m2m100_418M_tsn_en_rel_news / trainer_state.json
Davlan's picture
add MT model
f9ae74a
raw
history blame contribute delete
No virus
64.7 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"global_step": 261390,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.990435747350702e-05,
"loss": 2.7399,
"step": 500
},
{
"epoch": 0.01,
"learning_rate": 4.980871494701404e-05,
"loss": 2.1925,
"step": 1000
},
{
"epoch": 0.02,
"learning_rate": 4.971307242052106e-05,
"loss": 1.9923,
"step": 1500
},
{
"epoch": 0.02,
"learning_rate": 4.9617429894028086e-05,
"loss": 1.9085,
"step": 2000
},
{
"epoch": 0.03,
"learning_rate": 4.9521787367535106e-05,
"loss": 1.8605,
"step": 2500
},
{
"epoch": 0.03,
"learning_rate": 4.942614484104212e-05,
"loss": 1.7914,
"step": 3000
},
{
"epoch": 0.04,
"learning_rate": 4.9330502314549144e-05,
"loss": 1.7394,
"step": 3500
},
{
"epoch": 0.05,
"learning_rate": 4.9234859788056164e-05,
"loss": 1.7217,
"step": 4000
},
{
"epoch": 0.05,
"learning_rate": 4.913921726156318e-05,
"loss": 1.6862,
"step": 4500
},
{
"epoch": 0.06,
"learning_rate": 4.90435747350702e-05,
"loss": 1.6501,
"step": 5000
},
{
"epoch": 0.06,
"learning_rate": 4.894793220857722e-05,
"loss": 1.618,
"step": 5500
},
{
"epoch": 0.07,
"learning_rate": 4.885228968208425e-05,
"loss": 1.5957,
"step": 6000
},
{
"epoch": 0.07,
"learning_rate": 4.875664715559126e-05,
"loss": 1.5651,
"step": 6500
},
{
"epoch": 0.08,
"learning_rate": 4.8661004629098287e-05,
"loss": 1.5605,
"step": 7000
},
{
"epoch": 0.09,
"learning_rate": 4.8565362102605306e-05,
"loss": 1.5306,
"step": 7500
},
{
"epoch": 0.09,
"learning_rate": 4.8469719576112325e-05,
"loss": 1.5479,
"step": 8000
},
{
"epoch": 0.1,
"learning_rate": 4.8374077049619345e-05,
"loss": 1.4993,
"step": 8500
},
{
"epoch": 0.1,
"learning_rate": 4.8278434523126364e-05,
"loss": 1.5246,
"step": 9000
},
{
"epoch": 0.11,
"learning_rate": 4.818279199663338e-05,
"loss": 1.4854,
"step": 9500
},
{
"epoch": 0.11,
"learning_rate": 4.808714947014041e-05,
"loss": 1.4766,
"step": 10000
},
{
"epoch": 0.12,
"learning_rate": 4.799150694364742e-05,
"loss": 1.4782,
"step": 10500
},
{
"epoch": 0.13,
"learning_rate": 4.789586441715445e-05,
"loss": 1.4649,
"step": 11000
},
{
"epoch": 0.13,
"learning_rate": 4.780022189066147e-05,
"loss": 1.461,
"step": 11500
},
{
"epoch": 0.14,
"learning_rate": 4.7704579364168487e-05,
"loss": 1.4618,
"step": 12000
},
{
"epoch": 0.14,
"learning_rate": 4.7608936837675506e-05,
"loss": 1.437,
"step": 12500
},
{
"epoch": 0.15,
"learning_rate": 4.7513294311182525e-05,
"loss": 1.429,
"step": 13000
},
{
"epoch": 0.15,
"learning_rate": 4.741765178468955e-05,
"loss": 1.4249,
"step": 13500
},
{
"epoch": 0.16,
"learning_rate": 4.7322009258196564e-05,
"loss": 1.4216,
"step": 14000
},
{
"epoch": 0.17,
"learning_rate": 4.722636673170358e-05,
"loss": 1.3887,
"step": 14500
},
{
"epoch": 0.17,
"learning_rate": 4.713072420521061e-05,
"loss": 1.409,
"step": 15000
},
{
"epoch": 0.18,
"learning_rate": 4.703508167871763e-05,
"loss": 1.3975,
"step": 15500
},
{
"epoch": 0.18,
"learning_rate": 4.693943915222465e-05,
"loss": 1.4014,
