Davlan's picture
Upload trainer_state.json
f30098f
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"global_step": 114399,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.9781466621211725e-05,
"loss": 2.1728,
"step": 500
},
{
"epoch": 0.03,
"learning_rate": 4.956293324242345e-05,
"loss": 2.0108,
"step": 1000
},
{
"epoch": 0.04,
"learning_rate": 4.934439986363518e-05,
"loss": 1.9521,
"step": 1500
},
{
"epoch": 0.05,
"learning_rate": 4.91258664848469e-05,
"loss": 1.8771,
"step": 2000
},
{
"epoch": 0.07,
"learning_rate": 4.8907333106058624e-05,
"loss": 1.8337,
"step": 2500
},
{
"epoch": 0.08,
"learning_rate": 4.868879972727035e-05,
"loss": 1.8045,
"step": 3000
},
{
"epoch": 0.09,
"learning_rate": 4.847026634848207e-05,
"loss": 1.771,
"step": 3500
},
{
"epoch": 0.1,
"learning_rate": 4.825173296969379e-05,
"loss": 1.7447,
"step": 4000
},
{
"epoch": 0.12,
"learning_rate": 4.803319959090552e-05,
"loss": 1.7067,
"step": 4500
},
{
"epoch": 0.13,
"learning_rate": 4.7814666212117245e-05,
"loss": 1.6988,
"step": 5000
},
{
"epoch": 0.14,
"learning_rate": 4.759613283332897e-05,
"loss": 1.676,
"step": 5500
},
{
"epoch": 0.16,
"learning_rate": 4.737759945454069e-05,
"loss": 1.6483,
"step": 6000
},
{
"epoch": 0.17,
"learning_rate": 4.7159066075752414e-05,
"loss": 1.638,
"step": 6500
},
{
"epoch": 0.18,
"learning_rate": 4.694053269696414e-05,
"loss": 1.6235,
"step": 7000
},
{
"epoch": 0.2,
"learning_rate": 4.6721999318175866e-05,
"loss": 1.6096,
"step": 7500
},
{
"epoch": 0.21,
"learning_rate": 4.650346593938759e-05,
"loss": 1.6011,
"step": 8000
},
{
"epoch": 0.22,
"learning_rate": 4.628493256059931e-05,
"loss": 1.5781,
"step": 8500
},
{
"epoch": 0.24,
"learning_rate": 4.6066399181811035e-05,
"loss": 1.572,
"step": 9000
},
{
"epoch": 0.25,
"learning_rate": 4.584786580302276e-05,
"loss": 1.5612,
"step": 9500
},
{
"epoch": 0.26,
"learning_rate": 4.562933242423448e-05,
"loss": 1.5728,
"step": 10000
},
{
"epoch": 0.28,
"learning_rate": 4.5410799045446204e-05,
"loss": 1.5437,
"step": 10500
},
{
"epoch": 0.29,
"learning_rate": 4.519226566665793e-05,
"loss": 1.5245,
"step": 11000
},
{
"epoch": 0.3,
"learning_rate": 4.497373228786965e-05,
"loss": 1.5209,
"step": 11500
},
{
"epoch": 0.31,
"learning_rate": 4.475519890908137e-05,
"loss": 1.5211,
"step": 12000
},
{
"epoch": 0.33,
"learning_rate": 4.4536665530293096e-05,
"loss": 1.5035,
"step": 12500
},
{
"epoch": 0.34,
"learning_rate": 4.431813215150482e-05,
"loss": 1.4973,
"step": 13000
},
{
"epoch": 0.35,
"learning_rate": 4.409959877271655e-05,
"loss": 1.4881,
"step": 13500
},
{
"epoch": 0.37,
"learning_rate": 4.388106539392827e-05,
"loss": 1.474,
