m2m100_418M_en_zul_rel_news / trainer_state.json
Davlan's picture
add MT model
4c0674d
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"global_step": 200220,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.9875137348916196e-05,
"loss": 3.7531,
"step": 500
},
{
"epoch": 0.01,
"learning_rate": 4.975027469783239e-05,
"loss": 2.9041,
"step": 1000
},
{
"epoch": 0.02,
"learning_rate": 4.962541204674858e-05,
"loss": 2.5413,
"step": 1500
},
{
"epoch": 0.03,
"learning_rate": 4.950054939566477e-05,
"loss": 2.3266,
"step": 2000
},
{
"epoch": 0.04,
"learning_rate": 4.937568674458096e-05,
"loss": 2.1497,
"step": 2500
},
{
"epoch": 0.04,
"learning_rate": 4.9250824093497155e-05,
"loss": 2.0392,
"step": 3000
},
{
"epoch": 0.05,
"learning_rate": 4.912596144241335e-05,
"loss": 1.9413,
"step": 3500
},
{
"epoch": 0.06,
"learning_rate": 4.900109879132954e-05,
"loss": 1.8698,
"step": 4000
},
{
"epoch": 0.07,
"learning_rate": 4.8876236140245735e-05,
"loss": 1.8197,
"step": 4500
},
{
"epoch": 0.07,
"learning_rate": 4.875137348916192e-05,
"loss": 1.7706,
"step": 5000
},
{
"epoch": 0.08,
"learning_rate": 4.8626510838078115e-05,
"loss": 1.6978,
"step": 5500
},
{
"epoch": 0.09,
"learning_rate": 4.850164818699431e-05,
"loss": 1.6654,
"step": 6000
},
{
"epoch": 0.1,
"learning_rate": 4.83767855359105e-05,
"loss": 1.6313,
"step": 6500
},
{
"epoch": 0.1,
"learning_rate": 4.8251922884826695e-05,
"loss": 1.6022,
"step": 7000
},
{
"epoch": 0.11,
"learning_rate": 4.812706023374289e-05,
"loss": 1.558,
"step": 7500
},
{
"epoch": 0.12,
"learning_rate": 4.8002197582659075e-05,
"loss": 1.5389,
"step": 8000
},
{
"epoch": 0.13,
"learning_rate": 4.787733493157527e-05,
"loss": 1.5257,
"step": 8500
},
{
"epoch": 0.13,
"learning_rate": 4.775247228049146e-05,
"loss": 1.5091,
"step": 9000
},
{
"epoch": 0.14,
"learning_rate": 4.7627609629407655e-05,
"loss": 1.4909,
"step": 9500
},
{
"epoch": 0.15,
"learning_rate": 4.750274697832385e-05,
"loss": 1.479,
"step": 10000
},
{
"epoch": 0.16,
"learning_rate": 4.737788432724004e-05,
"loss": 1.4322,
"step": 10500
},
{
"epoch": 0.16,
"learning_rate": 4.725302167615623e-05,
"loss": 1.4256,
"step": 11000
},
{
"epoch": 0.17,
"learning_rate": 4.712815902507242e-05,
"loss": 1.4211,
"step": 11500
},
{
"epoch": 0.18,
"learning_rate": 4.7003296373988614e-05,
"loss": 1.3964,
"step": 12000
},
{
"epoch": 0.19,
"learning_rate": 4.687843372290481e-05,
"loss": 1.3892,
"step": 12500
},
{
"epoch": 0.19,
"learning_rate": 4.6753571071821e-05,
"loss": 1.3777,
"step": 13000
},
{
"epoch": 0.2,
"learning_rate": 4.6628708420737194e-05,
"loss": 1.3917,
"step": 13500
},
{
"epoch": 0.21,
"learning_rate": 4.650384576965338e-05,
"loss": 1.3443,
"step": 14000
},
{
"epoch": 0.22,
"learning_rate": 4.6378983118569574e-05,
"loss": 1.3544,
"step": 14500
},
{
"epoch": 0.22,
"learning_rate": 4.625412046748577e-05,
"loss": 1.3199,
"step": 15000
},
{
"epoch": 0.23,
"learning_rate": 4.612925781640196e-05,
"loss": 1.3102,
"step": 15500
},
{
"epoch": 0.24,
"learning_rate": 4.6004395165318154e-05,
"loss": 1.3007,
"step": 16000
},
{
"epoch": 0.25,
"learning_rate": 4.587953251423435e-05,
"loss": 1.3053,
"step": 16500
},
{
"epoch": 0.25,
"learning_rate": 4.5754669863150534e-05,
"loss": 1.2832,
"step": 17000
},
{
"epoch": 0.26,
"learning_rate": 4.562980721206673e-05,
"loss": 1.2878,
"step": 17500
},
{
"epoch": 0.27,
"learning_rate": 4.550494456098292e-05,
"loss": 1.2722,
"step": 18000
},
{
"epoch": 0.28,
"learning_rate": 4.538008190989911e-05,
"loss": 1.2699,
"step": 18500
},
{
"epoch": 0.28,
"learning_rate": 4.525521925881531e-05,
"loss": 1.2651,
"step": 19000
},
{
"epoch": 0.29,
"learning_rate": 4.51303566077315e-05,
"loss": 1.2586,
"step": 19500
},
{
"epoch": 0.3,
"learning_rate": 4.5005493956647686e-05,
"loss": 1.2421,
"step": 20000
},
{
"epoch": 0.31,
"learning_rate": 4.488063130556388e-05,
"loss": 1.2379,
"step": 20500
},
{
"epoch": 0.31,
"learning_rate": 4.475576865448007e-05,
"loss": 1.2221,
"step": 21000
},
{
"epoch": 0.32,
"learning_rate": 4.4630906003396266e-05,
"loss": 1.2206,
"step": 21500
},
{
"epoch": 0.33,
"learning_rate": 4.450604335231246e-05,
"loss": 1.2095,
"step": 22000
},
{
"epoch": 0.34,
"learning_rate": 4.438118070122865e-05,
"loss": 1.1966,
"step": 22500
},
{
"epoch": 0.34,
"learning_rate": 4.425631805014484e-05,
"loss": 1.2054,
"step": 23000
},
{
"epoch": 0.35,
"learning_rate": 4.413145539906103e-05,
"loss": 1.178,
"step": 23500
},
{
"epoch": 0.36,
"learning_rate": 4.4006592747977226e-05,
"loss": 1.1836,
"step": 24000
},
{
"epoch": 0.37,
"learning_rate": 4.388173009689342e-05,
"loss": 1.1883,
"step": 24500
},
{
"epoch": 0.37,
