BertHarmon / trainer_state.json
squish's picture
First version
4858446
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 100.0,
"global_step": 203000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.25,
"learning_rate": 4.9876847290640394e-05,
"loss": 2.3164,
"step": 500
},
{
"epoch": 0.49,
"learning_rate": 4.975369458128079e-05,
"loss": 2.0051,
"step": 1000
},
{
"epoch": 0.74,
"learning_rate": 4.9630541871921185e-05,
"loss": 1.8659,
"step": 1500
},
{
"epoch": 0.99,
"learning_rate": 4.950738916256158e-05,
"loss": 1.7825,
"step": 2000
},
{
"epoch": 1.23,
"learning_rate": 4.938423645320197e-05,
"loss": 1.723,
"step": 2500
},
{
"epoch": 1.48,
"learning_rate": 4.926108374384237e-05,
"loss": 1.6874,
"step": 3000
},
{
"epoch": 1.72,
"learning_rate": 4.913793103448276e-05,
"loss": 1.6599,
"step": 3500
},
{
"epoch": 1.97,
"learning_rate": 4.901477832512316e-05,
"loss": 1.631,
"step": 4000
},
{
"epoch": 2.22,
"learning_rate": 4.889162561576355e-05,
"loss": 1.6082,
"step": 4500
},
{
"epoch": 2.46,
"learning_rate": 4.876847290640394e-05,
"loss": 1.5941,
"step": 5000
},
{
"epoch": 2.71,
"learning_rate": 4.8645320197044334e-05,
"loss": 1.5733,
"step": 5500
},
{
"epoch": 2.96,
"learning_rate": 4.852216748768473e-05,
"loss": 1.5622,
"step": 6000
},
{
"epoch": 3.2,
"learning_rate": 4.8399014778325125e-05,
"loss": 1.5521,
"step": 6500
},
{
"epoch": 3.45,
"learning_rate": 4.827586206896552e-05,
"loss": 1.54,
"step": 7000
},
{
"epoch": 3.69,
"learning_rate": 4.8152709359605915e-05,
"loss": 1.5182,
"step": 7500
},
{
"epoch": 3.94,
"learning_rate": 4.802955665024631e-05,
"loss": 1.5135,
"step": 8000
},
{
"epoch": 4.19,
"learning_rate": 4.79064039408867e-05,
"loss": 1.5018,
"step": 8500
},
{
"epoch": 4.43,
"learning_rate": 4.77832512315271e-05,
"loss": 1.4859,
"step": 9000
},
{
"epoch": 4.68,
"learning_rate": 4.766009852216749e-05,
"loss": 1.4816,
"step": 9500
},
{
"epoch": 4.93,
"learning_rate": 4.753694581280788e-05,
"loss": 1.4653,
"step": 10000
},
{
"epoch": 5.17,
"learning_rate": 4.741379310344828e-05,
"loss": 1.4564,
"step": 10500
},
{
"epoch": 5.42,
"learning_rate": 4.729064039408867e-05,
"loss": 1.4477,
"step": 11000
},
{
"epoch": 5.67,
"learning_rate": 4.7167487684729064e-05,
"loss": 1.4335,
"step": 11500
},
{
"epoch": 5.91,
"learning_rate": 4.7044334975369456e-05,
"loss": 1.43,
"step": 12000
},
{
"epoch": 6.16,
"learning_rate": 4.6921182266009855e-05,
"loss": 1.418,
"step": 12500
},
{
"epoch": 6.4,
"learning_rate": 4.679802955665025e-05,
"loss": 1.407,
"step": 13000
},
{
"epoch": 6.65,
"learning_rate": 4.6674876847290645e-05,
"loss": 1.4022,
"step": 13500
},
{
"epoch": 6.9,
"learning_rate": 4.655172413793104e-05,
"loss": 1.3932,
"step": 14000
},
{
"epoch": 7.14,
"learning_rate": 4.642857142857143e-05,
"loss": 1.3844,
"step": 14500
},
{
"epoch": 7.39,
"learning_rate": 4.630541871921182e-05,
"loss": 1.3764,
"step": 15000
},
{
"epoch": 7.64,
"learning_rate": 4.618226600985222e-05,
"loss": 1.3708,
"step": 15500
},
{
"epoch": 7.88,
"learning_rate": 4.605911330049261e-05,
"loss": 1.3614,
"step": 16000
},
{
"epoch": 8.13,
"learning_rate": 4.593596059113301e-05,
"loss": 1.3515,
"step": 16500
},
{
"epoch": 8.37,
"learning_rate": 4.58128078817734e-05,
"loss": 1.3433,
"step": 17000
},
{
"epoch": 8.62,
"learning_rate": 4.5689655172413794e-05,
"loss": 1.3365,
"step": 17500
},
{
"epoch": 8.87,
"learning_rate": 4.5566502463054186e-05,
"loss": 1.332,
"step": 18000
},
{
"epoch": 9.11,
"learning_rate": 4.544334975369458e-05,
"loss": 1.328,
"step": 18500
},
{
"epoch": 9.36,
"learning_rate": 4.532019704433498e-05,
"loss": 1.3269,
"step": 19000
},
{
"epoch": 9.61,
"learning_rate": 4.519704433497537e-05,
"loss": 1.3189,
"step": 19500
},
{
"epoch": 9.85,
"learning_rate": 4.507389162561577e-05,
"loss": 1.3119,
"step": 20000
},
{
"epoch": 10.1,
"learning_rate": 4.495073891625616e-05,
"loss": 1.3072,
"step": 20500
},
{
"epoch": 10.34,
"learning_rate": 4.482758620689655e-05,
"loss": 1.3027,
"step": 21000
},
{
"epoch": 10.59,
"learning_rate": 4.4704433497536943e-05,
"loss": 1.3003,
"step": 21500
},
{
"epoch": 10.84,
"learning_rate": 4.458128078817734e-05,
"loss": 1.2961,
"step": 22000
},
{
"epoch": 11.08,
"learning_rate": 4.4458128078817734e-05,
"loss": 1.2894,
"step": 22500
},
{
"epoch": 11.33,
"learning_rate": 4.433497536945813e-05,
"loss": 1.291,
"step": 23000
},
{
"epoch": 11.58,
"learning_rate": 4.4211822660098525e-05,
"loss": 1.2802,
"step": 23500
},
{
"epoch": 11.82,
"learning_rate": 4.408866995073892e-05,
"loss": 1.2777,
"step": 24000
},
{
"epoch": 12.07,
"learning_rate": 4.396551724137931e-05,
"loss": 1.2764,
"step": 24500
},
{
"epoch": 12.32,
"learning_rate": 4.384236453201971e-05,
"loss": 1.2685,
"step": 25000
},
{