"step": 14000
},
{
"epoch": 0.38,
"learning_rate": 4.3662532015139994e-05,
"loss": 1.4652,
"step": 14500
},
{
"epoch": 0.39,
"learning_rate": 4.344399863635172e-05,
"loss": 1.4629,
"step": 15000
},
{
"epoch": 0.41,
"learning_rate": 4.322546525756344e-05,
"loss": 1.4809,
"step": 15500
},
{
"epoch": 0.42,
"learning_rate": 4.300693187877516e-05,
"loss": 1.4629,
"step": 16000
},
{
"epoch": 0.43,
"learning_rate": 4.278839849998689e-05,
"loss": 1.4502,
"step": 16500
},
{
"epoch": 0.45,
"learning_rate": 4.2569865121198615e-05,
"loss": 1.4361,
"step": 17000
},
{
"epoch": 0.46,
"learning_rate": 4.235133174241034e-05,
"loss": 1.4425,
"step": 17500
},
{
"epoch": 0.47,
"learning_rate": 4.213279836362206e-05,
"loss": 1.422,
"step": 18000
},
{
"epoch": 0.49,
"learning_rate": 4.1914264984833784e-05,
"loss": 1.4164,
"step": 18500
},
{
"epoch": 0.5,
"learning_rate": 4.169573160604551e-05,
"loss": 1.4208,
"step": 19000
},
{
"epoch": 0.51,
"learning_rate": 4.147719822725724e-05,
"loss": 1.4204,
"step": 19500
},
{
"epoch": 0.52,
"learning_rate": 4.125866484846896e-05,
"loss": 1.3949,
"step": 20000
},
{
"epoch": 0.54,
"learning_rate": 4.104013146968068e-05,
"loss": 1.4103,
"step": 20500
},
{
"epoch": 0.55,
"learning_rate": 4.0821598090892405e-05,
"loss": 1.4015,
"step": 21000
},
{
"epoch": 0.56,
"learning_rate": 4.060306471210413e-05,
"loss": 1.3844,
"step": 21500
},
{
"epoch": 0.58,
"learning_rate": 4.038453133331585e-05,
"loss": 1.3823,
"step": 22000
},
{
"epoch": 0.59,
"learning_rate": 4.016599795452758e-05,
"loss": 1.3842,
"step": 22500
},
{
"epoch": 0.6,
"learning_rate": 3.9947464575739304e-05,
"loss": 1.373,
"step": 23000
},
{
"epoch": 0.62,
"learning_rate": 3.972893119695103e-05,
"loss": 1.3756,
"step": 23500
},
{
"epoch": 0.63,
"learning_rate": 3.951039781816275e-05,
"loss": 1.3714,
"step": 24000
},
{
"epoch": 0.64,
"learning_rate": 3.929186443937447e-05,
"loss": 1.3614,
"step": 24500
},
{
"epoch": 0.66,
"learning_rate": 3.9073331060586195e-05,
"loss": 1.3703,
"step": 25000
},
{
"epoch": 0.67,
"learning_rate": 3.8854797681797925e-05,
"loss": 1.3494,
"step": 25500
},
{
"epoch": 0.68,
"learning_rate": 3.863626430300965e-05,
"loss": 1.3588,
"step": 26000
},
{
"epoch": 0.69,
"learning_rate": 3.841773092422137e-05,
"loss": 1.3514,
"step": 26500
},
{
"epoch": 0.71,
"learning_rate": 3.8199197545433094e-05,
"loss": 1.3591,
"step": 27000
},
{
"epoch": 0.72,
"learning_rate": 3.798066416664482e-05,
"loss": 1.3479,
"step": 27500
},
{
"epoch": 0.73,
"learning_rate": 3.776213078785654e-05,
"loss": 1.3524,
"step": 28000
},
{
"epoch": 0.75,
"learning_rate": 3.754359740906826e-05,
"loss": 1.3366,
"step": 28500
},
{