"step": 16000
},
{
"epoch": 0.19,
"learning_rate": 4.684379662573167e-05,
"loss": 1.3665,
"step": 16500
},
{
"epoch": 0.2,
"learning_rate": 4.674815409923869e-05,
"loss": 1.3719,
"step": 17000
},
{
"epoch": 0.2,
"learning_rate": 4.6652511572745706e-05,
"loss": 1.3685,
"step": 17500
},
{
"epoch": 0.21,
"learning_rate": 4.6556869046252725e-05,
"loss": 1.3719,
"step": 18000
},
{
"epoch": 0.21,
"learning_rate": 4.646122651975975e-05,
"loss": 1.3583,
"step": 18500
},
{
"epoch": 0.22,
"learning_rate": 4.636558399326677e-05,
"loss": 1.3596,
"step": 19000
},
{
"epoch": 0.22,
"learning_rate": 4.626994146677378e-05,
"loss": 1.3409,
"step": 19500
},
{
"epoch": 0.23,
"learning_rate": 4.617429894028081e-05,
"loss": 1.3693,
"step": 20000
},
{
"epoch": 0.24,
"learning_rate": 4.607865641378783e-05,
"loss": 1.3519,
"step": 20500
},
{
"epoch": 0.24,
"learning_rate": 4.598301388729485e-05,
"loss": 1.3474,
"step": 21000
},
{
"epoch": 0.25,
"learning_rate": 4.588737136080187e-05,
"loss": 1.3305,
"step": 21500
},
{
"epoch": 0.25,
"learning_rate": 4.579172883430889e-05,
"loss": 1.3335,
"step": 22000
},
{
"epoch": 0.26,
"learning_rate": 4.569608630781591e-05,
"loss": 1.324,
"step": 22500
},
{
"epoch": 0.26,
"learning_rate": 4.560044378132293e-05,
"loss": 1.309,
"step": 23000
},
{
"epoch": 0.27,
"learning_rate": 4.550480125482995e-05,
"loss": 1.3019,
"step": 23500
},
{
"epoch": 0.28,
"learning_rate": 4.540915872833697e-05,
"loss": 1.2999,
"step": 24000
},
{
"epoch": 0.28,
"learning_rate": 4.531351620184399e-05,
"loss": 1.3146,
"step": 24500
},
{
"epoch": 0.29,
"learning_rate": 4.521787367535101e-05,
"loss": 1.2995,
"step": 25000
},
{
"epoch": 0.29,
"learning_rate": 4.512223114885803e-05,
"loss": 1.2839,
"step": 25500
},
{
"epoch": 0.3,
"learning_rate": 4.502658862236505e-05,
"loss": 1.3184,
"step": 26000
},
{
"epoch": 0.3,
"learning_rate": 4.4930946095872074e-05,
"loss": 1.3031,
"step": 26500
},
{
"epoch": 0.31,
"learning_rate": 4.483530356937909e-05,
"loss": 1.3199,
"step": 27000
},
{
"epoch": 0.32,
"learning_rate": 4.473966104288611e-05,
"loss": 1.2845,
"step": 27500
},
{
"epoch": 0.32,
"learning_rate": 4.464401851639313e-05,
"loss": 1.2821,
"step": 28000
},
{
"epoch": 0.33,
"learning_rate": 4.454837598990015e-05,
"loss": 1.2603,
"step": 28500
},
{
"epoch": 0.33,
"learning_rate": 4.445273346340717e-05,
"loss": 1.2723,
"step": 29000
},
{
"epoch": 0.34,
"learning_rate": 4.435709093691419e-05,
"loss": 1.27,
"step": 29500
},
{
"epoch": 0.34,
"learning_rate": 4.4261448410421216e-05,
"loss": 1.2514,
"step": 30000
},
{
"epoch": 0.35,
"learning_rate": 4.416580588392823e-05,
"loss": 1.2363,
"step": 30500
},
{
"epoch": 0.36,
"learning_rate": 4.407016335743525e-05,
"loss": 1.2645,
"step": 31000
},
{
"epoch": 0.36,
"learning_rate": 4.3974520830942274e-05,
"loss": 1.2564,
"step": 31500
},
{
"epoch": 0.37,
"learning_rate": 4.3878878304449294e-05,
"loss": 1.2799,
"step": 32000
},
{
"epoch": 0.37,
"learning_rate": 4.378323577795631e-05,
"loss": 1.244,
"step": 32500
},
{
"epoch": 0.38,
"learning_rate": 4.368759325146333e-05,
"loss": 1.2425,
"step": 33000
},
{
"epoch": 0.38,