"epoch": 12.56,
"learning_rate": 4.37192118226601e-05,
"loss": 1.2678,
"step": 25500
},
{
"epoch": 12.81,
"learning_rate": 4.35960591133005e-05,
"loss": 1.2634,
"step": 26000
},
{
"epoch": 13.05,
"learning_rate": 4.347290640394089e-05,
"loss": 1.2644,
"step": 26500
},
{
"epoch": 13.3,
"learning_rate": 4.334975369458129e-05,
"loss": 1.2507,
"step": 27000
},
{
"epoch": 13.55,
"learning_rate": 4.3226600985221674e-05,
"loss": 1.2557,
"step": 27500
},
{
"epoch": 13.79,
"learning_rate": 4.3103448275862066e-05,
"loss": 1.248,
"step": 28000
},
{
"epoch": 14.04,
"learning_rate": 4.2980295566502464e-05,
"loss": 1.249,
"step": 28500
},
{
"epoch": 14.29,
"learning_rate": 4.2857142857142856e-05,
"loss": 1.2395,
"step": 29000
},
{
"epoch": 14.53,
"learning_rate": 4.2733990147783255e-05,
"loss": 1.2419,
"step": 29500
},
{
"epoch": 14.78,
"learning_rate": 4.261083743842365e-05,
"loss": 1.2405,
"step": 30000
},
{
"epoch": 15.02,
"learning_rate": 4.2487684729064046e-05,
"loss": 1.2395,
"step": 30500
},
{
"epoch": 15.27,
"learning_rate": 4.236453201970443e-05,
"loss": 1.2332,
"step": 31000
},
{
"epoch": 15.52,
"learning_rate": 4.224137931034483e-05,
"loss": 1.229,
"step": 31500
},
{
"epoch": 15.76,
"learning_rate": 4.211822660098522e-05,
"loss": 1.2324,
"step": 32000
},
{
"epoch": 16.01,
"learning_rate": 4.199507389162562e-05,
"loss": 1.2274,
"step": 32500
},
{
"epoch": 16.26,
"learning_rate": 4.187192118226601e-05,
"loss": 1.2168,
"step": 33000
},
{
"epoch": 16.5,
"learning_rate": 4.174876847290641e-05,
"loss": 1.2198,
"step": 33500
},
{
"epoch": 16.75,
"learning_rate": 4.1625615763546796e-05,
"loss": 1.2205,
"step": 34000
},
{
"epoch": 17.0,
"learning_rate": 4.1502463054187195e-05,
"loss": 1.2136,
"step": 34500
},
{
"epoch": 17.24,
"learning_rate": 4.1379310344827587e-05,
"loss": 1.2102,
"step": 35000
},
{
"epoch": 17.49,
"learning_rate": 4.1256157635467985e-05,
"loss": 1.2102,
"step": 35500
},
{
"epoch": 17.73,
"learning_rate": 4.113300492610838e-05,
"loss": 1.2068,
"step": 36000
},
{
"epoch": 17.98,
"learning_rate": 4.1009852216748776e-05,
"loss": 1.204,
"step": 36500
},
{
"epoch": 18.23,
"learning_rate": 4.088669950738917e-05,
"loss": 1.2014,
"step": 37000
},
{
"epoch": 18.47,
"learning_rate": 4.076354679802955e-05,
"loss": 1.1988,
"step": 37500
},
{
"epoch": 18.72,
"learning_rate": 4.064039408866995e-05,
"loss": 1.1986,
"step": 38000
},
{
"epoch": 18.97,
"learning_rate": 4.0517241379310344e-05,
"loss": 1.1939,
"step": 38500
},
{
"epoch": 19.21,
"learning_rate": 4.039408866995074e-05,
"loss": 1.1923,
"step": 39000
},
{
"epoch": 19.46,
"learning_rate": 4.0270935960591134e-05,
"loss": 1.1919,
"step": 39500
},
{
"epoch": 19.7,
"learning_rate": 4.014778325123153e-05,
"loss": 1.1883,
"step": 40000
},
{
"epoch": 19.95,
"learning_rate": 4.002463054187192e-05,
"loss": 1.1878,
"step": 40500
},
{
"epoch": 20.2,
"learning_rate": 3.990147783251232e-05,
"loss": 1.1851,
"step": 41000
},
{
"epoch": 20.44,
"learning_rate": 3.977832512315271e-05,
"loss": 1.184,
"step": 41500
},
{
"epoch": 20.69,
"learning_rate": 3.965517241379311e-05,
"loss": 1.1791,
"step": 42000
},
{
"epoch": 20.94,
"learning_rate": 3.95320197044335e-05,
"loss": 1.1833,
"step": 42500
},
{
"epoch": 21.18,
"learning_rate": 3.94088669950739e-05,
"loss": 1.1757,
"step": 43000
},
{
"epoch": 21.43,
"learning_rate": 3.928571428571429e-05,
"loss": 1.176,
"step": 43500
},
{
"epoch": 21.67,
"learning_rate": 3.916256157635468e-05,
"loss": 1.1756,
"step": 44000
},
{
"epoch": 21.92,
"learning_rate": 3.9039408866995074e-05,
"loss": 1.1698,
"step": 44500
},
{
"epoch": 22.17,
"learning_rate": 3.891625615763547e-05,
"loss": 1.1689,
"step": 45000
},
{
"epoch": 22.41,
"learning_rate": 3.8793103448275865e-05,
"loss": 1.1655,
"step": 45500
},
{
"epoch": 22.66,
"learning_rate": 3.866995073891626e-05,
"loss": 1.1627,
"step": 46000
},
{
"epoch": 22.91,
"learning_rate": 3.8546798029556655e-05,
"loss": 1.1638,
"step": 46500
},
{
"epoch": 23.15,
"learning_rate": 3.842364532019704e-05,
"loss": 1.1609,
"step": 47000
},
{
"epoch": 23.4,
"learning_rate": 3.830049261083744e-05,
"loss": 1.1555,
"step": 47500
},
{
"epoch": 23.65,
"learning_rate": 3.817733990147783e-05,
"loss": 1.1593,
"step": 48000
},
{
"epoch": 23.89,
"learning_rate": 3.805418719211823e-05,
"loss": 1.1501,
"step": 48500
},
{
"epoch": 24.14,
"learning_rate": 3.793103448275862e-05,
"loss": 1.152,
"step": 49000
},
{
"epoch": 24.38,
"learning_rate": 3.780788177339902e-05,
"loss": 1.1482,
"step": 49500
},
{
"epoch": 24.63,
"learning_rate": 3.768472906403941e-05,
"loss": 1.1492,
"step": 50000
},
{
"epoch": 24.88,
"learning_rate": 3.7561576354679804e-05,
"loss": 1.1495,
"step": 50500
},
{
"epoch": 25.12,
"learning_rate": 3.7438423645320196e-05,