"epoch": 0.76,
"learning_rate": 3.7325064030279986e-05,
"loss": 1.3498,
"step": 29000
},
{
"epoch": 0.77,
"learning_rate": 3.710653065149171e-05,
"loss": 1.3305,
"step": 29500
},
{
"epoch": 0.79,
"learning_rate": 3.688799727270343e-05,
"loss": 1.3263,
"step": 30000
},
{
"epoch": 0.8,
"learning_rate": 3.6669463893915154e-05,
"loss": 1.3203,
"step": 30500
},
{
"epoch": 0.81,
"learning_rate": 3.645093051512688e-05,
"loss": 1.3128,
"step": 31000
},
{
"epoch": 0.83,
"learning_rate": 3.623239713633861e-05,
"loss": 1.3157,
"step": 31500
},
{
"epoch": 0.84,
"learning_rate": 3.601386375755033e-05,
"loss": 1.3167,
"step": 32000
},
{
"epoch": 0.85,
"learning_rate": 3.579533037876205e-05,
"loss": 1.3075,
"step": 32500
},
{
"epoch": 0.87,
"learning_rate": 3.5576796999973776e-05,
"loss": 1.3077,
"step": 33000
},
{
"epoch": 0.88,
"learning_rate": 3.53582636211855e-05,
"loss": 1.3082,
"step": 33500
},
{
"epoch": 0.89,
"learning_rate": 3.513973024239722e-05,
"loss": 1.3128,
"step": 34000
},
{
"epoch": 0.9,
"learning_rate": 3.492119686360895e-05,
"loss": 1.3122,
"step": 34500
},
{
"epoch": 0.92,
"learning_rate": 3.4702663484820674e-05,
"loss": 1.2951,
"step": 35000
},
{
"epoch": 0.93,
"learning_rate": 3.44841301060324e-05,
"loss": 1.2928,
"step": 35500
},
{
"epoch": 0.94,
"learning_rate": 3.426559672724412e-05,
"loss": 1.2889,
"step": 36000
},
{
"epoch": 0.96,
"learning_rate": 3.404706334845584e-05,
"loss": 1.2825,
"step": 36500
},
{
"epoch": 0.97,
"learning_rate": 3.3828529969667566e-05,
"loss": 1.2809,
"step": 37000
},
{
"epoch": 0.98,
"learning_rate": 3.3609996590879295e-05,
"loss": 1.2806,
"step": 37500
},
{
"epoch": 1.0,
"learning_rate": 3.339146321209102e-05,
"loss": 1.2847,
"step": 38000
},
{
"epoch": 1.01,
"learning_rate": 3.317292983330274e-05,
"loss": 1.2637,
"step": 38500
},
{
"epoch": 1.02,
"learning_rate": 3.2954396454514464e-05,
"loss": 1.2698,
"step": 39000
},
{
"epoch": 1.04,
"learning_rate": 3.273586307572619e-05,
"loss": 1.2469,
"step": 39500
},
{
"epoch": 1.05,
"learning_rate": 3.251732969693791e-05,
"loss": 1.2516,
"step": 40000
},
{
"epoch": 1.06,
"learning_rate": 3.229879631814964e-05,
"loss": 1.2525,
"step": 40500
},
{
"epoch": 1.08,
"learning_rate": 3.208026293936136e-05,
"loss": 1.2573,
"step": 41000
},
{
"epoch": 1.09,
"learning_rate": 3.1861729560573085e-05,
"loss": 1.2516,
"step": 41500
},
{
"epoch": 1.1,
"learning_rate": 3.164319618178481e-05,
"loss": 1.2425,
"step": 42000
},
{
"epoch": 1.11,
"learning_rate": 3.142466280299653e-05,
"loss": 1.2487,
"step": 42500
},
{
"epoch": 1.13,
"learning_rate": 3.1206129424208254e-05,
"loss": 1.238,
"step": 43000
},
{
"epoch": 1.14,
