
{
"epoch": 3.64,
"learning_rate": 0.00044555555555555554,
"loss": 2.6316,
"step": 49500
},
{
"epoch": 3.68,
"learning_rate": 0.0004444444444444444,
"loss": 2.629,
"step": 50000
},
{
"epoch": 3.71,
"learning_rate": 0.00044333333333333334,
"loss": 2.6283,
"step": 50500
},
{
"epoch": 3.75,
"learning_rate": 0.00044222222222222227,
"loss": 2.6266,
"step": 51000
},
{
"epoch": 3.79,
"learning_rate": 0.00044111111111111114,
"loss": 2.6265,
"step": 51500
},
{
"epoch": 3.82,
"learning_rate": 0.00044,
"loss": 2.6228,
"step": 52000
},
{
"epoch": 3.86,
"learning_rate": 0.0004388888888888889,
"loss": 2.6223,
"step": 52500
},
{
"epoch": 3.9,
"learning_rate": 0.00043777777777777776,
"loss": 2.6245,
"step": 53000
},
{
"epoch": 3.93,
"learning_rate": 0.00043666666666666664,
"loss": 2.6204,
"step": 53500
},
{
"epoch": 3.97,
"learning_rate": 0.0004355555555555555,
"loss": 2.6199,
"step": 54000
},
{
"epoch": 4.01,
"learning_rate": 0.0004344444444444445,
"loss": 2.6167,
"step": 54500
},
{
"epoch": 4.04,
"learning_rate": 0.00043333333333333337,
"loss": 2.5796,
"step": 55000
},
{
"epoch": 4.08,
"learning_rate": 0.00043222222222222224,
"loss": 2.5805,
"step": 55500
},
{
"epoch": 4.12,
"learning_rate": 0.0004311111111111111,
"loss": 2.5849,
"step": 56000
},
{
"epoch": 4.15,
"learning_rate": 0.00043,
"loss": 2.5859,
"step": 56500
},
{
"epoch": 4.19,
"learning_rate": 0.00042888888888888886,
"loss": 2.5873,
"step": 57000
},
{
"epoch": 4.23,
"learning_rate": 0.0004277777777777778,
"loss": 2.5868,
"step": 57500
},
{
"epoch": 4.27,
"learning_rate": 0.0004266666666666667,
"loss": 2.5884,
"step": 58000
},
{
"epoch": 4.3,
"learning_rate": 0.0004255555555555556,
"loss": 2.5852,
"step": 58500
},
{
"epoch": 4.34,
"learning_rate": 0.00042444444444444447,
"loss": 2.5867,
"step": 59000
},
{
"epoch": 4.38,
"learning_rate": 0.00042333333333333334,
"loss": 2.5881,
"step": 59500
},
{
"epoch": 4.41,
"learning_rate": 0.0004222222222222222,
"loss": 2.5879,
"step": 60000
},
{
"epoch": 4.45,
"learning_rate": 0.0004211111111111111,
"loss": 2.586,
"step": 60500
},
{
"epoch": 4.49,
"learning_rate": 0.00042,
"loss": 2.5848,
"step": 61000
},
{
"epoch": 4.52,
"learning_rate": 0.0004188888888888889,
"loss": 2.5854,
"step": 61500
},
{
"epoch": 4.56,
"learning_rate": 0.0004177777777777778,
"loss": 2.587,
"step": 62000
},
{
"epoch": 4.6,
"learning_rate": 0.0004166666666666667,
"loss": 2.5863,
"step": 62500
},
{
"epoch": 4.63,
"learning_rate": 0.00041555555555555557,
"loss": 2.5831,
"step": 63000
},
{
"epoch": 4.67,
"learning_rate": 0.00041444444444444444,
"loss": 2.5808,
"step": 63500
},
{
"epoch": 4.71,
"learning_rate": 0.0004133333333333333,
"loss": 2.5834,
"step": 64000
},
{
"epoch": 4.74,
"learning_rate": 0.00041222222222222224,
"loss": 2.5814,
"step": 64500
},
{
"epoch": 4.78,
"learning_rate": 0.0004111111111111111,
"loss": 2.579,
"step": 65000
},
{
"epoch": 4.82,
"learning_rate": 0.00041,
"loss": 2.5842,
"step": 65500
},
{
"epoch": 4.85,
"learning_rate": 0.0004088888888888889,
"loss": 2.5786,
"step": 66000
},
{
"epoch": 4.89,
"learning_rate": 0.0004077777777777778,
"loss": 2.5814,
"step": 66500
},
{
"epoch": 4.93,
"learning_rate": 0.00040666666666666667,
"loss": 2.5786,
"step": 67000
},
{
"epoch": 4.96,
"learning_rate": 0.00040555555555555554,
"loss": 2.58,
"step": 67500
},
{
"epoch": 5.0,
"learning_rate": 0.00040444444444444447,
"loss": 2.5795,
"step": 68000
},
{
"epoch": 5.04,
"learning_rate": 0.00040333333333333334,
"loss": 2.5391,
"step": 68500
},
{
"epoch": 5.07,
"learning_rate": 0.0004022222222222222,
"loss": 2.5398,
"step": 69000
},
{
"epoch": 5.11,
"learning_rate": 0.0004011111111111111,
"loss": 2.5421,
"step": 69500
},
{
"epoch": 5.15,
"learning_rate": 0.0004,
"loss": 2.545,
"step": 70000
},
{
"epoch": 5.18,
"learning_rate": 0.0003988888888888889,
"loss": 2.5493,
"step": 70500
},
{
"epoch": 5.22,
"learning_rate": 0.00039777777777777777,
"loss": 2.5484,
"step": 71000
},
{
"epoch": 5.26,
"learning_rate": 0.0003966666666666667,
"loss": 2.5483,
"step": 71500
},
{
"epoch": 5.29,
"learning_rate": 0.00039555555555555557,
"loss": 2.5511,
"step": 72000
},
{
"epoch": 5.33,
"learning_rate": 0.00039444444444444444,
"loss": 2.5497,
"step": 72500
},
{
"epoch": 5.37,
"learning_rate": 0.0003933333333333333,
"loss": 2.5495,
"step": 73000
},
{
"epoch": 5.4,
"learning_rate": 0.00039222222222222225,
"loss": 2.5508,
"step": 73500
},
{
"epoch": 5.44,
"learning_rate": 0.0003911111111111111,
"loss": 2.5474,
"step": 74000
},
{
"epoch": 5.48,
"learning_rate": 0.00039000000000000005,
"loss": 2.5508,
"step": 74500
},
{
"epoch": 5.52,
"learning_rate": 0.0003888888888888889,