"loss": 1.1452,
"step": 51000
},
{
"epoch": 25.37,
"learning_rate": 3.7315270935960595e-05,
"loss": 1.1436,
"step": 51500
},
{
"epoch": 25.62,
"learning_rate": 3.719211822660099e-05,
"loss": 1.1429,
"step": 52000
},
{
"epoch": 25.86,
"learning_rate": 3.7068965517241385e-05,
"loss": 1.1409,
"step": 52500
},
{
"epoch": 26.11,
"learning_rate": 3.694581280788178e-05,
"loss": 1.1393,
"step": 53000
},
{
"epoch": 26.35,
"learning_rate": 3.682266009852217e-05,
"loss": 1.1333,
"step": 53500
},
{
"epoch": 26.6,
"learning_rate": 3.669950738916256e-05,
"loss": 1.1365,
"step": 54000
},
{
"epoch": 26.85,
"learning_rate": 3.657635467980296e-05,
"loss": 1.1314,
"step": 54500
},
{
"epoch": 27.09,
"learning_rate": 3.645320197044335e-05,
"loss": 1.1284,
"step": 55000
},
{
"epoch": 27.34,
"learning_rate": 3.6330049261083744e-05,
"loss": 1.1247,
"step": 55500
},
{
"epoch": 27.59,
"learning_rate": 3.620689655172414e-05,
"loss": 1.1283,
"step": 56000
},
{
"epoch": 27.83,
"learning_rate": 3.6083743842364534e-05,
"loss": 1.1269,
"step": 56500
},
{
"epoch": 28.08,
"learning_rate": 3.5960591133004926e-05,
"loss": 1.1239,
"step": 57000
},
{
"epoch": 28.33,
"learning_rate": 3.583743842364532e-05,
"loss": 1.1199,
"step": 57500
},
{
"epoch": 28.57,
"learning_rate": 3.571428571428572e-05,
"loss": 1.12,
"step": 58000
},
{
"epoch": 28.82,
"learning_rate": 3.559113300492611e-05,
"loss": 1.1196,
"step": 58500
},
{
"epoch": 29.06,
"learning_rate": 3.546798029556651e-05,
"loss": 1.1197,
"step": 59000
},
{
"epoch": 29.31,
"learning_rate": 3.53448275862069e-05,
"loss": 1.1121,
"step": 59500
},
{
"epoch": 29.56,
"learning_rate": 3.522167487684729e-05,
"loss": 1.1151,
"step": 60000
},
{
"epoch": 29.8,
"learning_rate": 3.5098522167487683e-05,
"loss": 1.1088,
"step": 60500
},
{
"epoch": 30.05,
"learning_rate": 3.497536945812808e-05,
"loss": 1.1123,
"step": 61000
},
{
"epoch": 30.3,
"learning_rate": 3.4852216748768474e-05,
"loss": 1.1064,
"step": 61500
},
{
"epoch": 30.54,
"learning_rate": 3.472906403940887e-05,
"loss": 1.1125,
"step": 62000
},
{
"epoch": 30.79,
"learning_rate": 3.4605911330049265e-05,
"loss": 1.1016,
"step": 62500
},
{
"epoch": 31.03,
"learning_rate": 3.4482758620689657e-05,
"loss": 1.1071,
"step": 63000
},
{
"epoch": 31.28,
"learning_rate": 3.435960591133005e-05,
"loss": 1.1043,
"step": 63500
},
{
"epoch": 31.53,
"learning_rate": 3.423645320197045e-05,
"loss": 1.0966,
"step": 64000
},
{
"epoch": 31.77,
"learning_rate": 3.411330049261084e-05,
"loss": 1.0971,
"step": 64500
},
{
"epoch": 32.02,
"learning_rate": 3.399014778325123e-05,
"loss": 1.0971,
"step": 65000
},
{
"epoch": 32.27,
"learning_rate": 3.386699507389163e-05,
"loss": 1.093,
"step": 65500
},
{
"epoch": 32.51,
"learning_rate": 3.374384236453202e-05,
"loss": 1.0926,
"step": 66000
},
{
"epoch": 32.76,
"learning_rate": 3.3620689655172414e-05,
"loss": 1.0959,
"step": 66500
},
{
"epoch": 33.0,
"learning_rate": 3.3497536945812806e-05,
"loss": 1.0963,
"step": 67000
},
{
"epoch": 33.25,
"learning_rate": 3.3374384236453204e-05,
"loss": 1.0886,
"step": 67500
},
{
"epoch": 33.5,
"learning_rate": 3.3251231527093596e-05,
"loss": 1.0844,
"step": 68000
},
{
"epoch": 33.74,
"learning_rate": 3.3128078817733995e-05,
"loss": 1.0889,
"step": 68500
},
{
"epoch": 33.99,
"learning_rate": 3.300492610837439e-05,
"loss": 1.0898,
"step": 69000
},
{
"epoch": 34.24,
"learning_rate": 3.288177339901478e-05,
"loss": 1.0824,
"step": 69500
},
{
"epoch": 34.48,
"learning_rate": 3.275862068965517e-05,
"loss": 1.0871,
"step": 70000
},
{
"epoch": 34.73,
"learning_rate": 3.263546798029557e-05,
"loss": 1.0813,
"step": 70500
},
{
"epoch": 34.98,
"learning_rate": 3.251231527093596e-05,
"loss": 1.079,
"step": 71000
},
{
"epoch": 35.22,
"learning_rate": 3.238916256157636e-05,
"loss": 1.0745,
"step": 71500
},
{
"epoch": 35.47,
"learning_rate": 3.226600985221675e-05,
"loss": 1.0752,
"step": 72000
},
{
"epoch": 35.71,
"learning_rate": 3.2142857142857144e-05,
"loss": 1.0769,
"step": 72500
},
{
"epoch": 35.96,
"learning_rate": 3.2019704433497536e-05,
"loss": 1.0739,
"step": 73000
},
{
"epoch": 36.21,
"learning_rate": 3.1896551724137935e-05,
"loss": 1.071,
"step": 73500
},
{
"epoch": 36.45,
"learning_rate": 3.1773399014778326e-05,
"loss": 1.0699,
"step": 74000
},
{
"epoch": 36.7,
"learning_rate": 3.165024630541872e-05,
"loss": 1.0685,
"step": 74500
},
{
"epoch": 36.95,
"learning_rate": 3.152709359605912e-05,
"loss": 1.0708,
"step": 75000
},
{
"epoch": 37.19,
"learning_rate": 3.140394088669951e-05,
"loss": 1.0636,
"step": 75500
},
{
"epoch": 37.44,
"learning_rate": 3.12807881773399e-05,
"loss": 1.0616,
"step": 76000
},
{
"epoch": 37.68,
"learning_rate": 3.115763546798029e-05,
"loss": 1.0658,