"learning_rate": 3.0987596045419984e-05,
"loss": 1.2293,
"step": 43500
},
{
"epoch": 1.15,
"learning_rate": 3.076906266663171e-05,
"loss": 1.2392,
"step": 44000
},
{
"epoch": 1.17,
"learning_rate": 3.055052928784343e-05,
"loss": 1.2382,
"step": 44500
},
{
"epoch": 1.18,
"learning_rate": 3.033199590905515e-05,
"loss": 1.2327,
"step": 45000
},
{
"epoch": 1.19,
"learning_rate": 3.0113462530266872e-05,
"loss": 1.2345,
"step": 45500
},
{
"epoch": 1.21,
"learning_rate": 2.9894929151478595e-05,
"loss": 1.2214,
"step": 46000
},
{
"epoch": 1.22,
"learning_rate": 2.9676395772690325e-05,
"loss": 1.2356,
"step": 46500
},
{
"epoch": 1.23,
"learning_rate": 2.9457862393902048e-05,
"loss": 1.2336,
"step": 47000
},
{
"epoch": 1.25,
"learning_rate": 2.923932901511377e-05,
"loss": 1.2354,
"step": 47500
},
{
"epoch": 1.26,
"learning_rate": 2.9020795636325493e-05,
"loss": 1.2396,
"step": 48000
},
{
"epoch": 1.27,
"learning_rate": 2.8802262257537216e-05,
"loss": 1.2234,
"step": 48500
},
{
"epoch": 1.28,
"learning_rate": 2.858372887874894e-05,
"loss": 1.2239,
"step": 49000
},
{
"epoch": 1.3,
"learning_rate": 2.8365195499960666e-05,
"loss": 1.2123,
"step": 49500
},
{
"epoch": 1.31,
"learning_rate": 2.8146662121172392e-05,
"loss": 1.2116,
"step": 50000
},
{
"epoch": 1.32,
"learning_rate": 2.7928128742384115e-05,
"loss": 1.2139,
"step": 50500
},
{
"epoch": 1.34,
"learning_rate": 2.7709595363595838e-05,
"loss": 1.2115,
"step": 51000
},
{
"epoch": 1.35,
"learning_rate": 2.749106198480756e-05,
"loss": 1.2051,
"step": 51500
},
{
"epoch": 1.36,
"learning_rate": 2.7272528606019284e-05,
"loss": 1.2152,
"step": 52000
},
{
"epoch": 1.38,
"learning_rate": 2.705399522723101e-05,
"loss": 1.2084,
"step": 52500
},
{
"epoch": 1.39,
"learning_rate": 2.6835461848442733e-05,
"loss": 1.1987,
"step": 53000
},
{
"epoch": 1.4,
"learning_rate": 2.6616928469654456e-05,
"loss": 1.1892,
"step": 53500
},
{
"epoch": 1.42,
"learning_rate": 2.639839509086618e-05,
"loss": 1.2074,
"step": 54000
},
{
"epoch": 1.43,
"learning_rate": 2.61798617120779e-05,
"loss": 1.1952,
"step": 54500
},
{
"epoch": 1.44,
"learning_rate": 2.596132833328963e-05,
"loss": 1.196,
"step": 55000
},
{
"epoch": 1.46,
"learning_rate": 2.5742794954501354e-05,
"loss": 1.182,
"step": 55500
},
{
"epoch": 1.47,
"learning_rate": 2.5524261575713077e-05,
"loss": 1.1845,
"step": 56000
},
{
"epoch": 1.48,
"learning_rate": 2.53057281969248e-05,
"loss": 1.1889,
"step": 56500
},
{
"epoch": 1.49,
"learning_rate": 2.5087194818136523e-05,
"loss": 1.177,
"step": 57000
},
{
"epoch": 1.51,
"learning_rate": 2.486866143934825e-05,
"loss": 1.1845,
"step": 57500
},
{
"epoch": 1.52,