"learning_rate": 4.375686744580961e-05,
"loss": 1.1954,
"step": 25000
},
{
"epoch": 0.38,
"learning_rate": 4.3632004794725806e-05,
"loss": 1.177,
"step": 25500
},
{
"epoch": 0.39,
"learning_rate": 4.350714214364199e-05,
"loss": 1.1746,
"step": 26000
},
{
"epoch": 0.4,
"learning_rate": 4.3382279492558186e-05,
"loss": 1.1831,
"step": 26500
},
{
"epoch": 0.4,
"learning_rate": 4.325741684147438e-05,
"loss": 1.1552,
"step": 27000
},
{
"epoch": 0.41,
"learning_rate": 4.313255419039057e-05,
"loss": 1.1582,
"step": 27500
},
{
"epoch": 0.42,
"learning_rate": 4.3007691539306765e-05,
"loss": 1.1679,
"step": 28000
},
{
"epoch": 0.43,
"learning_rate": 4.288282888822296e-05,
"loss": 1.1695,
"step": 28500
},
{
"epoch": 0.43,
"learning_rate": 4.2757966237139145e-05,
"loss": 1.1419,
"step": 29000
},
{
"epoch": 0.44,
"learning_rate": 4.263310358605534e-05,
"loss": 1.1501,
"step": 29500
},
{
"epoch": 0.45,
"learning_rate": 4.250824093497153e-05,
"loss": 1.1318,
"step": 30000
},
{
"epoch": 0.46,
"learning_rate": 4.2383378283887725e-05,
"loss": 1.124,
"step": 30500
},
{
"epoch": 0.46,
"learning_rate": 4.225851563280392e-05,
"loss": 1.1139,
"step": 31000
},
{
"epoch": 0.47,
"learning_rate": 4.213365298172011e-05,
"loss": 1.1172,
"step": 31500
},
{
"epoch": 0.48,
"learning_rate": 4.20087903306363e-05,
"loss": 1.1086,
"step": 32000
},
{
"epoch": 0.49,
"learning_rate": 4.188392767955249e-05,
"loss": 1.1248,
"step": 32500
},
{
"epoch": 0.49,
"learning_rate": 4.1759065028468685e-05,
"loss": 1.1001,
"step": 33000
},
{
"epoch": 0.5,
"learning_rate": 4.163420237738488e-05,
"loss": 1.1074,
"step": 33500
},
{
"epoch": 0.51,
"learning_rate": 4.150933972630107e-05,
"loss": 1.0772,
"step": 34000
},
{
"epoch": 0.52,
"learning_rate": 4.1384477075217265e-05,
"loss": 1.0942,
"step": 34500
},
{
"epoch": 0.52,
"learning_rate": 4.125961442413345e-05,
"loss": 1.1072,
"step": 35000
},
{
"epoch": 0.53,
"learning_rate": 4.1134751773049644e-05,
"loss": 1.0815,
"step": 35500
},
{
"epoch": 0.54,
"learning_rate": 4.100988912196584e-05,
"loss": 1.0693,
"step": 36000
},
{
"epoch": 0.55,
"learning_rate": 4.088502647088203e-05,
"loss": 1.0837,
"step": 36500
},
{
"epoch": 0.55,
"learning_rate": 4.0760163819798224e-05,
"loss": 1.0951,
"step": 37000
},
{
"epoch": 0.56,
"learning_rate": 4.063530116871442e-05,
"loss": 1.0666,
"step": 37500
},
{
"epoch": 0.57,
"learning_rate": 4.0510438517630604e-05,
"loss": 1.0808,
"step": 38000
},
{
"epoch": 0.58,
"learning_rate": 4.03855758665468e-05,
"loss": 1.075,
"step": 38500
},
{
"epoch": 0.58,
"learning_rate": 4.026071321546299e-05,
"loss": 1.0683,
"step": 39000
},
{
"epoch": 0.59,
"learning_rate": 4.0135850564379184e-05,
"loss": 1.0633,
"step": 39500
},
{
"epoch": 0.6,
"learning_rate": 4.001098791329538e-05,
"loss": 1.0814,
"step": 40000
},
{
"epoch": 0.61,
"learning_rate": 3.988612526221157e-05,
"loss": 1.0485,
"step": 40500
},
{
"epoch": 0.61,
"learning_rate": 3.976126261112776e-05,
"loss": 1.0483,
"step": 41000
},
{
"epoch": 0.62,
"learning_rate": 3.963639996004395e-05,
"loss": 1.0656,
"step": 41500
},
{
"epoch": 0.63,
"learning_rate": 3.9511537308960144e-05,
"loss": 1.07,
"step": 42000
},
{
"epoch": 0.64,
"learning_rate": 3.938667465787634e-05,
"loss": 1.0441,
"step": 42500
},
{
"epoch": 0.64,
"learning_rate": 3.926181200679253e-05,
"loss": 1.0494,
"step": 43000
},
{
"epoch": 0.65,
"learning_rate": 3.9136949355708724e-05,
"loss": 1.0482,
"step": 43500
},
{
"epoch": 0.66,
"learning_rate": 3.901208670462492e-05,
"loss": 1.065,
"step": 44000
},
{
"epoch": 0.67,
"learning_rate": 3.88872240535411e-05,
"loss": 1.0641,
"step": 44500
},
{
"epoch": 0.67,
"learning_rate": 3.8762361402457297e-05,
"loss": 1.0439,
"step": 45000
},
{
"epoch": 0.68,
"learning_rate": 3.863749875137349e-05,
"loss": 1.0377,
"step": 45500
},
{
"epoch": 0.69,
"learning_rate": 3.851263610028968e-05,
"loss": 1.0256,
"step": 46000
},
{
"epoch": 0.7,
"learning_rate": 3.8387773449205876e-05,
"loss": 1.0327,
"step": 46500
},
{
"epoch": 0.7,
"learning_rate": 3.826291079812207e-05,
"loss": 1.0272,
"step": 47000
},
{
"epoch": 0.71,
"learning_rate": 3.8138048147038256e-05,
"loss": 1.0348,
"step": 47500
},
{
"epoch": 0.72,
"learning_rate": 3.801318549595445e-05,
"loss": 1.0194,
"step": 48000
},
{
"epoch": 0.73,
"learning_rate": 3.788832284487064e-05,
"loss": 1.0178,
"step": 48500
},
{
"epoch": 0.73,
"learning_rate": 3.7763460193786836e-05,
"loss": 1.0144,
"step": 49000
},
{
"epoch": 0.74,
"learning_rate": 3.763859754270303e-05,
"loss": 1.0073,
"step": 49500
},
{
"epoch": 0.75,
"learning_rate": 3.751373489161922e-05,
"loss": 1.0012,
"step": 50000
},
{
"epoch": 0.76,