"loss": 2.5512,
"step": 75000
},
{
"epoch": 5.55,
"learning_rate": 0.0003877777777777778,
"loss": 2.5488,
"step": 75500
},
{
"epoch": 5.59,
"learning_rate": 0.00038666666666666667,
"loss": 2.5511,
"step": 76000
},
{
"epoch": 5.63,
"learning_rate": 0.00038555555555555554,
"loss": 2.5507,
"step": 76500
},
{
"epoch": 5.66,
"learning_rate": 0.0003844444444444444,
"loss": 2.5507,
"step": 77000
},
{
"epoch": 5.7,
"learning_rate": 0.00038333333333333334,
"loss": 2.5502,
"step": 77500
},
{
"epoch": 5.74,
"learning_rate": 0.0003822222222222223,
"loss": 2.5464,
"step": 78000
},
{
"epoch": 5.77,
"learning_rate": 0.00038111111111111115,
"loss": 2.5483,
"step": 78500
},
{
"epoch": 5.81,
"learning_rate": 0.00038,
"loss": 2.5463,
"step": 79000
},
{
"epoch": 5.85,
"learning_rate": 0.0003788888888888889,
"loss": 2.5474,
"step": 79500
},
{
"epoch": 5.88,
"learning_rate": 0.00037777777777777777,
"loss": 2.5482,
"step": 80000
},
{
"epoch": 5.92,
"learning_rate": 0.00037666666666666664,
"loss": 2.548,
"step": 80500
},
{
"epoch": 5.96,
"learning_rate": 0.0003755555555555555,
"loss": 2.5435,
"step": 81000
},
{
"epoch": 5.99,
"learning_rate": 0.0003744444444444445,
"loss": 2.5452,
"step": 81500
},
{
"epoch": 6.03,
"learning_rate": 0.0003733333333333334,
"loss": 2.5167,
"step": 82000
},
{
"epoch": 6.07,
"learning_rate": 0.00037222222222222225,
"loss": 2.5106,
"step": 82500
},
{
"epoch": 6.1,
"learning_rate": 0.0003711111111111111,
"loss": 2.5125,
"step": 83000
},
{
"epoch": 6.14,
"learning_rate": 0.00037,
"loss": 2.5132,
"step": 83500
},
{
"epoch": 6.18,
"learning_rate": 0.00036888888888888887,
"loss": 2.5186,
"step": 84000
},
{
"epoch": 6.21,
"learning_rate": 0.00036777777777777774,
"loss": 2.5191,
"step": 84500
},
{
"epoch": 6.25,
"learning_rate": 0.00036666666666666667,
"loss": 2.5202,
"step": 85000
},
{
"epoch": 6.29,
"learning_rate": 0.0003655555555555556,
"loss": 2.5169,
"step": 85500
},
{
"epoch": 6.32,
"learning_rate": 0.00036444444444444447,
"loss": 2.5171,
"step": 86000
},
{
"epoch": 6.36,
"learning_rate": 0.00036333333333333335,
"loss": 2.5197,
"step": 86500
},
{
"epoch": 6.4,
"learning_rate": 0.0003622222222222222,
"loss": 2.5188,
"step": 87000
},
{
"epoch": 6.43,
"learning_rate": 0.0003611111111111111,
"loss": 2.5198,
"step": 87500
},
{
"epoch": 6.47,
"learning_rate": 0.00035999999999999997,
"loss": 2.5223,
"step": 88000
},
{
"epoch": 6.51,
"learning_rate": 0.0003588888888888889,
"loss": 2.522,
"step": 88500
},
{
"epoch": 6.54,
"learning_rate": 0.00035777777777777777,
"loss": 2.5226,
"step": 89000
},
{
"epoch": 6.58,
"learning_rate": 0.0003566666666666667,
"loss": 2.5218,
"step": 89500
},
{
"epoch": 6.62,
"learning_rate": 0.00035555555555555557,
"loss": 2.5222,
"step": 90000
},
{
"epoch": 6.65,
"learning_rate": 0.00035444444444444445,
"loss": 2.5184,
"step": 90500
},
{
"epoch": 6.69,
"learning_rate": 0.0003533333333333333,
"loss": 2.5206,
"step": 91000
},
{
"epoch": 6.73,
"learning_rate": 0.00035222222222222225,
"loss": 2.5247,
"step": 91500
},
{
"epoch": 6.77,
"learning_rate": 0.0003511111111111111,
"loss": 2.5244,
"step": 92000
},
{
"epoch": 6.8,
"learning_rate": 0.00035,
"loss": 2.5228,
"step": 92500
},
{
"epoch": 6.84,
"learning_rate": 0.0003488888888888889,
"loss": 2.5208,
"step": 93000
},
{
"epoch": 6.88,
"learning_rate": 0.0003477777777777778,
"loss": 2.5208,
"step": 93500
},
{
"epoch": 6.91,
"learning_rate": 0.00034666666666666667,
"loss": 2.5202,
"step": 94000
},
{
"epoch": 6.95,
"learning_rate": 0.00034555555555555555,
"loss": 2.5219,
"step": 94500
},
{
"epoch": 6.99,
"learning_rate": 0.0003444444444444445,
"loss": 2.5204,
"step": 95000
},
{
"epoch": 7.02,
"learning_rate": 0.00034333333333333335,
"loss": 2.4973,
"step": 95500
},
{
"epoch": 7.06,
"learning_rate": 0.0003422222222222222,
"loss": 2.4822,
"step": 96000
},
{
"epoch": 7.1,
"learning_rate": 0.0003411111111111111,
"loss": 2.4842,
"step": 96500
},
{
"epoch": 7.13,
"learning_rate": 0.00034,
"loss": 2.4887,
"step": 97000
},
{
"epoch": 7.17,
"learning_rate": 0.0003388888888888889,
"loss": 2.4893,
"step": 97500
},
{
"epoch": 7.21,
"learning_rate": 0.00033777777777777777,
"loss": 2.4901,
"step": 98000
},
{
"epoch": 7.24,
"learning_rate": 0.0003366666666666667,
"loss": 2.4918,
"step": 98500
},
{
"epoch": 7.28,
"learning_rate": 0.0003355555555555556,
"loss": 2.4945,
"step": 99000
},
{
"epoch": 7.32,
"learning_rate": 0.00033444444444444445,
"loss": 2.4978,
"step": 99500
},
{
"epoch": 7.35,
"learning_rate": 0.0003333333333333333,
"loss": 2.498,
"step": 100000
},
{
"epoch": 7.39,
"learning_rate": 0.0003322222222222222,
"loss": 2.4966,