"learning_rate": 2.4650128060559972e-05,
"loss": 1.1865,
"step": 58000
},
{
"epoch": 1.53,
"learning_rate": 2.4431594681771695e-05,
"loss": 1.1766,
"step": 58500
},
{
"epoch": 1.55,
"learning_rate": 2.421306130298342e-05,
"loss": 1.1784,
"step": 59000
},
{
"epoch": 1.56,
"learning_rate": 2.3994527924195144e-05,
"loss": 1.1728,
"step": 59500
},
{
"epoch": 1.57,
"learning_rate": 2.3775994545406867e-05,
"loss": 1.1775,
"step": 60000
},
{
"epoch": 1.59,
"learning_rate": 2.355746116661859e-05,
"loss": 1.1767,
"step": 60500
},
{
"epoch": 1.6,
"learning_rate": 2.3338927787830313e-05,
"loss": 1.1809,
"step": 61000
},
{
"epoch": 1.61,
"learning_rate": 2.3120394409042036e-05,
"loss": 1.1744,
"step": 61500
},
{
"epoch": 1.63,
"learning_rate": 2.2901861030253762e-05,
"loss": 1.1694,
"step": 62000
},
{
"epoch": 1.64,
"learning_rate": 2.2683327651465485e-05,
"loss": 1.1568,
"step": 62500
},
{
"epoch": 1.65,
"learning_rate": 2.2464794272677208e-05,
"loss": 1.158,
"step": 63000
},
{
"epoch": 1.67,
"learning_rate": 2.2246260893888934e-05,
"loss": 1.1526,
"step": 63500
},
{
"epoch": 1.68,
"learning_rate": 2.2027727515100657e-05,
"loss": 1.1663,
"step": 64000
},
{
"epoch": 1.69,
"learning_rate": 2.180919413631238e-05,
"loss": 1.1625,
"step": 64500
},
{
"epoch": 1.7,
"learning_rate": 2.1590660757524106e-05,
"loss": 1.1565,
"step": 65000
},
{
"epoch": 1.72,
"learning_rate": 2.137212737873583e-05,
"loss": 1.1625,
"step": 65500
},
{
"epoch": 1.73,
"learning_rate": 2.1153593999947552e-05,
"loss": 1.1556,
"step": 66000
},
{
"epoch": 1.74,
"learning_rate": 2.093506062115928e-05,
"loss": 1.1553,
"step": 66500
},
{
"epoch": 1.76,
"learning_rate": 2.0716527242371e-05,
"loss": 1.1529,
"step": 67000
},
{
"epoch": 1.77,
"learning_rate": 2.0497993863582724e-05,
"loss": 1.1503,
"step": 67500
},
{
"epoch": 1.78,
"learning_rate": 2.027946048479445e-05,
"loss": 1.1627,
"step": 68000
},
{
"epoch": 1.8,
"learning_rate": 2.0060927106006174e-05,
"loss": 1.1392,
"step": 68500
},
{
"epoch": 1.81,
"learning_rate": 1.9842393727217896e-05,
"loss": 1.1399,
"step": 69000
},
{
"epoch": 1.82,
"learning_rate": 1.962386034842962e-05,
"loss": 1.1443,
"step": 69500
},
{
"epoch": 1.84,
"learning_rate": 1.9405326969641342e-05,
"loss": 1.129,
"step": 70000
},
{
"epoch": 1.85,
"learning_rate": 1.9186793590853065e-05,
"loss": 1.1355,
"step": 70500
},
{
"epoch": 1.86,
"learning_rate": 1.896826021206479e-05,
"loss": 1.1401,
"step": 71000
},
{
"epoch": 1.88,
"learning_rate": 1.8749726833276514e-05,
"loss": 1.1303,
"step": 71500
},
{
"epoch": 1.89,
"learning_rate": 1.8531193454488237e-05,
"loss": 1.1327,
"step": 72000
},
{
"epoch": 1.9,
"learning_rate": 1.8312660075699964e-05,