"step": 76500
},
{
"epoch": 37.93,
"learning_rate": 3.103448275862069e-05,
"loss": 1.0651,
"step": 77000
},
{
"epoch": 38.18,
"learning_rate": 3.0911330049261084e-05,
"loss": 1.0605,
"step": 77500
},
{
"epoch": 38.42,
"learning_rate": 3.078817733990148e-05,
"loss": 1.0576,
"step": 78000
},
{
"epoch": 38.67,
"learning_rate": 3.0665024630541874e-05,
"loss": 1.0626,
"step": 78500
},
{
"epoch": 38.92,
"learning_rate": 3.0541871921182266e-05,
"loss": 1.0574,
"step": 79000
},
{
"epoch": 39.16,
"learning_rate": 3.041871921182266e-05,
"loss": 1.0521,
"step": 79500
},
{
"epoch": 39.41,
"learning_rate": 3.0295566502463057e-05,
"loss": 1.0514,
"step": 80000
},
{
"epoch": 39.66,
"learning_rate": 3.017241379310345e-05,
"loss": 1.0535,
"step": 80500
},
{
"epoch": 39.9,
"learning_rate": 3.0049261083743847e-05,
"loss": 1.0526,
"step": 81000
},
{
"epoch": 40.15,
"learning_rate": 2.9926108374384236e-05,
"loss": 1.0499,
"step": 81500
},
{
"epoch": 40.39,
"learning_rate": 2.9802955665024635e-05,
"loss": 1.0482,
"step": 82000
},
{
"epoch": 40.64,
"learning_rate": 2.9679802955665027e-05,
"loss": 1.0472,
"step": 82500
},
{
"epoch": 40.89,
"learning_rate": 2.9556650246305422e-05,
"loss": 1.0486,
"step": 83000
},
{
"epoch": 41.13,
"learning_rate": 2.9433497536945814e-05,
"loss": 1.0422,
"step": 83500
},
{
"epoch": 41.38,
"learning_rate": 2.9310344827586206e-05,
"loss": 1.0435,
"step": 84000
},
{
"epoch": 41.63,
"learning_rate": 2.9187192118226604e-05,
"loss": 1.0405,
"step": 84500
},
{
"epoch": 41.87,
"learning_rate": 2.9064039408866993e-05,
"loss": 1.0426,
"step": 85000
},
{
"epoch": 42.12,
"learning_rate": 2.894088669950739e-05,
"loss": 1.0426,
"step": 85500
},
{
"epoch": 42.36,
"learning_rate": 2.8817733990147784e-05,
"loss": 1.0376,
"step": 86000
},
{
"epoch": 42.61,
"learning_rate": 2.869458128078818e-05,
"loss": 1.0333,
"step": 86500
},
{
"epoch": 42.86,
"learning_rate": 2.857142857142857e-05,
"loss": 1.0337,
"step": 87000
},
{
"epoch": 43.1,
"learning_rate": 2.844827586206897e-05,
"loss": 1.0322,
"step": 87500
},
{
"epoch": 43.35,
"learning_rate": 2.8325123152709358e-05,
"loss": 1.0307,
"step": 88000
},
{
"epoch": 43.6,
"learning_rate": 2.8201970443349757e-05,
"loss": 1.0311,
"step": 88500
},
{
"epoch": 43.84,
"learning_rate": 2.807881773399015e-05,
"loss": 1.0299,
"step": 89000
},
{
"epoch": 44.09,
"learning_rate": 2.7955665024630544e-05,
"loss": 1.0279,
"step": 89500
},
{
"epoch": 44.33,
"learning_rate": 2.7832512315270936e-05,
"loss": 1.0219,
"step": 90000
},
{
"epoch": 44.58,
"learning_rate": 2.7709359605911335e-05,
"loss": 1.0234,
"step": 90500
},
{
"epoch": 44.83,
"learning_rate": 2.7586206896551727e-05,
"loss": 1.0282,
"step": 91000
},
{
"epoch": 45.07,
"learning_rate": 2.7463054187192122e-05,
"loss": 1.0206,
"step": 91500
},
{
"epoch": 45.32,
"learning_rate": 2.7339901477832514e-05,
"loss": 1.0204,
"step": 92000
},
{
"epoch": 45.57,
"learning_rate": 2.7216748768472906e-05,
"loss": 1.0211,
"step": 92500
},
{
"epoch": 45.81,
"learning_rate": 2.70935960591133e-05,
"loss": 1.0212,
"step": 93000
},
{
"epoch": 46.06,
"learning_rate": 2.6970443349753693e-05,
"loss": 1.0124,
"step": 93500
},
{
"epoch": 46.31,
"learning_rate": 2.6847290640394092e-05,
"loss": 1.0156,
"step": 94000
},
{
"epoch": 46.55,
"learning_rate": 2.672413793103448e-05,
"loss": 1.0127,
"step": 94500
},
{
"epoch": 46.8,
"learning_rate": 2.660098522167488e-05,
"loss": 1.016,
"step": 95000
},
{
"epoch": 47.04,
"learning_rate": 2.647783251231527e-05,
"loss": 1.0183,
"step": 95500
},
{
"epoch": 47.29,
"learning_rate": 2.6354679802955666e-05,
"loss": 1.014,
"step": 96000
},
{
"epoch": 47.54,
"learning_rate": 2.6231527093596058e-05,
"loss": 1.01,
"step": 96500
},
{
"epoch": 47.78,
"learning_rate": 2.6108374384236457e-05,
"loss": 1.0105,
"step": 97000
},
{
"epoch": 48.03,
"learning_rate": 2.598522167487685e-05,
"loss": 1.0096,
"step": 97500
},
{
"epoch": 48.28,
"learning_rate": 2.5862068965517244e-05,
"loss": 1.0056,
"step": 98000
},
{
"epoch": 48.52,
"learning_rate": 2.5738916256157636e-05,
"loss": 1.0064,
"step": 98500
},
{
"epoch": 48.77,
"learning_rate": 2.561576354679803e-05,
"loss": 1.0081,
"step": 99000
},
{
"epoch": 49.01,
"learning_rate": 2.5492610837438423e-05,
"loss": 1.0014,
"step": 99500
},
{
"epoch": 49.26,
"learning_rate": 2.5369458128078822e-05,
"loss": 1.0005,
"step": 100000
},
{
"epoch": 49.51,
"learning_rate": 2.5246305418719214e-05,
"loss": 0.9986,
"step": 100500
},
{
"epoch": 49.75,
"learning_rate": 2.512315270935961e-05,
"loss": 1.003,
"step": 101000
},
{
"epoch": 50.0,
"learning_rate": 2.5e-05,
"loss": 1.0022,
"step": 101500
},
{
"epoch": 50.25,
"learning_rate": 2.4876847290640397e-05,
"loss": 0.9967,
"step": 102000
},
{
"epoch": 50.49,