"step": 100500
},
{
"epoch": 7.43,
"learning_rate": 0.0003311111111111111,
"loss": 2.4983,
"step": 101000
},
{
"epoch": 7.46,
"learning_rate": 0.00033,
"loss": 2.4984,
"step": 101500
},
{
"epoch": 7.5,
"learning_rate": 0.0003288888888888889,
"loss": 2.4961,
"step": 102000
},
{
"epoch": 7.54,
"learning_rate": 0.0003277777777777778,
"loss": 2.4996,
"step": 102500
},
{
"epoch": 7.57,
"learning_rate": 0.0003266666666666667,
"loss": 2.4984,
"step": 103000
},
{
"epoch": 7.61,
"learning_rate": 0.00032555555555555555,
"loss": 2.4992,
"step": 103500
},
{
"epoch": 7.65,
"learning_rate": 0.0003244444444444444,
"loss": 2.4985,
"step": 104000
},
{
"epoch": 7.68,
"learning_rate": 0.0003233333333333333,
"loss": 2.5009,
"step": 104500
},
{
"epoch": 7.72,
"learning_rate": 0.0003222222222222222,
"loss": 2.4971,
"step": 105000
},
{
"epoch": 7.76,
"learning_rate": 0.00032111111111111115,
"loss": 2.4994,
"step": 105500
},
{
"epoch": 7.79,
"learning_rate": 0.00032,
"loss": 2.4963,
"step": 106000
},
{
"epoch": 7.83,
"learning_rate": 0.0003188888888888889,
"loss": 2.4976,
"step": 106500
},
{
"epoch": 7.87,
"learning_rate": 0.0003177777777777778,
"loss": 2.4977,
"step": 107000
},
{
"epoch": 7.9,
"learning_rate": 0.00031666666666666665,
"loss": 2.5017,
"step": 107500
},
{
"epoch": 7.94,
"learning_rate": 0.0003155555555555555,
"loss": 2.4991,
"step": 108000
},
{
"epoch": 7.98,
"learning_rate": 0.0003144444444444445,
"loss": 2.5011,
"step": 108500
},
{
"epoch": 8.02,
"learning_rate": 0.0003133333333333334,
"loss": 2.4842,
"step": 109000
},
{
"epoch": 8.05,
"learning_rate": 0.00031222222222222225,
"loss": 2.4579,
"step": 109500
},
{
"epoch": 8.09,
"learning_rate": 0.0003111111111111111,
"loss": 2.4613,
"step": 110000
},
{
"epoch": 8.13,
"learning_rate": 0.00031,
"loss": 2.4681,
"step": 110500
},
{
"epoch": 8.16,
"learning_rate": 0.0003088888888888889,
"loss": 2.4665,
"step": 111000
},
{
"epoch": 8.2,
"learning_rate": 0.00030777777777777775,
"loss": 2.4709,
"step": 111500
},
{
"epoch": 8.24,
"learning_rate": 0.0003066666666666667,
"loss": 2.47,
"step": 112000
},
{
"epoch": 8.27,
"learning_rate": 0.0003055555555555556,
"loss": 2.4714,
"step": 112500
},
{
"epoch": 8.31,
"learning_rate": 0.0003044444444444445,
"loss": 2.473,
"step": 113000
},
{
"epoch": 8.35,
"learning_rate": 0.00030333333333333335,
"loss": 2.4769,
"step": 113500
},
{
"epoch": 8.38,
"learning_rate": 0.0003022222222222222,
"loss": 2.4753,
"step": 114000
},
{
"epoch": 8.42,
"learning_rate": 0.0003011111111111111,
"loss": 2.4772,
"step": 114500
},
{
"epoch": 8.46,
"learning_rate": 0.0003,
"loss": 2.4789,
"step": 115000
},
{
"epoch": 8.49,
"learning_rate": 0.0002988888888888889,
"loss": 2.4787,
"step": 115500
},
{
"epoch": 8.53,
"learning_rate": 0.0002977777777777778,
"loss": 2.4789,
"step": 116000
},
{
"epoch": 8.57,
"learning_rate": 0.0002966666666666667,
"loss": 2.4771,
"step": 116500
},
{
"epoch": 8.6,
"learning_rate": 0.0002955555555555556,
"loss": 2.48,
"step": 117000
},
{
"epoch": 8.64,
"learning_rate": 0.00029444444444444445,
"loss": 2.477,
"step": 117500
},
{
"epoch": 8.68,
"learning_rate": 0.0002933333333333333,
"loss": 2.4761,
"step": 118000
},
{
"epoch": 8.71,
"learning_rate": 0.0002922222222222222,
"loss": 2.4798,
"step": 118500
},
{
"epoch": 8.75,
"learning_rate": 0.00029111111111111113,
"loss": 2.4795,
"step": 119000
},
{
"epoch": 8.79,
"learning_rate": 0.00029,
"loss": 2.4845,
"step": 119500
},
{
"epoch": 8.82,
"learning_rate": 0.0002888888888888889,
"loss": 2.48,
"step": 120000
},
{
"epoch": 8.86,
"learning_rate": 0.0002877777777777778,
"loss": 2.4808,
"step": 120500
},
{
"epoch": 8.9,
"learning_rate": 0.0002866666666666667,
"loss": 2.4786,
"step": 121000
},
{
"epoch": 8.93,
"learning_rate": 0.00028555555555555555,
"loss": 2.4771,
"step": 121500
},
{
"epoch": 8.97,
"learning_rate": 0.0002844444444444444,
"loss": 2.4775,
"step": 122000
},
{
"epoch": 9.01,
"learning_rate": 0.00028333333333333335,
"loss": 2.4704,
"step": 122500
},
{
"epoch": 9.04,
"learning_rate": 0.00028222222222222223,
"loss": 2.4411,
"step": 123000
},
{
"epoch": 9.08,
"learning_rate": 0.0002811111111111111,
"loss": 2.4426,
"step": 123500
},
{
"epoch": 9.12,
"learning_rate": 0.00028000000000000003,
"loss": 2.449,
"step": 124000
},
{
"epoch": 9.16,
"learning_rate": 0.0002788888888888889,
"loss": 2.4474,
"step": 124500
},
{
"epoch": 9.19,
"learning_rate": 0.0002777777777777778,
"loss": 2.4491,
"step": 125000
},
{
"epoch": 9.23,
"learning_rate": 0.00027666666666666665,
"loss": 2.4512,
"step": 125500
},
{
"epoch": 9.27,
"learning_rate": 0.0002755555555555556,
"loss": 2.4531,
"step": 126000