"learning_rate": 3.738887224053541e-05,
"loss": 1.0055,
"step": 50500
},
{
"epoch": 0.76,
"learning_rate": 3.72640095894516e-05,
"loss": 1.0066,
"step": 51000
},
{
"epoch": 0.77,
"learning_rate": 3.7139146938367796e-05,
"loss": 1.0111,
"step": 51500
},
{
"epoch": 0.78,
"learning_rate": 3.701428428728399e-05,
"loss": 1.0177,
"step": 52000
},
{
"epoch": 0.79,
"learning_rate": 3.688942163620018e-05,
"loss": 1.0059,
"step": 52500
},
{
"epoch": 0.79,
"learning_rate": 3.6764558985116376e-05,
"loss": 0.9872,
"step": 53000
},
{
"epoch": 0.8,
"learning_rate": 3.663969633403256e-05,
"loss": 1.0135,
"step": 53500
},
{
"epoch": 0.81,
"learning_rate": 3.6514833682948755e-05,
"loss": 0.9899,
"step": 54000
},
{
"epoch": 0.82,
"learning_rate": 3.638997103186495e-05,
"loss": 0.9918,
"step": 54500
},
{
"epoch": 0.82,
"learning_rate": 3.626510838078114e-05,
"loss": 1.0154,
"step": 55000
},
{
"epoch": 0.83,
"learning_rate": 3.6140245729697335e-05,
"loss": 0.9957,
"step": 55500
},
{
"epoch": 0.84,
"learning_rate": 3.601538307861353e-05,
"loss": 0.9839,
"step": 56000
},
{
"epoch": 0.85,
"learning_rate": 3.5890520427529715e-05,
"loss": 0.999,
"step": 56500
},
{
"epoch": 0.85,
"learning_rate": 3.576565777644591e-05,
"loss": 0.9836,
"step": 57000
},
{
"epoch": 0.86,
"learning_rate": 3.56407951253621e-05,
"loss": 0.9774,
"step": 57500
},
{
"epoch": 0.87,
"learning_rate": 3.5515932474278295e-05,
"loss": 0.9709,
"step": 58000
},
{
"epoch": 0.88,
"learning_rate": 3.539106982319449e-05,
"loss": 0.985,
"step": 58500
},
{
"epoch": 0.88,
"learning_rate": 3.526620717211068e-05,
"loss": 0.977,
"step": 59000
},
{
"epoch": 0.89,
"learning_rate": 3.514134452102687e-05,
"loss": 0.9578,
"step": 59500
},
{
"epoch": 0.9,
"learning_rate": 3.501648186994306e-05,
"loss": 0.964,
"step": 60000
},
{
"epoch": 0.91,
"learning_rate": 3.4891619218859255e-05,
"loss": 0.9739,
"step": 60500
},
{
"epoch": 0.91,
"learning_rate": 3.476675656777545e-05,
"loss": 0.9647,
"step": 61000
},
{
"epoch": 0.92,
"learning_rate": 3.464189391669164e-05,
"loss": 0.9653,
"step": 61500
},
{
"epoch": 0.93,
"learning_rate": 3.4517031265607834e-05,
"loss": 0.9643,
"step": 62000
},
{
"epoch": 0.94,
"learning_rate": 3.439216861452402e-05,
"loss": 0.9644,
"step": 62500
},
{
"epoch": 0.94,
"learning_rate": 3.4267305963440214e-05,
"loss": 0.9384,
"step": 63000
},
{
"epoch": 0.95,
"learning_rate": 3.414244331235641e-05,
"loss": 0.969,
"step": 63500
},
{
"epoch": 0.96,
"learning_rate": 3.40175806612726e-05,
"loss": 0.9602,
"step": 64000
},
{
"epoch": 0.97,
"learning_rate": 3.3892718010188794e-05,
"loss": 0.9549,
"step": 64500
},
{
"epoch": 0.97,
"learning_rate": 3.376785535910499e-05,
"loss": 0.9607,
"step": 65000
},
{
"epoch": 0.98,
"learning_rate": 3.3642992708021174e-05,
"loss": 0.9616,
"step": 65500
},
{
"epoch": 0.99,
"learning_rate": 3.351813005693737e-05,
"loss": 0.9497,
"step": 66000
},
{
"epoch": 1.0,
"learning_rate": 3.339326740585356e-05,
"loss": 0.9618,
"step": 66500
},
{
"epoch": 1.0,
"learning_rate": 3.3268404754769754e-05,
"loss": 0.9013,
"step": 67000
},
{
"epoch": 1.01,
"learning_rate": 3.314354210368595e-05,
"loss": 0.8594,
"step": 67500
},
{
"epoch": 1.02,
"learning_rate": 3.301867945260214e-05,
"loss": 0.85,
"step": 68000
},
{
"epoch": 1.03,
"learning_rate": 3.289381680151833e-05,
"loss": 0.8586,
"step": 68500
},
{
"epoch": 1.03,
"learning_rate": 3.276895415043452e-05,
"loss": 0.864,
"step": 69000
},
{
"epoch": 1.04,
"learning_rate": 3.2644091499350713e-05,
"loss": 0.8557,
"step": 69500
},
{
"epoch": 1.05,
"learning_rate": 3.251922884826691e-05,
"loss": 0.8625,
"step": 70000
},
{
"epoch": 1.06,
"learning_rate": 3.23943661971831e-05,
"loss": 0.8501,
"step": 70500
},
{
"epoch": 1.06,
"learning_rate": 3.226950354609929e-05,
"loss": 0.8485,
"step": 71000
},
{
"epoch": 1.07,
"learning_rate": 3.214464089501548e-05,
"loss": 0.8504,
"step": 71500
},
{
"epoch": 1.08,
"learning_rate": 3.201977824393167e-05,
"loss": 0.8533,
"step": 72000
},
{
"epoch": 1.09,
"learning_rate": 3.1894915592847866e-05,
"loss": 0.8528,
"step": 72500
},
{
"epoch": 1.09,
"learning_rate": 3.177005294176406e-05,
"loss": 0.8327,
"step": 73000
},
{
"epoch": 1.1,
"learning_rate": 3.164519029068025e-05,
"loss": 0.8454,
"step": 73500
},
{
"epoch": 1.11,
"learning_rate": 3.1520327639596446e-05,
"loss": 0.8416,
"step": 74000
},
{
"epoch": 1.12,
"learning_rate": 3.139546498851263e-05,
"loss": 0.8413,
"step": 74500
},
{
"epoch": 1.12,
"learning_rate": 3.1270602337428826e-05,
"loss": 0.8443,
"step": 75000
},
{
"epoch": 1.13,
"learning_rate": 3.114573968634502e-05,
"loss": 0.8576,
"step": 75500
},
{
"epoch": 1.14,