"loss": 1.1316,
"step": 72500
},
{
"epoch": 1.91,
"learning_rate": 1.8094126696911687e-05,
"loss": 1.1488,
"step": 73000
},
{
"epoch": 1.93,
"learning_rate": 1.787559331812341e-05,
"loss": 1.1397,
"step": 73500
},
{
"epoch": 1.94,
"learning_rate": 1.7657059939335136e-05,
"loss": 1.129,
"step": 74000
},
{
"epoch": 1.95,
"learning_rate": 1.743852656054686e-05,
"loss": 1.1278,
"step": 74500
},
{
"epoch": 1.97,
"learning_rate": 1.721999318175858e-05,
"loss": 1.1146,
"step": 75000
},
{
"epoch": 1.98,
"learning_rate": 1.7001459802970308e-05,
"loss": 1.1195,
"step": 75500
},
{
"epoch": 1.99,
"learning_rate": 1.678292642418203e-05,
"loss": 1.1094,
"step": 76000
},
{
"epoch": 2.01,
"learning_rate": 1.6564393045393754e-05,
"loss": 1.1165,
"step": 76500
},
{
"epoch": 2.02,
"learning_rate": 1.634585966660548e-05,
"loss": 1.103,
"step": 77000
},
{
"epoch": 2.03,
"learning_rate": 1.6127326287817203e-05,
"loss": 1.105,
"step": 77500
},
{
"epoch": 2.05,
"learning_rate": 1.5908792909028926e-05,
"loss": 1.0976,
"step": 78000
},
{
"epoch": 2.06,
"learning_rate": 1.569025953024065e-05,
"loss": 1.1125,
"step": 78500
},
{
"epoch": 2.07,
"learning_rate": 1.547172615145237e-05,
"loss": 1.0944,
"step": 79000
},
{
"epoch": 2.08,
"learning_rate": 1.5253192772664096e-05,
"loss": 1.1034,
"step": 79500
},
{
"epoch": 2.1,
"learning_rate": 1.5034659393875823e-05,
"loss": 1.0957,
"step": 80000
},
{
"epoch": 2.11,
"learning_rate": 1.4816126015087545e-05,
"loss": 1.1147,
"step": 80500
},
{
"epoch": 2.12,
"learning_rate": 1.459759263629927e-05,
"loss": 1.0976,
"step": 81000
},
{
"epoch": 2.14,
"learning_rate": 1.4379059257510993e-05,
"loss": 1.1009,
"step": 81500
},
{
"epoch": 2.15,
"learning_rate": 1.4160525878722716e-05,
"loss": 1.0959,
"step": 82000
},
{
"epoch": 2.16,
"learning_rate": 1.3941992499934442e-05,
"loss": 1.085,
"step": 82500
},
{
"epoch": 2.18,
"learning_rate": 1.3723459121146165e-05,
"loss": 1.0898,
"step": 83000
},
{
"epoch": 2.19,
"learning_rate": 1.3504925742357888e-05,
"loss": 1.0855,
"step": 83500
},
{
"epoch": 2.2,
"learning_rate": 1.3286392363569614e-05,
"loss": 1.0875,
"step": 84000
},
{
"epoch": 2.22,
"learning_rate": 1.3067858984781337e-05,
"loss": 1.0904,
"step": 84500
},
{
"epoch": 2.23,
"learning_rate": 1.284932560599306e-05,
"loss": 1.0889,
"step": 85000
},
{
"epoch": 2.24,
"learning_rate": 1.2630792227204785e-05,
"loss": 1.0947,
"step": 85500
},
{
"epoch": 2.26,
"learning_rate": 1.2412258848416508e-05,
"loss": 1.0895,
"step": 86000
},
{
"epoch": 2.27,
"learning_rate": 1.219372546962823e-05,
"loss": 1.084,
"step": 86500
},
{
"epoch": 2.28,
"learning_rate": 1.1975192090839955e-05,