"learning_rate": 2.475369458128079e-05,
"loss": 0.9949,
"step": 102500
},
{
"epoch": 50.74,
"learning_rate": 2.4630541871921184e-05,
"loss": 0.9946,
"step": 103000
},
{
"epoch": 50.99,
"learning_rate": 2.450738916256158e-05,
"loss": 0.9998,
"step": 103500
},
{
"epoch": 51.23,
"learning_rate": 2.438423645320197e-05,
"loss": 0.9916,
"step": 104000
},
{
"epoch": 51.48,
"learning_rate": 2.4261083743842366e-05,
"loss": 0.9964,
"step": 104500
},
{
"epoch": 51.72,
"learning_rate": 2.413793103448276e-05,
"loss": 0.9878,
"step": 105000
},
{
"epoch": 51.97,
"learning_rate": 2.4014778325123154e-05,
"loss": 0.9953,
"step": 105500
},
{
"epoch": 52.22,
"learning_rate": 2.389162561576355e-05,
"loss": 0.9853,
"step": 106000
},
{
"epoch": 52.46,
"learning_rate": 2.376847290640394e-05,
"loss": 0.9878,
"step": 106500
},
{
"epoch": 52.71,
"learning_rate": 2.3645320197044336e-05,
"loss": 0.987,
"step": 107000
},
{
"epoch": 52.96,
"learning_rate": 2.3522167487684728e-05,
"loss": 0.9866,
"step": 107500
},
{
"epoch": 53.2,
"learning_rate": 2.3399014778325123e-05,
"loss": 0.9821,
"step": 108000
},
{
"epoch": 53.45,
"learning_rate": 2.327586206896552e-05,
"loss": 0.9826,
"step": 108500
},
{
"epoch": 53.69,
"learning_rate": 2.315270935960591e-05,
"loss": 0.9806,
"step": 109000
},
{
"epoch": 53.94,
"learning_rate": 2.3029556650246306e-05,
"loss": 0.9836,
"step": 109500
},
{
"epoch": 54.19,
"learning_rate": 2.29064039408867e-05,
"loss": 0.9807,
"step": 110000
},
{
"epoch": 54.43,
"learning_rate": 2.2783251231527093e-05,
"loss": 0.9754,
"step": 110500
},
{
"epoch": 54.68,
"learning_rate": 2.266009852216749e-05,
"loss": 0.9767,
"step": 111000
},
{
"epoch": 54.93,
"learning_rate": 2.2536945812807884e-05,
"loss": 0.9777,
"step": 111500
},
{
"epoch": 55.17,
"learning_rate": 2.2413793103448276e-05,
"loss": 0.9746,
"step": 112000
},
{
"epoch": 55.42,
"learning_rate": 2.229064039408867e-05,
"loss": 0.974,
"step": 112500
},
{
"epoch": 55.67,
"learning_rate": 2.2167487684729066e-05,
"loss": 0.9755,
"step": 113000
},
{
"epoch": 55.91,
"learning_rate": 2.204433497536946e-05,
"loss": 0.9743,
"step": 113500
},
{
"epoch": 56.16,
"learning_rate": 2.1921182266009854e-05,
"loss": 0.9691,
"step": 114000
},
{
"epoch": 56.4,
"learning_rate": 2.179802955665025e-05,
"loss": 0.9699,
"step": 114500
},
{
"epoch": 56.65,
"learning_rate": 2.1674876847290644e-05,
"loss": 0.9742,
"step": 115000
},
{
"epoch": 56.9,
"learning_rate": 2.1551724137931033e-05,
"loss": 0.9672,
"step": 115500
},
{
"epoch": 57.14,
"learning_rate": 2.1428571428571428e-05,
"loss": 0.9701,
"step": 116000
},
{
"epoch": 57.39,
"learning_rate": 2.1305418719211823e-05,
"loss": 0.971,
"step": 116500
},
{
"epoch": 57.64,
"learning_rate": 2.1182266009852215e-05,
"loss": 0.9684,
"step": 117000
},
{
"epoch": 57.88,
"learning_rate": 2.105911330049261e-05,
"loss": 0.9674,
"step": 117500
},
{
"epoch": 58.13,
"learning_rate": 2.0935960591133006e-05,
"loss": 0.966,
"step": 118000
},
{
"epoch": 58.37,
"learning_rate": 2.0812807881773398e-05,
"loss": 0.9612,
"step": 118500
},
{
"epoch": 58.62,
"learning_rate": 2.0689655172413793e-05,
"loss": 0.9661,
"step": 119000
},
{
"epoch": 58.87,
"learning_rate": 2.056650246305419e-05,
"loss": 0.9628,
"step": 119500
},
{
"epoch": 59.11,
"learning_rate": 2.0443349753694584e-05,
"loss": 0.9589,
"step": 120000
},
{
"epoch": 59.36,
"learning_rate": 2.0320197044334976e-05,
"loss": 0.9559,
"step": 120500
},
{
"epoch": 59.61,
"learning_rate": 2.019704433497537e-05,
"loss": 0.9599,
"step": 121000
},
{
"epoch": 59.85,
"learning_rate": 2.0073891625615766e-05,
"loss": 0.96,
"step": 121500
},
{
"epoch": 60.1,
"learning_rate": 1.995073891625616e-05,
"loss": 0.9556,
"step": 122000
},
{
"epoch": 60.34,
"learning_rate": 1.9827586206896554e-05,
"loss": 0.9537,
"step": 122500
},
{
"epoch": 60.59,
"learning_rate": 1.970443349753695e-05,
"loss": 0.9548,
"step": 123000
},
{
"epoch": 60.84,
"learning_rate": 1.958128078817734e-05,
"loss": 0.956,
"step": 123500
},
{
"epoch": 61.08,
"learning_rate": 1.9458128078817736e-05,
"loss": 0.9535,
"step": 124000
},
{
"epoch": 61.33,
"learning_rate": 1.933497536945813e-05,
"loss": 0.9515,
"step": 124500
},
{
"epoch": 61.58,
"learning_rate": 1.921182266009852e-05,
"loss": 0.9514,
"step": 125000
},
{
"epoch": 61.82,
"learning_rate": 1.9088669950738915e-05,
"loss": 0.954,
"step": 125500
},
{
"epoch": 62.07,
"learning_rate": 1.896551724137931e-05,
"loss": 0.9512,
"step": 126000
},
{
"epoch": 62.32,
"learning_rate": 1.8842364532019706e-05,
"loss": 0.9458,
"step": 126500
},
{
"epoch": 62.56,
"learning_rate": 1.8719211822660098e-05,
"loss": 0.9492,
"step": 127000
},
{
"epoch": 62.81,
"learning_rate": 1.8596059113300493e-05,
"loss": 0.9438,