"learning_rate": 3.102087703526121e-05,
"loss": 0.8517,
"step": 76000
},
{
"epoch": 1.15,
"learning_rate": 3.0896014384177406e-05,
"loss": 0.8397,
"step": 76500
},
{
"epoch": 1.15,
"learning_rate": 3.07711517330936e-05,
"loss": 0.8555,
"step": 77000
},
{
"epoch": 1.16,
"learning_rate": 3.0646289082009786e-05,
"loss": 0.8403,
"step": 77500
},
{
"epoch": 1.17,
"learning_rate": 3.052142643092598e-05,
"loss": 0.8531,
"step": 78000
},
{
"epoch": 1.18,
"learning_rate": 3.0396563779842176e-05,
"loss": 0.8494,
"step": 78500
},
{
"epoch": 1.18,
"learning_rate": 3.027170112875837e-05,
"loss": 0.8397,
"step": 79000
},
{
"epoch": 1.19,
"learning_rate": 3.014683847767456e-05,
"loss": 0.8535,
"step": 79500
},
{
"epoch": 1.2,
"learning_rate": 3.0021975826590752e-05,
"loss": 0.8388,
"step": 80000
},
{
"epoch": 1.21,
"learning_rate": 2.9897113175506942e-05,
"loss": 0.8364,
"step": 80500
},
{
"epoch": 1.21,
"learning_rate": 2.9772250524423135e-05,
"loss": 0.8333,
"step": 81000
},
{
"epoch": 1.22,
"learning_rate": 2.964738787333933e-05,
"loss": 0.8416,
"step": 81500
},
{
"epoch": 1.23,
"learning_rate": 2.9522525222255522e-05,
"loss": 0.8578,
"step": 82000
},
{
"epoch": 1.24,
"learning_rate": 2.9397662571171712e-05,
"loss": 0.8357,
"step": 82500
},
{
"epoch": 1.24,
"learning_rate": 2.9272799920087905e-05,
"loss": 0.8309,
"step": 83000
},
{
"epoch": 1.25,
"learning_rate": 2.9147937269004095e-05,
"loss": 0.8379,
"step": 83500
},
{
"epoch": 1.26,
"learning_rate": 2.902307461792029e-05,
"loss": 0.8419,
"step": 84000
},
{
"epoch": 1.27,
"learning_rate": 2.889821196683648e-05,
"loss": 0.8317,
"step": 84500
},
{
"epoch": 1.27,
"learning_rate": 2.8773349315752675e-05,
"loss": 0.8287,
"step": 85000
},
{
"epoch": 1.28,
"learning_rate": 2.8648486664668868e-05,
"loss": 0.8308,
"step": 85500
},
{
"epoch": 1.29,
"learning_rate": 2.8523624013585058e-05,
"loss": 0.8396,
"step": 86000
},
{
"epoch": 1.3,
"learning_rate": 2.839876136250125e-05,
"loss": 0.8418,
"step": 86500
},
{
"epoch": 1.3,
"learning_rate": 2.827389871141744e-05,
"loss": 0.8404,
"step": 87000
},
{
"epoch": 1.31,
"learning_rate": 2.8149036060333635e-05,
"loss": 0.8323,
"step": 87500
},
{
"epoch": 1.32,
"learning_rate": 2.8024173409249828e-05,
"loss": 0.8432,
"step": 88000
},
{
"epoch": 1.33,
"learning_rate": 2.789931075816602e-05,
"loss": 0.8431,
"step": 88500
},
{
"epoch": 1.33,
"learning_rate": 2.777444810708221e-05,
"loss": 0.8356,
"step": 89000
},
{
"epoch": 1.34,
"learning_rate": 2.7649585455998404e-05,
"loss": 0.8334,
"step": 89500
},
{
"epoch": 1.35,
"learning_rate": 2.7524722804914594e-05,
"loss": 0.8314,
"step": 90000
},
{
"epoch": 1.36,
"learning_rate": 2.7399860153830787e-05,
"loss": 0.8383,
"step": 90500
},
{
"epoch": 1.36,
"learning_rate": 2.727499750274698e-05,
"loss": 0.8346,
"step": 91000
},
{
"epoch": 1.37,
"learning_rate": 2.7150134851663174e-05,
"loss": 0.8259,
"step": 91500
},
{
"epoch": 1.38,
"learning_rate": 2.7025272200579367e-05,
"loss": 0.8377,
"step": 92000
},
{
"epoch": 1.39,
"learning_rate": 2.6900409549495557e-05,
"loss": 0.8171,
"step": 92500
},
{
"epoch": 1.39,
"learning_rate": 2.6775546898411747e-05,
"loss": 0.8252,
"step": 93000
},
{
"epoch": 1.4,
"learning_rate": 2.665068424732794e-05,
"loss": 0.8242,
"step": 93500
},
{
"epoch": 1.41,
"learning_rate": 2.6525821596244134e-05,
"loss": 0.8202,
"step": 94000
},
{
"epoch": 1.42,
"learning_rate": 2.6400958945160327e-05,
"loss": 0.8229,
"step": 94500
},
{
"epoch": 1.42,
"learning_rate": 2.627609629407652e-05,
"loss": 0.8258,
"step": 95000
},
{
"epoch": 1.43,
"learning_rate": 2.615123364299271e-05,
"loss": 0.8152,
"step": 95500
},
{
"epoch": 1.44,
"learning_rate": 2.60263709919089e-05,
"loss": 0.8174,
"step": 96000
},
{
"epoch": 1.45,
"learning_rate": 2.5901508340825093e-05,
"loss": 0.8291,
"step": 96500
},
{
"epoch": 1.45,
"learning_rate": 2.5776645689741287e-05,
"loss": 0.8151,
"step": 97000
},
{
"epoch": 1.46,
"learning_rate": 2.565178303865748e-05,
"loss": 0.8144,
"step": 97500
},
{
"epoch": 1.47,
"learning_rate": 2.5526920387573673e-05,
"loss": 0.8142,
"step": 98000
},
{
"epoch": 1.48,
"learning_rate": 2.5402057736489867e-05,
"loss": 0.8134,
"step": 98500
},
{
"epoch": 1.48,
"learning_rate": 2.5277195085406053e-05,
"loss": 0.8164,
"step": 99000
},
{
"epoch": 1.49,
"learning_rate": 2.5152332434322246e-05,
"loss": 0.8202,
"step": 99500
},
{
"epoch": 1.5,
"learning_rate": 2.502746978323844e-05,
"loss": 0.8164,
"step": 100000
},
{
"epoch": 1.51,
"learning_rate": 2.4902607132154633e-05,
"loss": 0.8013,
"step": 100500
},
{
"epoch": 1.51,
"learning_rate": 2.4777744481070823e-05,
"loss": 0.8098,
"step": 101000
},
{
"epoch": 1.52,