},
{
"epoch": 9.3,
"learning_rate": 0.00027444444444444445,
"loss": 2.4534,
"step": 126500
},
{
"epoch": 9.34,
"learning_rate": 0.00027333333333333333,
"loss": 2.4576,
"step": 127000
},
{
"epoch": 9.38,
"learning_rate": 0.0002722222222222222,
"loss": 2.4584,
"step": 127500
},
{
"epoch": 9.41,
"learning_rate": 0.00027111111111111113,
"loss": 2.4542,
"step": 128000
},
{
"epoch": 9.45,
"learning_rate": 0.00027,
"loss": 2.4572,
"step": 128500
},
{
"epoch": 9.49,
"learning_rate": 0.00026888888888888893,
"loss": 2.4619,
"step": 129000
},
{
"epoch": 9.52,
"learning_rate": 0.0002677777777777778,
"loss": 2.4578,
"step": 129500
},
{
"epoch": 9.56,
"learning_rate": 0.0002666666666666667,
"loss": 2.4596,
"step": 130000
},
{
"epoch": 9.6,
"learning_rate": 0.00026555555555555555,
"loss": 2.4569,
"step": 130500
},
{
"epoch": 9.63,
"learning_rate": 0.00026444444444444443,
"loss": 2.4611,
"step": 131000
},
{
"epoch": 9.67,
"learning_rate": 0.0002633333333333333,
"loss": 2.4605,
"step": 131500
},
{
"epoch": 9.71,
"learning_rate": 0.00026222222222222223,
"loss": 2.4632,
"step": 132000
},
{
"epoch": 9.74,
"learning_rate": 0.00026111111111111116,
"loss": 2.4603,
"step": 132500
},
{
"epoch": 9.78,
"learning_rate": 0.00026000000000000003,
"loss": 2.4609,
"step": 133000
},
{
"epoch": 9.82,
"learning_rate": 0.0002588888888888889,
"loss": 2.4605,
"step": 133500
},
{
"epoch": 9.85,
"learning_rate": 0.0002577777777777778,
"loss": 2.4623,
"step": 134000
},
{
"epoch": 9.89,
"learning_rate": 0.00025666666666666665,
"loss": 2.4634,
"step": 134500
},
{
"epoch": 9.93,
"learning_rate": 0.00025555555555555553,
"loss": 2.4601,
"step": 135000
},
{
"epoch": 9.96,
"learning_rate": 0.0002544444444444444,
"loss": 2.4608,
"step": 135500
},
{
"epoch": 10.0,
"learning_rate": 0.0002533333333333334,
"loss": 2.4625,
"step": 136000
},
{
"epoch": 10.04,
"learning_rate": 0.00025222222222222226,
"loss": 2.423,
"step": 136500
},
{
"epoch": 10.07,
"learning_rate": 0.00025111111111111113,
"loss": 2.4245,
"step": 137000
},
{
"epoch": 10.11,
"learning_rate": 0.00025,
"loss": 2.4306,
"step": 137500
},
{
"epoch": 10.15,
"learning_rate": 0.0002488888888888889,
"loss": 2.4346,
"step": 138000
},
{
"epoch": 10.18,
"learning_rate": 0.0002477777777777778,
"loss": 2.436,
"step": 138500
},
{
"epoch": 10.22,
"learning_rate": 0.0002466666666666667,
"loss": 2.4336,
"step": 139000
},
{
"epoch": 10.26,
"learning_rate": 0.00024555555555555556,
"loss": 2.4358,
"step": 139500
},
{
"epoch": 10.29,
"learning_rate": 0.00024444444444444443,
"loss": 2.4363,
"step": 140000
},
{
"epoch": 10.33,
"learning_rate": 0.00024333333333333336,
"loss": 2.4386,
"step": 140500
},
{
"epoch": 10.37,
"learning_rate": 0.00024222222222222223,
"loss": 2.4396,
"step": 141000
},
{
"epoch": 10.41,
"learning_rate": 0.0002411111111111111,
"loss": 2.4395,
"step": 141500
},
{
"epoch": 10.44,
"learning_rate": 0.00024,
"loss": 2.4379,
"step": 142000
},
{
"epoch": 10.48,
"learning_rate": 0.0002388888888888889,
"loss": 2.4396,
"step": 142500
},
{
"epoch": 10.52,
"learning_rate": 0.00023777777777777778,
"loss": 2.4418,
"step": 143000
},
{
"epoch": 10.55,
"learning_rate": 0.00023666666666666668,
"loss": 2.4415,
"step": 143500
},
{
"epoch": 10.59,
"learning_rate": 0.00023555555555555556,
"loss": 2.442,
"step": 144000
},
{
"epoch": 10.63,
"learning_rate": 0.00023444444444444446,
"loss": 2.4421,
"step": 144500
},
{
"epoch": 10.66,
"learning_rate": 0.00023333333333333333,
"loss": 2.4426,
"step": 145000
},
{
"epoch": 10.7,
"learning_rate": 0.00023222222222222223,
"loss": 2.443,
"step": 145500
},
{
"epoch": 10.74,
"learning_rate": 0.0002311111111111111,
"loss": 2.4423,
"step": 146000
},
{
"epoch": 10.77,
"learning_rate": 0.00023,
"loss": 2.4454,
"step": 146500
},
{
"epoch": 10.81,
"learning_rate": 0.0002288888888888889,
"loss": 2.4419,
"step": 147000
},
{
"epoch": 10.85,
"learning_rate": 0.00022777777777777778,
"loss": 2.4415,
"step": 147500
},
{
"epoch": 10.88,
"learning_rate": 0.00022666666666666666,
"loss": 2.4434,
"step": 148000
},
{
"epoch": 10.92,
"learning_rate": 0.00022555555555555556,
"loss": 2.4447,
"step": 148500
},
{
"epoch": 10.96,
"learning_rate": 0.00022444444444444446,
"loss": 2.4437,
"step": 149000
},
{
"epoch": 10.99,
"learning_rate": 0.00022333333333333333,
"loss": 2.4475,
"step": 149500
},
{
"epoch": 11.03,
"learning_rate": 0.0002222222222222222,
"loss": 2.4159,
"step": 150000
},
{
"epoch": 11.07,
"learning_rate": 0.00022111111111111113,
"loss": 2.4095,
"step": 150500
},
{
"epoch": 11.1,
"learning_rate": 0.00022,
"loss": 2.4117,
"step": 151000
},
{
"epoch": 11.14,