"step": 127500
},
{
"epoch": 63.05,
"learning_rate": 1.847290640394089e-05,
"loss": 0.9456,
"step": 128000
},
{
"epoch": 63.3,
"learning_rate": 1.834975369458128e-05,
"loss": 0.9424,
"step": 128500
},
{
"epoch": 63.55,
"learning_rate": 1.8226600985221676e-05,
"loss": 0.9447,
"step": 129000
},
{
"epoch": 63.79,
"learning_rate": 1.810344827586207e-05,
"loss": 0.9459,
"step": 129500
},
{
"epoch": 64.04,
"learning_rate": 1.7980295566502463e-05,
"loss": 0.947,
"step": 130000
},
{
"epoch": 64.29,
"learning_rate": 1.785714285714286e-05,
"loss": 0.9373,
"step": 130500
},
{
"epoch": 64.53,
"learning_rate": 1.7733990147783254e-05,
"loss": 0.9417,
"step": 131000
},
{
"epoch": 64.78,
"learning_rate": 1.7610837438423646e-05,
"loss": 0.9481,
"step": 131500
},
{
"epoch": 65.02,
"learning_rate": 1.748768472906404e-05,
"loss": 0.9422,
"step": 132000
},
{
"epoch": 65.27,
"learning_rate": 1.7364532019704436e-05,
"loss": 0.9333,
"step": 132500
},
{
"epoch": 65.52,
"learning_rate": 1.7241379310344828e-05,
"loss": 0.9369,
"step": 133000
},
{
"epoch": 65.76,
"learning_rate": 1.7118226600985224e-05,
"loss": 0.9407,
"step": 133500
},
{
"epoch": 66.01,
"learning_rate": 1.6995073891625616e-05,
"loss": 0.9374,
"step": 134000
},
{
"epoch": 66.26,
"learning_rate": 1.687192118226601e-05,
"loss": 0.9338,
"step": 134500
},
{
"epoch": 66.5,
"learning_rate": 1.6748768472906403e-05,
"loss": 0.934,
"step": 135000
},
{
"epoch": 66.75,
"learning_rate": 1.6625615763546798e-05,
"loss": 0.9339,
"step": 135500
},
{
"epoch": 67.0,
"learning_rate": 1.6502463054187193e-05,
"loss": 0.9358,
"step": 136000
},
{
"epoch": 67.24,
"learning_rate": 1.6379310344827585e-05,
"loss": 0.9305,
"step": 136500
},
{
"epoch": 67.49,
"learning_rate": 1.625615763546798e-05,
"loss": 0.9288,
"step": 137000
},
{
"epoch": 67.73,
"learning_rate": 1.6133004926108376e-05,
"loss": 0.9279,
"step": 137500
},
{
"epoch": 67.98,
"learning_rate": 1.6009852216748768e-05,
"loss": 0.9305,
"step": 138000
},
{
"epoch": 68.23,
"learning_rate": 1.5886699507389163e-05,
"loss": 0.9269,
"step": 138500
},
{
"epoch": 68.47,
"learning_rate": 1.576354679802956e-05,
"loss": 0.9291,
"step": 139000
},
{
"epoch": 68.72,
"learning_rate": 1.564039408866995e-05,
"loss": 0.9263,
"step": 139500
},
{
"epoch": 68.97,
"learning_rate": 1.5517241379310346e-05,
"loss": 0.9271,
"step": 140000
},
{
"epoch": 69.21,
"learning_rate": 1.539408866995074e-05,
"loss": 0.9241,
"step": 140500
},
{
"epoch": 69.46,
"learning_rate": 1.5270935960591133e-05,
"loss": 0.9268,
"step": 141000
},
{
"epoch": 69.7,
"learning_rate": 1.5147783251231528e-05,
"loss": 0.926,
"step": 141500
},
{
"epoch": 69.95,
"learning_rate": 1.5024630541871924e-05,
"loss": 0.9231,
"step": 142000
},
{
"epoch": 70.2,
"learning_rate": 1.4901477832512317e-05,
"loss": 0.923,
"step": 142500
},
{
"epoch": 70.44,
"learning_rate": 1.4778325123152711e-05,
"loss": 0.9211,
"step": 143000
},
{
"epoch": 70.69,
"learning_rate": 1.4655172413793103e-05,
"loss": 0.9243,
"step": 143500
},
{
"epoch": 70.94,
"learning_rate": 1.4532019704433496e-05,
"loss": 0.9218,
"step": 144000
},
{
"epoch": 71.18,
"learning_rate": 1.4408866995073892e-05,
"loss": 0.9178,
"step": 144500
},
{
"epoch": 71.43,
"learning_rate": 1.4285714285714285e-05,
"loss": 0.9153,
"step": 145000
},
{
"epoch": 71.67,
"learning_rate": 1.4162561576354679e-05,
"loss": 0.9182,
"step": 145500
},
{
"epoch": 71.92,
"learning_rate": 1.4039408866995074e-05,
"loss": 0.9184,
"step": 146000
},
{
"epoch": 72.17,
"learning_rate": 1.3916256157635468e-05,
"loss": 0.9186,
"step": 146500
},
{
"epoch": 72.41,
"learning_rate": 1.3793103448275863e-05,
"loss": 0.914,
"step": 147000
},
{
"epoch": 72.66,
"learning_rate": 1.3669950738916257e-05,
"loss": 0.914,
"step": 147500
},
{
"epoch": 72.91,
"learning_rate": 1.354679802955665e-05,
"loss": 0.9161,
"step": 148000
},
{
"epoch": 73.15,
"learning_rate": 1.3423645320197046e-05,
"loss": 0.9138,
"step": 148500
},
{
"epoch": 73.4,
"learning_rate": 1.330049261083744e-05,
"loss": 0.9106,
"step": 149000
},
{
"epoch": 73.65,
"learning_rate": 1.3177339901477833e-05,
"loss": 0.9115,
"step": 149500
},
{
"epoch": 73.89,
"learning_rate": 1.3054187192118228e-05,
"loss": 0.9135,
"step": 150000
},
{
"epoch": 74.14,
"learning_rate": 1.2931034482758622e-05,
"loss": 0.9116,
"step": 150500
},
{
"epoch": 74.38,
"learning_rate": 1.2807881773399016e-05,
"loss": 0.9109,
"step": 151000
},
{
"epoch": 74.63,
"learning_rate": 1.2684729064039411e-05,
"loss": 0.9097,
"step": 151500
},
{
"epoch": 74.88,
"learning_rate": 1.2561576354679805e-05,
"loss": 0.9099,
"step": 152000
},
{
"epoch": 75.12,
"learning_rate": 1.2438423645320198e-05,
"loss": 0.9106,
"step": 152500
},
{
"epoch": 75.37,
"learning_rate": 1.2315270935960592e-05,