"learning_rate": 4.359195072497035e-05,
"loss": 1.2347,
"step": 33500
},
{
"epoch": 0.39,
"learning_rate": 4.349630819847738e-05,
"loss": 1.2284,
"step": 34000
},
{
"epoch": 0.4,
"learning_rate": 4.340066567198439e-05,
"loss": 1.2402,
"step": 34500
},
{
"epoch": 0.4,
"learning_rate": 4.3305023145491416e-05,
"loss": 1.2206,
"step": 35000
},
{
"epoch": 0.41,
"learning_rate": 4.3209380618998436e-05,
"loss": 1.2239,
"step": 35500
},
{
"epoch": 0.41,
"learning_rate": 4.311373809250545e-05,
"loss": 1.2436,
"step": 36000
},
{
"epoch": 0.42,
"learning_rate": 4.3018095566012474e-05,
"loss": 1.2062,
"step": 36500
},
{
"epoch": 0.42,
"learning_rate": 4.2922453039519494e-05,
"loss": 1.2363,
"step": 37000
},
{
"epoch": 0.43,
"learning_rate": 4.282681051302651e-05,
"loss": 1.2323,
"step": 37500
},
{
"epoch": 0.44,
"learning_rate": 4.273116798653353e-05,
"loss": 1.2228,
"step": 38000
},
{
"epoch": 0.44,
"learning_rate": 4.263552546004055e-05,
"loss": 1.2277,
"step": 38500
},
{
"epoch": 0.45,
"learning_rate": 4.253988293354758e-05,
"loss": 1.2297,
"step": 39000
},
{
"epoch": 0.45,
"learning_rate": 4.24442404070546e-05,
"loss": 1.2231,
"step": 39500
},
{
"epoch": 0.46,
"learning_rate": 4.2348597880561616e-05,
"loss": 1.2199,
"step": 40000
},
{
"epoch": 0.46,
"learning_rate": 4.2252955354068636e-05,
"loss": 1.2224,
"step": 40500
},
{
"epoch": 0.47,
"learning_rate": 4.2157312827575655e-05,
"loss": 1.1919,
"step": 41000
},
{
"epoch": 0.48,
"learning_rate": 4.2061670301082674e-05,
"loss": 1.1953,
"step": 41500
},
{
"epoch": 0.48,
"learning_rate": 4.1966027774589694e-05,
"loss": 1.2127,
"step": 42000
},
{
"epoch": 0.49,
"learning_rate": 4.187038524809671e-05,
"loss": 1.2071,
"step": 42500
},
{
"epoch": 0.49,
"learning_rate": 4.177474272160374e-05,
"loss": 1.1869,
"step": 43000
},
{
"epoch": 0.5,
"learning_rate": 4.167910019511075e-05,
"loss": 1.1954,
"step": 43500
},
{
"epoch": 0.5,
"learning_rate": 4.158345766861778e-05,
"loss": 1.2254,
"step": 44000
},
{
"epoch": 0.51,
"learning_rate": 4.14878151421248e-05,
"loss": 1.1943,
"step": 44500
},
{
"epoch": 0.52,
"learning_rate": 4.1392172615631816e-05,
"loss": 1.1864,
"step": 45000
},
{
"epoch": 0.52,
"learning_rate": 4.1296530089138836e-05,
"loss": 1.1929,
"step": 45500
},
{
"epoch": 0.53,
"learning_rate": 4.1200887562645855e-05,
"loss": 1.187,
"step": 46000
},
{
"epoch": 0.53,
"learning_rate": 4.110524503615288e-05,
"loss": 1.17,
"step": 46500
},
{
"epoch": 0.54,
"learning_rate": 4.1009602509659894e-05,
"loss": 1.1931,
"step": 47000
},
{
"epoch": 0.55,
"learning_rate": 4.091395998316691e-05,
"loss": 1.1988,
"step": 47500
},
{
"epoch": 0.55,
"learning_rate": 4.081831745667394e-05,
"loss": 1.2007,
"step": 48000
},
{
"epoch": 0.56,
"learning_rate": 4.072267493018096e-05,
"loss": 1.1844,
"step": 48500
},
{
"epoch": 0.56,
"learning_rate": 4.062703240368798e-05,
"loss": 1.1685,
"step": 49000
},
{
"epoch": 0.57,
"learning_rate": 4.0531389877195e-05,
"loss": 1.1781,
"step": 49500
},
{
"epoch": 0.57,
"learning_rate": 4.0435747350702016e-05,
"loss": 1.1566,
"step": 50000
},
{
"epoch": 0.58,
"learning_rate": 4.034010482420904e-05,