"learning_rate": 2.4652881829987016e-05,
"loss": 0.8132,
"step": 101500
},
{
"epoch": 1.53,
"learning_rate": 2.452801917890321e-05,
"loss": 0.8091,
"step": 102000
},
{
"epoch": 1.54,
"learning_rate": 2.44031565278194e-05,
"loss": 0.8059,
"step": 102500
},
{
"epoch": 1.54,
"learning_rate": 2.4278293876735593e-05,
"loss": 0.8035,
"step": 103000
},
{
"epoch": 1.55,
"learning_rate": 2.4153431225651786e-05,
"loss": 0.8123,
"step": 103500
},
{
"epoch": 1.56,
"learning_rate": 2.4028568574567976e-05,
"loss": 0.8141,
"step": 104000
},
{
"epoch": 1.57,
"learning_rate": 2.390370592348417e-05,
"loss": 0.8042,
"step": 104500
},
{
"epoch": 1.57,
"learning_rate": 2.3778843272400362e-05,
"loss": 0.8166,
"step": 105000
},
{
"epoch": 1.58,
"learning_rate": 2.3653980621316552e-05,
"loss": 0.8021,
"step": 105500
},
{
"epoch": 1.59,
"learning_rate": 2.3529117970232746e-05,
"loss": 0.798,
"step": 106000
},
{
"epoch": 1.6,
"learning_rate": 2.340425531914894e-05,
"loss": 0.809,
"step": 106500
},
{
"epoch": 1.6,
"learning_rate": 2.3279392668065132e-05,
"loss": 0.8161,
"step": 107000
},
{
"epoch": 1.61,
"learning_rate": 2.3154530016981322e-05,
"loss": 0.8093,
"step": 107500
},
{
"epoch": 1.62,
"learning_rate": 2.3029667365897515e-05,
"loss": 0.7977,
"step": 108000
},
{
"epoch": 1.63,
"learning_rate": 2.290480471481371e-05,
"loss": 0.7985,
"step": 108500
},
{
"epoch": 1.63,
"learning_rate": 2.27799420637299e-05,
"loss": 0.8036,
"step": 109000
},
{
"epoch": 1.64,
"learning_rate": 2.2655079412646092e-05,
"loss": 0.7973,
"step": 109500
},
{
"epoch": 1.65,
"learning_rate": 2.2530216761562285e-05,
"loss": 0.802,
"step": 110000
},
{
"epoch": 1.66,
"learning_rate": 2.2405354110478475e-05,
"loss": 0.7877,
"step": 110500
},
{
"epoch": 1.66,
"learning_rate": 2.2280491459394668e-05,
"loss": 0.8018,
"step": 111000
},
{
"epoch": 1.67,
"learning_rate": 2.215562880831086e-05,
"loss": 0.7984,
"step": 111500
},
{
"epoch": 1.68,
"learning_rate": 2.203076615722705e-05,
"loss": 0.794,
"step": 112000
},
{
"epoch": 1.69,
"learning_rate": 2.1905903506143245e-05,
"loss": 0.7962,
"step": 112500
},
{
"epoch": 1.69,
"learning_rate": 2.1781040855059438e-05,
"loss": 0.8085,
"step": 113000
},
{
"epoch": 1.7,
"learning_rate": 2.1656178203975628e-05,
"loss": 0.7877,
"step": 113500
},
{
"epoch": 1.71,
"learning_rate": 2.153131555289182e-05,
"loss": 0.7943,
"step": 114000
},
{
"epoch": 1.72,
"learning_rate": 2.1406452901808014e-05,
"loss": 0.7982,
"step": 114500
},
{
"epoch": 1.72,
"learning_rate": 2.1281590250724204e-05,
"loss": 0.8015,
"step": 115000
},
{
"epoch": 1.73,
"learning_rate": 2.1156727599640398e-05,
"loss": 0.793,
"step": 115500
},
{
"epoch": 1.74,
"learning_rate": 2.103186494855659e-05,
"loss": 0.7978,
"step": 116000
},
{
"epoch": 1.75,
"learning_rate": 2.090700229747278e-05,
"loss": 0.796,
"step": 116500
},
{
"epoch": 1.75,
"learning_rate": 2.0782139646388974e-05,
"loss": 0.7805,
"step": 117000
},
{
"epoch": 1.76,
"learning_rate": 2.0657276995305167e-05,
"loss": 0.7859,
"step": 117500
},
{
"epoch": 1.77,
"learning_rate": 2.0532414344221357e-05,
"loss": 0.7747,
"step": 118000
},
{
"epoch": 1.78,
"learning_rate": 2.040755169313755e-05,
"loss": 0.7859,
"step": 118500
},
{
"epoch": 1.78,
"learning_rate": 2.0282689042053744e-05,
"loss": 0.8023,
"step": 119000
},
{
"epoch": 1.79,
"learning_rate": 2.0157826390969934e-05,
"loss": 0.7787,
"step": 119500
},
{
"epoch": 1.8,
"learning_rate": 2.0032963739886127e-05,
"loss": 0.7812,
"step": 120000
},
{
"epoch": 1.81,
"learning_rate": 1.990810108880232e-05,
"loss": 0.783,
"step": 120500
},
{
"epoch": 1.81,
"learning_rate": 1.978323843771851e-05,
"loss": 0.7944,
"step": 121000
},
{
"epoch": 1.82,
"learning_rate": 1.9658375786634704e-05,
"loss": 0.772,
"step": 121500
},
{
"epoch": 1.83,
"learning_rate": 1.9533513135550897e-05,
"loss": 0.7822,
"step": 122000
},
{
"epoch": 1.84,
"learning_rate": 1.9408650484467087e-05,
"loss": 0.7723,
"step": 122500
},
{
"epoch": 1.84,
"learning_rate": 1.928378783338328e-05,
"loss": 0.7754,
"step": 123000
},
{
"epoch": 1.85,
"learning_rate": 1.9158925182299473e-05,
"loss": 0.7796,
"step": 123500
},
{
"epoch": 1.86,
"learning_rate": 1.9034062531215663e-05,
"loss": 0.7674,
"step": 124000
},
{
"epoch": 1.87,
"learning_rate": 1.8909199880131856e-05,
"loss": 0.7939,
"step": 124500
},
{
"epoch": 1.87,
"learning_rate": 1.878433722904805e-05,
"loss": 0.7741,
"step": 125000
},
{
"epoch": 1.88,
"learning_rate": 1.865947457796424e-05,
"loss": 0.7817,
"step": 125500
},
{
"epoch": 1.89,
"learning_rate": 1.8534611926880433e-05,
"loss": 0.7828,
"step": 126000
},
{
"epoch": 1.9,
"learning_rate": 1.8409749275796626e-05,