"learning_rate": 0.00021888888888888888,
"loss": 2.4166,
"step": 151500
},
{
"epoch": 11.18,
"learning_rate": 0.00021777777777777776,
"loss": 2.4161,
"step": 152000
},
{
"epoch": 11.21,
"learning_rate": 0.00021666666666666668,
"loss": 2.416,
"step": 152500
},
{
"epoch": 11.25,
"learning_rate": 0.00021555555555555556,
"loss": 2.4175,
"step": 153000
},
{
"epoch": 11.29,
"learning_rate": 0.00021444444444444443,
"loss": 2.4184,
"step": 153500
},
{
"epoch": 11.32,
"learning_rate": 0.00021333333333333336,
"loss": 2.4206,
"step": 154000
},
{
"epoch": 11.36,
"learning_rate": 0.00021222222222222223,
"loss": 2.4262,
"step": 154500
},
{
"epoch": 11.4,
"learning_rate": 0.0002111111111111111,
"loss": 2.4236,
"step": 155000
},
{
"epoch": 11.43,
"learning_rate": 0.00021,
"loss": 2.4249,
"step": 155500
},
{
"epoch": 11.47,
"learning_rate": 0.0002088888888888889,
"loss": 2.4244,
"step": 156000
},
{
"epoch": 11.51,
"learning_rate": 0.00020777777777777778,
"loss": 2.4235,
"step": 156500
},
{
"epoch": 11.54,
"learning_rate": 0.00020666666666666666,
"loss": 2.4211,
"step": 157000
},
{
"epoch": 11.58,
"learning_rate": 0.00020555555555555556,
"loss": 2.429,
"step": 157500
},
{
"epoch": 11.62,
"learning_rate": 0.00020444444444444446,
"loss": 2.4279,
"step": 158000
},
{
"epoch": 11.66,
"learning_rate": 0.00020333333333333333,
"loss": 2.4269,
"step": 158500
},
{
"epoch": 11.69,
"learning_rate": 0.00020222222222222223,
"loss": 2.4294,
"step": 159000
},
{
"epoch": 11.73,
"learning_rate": 0.0002011111111111111,
"loss": 2.4267,
"step": 159500
},
{
"epoch": 11.77,
"learning_rate": 0.0002,
"loss": 2.4241,
"step": 160000
},
{
"epoch": 11.8,
"learning_rate": 0.00019888888888888888,
"loss": 2.4278,
"step": 160500
},
{
"epoch": 11.84,
"learning_rate": 0.00019777777777777778,
"loss": 2.429,
"step": 161000
},
{
"epoch": 11.88,
"learning_rate": 0.00019666666666666666,
"loss": 2.4277,
"step": 161500
},
{
"epoch": 11.91,
"learning_rate": 0.00019555555555555556,
"loss": 2.4281,
"step": 162000
},
{
"epoch": 11.95,
"learning_rate": 0.00019444444444444446,
"loss": 2.4282,
"step": 162500
},
{
"epoch": 11.99,
"learning_rate": 0.00019333333333333333,
"loss": 2.4299,
"step": 163000
},
{
"epoch": 12.02,
"learning_rate": 0.0001922222222222222,
"loss": 2.4077,
"step": 163500
},
{
"epoch": 12.06,
"learning_rate": 0.00019111111111111114,
"loss": 2.3924,
"step": 164000
},
{
"epoch": 12.1,
"learning_rate": 0.00019,
"loss": 2.3967,
"step": 164500
},
{
"epoch": 12.13,
"learning_rate": 0.00018888888888888888,
"loss": 2.3987,
"step": 165000
},
{
"epoch": 12.17,
"learning_rate": 0.00018777777777777776,
"loss": 2.4014,
"step": 165500
},
{
"epoch": 12.21,
"learning_rate": 0.0001866666666666667,
"loss": 2.4003,
"step": 166000
},
{
"epoch": 12.24,
"learning_rate": 0.00018555555555555556,
"loss": 2.4062,
"step": 166500
},
{
"epoch": 12.28,
"learning_rate": 0.00018444444444444443,
"loss": 2.4038,
"step": 167000
},
{
"epoch": 12.32,
"learning_rate": 0.00018333333333333334,
"loss": 2.406,
"step": 167500
},
{
"epoch": 12.35,
"learning_rate": 0.00018222222222222224,
"loss": 2.4058,
"step": 168000
},
{
"epoch": 12.39,
"learning_rate": 0.0001811111111111111,
"loss": 2.406,
"step": 168500
},
{
"epoch": 12.43,
"learning_rate": 0.00017999999999999998,
"loss": 2.4058,
"step": 169000
},
{
"epoch": 12.46,
"learning_rate": 0.00017888888888888889,
"loss": 2.4086,
"step": 169500
},
{
"epoch": 12.5,
"learning_rate": 0.00017777777777777779,
"loss": 2.41,
"step": 170000
},
{
"epoch": 12.54,
"learning_rate": 0.00017666666666666666,
"loss": 2.4093,
"step": 170500
},
{
"epoch": 12.57,
"learning_rate": 0.00017555555555555556,
"loss": 2.411,
"step": 171000
},
{
"epoch": 12.61,
"learning_rate": 0.00017444444444444446,
"loss": 2.4099,
"step": 171500
},
{
"epoch": 12.65,
"learning_rate": 0.00017333333333333334,
"loss": 2.4121,
"step": 172000
},
{
"epoch": 12.68,
"learning_rate": 0.00017222222222222224,
"loss": 2.4129,
"step": 172500
},
{
"epoch": 12.72,
"learning_rate": 0.0001711111111111111,
"loss": 2.4139,
"step": 173000
},
{
"epoch": 12.76,
"learning_rate": 0.00017,
"loss": 2.4121,
"step": 173500
},
{
"epoch": 12.8,
"learning_rate": 0.00016888888888888889,
"loss": 2.412,
"step": 174000
},
{
"epoch": 12.83,
"learning_rate": 0.0001677777777777778,
"loss": 2.408,
"step": 174500
},
{
"epoch": 12.87,
"learning_rate": 0.00016666666666666666,
"loss": 2.4135,
"step": 175000
},
{
"epoch": 12.91,
"learning_rate": 0.00016555555555555556,
"loss": 2.411,
"step": 175500
},
{
"epoch": 12.94,
"learning_rate": 0.00016444444444444446,
"loss": 2.4108,
"step": 176000
},
{
"epoch": 12.98,
"learning_rate": 0.00016333333333333334,