"loss": 0.9098,
"step": 153000
},
{
"epoch": 75.62,
"learning_rate": 1.2192118226600986e-05,
"loss": 0.9098,
"step": 153500
},
{
"epoch": 75.86,
"learning_rate": 1.206896551724138e-05,
"loss": 0.9092,
"step": 154000
},
{
"epoch": 76.11,
"learning_rate": 1.1945812807881774e-05,
"loss": 0.9071,
"step": 154500
},
{
"epoch": 76.35,
"learning_rate": 1.1822660098522168e-05,
"loss": 0.9045,
"step": 155000
},
{
"epoch": 76.6,
"learning_rate": 1.1699507389162562e-05,
"loss": 0.9051,
"step": 155500
},
{
"epoch": 76.85,
"learning_rate": 1.1576354679802955e-05,
"loss": 0.9061,
"step": 156000
},
{
"epoch": 77.09,
"learning_rate": 1.145320197044335e-05,
"loss": 0.9023,
"step": 156500
},
{
"epoch": 77.34,
"learning_rate": 1.1330049261083744e-05,
"loss": 0.9014,
"step": 157000
},
{
"epoch": 77.59,
"learning_rate": 1.1206896551724138e-05,
"loss": 0.9047,
"step": 157500
},
{
"epoch": 77.83,
"learning_rate": 1.1083743842364533e-05,
"loss": 0.9004,
"step": 158000
},
{
"epoch": 78.08,
"learning_rate": 1.0960591133004927e-05,
"loss": 0.9053,
"step": 158500
},
{
"epoch": 78.33,
"learning_rate": 1.0837438423645322e-05,
"loss": 0.9031,
"step": 159000
},
{
"epoch": 78.57,
"learning_rate": 1.0714285714285714e-05,
"loss": 0.9027,
"step": 159500
},
{
"epoch": 78.82,
"learning_rate": 1.0591133004926108e-05,
"loss": 0.8996,
"step": 160000
},
{
"epoch": 79.06,
"learning_rate": 1.0467980295566503e-05,
"loss": 0.9031,
"step": 160500
},
{
"epoch": 79.31,
"learning_rate": 1.0344827586206897e-05,
"loss": 0.9,
"step": 161000
},
{
"epoch": 79.56,
"learning_rate": 1.0221674876847292e-05,
"loss": 0.9001,
"step": 161500
},
{
"epoch": 79.8,
"learning_rate": 1.0098522167487686e-05,
"loss": 0.8986,
"step": 162000
},
{
"epoch": 80.05,
"learning_rate": 9.97536945812808e-06,
"loss": 0.8956,
"step": 162500
},
{
"epoch": 80.3,
"learning_rate": 9.852216748768475e-06,
"loss": 0.8961,
"step": 163000
},
{
"epoch": 80.54,
"learning_rate": 9.729064039408868e-06,
"loss": 0.8973,
"step": 163500
},
{
"epoch": 80.79,
"learning_rate": 9.60591133004926e-06,
"loss": 0.8996,
"step": 164000
},
{
"epoch": 81.03,
"learning_rate": 9.482758620689655e-06,
"loss": 0.8954,
"step": 164500
},
{
"epoch": 81.28,
"learning_rate": 9.359605911330049e-06,
"loss": 0.8925,
"step": 165000
},
{
"epoch": 81.53,
"learning_rate": 9.236453201970444e-06,
"loss": 0.8986,
"step": 165500
},
{
"epoch": 81.77,
"learning_rate": 9.113300492610838e-06,
"loss": 0.8943,
"step": 166000
},
{
"epoch": 82.02,
"learning_rate": 8.990147783251232e-06,
"loss": 0.8972,
"step": 166500
},
{
"epoch": 82.27,
"learning_rate": 8.866995073891627e-06,
"loss": 0.8944,
"step": 167000
},
{
"epoch": 82.51,
"learning_rate": 8.74384236453202e-06,
"loss": 0.8898,
"step": 167500
},
{
"epoch": 82.76,
"learning_rate": 8.620689655172414e-06,
"loss": 0.8929,
"step": 168000
},
{
"epoch": 83.0,
"learning_rate": 8.497536945812808e-06,
"loss": 0.8932,
"step": 168500
},
{
"epoch": 83.25,
"learning_rate": 8.374384236453201e-06,
"loss": 0.8929,
"step": 169000
},
{
"epoch": 83.5,
"learning_rate": 8.251231527093597e-06,
"loss": 0.8909,
"step": 169500
},
{
"epoch": 83.74,
"learning_rate": 8.12807881773399e-06,
"loss": 0.889,
"step": 170000
},
{
"epoch": 83.99,
"learning_rate": 8.004926108374384e-06,
"loss": 0.8903,
"step": 170500
},
{
"epoch": 84.24,
"learning_rate": 7.88177339901478e-06,
"loss": 0.886,
"step": 171000
},
{
"epoch": 84.48,
"learning_rate": 7.758620689655173e-06,
"loss": 0.8935,
"step": 171500
},
{
"epoch": 84.73,
"learning_rate": 7.635467980295567e-06,
"loss": 0.8919,
"step": 172000
},
{
"epoch": 84.98,
"learning_rate": 7.512315270935962e-06,
"loss": 0.8903,
"step": 172500
},
{
"epoch": 85.22,
"learning_rate": 7.3891625615763555e-06,
"loss": 0.8864,
"step": 173000
},
{
"epoch": 85.47,
"learning_rate": 7.266009852216748e-06,
"loss": 0.8856,
"step": 173500
},
{
"epoch": 85.71,
"learning_rate": 7.142857142857143e-06,
"loss": 0.8862,
"step": 174000
},
{
"epoch": 85.96,
"learning_rate": 7.019704433497537e-06,
"loss": 0.8874,
"step": 174500
},
{
"epoch": 86.21,
"learning_rate": 6.896551724137932e-06,
"loss": 0.8838,
"step": 175000
},
{
"epoch": 86.45,
"learning_rate": 6.773399014778325e-06,
"loss": 0.8864,
"step": 175500
},
{
"epoch": 86.7,
"learning_rate": 6.65024630541872e-06,
"loss": 0.8805,
"step": 176000
},
{
"epoch": 86.95,
"learning_rate": 6.527093596059114e-06,
"loss": 0.8843,
"step": 176500
},
{
"epoch": 87.19,
"learning_rate": 6.403940886699508e-06,
"loss": 0.8831,
"step": 177000
},
{
"epoch": 87.44,
"learning_rate": 6.280788177339902e-06,
"loss": 0.8841,
"step": 177500
},
{
"epoch": 87.68,
"learning_rate": 6.157635467980296e-06,
"loss": 0.8849,
"step": 178000
},
{
"epoch": 87.93,