"loss": 0.7731,
"step": 126500
},
{
"epoch": 1.9,
"learning_rate": 1.8284886624712816e-05,
"loss": 0.7683,
"step": 127000
},
{
"epoch": 1.91,
"learning_rate": 1.816002397362901e-05,
"loss": 0.7797,
"step": 127500
},
{
"epoch": 1.92,
"learning_rate": 1.8035161322545203e-05,
"loss": 0.7831,
"step": 128000
},
{
"epoch": 1.93,
"learning_rate": 1.7910298671461393e-05,
"loss": 0.7736,
"step": 128500
},
{
"epoch": 1.93,
"learning_rate": 1.7785436020377586e-05,
"loss": 0.77,
"step": 129000
},
{
"epoch": 1.94,
"learning_rate": 1.766057336929378e-05,
"loss": 0.7739,
"step": 129500
},
{
"epoch": 1.95,
"learning_rate": 1.753571071820997e-05,
"loss": 0.7743,
"step": 130000
},
{
"epoch": 1.96,
"learning_rate": 1.7410848067126162e-05,
"loss": 0.7706,
"step": 130500
},
{
"epoch": 1.96,
"learning_rate": 1.7285985416042356e-05,
"loss": 0.7727,
"step": 131000
},
{
"epoch": 1.97,
"learning_rate": 1.7161122764958546e-05,
"loss": 0.7736,
"step": 131500
},
{
"epoch": 1.98,
"learning_rate": 1.703626011387474e-05,
"loss": 0.7618,
"step": 132000
},
{
"epoch": 1.99,
"learning_rate": 1.6911397462790932e-05,
"loss": 0.7685,
"step": 132500
},
{
"epoch": 1.99,
"learning_rate": 1.6786534811707122e-05,
"loss": 0.7821,
"step": 133000
},
{
"epoch": 2.0,
"learning_rate": 1.6661672160623315e-05,
"loss": 0.7712,
"step": 133500
},
{
"epoch": 2.01,
"learning_rate": 1.653680950953951e-05,
"loss": 0.6751,
"step": 134000
},
{
"epoch": 2.02,
"learning_rate": 1.64119468584557e-05,
"loss": 0.6709,
"step": 134500
},
{
"epoch": 2.02,
"learning_rate": 1.6287084207371892e-05,
"loss": 0.6627,
"step": 135000
},
{
"epoch": 2.03,
"learning_rate": 1.6162221556288085e-05,
"loss": 0.6725,
"step": 135500
},
{
"epoch": 2.04,
"learning_rate": 1.6037358905204275e-05,
"loss": 0.672,
"step": 136000
},
{
"epoch": 2.05,
"learning_rate": 1.5912496254120468e-05,
"loss": 0.6679,
"step": 136500
},
{
"epoch": 2.05,
"learning_rate": 1.578763360303666e-05,
"loss": 0.6662,
"step": 137000
},
{
"epoch": 2.06,
"learning_rate": 1.566277095195285e-05,
"loss": 0.673,
"step": 137500
},
{
"epoch": 2.07,
"learning_rate": 1.5537908300869045e-05,
"loss": 0.6771,
"step": 138000
},
{
"epoch": 2.08,
"learning_rate": 1.5413045649785238e-05,
"loss": 0.6617,
"step": 138500
},
{
"epoch": 2.08,
"learning_rate": 1.5288182998701428e-05,
"loss": 0.6749,
"step": 139000
},
{
"epoch": 2.09,
"learning_rate": 1.5163320347617621e-05,
"loss": 0.6715,
"step": 139500
},
{
"epoch": 2.1,
"learning_rate": 1.5038457696533815e-05,
"loss": 0.6622,
"step": 140000
},
{
"epoch": 2.11,
"learning_rate": 1.4913595045450004e-05,
"loss": 0.6756,
"step": 140500
},
{
"epoch": 2.11,
"learning_rate": 1.4788732394366198e-05,
"loss": 0.6785,
"step": 141000
},
{
"epoch": 2.12,
"learning_rate": 1.4663869743282391e-05,
"loss": 0.6577,
"step": 141500
},
{
"epoch": 2.13,
"learning_rate": 1.4539007092198581e-05,
"loss": 0.6629,
"step": 142000
},
{
"epoch": 2.14,
"learning_rate": 1.4414144441114774e-05,
"loss": 0.6707,
"step": 142500
},
{
"epoch": 2.14,
"learning_rate": 1.4289281790030967e-05,
"loss": 0.6603,
"step": 143000
},
{
"epoch": 2.15,
"learning_rate": 1.4164419138947157e-05,
"loss": 0.6583,
"step": 143500
},
{
"epoch": 2.16,
"learning_rate": 1.403955648786335e-05,
"loss": 0.6674,
"step": 144000
},
{
"epoch": 2.17,
"learning_rate": 1.3914693836779544e-05,
"loss": 0.6591,
"step": 144500
},
{
"epoch": 2.17,
"learning_rate": 1.3789831185695734e-05,
"loss": 0.6607,
"step": 145000
},
{
"epoch": 2.18,
"learning_rate": 1.3664968534611927e-05,
"loss": 0.6615,
"step": 145500
},
{
"epoch": 2.19,
"learning_rate": 1.354010588352812e-05,
"loss": 0.6635,
"step": 146000
},
{
"epoch": 2.2,
"learning_rate": 1.341524323244431e-05,
"loss": 0.6647,
"step": 146500
},
{
"epoch": 2.2,
"learning_rate": 1.3290380581360504e-05,
"loss": 0.6744,
"step": 147000
},
{
"epoch": 2.21,
"learning_rate": 1.3165517930276697e-05,
"loss": 0.6636,
"step": 147500
},
{
"epoch": 2.22,
"learning_rate": 1.3040655279192887e-05,
"loss": 0.6746,
"step": 148000
},
{
"epoch": 2.23,
"learning_rate": 1.291579262810908e-05,
"loss": 0.6613,
"step": 148500
},
{
"epoch": 2.23,
"learning_rate": 1.2790929977025273e-05,
"loss": 0.6655,
"step": 149000
},
{
"epoch": 2.24,
"learning_rate": 1.2666067325941463e-05,
"loss": 0.6717,
"step": 149500
},
{
"epoch": 2.25,
"learning_rate": 1.2541204674857657e-05,
"loss": 0.6482,
"step": 150000
},
{
"epoch": 2.26,
"learning_rate": 1.241634202377385e-05,
"loss": 0.655,
"step": 150500
},
{
"epoch": 2.26,
"learning_rate": 1.2291479372690041e-05,
"loss": 0.6633,
"step": 151000
},
{
"epoch": 2.27,
"learning_rate": 1.2166616721606233e-05,