"loss": 1.0783,
"step": 87000
},
{
"epoch": 2.29,
"learning_rate": 1.175665871205168e-05,
"loss": 1.0837,
"step": 87500
},
{
"epoch": 2.31,
"learning_rate": 1.1538125333263403e-05,
"loss": 1.0832,
"step": 88000
},
{
"epoch": 2.32,
"learning_rate": 1.1319591954475127e-05,
"loss": 1.0814,
"step": 88500
},
{
"epoch": 2.33,
"learning_rate": 1.1101058575686852e-05,
"loss": 1.0842,
"step": 89000
},
{
"epoch": 2.35,
"learning_rate": 1.0882525196898575e-05,
"loss": 1.0766,
"step": 89500
},
{
"epoch": 2.36,
"learning_rate": 1.0663991818110298e-05,
"loss": 1.0814,
"step": 90000
},
{
"epoch": 2.37,
"learning_rate": 1.0445458439322022e-05,
"loss": 1.0742,
"step": 90500
},
{
"epoch": 2.39,
"learning_rate": 1.0226925060533745e-05,
"loss": 1.0761,
"step": 91000
},
{
"epoch": 2.4,
"learning_rate": 1.000839168174547e-05,
"loss": 1.0693,
"step": 91500
},
{
"epoch": 2.41,
"learning_rate": 9.789858302957194e-06,
"loss": 1.0651,
"step": 92000
},
{
"epoch": 2.43,
"learning_rate": 9.571324924168917e-06,
"loss": 1.0774,
"step": 92500
},
{
"epoch": 2.44,
"learning_rate": 9.352791545380642e-06,
"loss": 1.0595,
"step": 93000
},
{
"epoch": 2.45,
"learning_rate": 9.134258166592367e-06,
"loss": 1.0623,
"step": 93500
},
{
"epoch": 2.47,
"learning_rate": 8.91572478780409e-06,
"loss": 1.0658,
"step": 94000
},
{
"epoch": 2.48,
"learning_rate": 8.697191409015812e-06,
"loss": 1.0673,
"step": 94500
},
{
"epoch": 2.49,
"learning_rate": 8.478658030227537e-06,
"loss": 1.063,
"step": 95000
},
{
"epoch": 2.5,
"learning_rate": 8.260124651439262e-06,
"loss": 1.0639,
"step": 95500
},
{
"epoch": 2.52,
"learning_rate": 8.041591272650985e-06,
"loss": 1.0542,
"step": 96000
},
{
"epoch": 2.53,
"learning_rate": 7.823057893862709e-06,
"loss": 1.0651,
"step": 96500
},
{
"epoch": 2.54,
"learning_rate": 7.604524515074434e-06,
"loss": 1.0594,
"step": 97000
},
{
"epoch": 2.56,
"learning_rate": 7.385991136286157e-06,
"loss": 1.0646,
"step": 97500
},
{
"epoch": 2.57,
"learning_rate": 7.16745775749788e-06,
"loss": 1.0554,
"step": 98000
},
{
"epoch": 2.58,
"learning_rate": 6.948924378709605e-06,
"loss": 1.0653,
"step": 98500
},
{
"epoch": 2.6,
"learning_rate": 6.730390999921328e-06,
"loss": 1.0483,
"step": 99000
},
{
"epoch": 2.61,
"learning_rate": 6.5118576211330525e-06,
"loss": 1.063,
"step": 99500
},
{
"epoch": 2.62,
"learning_rate": 6.293324242344776e-06,
"loss": 1.0314,
"step": 100000
},
{
"epoch": 2.64,
"learning_rate": 6.074790863556499e-06,
"loss": 1.0514,
"step": 100500
},
{
"epoch": 2.65,
"learning_rate": 5.856257484768224e-06,
"loss": 1.044,
"step": 101000
},
{
"epoch": 2.66,
"learning_rate": 5.637724105979948e-06,
"loss": 1.0417,