"learning_rate": 6.03448275862069e-06,
"loss": 0.8809,
"step": 178500
},
{
"epoch": 88.18,
"learning_rate": 5.911330049261084e-06,
"loss": 0.882,
"step": 179000
},
{
"epoch": 88.42,
"learning_rate": 5.788177339901478e-06,
"loss": 0.8855,
"step": 179500
},
{
"epoch": 88.67,
"learning_rate": 5.665024630541872e-06,
"loss": 0.8832,
"step": 180000
},
{
"epoch": 88.92,
"learning_rate": 5.541871921182267e-06,
"loss": 0.881,
"step": 180500
},
{
"epoch": 89.16,
"learning_rate": 5.418719211822661e-06,
"loss": 0.8789,
"step": 181000
},
{
"epoch": 89.41,
"learning_rate": 5.295566502463054e-06,
"loss": 0.8787,
"step": 181500
},
{
"epoch": 89.66,
"learning_rate": 5.172413793103448e-06,
"loss": 0.8833,
"step": 182000
},
{
"epoch": 89.9,
"learning_rate": 5.049261083743843e-06,
"loss": 0.8793,
"step": 182500
},
{
"epoch": 90.15,
"learning_rate": 4.926108374384237e-06,
"loss": 0.8768,
"step": 183000
},
{
"epoch": 90.39,
"learning_rate": 4.80295566502463e-06,
"loss": 0.8789,
"step": 183500
},
{
"epoch": 90.64,
"learning_rate": 4.6798029556650245e-06,
"loss": 0.8754,
"step": 184000
},
{
"epoch": 90.89,
"learning_rate": 4.556650246305419e-06,
"loss": 0.8768,
"step": 184500
},
{
"epoch": 91.13,
"learning_rate": 4.4334975369458135e-06,
"loss": 0.8786,
"step": 185000
},
{
"epoch": 91.38,
"learning_rate": 4.310344827586207e-06,
"loss": 0.8762,
"step": 185500
},
{
"epoch": 91.63,
"learning_rate": 4.187192118226601e-06,
"loss": 0.8744,
"step": 186000
},
{
"epoch": 91.87,
"learning_rate": 4.064039408866995e-06,
"loss": 0.8771,
"step": 186500
},
{
"epoch": 92.12,
"learning_rate": 3.94088669950739e-06,
"loss": 0.8752,
"step": 187000
},
{
"epoch": 92.36,
"learning_rate": 3.817733990147783e-06,
"loss": 0.8762,
"step": 187500
},
{
"epoch": 92.61,
"learning_rate": 3.6945812807881777e-06,
"loss": 0.8773,
"step": 188000
},
{
"epoch": 92.86,
"learning_rate": 3.5714285714285714e-06,
"loss": 0.8774,
"step": 188500
},
{
"epoch": 93.1,
"learning_rate": 3.448275862068966e-06,
"loss": 0.8739,
"step": 189000
},
{
"epoch": 93.35,
"learning_rate": 3.32512315270936e-06,
"loss": 0.8756,
"step": 189500
},
{
"epoch": 93.6,
"learning_rate": 3.201970443349754e-06,
"loss": 0.8725,
"step": 190000
},
{
"epoch": 93.84,
"learning_rate": 3.078817733990148e-06,
"loss": 0.8806,
"step": 190500
},
{
"epoch": 94.09,
"learning_rate": 2.955665024630542e-06,
"loss": 0.8739,
"step": 191000
},
{
"epoch": 94.33,
"learning_rate": 2.832512315270936e-06,
"loss": 0.869,
"step": 191500
},
{
"epoch": 94.58,
"learning_rate": 2.7093596059113305e-06,
"loss": 0.8755,
"step": 192000
},
{
"epoch": 94.83,
"learning_rate": 2.586206896551724e-06,
"loss": 0.8733,
"step": 192500
},
{
"epoch": 95.07,
"learning_rate": 2.4630541871921186e-06,
"loss": 0.8742,
"step": 193000
},
{
"epoch": 95.32,
"learning_rate": 2.3399014778325123e-06,
"loss": 0.8729,
"step": 193500
},
{
"epoch": 95.57,
"learning_rate": 2.2167487684729067e-06,
"loss": 0.8736,
"step": 194000
},
{
"epoch": 95.81,
"learning_rate": 2.0935960591133003e-06,
"loss": 0.8756,
"step": 194500
},
{
"epoch": 96.06,
"learning_rate": 1.970443349753695e-06,
"loss": 0.8696,
"step": 195000
},
{
"epoch": 96.31,
"learning_rate": 1.8472906403940889e-06,
"loss": 0.8692,
"step": 195500
},
{
"epoch": 96.55,
"learning_rate": 1.724137931034483e-06,
"loss": 0.8707,
"step": 196000
},
{
"epoch": 96.8,
"learning_rate": 1.600985221674877e-06,
"loss": 0.8748,
"step": 196500
},
{
"epoch": 97.04,
"learning_rate": 1.477832512315271e-06,
"loss": 0.8698,
"step": 197000
},
{
"epoch": 97.29,
"learning_rate": 1.3546798029556653e-06,
"loss": 0.8698,
"step": 197500
},
{
"epoch": 97.54,
"learning_rate": 1.2315270935960593e-06,
"loss": 0.8697,
"step": 198000
},
{
"epoch": 97.78,
"learning_rate": 1.1083743842364534e-06,
"loss": 0.8721,
"step": 198500
},
{
"epoch": 98.03,
"learning_rate": 9.852216748768474e-07,
"loss": 0.8695,
"step": 199000
},
{
"epoch": 98.28,
"learning_rate": 8.620689655172415e-07,
"loss": 0.8677,
"step": 199500
},
{
"epoch": 98.52,
"learning_rate": 7.389162561576355e-07,
"loss": 0.8715,
"step": 200000
},
{
"epoch": 98.77,
"learning_rate": 6.157635467980297e-07,
"loss": 0.8714,
"step": 200500
},
{
"epoch": 99.01,
"learning_rate": 4.926108374384237e-07,
"loss": 0.8703,
"step": 201000
},
{
"epoch": 99.26,
"learning_rate": 3.6945812807881775e-07,
"loss": 0.8681,
"step": 201500
},
{
"epoch": 99.51,
"learning_rate": 2.4630541871921185e-07,
"loss": 0.8684,
"step": 202000
},
{
"epoch": 99.75,
"learning_rate": 1.2315270935960593e-07,
"loss": 0.8731,
"step": 202500
},
{
"epoch": 100.0,
"learning_rate": 0.0,
"loss": 0.8699,
"step": 203000
}
],
"max_steps": 203000,
"num_train_epochs": 100,
"total_flos": 3.4424249629999104e+18,
"trial_name": null,
"trial_params": null
}