"loss": 0.6691,
"step": 151500
},
{
"epoch": 2.28,
"learning_rate": 1.2041754070522426e-05,
"loss": 0.6635,
"step": 152000
},
{
"epoch": 2.28,
"learning_rate": 1.1916891419438618e-05,
"loss": 0.6657,
"step": 152500
},
{
"epoch": 2.29,
"learning_rate": 1.179202876835481e-05,
"loss": 0.662,
"step": 153000
},
{
"epoch": 2.3,
"learning_rate": 1.1667166117271003e-05,
"loss": 0.6477,
"step": 153500
},
{
"epoch": 2.31,
"learning_rate": 1.1542303466187194e-05,
"loss": 0.6649,
"step": 154000
},
{
"epoch": 2.31,
"learning_rate": 1.1417440815103386e-05,
"loss": 0.6545,
"step": 154500
},
{
"epoch": 2.32,
"learning_rate": 1.129257816401958e-05,
"loss": 0.6579,
"step": 155000
},
{
"epoch": 2.33,
"learning_rate": 1.116771551293577e-05,
"loss": 0.6722,
"step": 155500
},
{
"epoch": 2.34,
"learning_rate": 1.1042852861851962e-05,
"loss": 0.6594,
"step": 156000
},
{
"epoch": 2.34,
"learning_rate": 1.0917990210768156e-05,
"loss": 0.6604,
"step": 156500
},
{
"epoch": 2.35,
"learning_rate": 1.0793127559684347e-05,
"loss": 0.6547,
"step": 157000
},
{
"epoch": 2.36,
"learning_rate": 1.0668264908600539e-05,
"loss": 0.6678,
"step": 157500
},
{
"epoch": 2.37,
"learning_rate": 1.0543402257516732e-05,
"loss": 0.6547,
"step": 158000
},
{
"epoch": 2.37,
"learning_rate": 1.0418539606432924e-05,
"loss": 0.6576,
"step": 158500
},
{
"epoch": 2.38,
"learning_rate": 1.0293676955349115e-05,
"loss": 0.6642,
"step": 159000
},
{
"epoch": 2.39,
"learning_rate": 1.0168814304265309e-05,
"loss": 0.6558,
"step": 159500
},
{
"epoch": 2.4,
"learning_rate": 1.00439516531815e-05,
"loss": 0.6534,
"step": 160000
},
{
"epoch": 2.4,
"learning_rate": 9.919089002097694e-06,
"loss": 0.6602,
"step": 160500
},
{
"epoch": 2.41,
"learning_rate": 9.794226351013885e-06,
"loss": 0.6471,
"step": 161000
},
{
"epoch": 2.42,
"learning_rate": 9.669363699930077e-06,
"loss": 0.6636,
"step": 161500
},
{
"epoch": 2.43,
"learning_rate": 9.54450104884627e-06,
"loss": 0.6533,
"step": 162000
},
{
"epoch": 2.43,
"learning_rate": 9.419638397762462e-06,
"loss": 0.6559,
"step": 162500
},
{
"epoch": 2.44,
"learning_rate": 9.294775746678653e-06,
"loss": 0.6645,
"step": 163000
},
{
"epoch": 2.45,
"learning_rate": 9.169913095594847e-06,
"loss": 0.6521,
"step": 163500
},
{
"epoch": 2.46,
"learning_rate": 9.045050444511038e-06,
"loss": 0.6454,
"step": 164000
},
{
"epoch": 2.46,
"learning_rate": 8.92018779342723e-06,
"loss": 0.6482,
"step": 164500
},
{
"epoch": 2.47,
"learning_rate": 8.795325142343423e-06,
"loss": 0.6569,
"step": 165000
},
{
"epoch": 2.48,
"learning_rate": 8.670462491259615e-06,
"loss": 0.6583,
"step": 165500
},
{
"epoch": 2.49,
"learning_rate": 8.545599840175806e-06,
"loss": 0.6541,
"step": 166000
},
{
"epoch": 2.49,
"learning_rate": 8.420737189092e-06,
"loss": 0.649,
"step": 166500
},
{
"epoch": 2.5,
"learning_rate": 8.295874538008191e-06,
"loss": 0.6499,
"step": 167000
},
{
"epoch": 2.51,
"learning_rate": 8.171011886924383e-06,
"loss": 0.6679,
"step": 167500
},
{
"epoch": 2.52,
"learning_rate": 8.046149235840576e-06,
"loss": 0.6361,
"step": 168000
},
{
"epoch": 2.52,
"learning_rate": 7.921286584756768e-06,
"loss": 0.6512,
"step": 168500
},
{
"epoch": 2.53,
"learning_rate": 7.796423933672959e-06,
"loss": 0.6354,
"step": 169000
},
{
"epoch": 2.54,
"learning_rate": 7.671561282589152e-06,
"loss": 0.6456,
"step": 169500
},
{
"epoch": 2.55,
"learning_rate": 7.546698631505345e-06,
"loss": 0.6523,
"step": 170000
},
{
"epoch": 2.55,
"learning_rate": 7.4218359804215364e-06,
"loss": 0.6512,
"step": 170500
},
{
"epoch": 2.56,
"learning_rate": 7.29697332933773e-06,
"loss": 0.6402,
"step": 171000
},
{
"epoch": 2.57,
"learning_rate": 7.172110678253921e-06,
"loss": 0.6648,
"step": 171500
},
{
"epoch": 2.58,
"learning_rate": 7.047248027170113e-06,
"loss": 0.6544,
"step": 172000
},
{
"epoch": 2.58,
"learning_rate": 6.922385376086306e-06,
"loss": 0.6548,
"step": 172500
},
{
"epoch": 2.59,
"learning_rate": 6.797522725002498e-06,
"loss": 0.6477,
"step": 173000
},
{
"epoch": 2.6,
"learning_rate": 6.672660073918689e-06,
"loss": 0.64,
"step": 173500
},
{
"epoch": 2.61,
"learning_rate": 6.547797422834883e-06,
"loss": 0.6502,
"step": 174000
},
{
"epoch": 2.61,
"learning_rate": 6.422934771751074e-06,
"loss": 0.6518,
"step": 174500
},
{
"epoch": 2.62,
"learning_rate": 6.298072120667266e-06,
"loss": 0.6559,
"step": 175000
},
{
"epoch": 2.63,
"learning_rate": 6.173209469583458e-06,
"loss": 0.6492,
"step": 175500
},
{
"epoch": 2.64,
"learning_rate": 6.048346818499651e-06,
"loss": 0.6395,
"step": 176000
},
{
"epoch": 2.64,
"learning_rate": 5.923484167415843e-06,
"loss": 0.6527,
"step": 176500
},
{
"epoch": 2.65,