"step": 101500
},
{
"epoch": 2.67,
"learning_rate": 5.419190727191672e-06,
"loss": 1.0528,
"step": 102000
},
{
"epoch": 2.69,
"learning_rate": 5.200657348403395e-06,
"loss": 1.0521,
"step": 102500
},
{
"epoch": 2.7,
"learning_rate": 4.982123969615119e-06,
"loss": 1.0451,
"step": 103000
},
{
"epoch": 2.71,
"learning_rate": 4.7635905908268435e-06,
"loss": 1.0449,
"step": 103500
},
{
"epoch": 2.73,
"learning_rate": 4.545057212038567e-06,
"loss": 1.0481,
"step": 104000
},
{
"epoch": 2.74,
"learning_rate": 4.326523833250291e-06,
"loss": 1.044,
"step": 104500
},
{
"epoch": 2.75,
"learning_rate": 4.107990454462015e-06,
"loss": 1.0465,
"step": 105000
},
{
"epoch": 2.77,
"learning_rate": 3.8894570756737385e-06,
"loss": 1.0466,
"step": 105500
},
{
"epoch": 2.78,
"learning_rate": 3.6709236968854623e-06,
"loss": 1.0337,
"step": 106000
},
{
"epoch": 2.79,
"learning_rate": 3.4523903180971864e-06,
"loss": 1.0418,
"step": 106500
},
{
"epoch": 2.81,
"learning_rate": 3.23385693930891e-06,
"loss": 1.0527,
"step": 107000
},
{
"epoch": 2.82,
"learning_rate": 3.0153235605206344e-06,
"loss": 1.0388,
"step": 107500
},
{
"epoch": 2.83,
"learning_rate": 2.7967901817323577e-06,
"loss": 1.0464,
"step": 108000
},
{
"epoch": 2.85,
"learning_rate": 2.578256802944082e-06,
"loss": 1.0339,
"step": 108500
},
{
"epoch": 2.86,
"learning_rate": 2.3597234241558057e-06,
"loss": 1.0395,
"step": 109000
},
{
"epoch": 2.87,
"learning_rate": 2.1411900453675294e-06,
"loss": 1.0362,
"step": 109500
},
{
"epoch": 2.88,
"learning_rate": 1.922656666579253e-06,
"loss": 1.0408,
"step": 110000
},
{
"epoch": 2.9,
"learning_rate": 1.7041232877909774e-06,
"loss": 1.0354,
"step": 110500
},
{
"epoch": 2.91,
"learning_rate": 1.4855899090027011e-06,
"loss": 1.0315,
"step": 111000
},
{
"epoch": 2.92,
"learning_rate": 1.2670565302144249e-06,
"loss": 1.0385,
"step": 111500
},
{
"epoch": 2.94,
"learning_rate": 1.0485231514261489e-06,
"loss": 1.0426,
"step": 112000
},
{
"epoch": 2.95,
"learning_rate": 8.299897726378727e-07,
"loss": 1.0348,
"step": 112500
},
{
"epoch": 2.96,
"learning_rate": 6.114563938495966e-07,
"loss": 1.0484,
"step": 113000
},
{
"epoch": 2.98,
"learning_rate": 3.929230150613205e-07,
"loss": 1.0376,
"step": 113500
},
{
"epoch": 2.99,
"learning_rate": 1.7438963627304437e-07,
"loss": 1.037,
"step": 114000
},
{
"epoch": 3.0,
"step": 114399,
"total_flos": 1.0669987784955003e+18,
"train_loss": 1.2423339333696508,
"train_runtime": 126615.6451,
"train_samples_per_second": 9.035,
"train_steps_per_second": 0.904
}
],
"max_steps": 114399,
"num_train_epochs": 3,
"total_flos": 1.0669987784955003e+18,
"trial_name": null,
"trial_params": null
}