"loss": 2.4105,
"step": 176500
},
{
"epoch": 13.02,
"learning_rate": 0.0001622222222222222,
"loss": 2.4026,
"step": 177000
},
{
"epoch": 13.05,
"learning_rate": 0.0001611111111111111,
"loss": 2.38,
"step": 177500
},
{
"epoch": 13.09,
"learning_rate": 0.00016,
"loss": 2.3798,
"step": 178000
},
{
"epoch": 13.13,
"learning_rate": 0.0001588888888888889,
"loss": 2.3848,
"step": 178500
},
{
"epoch": 13.16,
"learning_rate": 0.00015777777777777776,
"loss": 2.3865,
"step": 179000
},
{
"epoch": 13.2,
"learning_rate": 0.0001566666666666667,
"loss": 2.3872,
"step": 179500
},
{
"epoch": 13.24,
"learning_rate": 0.00015555555555555556,
"loss": 2.3894,
"step": 180000
},
{
"epoch": 13.27,
"learning_rate": 0.00015444444444444444,
"loss": 2.392,
"step": 180500
},
{
"epoch": 13.31,
"learning_rate": 0.00015333333333333334,
"loss": 2.3872,
"step": 181000
},
{
"epoch": 13.35,
"learning_rate": 0.00015222222222222224,
"loss": 2.3893,
"step": 181500
},
{
"epoch": 13.38,
"learning_rate": 0.0001511111111111111,
"loss": 2.391,
"step": 182000
},
{
"epoch": 13.42,
"learning_rate": 0.00015,
"loss": 2.3948,
"step": 182500
},
{
"epoch": 13.46,
"learning_rate": 0.0001488888888888889,
"loss": 2.3915,
"step": 183000
},
{
"epoch": 13.49,
"learning_rate": 0.0001477777777777778,
"loss": 2.3893,
"step": 183500
},
{
"epoch": 13.53,
"learning_rate": 0.00014666666666666666,
"loss": 2.3917,
"step": 184000
},
{
"epoch": 13.57,
"learning_rate": 0.00014555555555555556,
"loss": 2.395,
"step": 184500
},
{
"epoch": 13.6,
"learning_rate": 0.00014444444444444444,
"loss": 2.3926,
"step": 185000
},
{
"epoch": 13.64,
"learning_rate": 0.00014333333333333334,
"loss": 2.3938,
"step": 185500
},
{
"epoch": 13.68,
"learning_rate": 0.0001422222222222222,
"loss": 2.3934,
"step": 186000
},
{
"epoch": 13.71,
"learning_rate": 0.00014111111111111111,
"loss": 2.3982,
"step": 186500
},
{
"epoch": 13.75,
"learning_rate": 0.00014000000000000001,
"loss": 2.397,
"step": 187000
},
{
"epoch": 13.79,
"learning_rate": 0.0001388888888888889,
"loss": 2.3969,
"step": 187500
},
{
"epoch": 13.82,
"learning_rate": 0.0001377777777777778,
"loss": 2.3969,
"step": 188000
},
{
"epoch": 13.86,
"learning_rate": 0.00013666666666666666,
"loss": 2.3997,
"step": 188500
},
{
"epoch": 13.9,
"learning_rate": 0.00013555555555555556,
"loss": 2.3966,
"step": 189000
},
{
"epoch": 13.93,
"learning_rate": 0.00013444444444444447,
"loss": 2.3976,
"step": 189500
},
{
"epoch": 13.97,
"learning_rate": 0.00013333333333333334,
"loss": 2.3981,
"step": 190000
},
{
"epoch": 14.01,
"learning_rate": 0.00013222222222222221,
"loss": 2.3907,
"step": 190500
},
{
"epoch": 14.05,
"learning_rate": 0.00013111111111111111,
"loss": 2.3665,
"step": 191000
},
{
"epoch": 14.08,
"learning_rate": 0.00013000000000000002,
"loss": 2.3677,
"step": 191500
},
{
"epoch": 14.12,
"learning_rate": 0.0001288888888888889,
"loss": 2.3667,
"step": 192000
},
{
"epoch": 14.16,
"learning_rate": 0.00012777777777777776,
"loss": 2.3746,
"step": 192500
},
{
"epoch": 14.19,
"learning_rate": 0.0001266666666666667,
"loss": 2.3693,
"step": 193000
},
{
"epoch": 14.23,
"learning_rate": 0.00012555555555555557,
"loss": 2.3721,
"step": 193500
},
{
"epoch": 14.27,
"learning_rate": 0.00012444444444444444,
"loss": 2.3729,
"step": 194000
},
{
"epoch": 14.3,
"learning_rate": 0.00012333333333333334,
"loss": 2.3782,
"step": 194500
},
{
"epoch": 14.34,
"learning_rate": 0.00012222222222222221,
"loss": 2.3751,
"step": 195000
},
{
"epoch": 14.38,
"learning_rate": 0.00012111111111111112,
"loss": 2.3777,
"step": 195500
},
{
"epoch": 14.41,
"learning_rate": 0.00012,
"loss": 2.3767,
"step": 196000
},
{
"epoch": 14.45,
"learning_rate": 0.00011888888888888889,
"loss": 2.3761,
"step": 196500
},
{
"epoch": 14.49,
"learning_rate": 0.00011777777777777778,
"loss": 2.3787,
"step": 197000
},
{
"epoch": 14.52,
"learning_rate": 0.00011666666666666667,
"loss": 2.3796,
"step": 197500
},
{
"epoch": 14.56,
"learning_rate": 0.00011555555555555555,
"loss": 2.3796,
"step": 198000
},
{
"epoch": 14.6,
"learning_rate": 0.00011444444444444445,
"loss": 2.3791,
"step": 198500
},
{
"epoch": 14.63,
"learning_rate": 0.00011333333333333333,
"loss": 2.3789,
"step": 199000
},
{
"epoch": 14.67,
"learning_rate": 0.00011222222222222223,
"loss": 2.3818,
"step": 199500
},
{
"epoch": 14.71,
"learning_rate": 0.0001111111111111111,
"loss": 2.3813,
"step": 200000
},
{
"epoch": 14.74,
"learning_rate": 0.00011,
"loss": 2.3815,
"step": 200500
},
{
"epoch": 14.78,
"learning_rate": 0.00010888888888888888,
"loss": 2.3779,
"step": 201000
},
{
"epoch": 14.82,
"learning_rate": 0.00010777777777777778,