"learning_rate": 5.798621516332036e-06,
"loss": 0.6505,
"step": 177000
},
{
"epoch": 2.66,
"learning_rate": 5.673758865248227e-06,
"loss": 0.6385,
"step": 177500
},
{
"epoch": 2.67,
"learning_rate": 5.54889621416442e-06,
"loss": 0.6353,
"step": 178000
},
{
"epoch": 2.67,
"learning_rate": 5.424033563080612e-06,
"loss": 0.6428,
"step": 178500
},
{
"epoch": 2.68,
"learning_rate": 5.299170911996804e-06,
"loss": 0.6349,
"step": 179000
},
{
"epoch": 2.69,
"learning_rate": 5.174308260912996e-06,
"loss": 0.6387,
"step": 179500
},
{
"epoch": 2.7,
"learning_rate": 5.0494456098291886e-06,
"loss": 0.6405,
"step": 180000
},
{
"epoch": 2.7,
"learning_rate": 4.92458295874538e-06,
"loss": 0.6471,
"step": 180500
},
{
"epoch": 2.71,
"learning_rate": 4.799720307661573e-06,
"loss": 0.644,
"step": 181000
},
{
"epoch": 2.72,
"learning_rate": 4.674857656577765e-06,
"loss": 0.6441,
"step": 181500
},
{
"epoch": 2.73,
"learning_rate": 4.549995005493957e-06,
"loss": 0.6438,
"step": 182000
},
{
"epoch": 2.73,
"learning_rate": 4.425132354410149e-06,
"loss": 0.6464,
"step": 182500
},
{
"epoch": 2.74,
"learning_rate": 4.3002697033263415e-06,
"loss": 0.6328,
"step": 183000
},
{
"epoch": 2.75,
"learning_rate": 4.175407052242533e-06,
"loss": 0.6373,
"step": 183500
},
{
"epoch": 2.76,
"learning_rate": 4.0505444011587255e-06,
"loss": 0.6333,
"step": 184000
},
{
"epoch": 2.76,
"learning_rate": 3.925681750074918e-06,
"loss": 0.6367,
"step": 184500
},
{
"epoch": 2.77,
"learning_rate": 3.8008190989911096e-06,
"loss": 0.635,
"step": 185000
},
{
"epoch": 2.78,
"learning_rate": 3.675956447907302e-06,
"loss": 0.6364,
"step": 185500
},
{
"epoch": 2.79,
"learning_rate": 3.5510937968234945e-06,
"loss": 0.6384,
"step": 186000
},
{
"epoch": 2.79,
"learning_rate": 3.426231145739686e-06,
"loss": 0.626,
"step": 186500
},
{
"epoch": 2.8,
"learning_rate": 3.3013684946558785e-06,
"loss": 0.6335,
"step": 187000
},
{
"epoch": 2.81,
"learning_rate": 3.176505843572071e-06,
"loss": 0.6423,
"step": 187500
},
{
"epoch": 2.82,
"learning_rate": 3.051643192488263e-06,
"loss": 0.6466,
"step": 188000
},
{
"epoch": 2.82,
"learning_rate": 2.926780541404455e-06,
"loss": 0.6326,
"step": 188500
},
{
"epoch": 2.83,
"learning_rate": 2.8019178903206474e-06,
"loss": 0.6382,
"step": 189000
},
{
"epoch": 2.84,
"learning_rate": 2.6770552392368394e-06,
"loss": 0.638,
"step": 189500
},
{
"epoch": 2.85,
"learning_rate": 2.552192588153032e-06,
"loss": 0.6279,
"step": 190000
},
{
"epoch": 2.85,
"learning_rate": 2.427329937069224e-06,
"loss": 0.6362,
"step": 190500
},
{
"epoch": 2.86,
"learning_rate": 2.302467285985416e-06,
"loss": 0.6331,
"step": 191000
},
{
"epoch": 2.87,
"learning_rate": 2.1776046349016083e-06,
"loss": 0.6179,
"step": 191500
},
{
"epoch": 2.88,
"learning_rate": 2.0527419838178003e-06,
"loss": 0.6424,
"step": 192000
},
{
"epoch": 2.88,
"learning_rate": 1.927879332733993e-06,
"loss": 0.6323,
"step": 192500
},
{
"epoch": 2.89,
"learning_rate": 1.803016681650185e-06,
"loss": 0.6359,
"step": 193000
},
{
"epoch": 2.9,
"learning_rate": 1.678154030566377e-06,
"loss": 0.6377,
"step": 193500
},
{
"epoch": 2.91,
"learning_rate": 1.5532913794825693e-06,
"loss": 0.6346,
"step": 194000
},
{
"epoch": 2.91,
"learning_rate": 1.4284287283987615e-06,
"loss": 0.6344,
"step": 194500
},
{
"epoch": 2.92,
"learning_rate": 1.3035660773149535e-06,
"loss": 0.627,
"step": 195000
},
{
"epoch": 2.93,
"learning_rate": 1.1787034262311457e-06,
"loss": 0.6335,
"step": 195500
},
{
"epoch": 2.94,
"learning_rate": 1.053840775147338e-06,
"loss": 0.6388,
"step": 196000
},
{
"epoch": 2.94,
"learning_rate": 9.289781240635301e-07,
"loss": 0.632,
"step": 196500
},
{
"epoch": 2.95,
"learning_rate": 8.041154729797223e-07,
"loss": 0.6351,
"step": 197000
},
{
"epoch": 2.96,
"learning_rate": 6.792528218959145e-07,
"loss": 0.6242,
"step": 197500
},
{
"epoch": 2.97,
"learning_rate": 5.543901708121067e-07,
"loss": 0.6258,
"step": 198000
},
{
"epoch": 2.97,
"learning_rate": 4.295275197282989e-07,
"loss": 0.6375,
"step": 198500
},
{
"epoch": 2.98,
"learning_rate": 3.0466486864449107e-07,
"loss": 0.6256,
"step": 199000
},
{
"epoch": 2.99,
"learning_rate": 1.7980221756068325e-07,
"loss": 0.6334,
"step": 199500
},
{
"epoch": 3.0,
"learning_rate": 5.493956647687544e-08,
"loss": 0.6239,
"step": 200000
},
{
"epoch": 3.0,
"step": 200220,
"total_flos": 2.2322905274597376e+17,
"train_loss": 0.8990719857052506,
"train_runtime": 61304.5877,
"train_samples_per_second": 32.66,
"train_steps_per_second": 3.266
}
],
"max_steps": 200220,
"num_train_epochs": 3,
"total_flos": 2.2322905274597376e+17,
"trial_name": null,
"trial_params": null
}