"loss": 2.3818,
"step": 201500
},
{
"epoch": 14.85,
"learning_rate": 0.00010666666666666668,
"loss": 2.3796,
"step": 202000
},
{
"epoch": 14.89,
"learning_rate": 0.00010555555555555555,
"loss": 2.3822,
"step": 202500
},
{
"epoch": 14.93,
"learning_rate": 0.00010444444444444445,
"loss": 2.3829,
"step": 203000
},
{
"epoch": 14.96,
"learning_rate": 0.00010333333333333333,
"loss": 2.3798,
"step": 203500
},
{
"epoch": 15.0,
"learning_rate": 0.00010222222222222223,
"loss": 2.3833,
"step": 204000
},
{
"epoch": 15.04,
"learning_rate": 0.00010111111111111112,
"loss": 2.3523,
"step": 204500
},
{
"epoch": 15.07,
"learning_rate": 0.0001,
"loss": 2.3534,
"step": 205000
},
{
"epoch": 15.11,
"learning_rate": 9.888888888888889e-05,
"loss": 2.3552,
"step": 205500
},
{
"epoch": 15.15,
"learning_rate": 9.777777777777778e-05,
"loss": 2.3573,
"step": 206000
},
{
"epoch": 15.18,
"learning_rate": 9.666666666666667e-05,
"loss": 2.3597,
"step": 206500
},
{
"epoch": 15.22,
"learning_rate": 9.555555555555557e-05,
"loss": 2.3594,
"step": 207000
},
{
"epoch": 15.26,
"learning_rate": 9.444444444444444e-05,
"loss": 2.3626,
"step": 207500
},
{
"epoch": 15.3,
"learning_rate": 9.333333333333334e-05,
"loss": 2.3613,
"step": 208000
},
{
"epoch": 15.33,
"learning_rate": 9.222222222222222e-05,
"loss": 2.3594,
"step": 208500
},
{
"epoch": 15.37,
"learning_rate": 9.111111111111112e-05,
"loss": 2.3609,
"step": 209000
},
{
"epoch": 15.41,
"learning_rate": 8.999999999999999e-05,
"loss": 2.3618,
"step": 209500
},
{
"epoch": 15.44,
"learning_rate": 8.888888888888889e-05,
"loss": 2.3643,
"step": 210000
},
{
"epoch": 15.48,
"learning_rate": 8.777777777777778e-05,
"loss": 2.3646,
"step": 210500
},
{
"epoch": 15.52,
"learning_rate": 8.666666666666667e-05,
"loss": 2.3599,
"step": 211000
},
{
"epoch": 15.55,
"learning_rate": 8.555555555555556e-05,
"loss": 2.3623,
"step": 211500
},
{
"epoch": 15.59,
"learning_rate": 8.444444444444444e-05,
"loss": 2.364,
"step": 212000
},
{
"epoch": 15.63,
"learning_rate": 8.333333333333333e-05,
"loss": 2.3654,
"step": 212500
},
{
"epoch": 15.66,
"learning_rate": 8.222222222222223e-05,
"loss": 2.3626,
"step": 213000
},
{
"epoch": 15.7,
"learning_rate": 8.11111111111111e-05,
"loss": 2.3663,
"step": 213500
},
{
"epoch": 15.74,
"learning_rate": 8e-05,
"loss": 2.3669,
"step": 214000
},
{
"epoch": 15.77,
"learning_rate": 7.888888888888888e-05,
"loss": 2.3662,
"step": 214500
},
{
"epoch": 15.81,
"learning_rate": 7.777777777777778e-05,
"loss": 2.3641,
"step": 215000
},
{
"epoch": 15.85,
"learning_rate": 7.666666666666667e-05,
"loss": 2.365,
"step": 215500
},
{
"epoch": 15.88,
"learning_rate": 7.555555555555556e-05,
"loss": 2.3685,
"step": 216000
},
{
"epoch": 15.92,
"learning_rate": 7.444444444444444e-05,
"loss": 2.3667,
"step": 216500
},
{
"epoch": 15.96,
"learning_rate": 7.333333333333333e-05,
"loss": 2.3641,
"step": 217000
},
{
"epoch": 15.99,
"learning_rate": 7.222222222222222e-05,
"loss": 2.3651,
"step": 217500
},
{
"epoch": 16.03,
"learning_rate": 7.11111111111111e-05,
"loss": 2.3486,
"step": 218000
},
{
"epoch": 16.07,
"learning_rate": 7.000000000000001e-05,
"loss": 2.3431,
"step": 218500
},
{
"epoch": 16.1,
"learning_rate": 6.88888888888889e-05,
"loss": 2.3439,
"step": 219000
},
{
"epoch": 16.14,
"learning_rate": 6.777777777777778e-05,
"loss": 2.3438,
"step": 219500
},
{
"epoch": 16.18,
"learning_rate": 6.666666666666667e-05,
"loss": 2.3426,
"step": 220000
},
{
"epoch": 16.21,
"learning_rate": 6.555555555555556e-05,
"loss": 2.3463,
"step": 220500
},
{
"epoch": 16.25,
"learning_rate": 6.444444444444444e-05,
"loss": 2.3486,
"step": 221000
},
{
"epoch": 16.29,
"learning_rate": 6.333333333333335e-05,
"loss": 2.3476,
"step": 221500
},
{
"epoch": 16.32,
"learning_rate": 6.222222222222222e-05,
"loss": 2.3469,
"step": 222000
},
{
"epoch": 16.36,
"learning_rate": 6.111111111111111e-05,
"loss": 2.3519,
"step": 222500
},
{
"epoch": 16.4,
"learning_rate": 6e-05,
"loss": 2.3485,
"step": 223000
},
{
"epoch": 16.44,
"learning_rate": 5.888888888888889e-05,
"loss": 2.3464,
"step": 223500
},
{
"epoch": 16.47,
"learning_rate": 5.7777777777777776e-05,
"loss": 2.3524,
"step": 224000
},
{
"epoch": 16.51,
"learning_rate": 5.6666666666666664e-05,
"loss": 2.3482,
"step": 224500
},
{
"epoch": 16.55,
"learning_rate": 5.555555555555555e-05,
"loss": 2.3485,
"step": 225000
},
{
"epoch": 16.58,
"learning_rate": 5.444444444444444e-05,
"loss": 2.3476,
"step": 225500
},
{
"epoch": 16.62,
"learning_rate": 5.333333333333334e-05,
"loss": 2.3452,
"step": 226000
},
{
"epoch": 16.66,
"learning_rate": 5.222222222222223e-05,
"loss": 2.3491,
"step": 226500
},
{
"epoch": 16.69,
"learning_rate": 5.1111111111111115e-05,
"loss": 2.3504,
"step": 227000
},
{
"epoch": 16.73,
"learning_rate": 5e-05,
"loss": 2.3517,
"step": 227500
},
{
"epoch": 16.77,
"learning_rate": 4.888888888888889e-05,
"loss": 2.3508,
"step": 228000
},
{
"epoch": 16.8,
"learning_rate": 4.7777777777777784e-05,
"loss": 2.3522,
"step": 228500
},
{
"epoch": 16.84,
"learning_rate": 4.666666666666667e-05,
"loss": 2.3494,
"step": 229000
},
{
"epoch": 16.88,
"learning_rate": 4.555555555555556e-05,
"loss": 2.3508,
"step": 229500
},
{
"epoch": 16.91,
"learning_rate": 4.4444444444444447e-05,
"loss": 2.349,
"step": 230000
},
{
"epoch": 16.95,
"learning_rate": 4.3333333333333334e-05,
"loss": 2.3499,
"step": 230500
},
{
"epoch": 16.99,
"learning_rate": 4.222222222222222e-05,
"loss": 2.3474,
"step": 231000
},
{
"epoch": 17.02,
"learning_rate": 4.1111111111111116e-05,
"loss": 2.3398,
"step": 231500
},
{
"epoch": 17.06,
"learning_rate": 4e-05,
"loss": 2.3326,
"step": 232000
},
{
"epoch": 17.1,
"learning_rate": 3.888888888888889e-05,
"loss": 2.332,
"step": 232500
},
{
"epoch": 17.13,
"learning_rate": 3.777777777777778e-05,
"loss": 2.3296,
"step": 233000
},
{
"epoch": 17.17,
"learning_rate": 3.6666666666666666e-05,
"loss": 2.3336,
"step": 233500
},
{
"epoch": 17.21,
"learning_rate": 3.555555555555555e-05,
"loss": 2.3328,
"step": 234000
},
{
"epoch": 17.24,
"learning_rate": 3.444444444444445e-05,
"loss": 2.3386,
"step": 234500
},
{
"epoch": 17.28,
"learning_rate": 3.3333333333333335e-05,
"loss": 2.3343,
"step": 235000
},
{
"epoch": 17.32,
"learning_rate": 3.222222222222222e-05,
"loss": 2.3354,
"step": 235500
},
{
"epoch": 17.35,
"learning_rate": 3.111111111111111e-05,
"loss": 2.3331,
"step": 236000
},
{
"epoch": 17.39,
"learning_rate": 3e-05,
"loss": 2.3351,
"step": 236500
},
{
"epoch": 17.43,
"learning_rate": 2.8888888888888888e-05,
"loss": 2.3319,
"step": 237000
},
{
"epoch": 17.46,
"learning_rate": 2.7777777777777776e-05,
"loss": 2.3351,
"step": 237500
},
{
"epoch": 17.5,
"learning_rate": 2.666666666666667e-05,
"loss": 2.3342,
"step": 238000
},
{
"epoch": 17.54,
"learning_rate": 2.5555555555555557e-05,
"loss": 2.3358,
"step": 238500
},
{
"epoch": 17.57,
"learning_rate": 2.4444444444444445e-05,
"loss": 2.3334,
"step": 239000
},
{
"epoch": 17.61,
"learning_rate": 2.3333333333333336e-05,
"loss": 2.3349,
"step": 239500
},
{
"epoch": 17.65,
"learning_rate": 2.2222222222222223e-05,
"loss": 2.3337,
"step": 240000
},
{
"epoch": 17.69,
"learning_rate": 2.111111111111111e-05,
"loss": 2.3357,
"step": 240500
},
{
"epoch": 17.72,
"learning_rate": 2e-05,
"loss": 2.3369,
"step": 241000
},
{
"epoch": 17.76,
"learning_rate": 1.888888888888889e-05,
"loss": 2.3365,
"step": 241500
},
{
"epoch": 17.8,
"learning_rate": 1.7777777777777777e-05,
"loss": 2.3347,
"step": 242000
},
{
"epoch": 17.83,
"learning_rate": 1.6666666666666667e-05,
"loss": 2.336,
"step": 242500
},
{
"epoch": 17.87,
"learning_rate": 1.5555555555555555e-05,
"loss": 2.334,
"step": 243000
},
{
"epoch": 17.91,
"learning_rate": 1.4444444444444444e-05,
"loss": 2.3343,
"step": 243500
},
{
"epoch": 17.94,
"learning_rate": 1.3333333333333335e-05,
"loss": 2.3332,
"step": 244000
},
{
"epoch": 17.98,
"learning_rate": 1.2222222222222222e-05,
"loss": 2.3376,
"step": 244500
},
{
"epoch": 18.02,
"learning_rate": 1.1111111111111112e-05,
"loss": 2.3319,
"step": 245000
},
{
"epoch": 18.05,
"learning_rate": 1e-05,
"loss": 2.3224,
"step": 245500
},
{
"epoch": 18.09,
"learning_rate": 8.888888888888888e-06,
"loss": 2.3249,
"step": 246000
},
{
"epoch": 18.13,
"learning_rate": 7.777777777777777e-06,
"loss": 2.3241,
"step": 246500
},
{
"epoch": 18.16,
"learning_rate": 6.6666666666666675e-06,
"loss": 2.3222,
"step": 247000
},
{
"epoch": 18.2,
"learning_rate": 5.555555555555556e-06,
"loss": 2.3221,
"step": 247500
},
{
"epoch": 18.24,
"learning_rate": 4.444444444444444e-06,
"loss": 2.324,
"step": 248000
},
{
"epoch": 18.27,
"learning_rate": 3.3333333333333337e-06,
"loss": 2.3234,
"step": 248500
},
{
"epoch": 18.31,
"learning_rate": 2.222222222222222e-06,
"loss": 2.3225,
"step": 249000
},
{
"epoch": 18.35,
"learning_rate": 1.111111111111111e-06,
"loss": 2.3234,
"step": 249500
},
{
"epoch": 18.38,
"learning_rate": 0.0,
"loss": 2.3237,
"step": 250000
},
{
"epoch": 18.38,
"step": 250000,
"total_flos": 3.482052847164457e+19,
"train_loss": 0.7083540009765625,
"train_runtime": 869248.5122,
"train_samples_per_second": 73.627,
"train_steps_per_second": 0.288
}
],
"max_steps": 250000,
"num_train_epochs": 19,
"total_flos": 3.482052847164457e+19,
"trial_name": null,
"trial_params": null
}