"loss": 1.1747,
"step": 50500
},
{
"epoch": 0.59,
"learning_rate": 4.0244462297716055e-05,
"loss": 1.1879,
"step": 51000
},
{
"epoch": 0.59,
"learning_rate": 4.014881977122308e-05,
"loss": 1.161,
"step": 51500
},
{
"epoch": 0.6,
"learning_rate": 4.00531772447301e-05,
"loss": 1.1593,
"step": 52000
},
{
"epoch": 0.6,
"learning_rate": 3.995753471823711e-05,
"loss": 1.1584,
"step": 52500
},
{
"epoch": 0.61,
"learning_rate": 3.986189219174414e-05,
"loss": 1.1743,
"step": 53000
},
{
"epoch": 0.61,
"learning_rate": 3.976624966525116e-05,
"loss": 1.1564,
"step": 53500
},
{
"epoch": 0.62,
"learning_rate": 3.967060713875818e-05,
"loss": 1.1662,
"step": 54000
},
{
"epoch": 0.63,
"learning_rate": 3.95749646122652e-05,
"loss": 1.1691,
"step": 54500
},
{
"epoch": 0.63,
"learning_rate": 3.9479322085772217e-05,
"loss": 1.1628,
"step": 55000
},
{
"epoch": 0.64,
"learning_rate": 3.938367955927924e-05,
"loss": 1.1515,
"step": 55500
},
{
"epoch": 0.64,
"learning_rate": 3.928803703278626e-05,
"loss": 1.1561,
"step": 56000
},
{
"epoch": 0.65,
"learning_rate": 3.919239450629328e-05,
"loss": 1.1771,
"step": 56500
},
{
"epoch": 0.65,
"learning_rate": 3.90967519798003e-05,
"loss": 1.1506,
"step": 57000
},
{
"epoch": 0.66,
"learning_rate": 3.900110945330732e-05,
"loss": 1.1294,
"step": 57500
},
{
"epoch": 0.67,
"learning_rate": 3.890546692681434e-05,
"loss": 1.1396,
"step": 58000
},
{
"epoch": 0.67,
"learning_rate": 3.880982440032136e-05,
"loss": 1.1377,
"step": 58500
},
{
"epoch": 0.68,
"learning_rate": 3.871418187382838e-05,
"loss": 1.1266,
"step": 59000
},
{
"epoch": 0.68,
"learning_rate": 3.8618539347335404e-05,
"loss": 1.1561,
"step": 59500
},
{
"epoch": 0.69,
"learning_rate": 3.8522896820842417e-05,
"loss": 1.1461,
"step": 60000
},
{
"epoch": 0.69,
"learning_rate": 3.842725429434944e-05,
"loss": 1.1404,
"step": 60500
},
{
"epoch": 0.7,
"learning_rate": 3.833161176785646e-05,
"loss": 1.1372,
"step": 61000
},
{
"epoch": 0.71,
"learning_rate": 3.823596924136348e-05,
"loss": 1.1424,
"step": 61500
},
{
"epoch": 0.71,
"learning_rate": 3.81403267148705e-05,
"loss": 1.1142,
"step": 62000
},
{
"epoch": 0.72,
"learning_rate": 3.804468418837752e-05,
"loss": 1.136,
"step": 62500
},
{
"epoch": 0.72,
"learning_rate": 3.7949041661884546e-05,
"loss": 1.1272,
"step": 63000
},
{
"epoch": 0.73,
"learning_rate": 3.785339913539156e-05,
"loss": 1.1305,
"step": 63500
},
{
"epoch": 0.73,
"learning_rate": 3.775775660889858e-05,
"loss": 1.1445,
"step": 64000
},
{
"epoch": 0.74,
"learning_rate": 3.7662114082405604e-05,
"loss": 1.1253,
"step": 64500
},
{
"epoch": 0.75,
"learning_rate": 3.7566471555912623e-05,
"loss": 1.1431,
"step": 65000
},
{
"epoch": 0.75,
"learning_rate": 3.747082902941964e-05,
"loss": 1.1177,
"step": 65500
},
{
"epoch": 0.76,
"learning_rate": 3.737518650292666e-05,
"loss": 1.1116,
"step": 66000
},
{
"epoch": 0.76,
"learning_rate": 3.727954397643368e-05,
"loss": 1.1188,
"step": 66500
},
{
"epoch": 0.77,
"learning_rate": 3.718390144994071e-05,
"loss": 1.1282,
"step": 67000
},
{
"epoch": 0.77,
"learning_rate": 3.708825892344772e-05,
"loss": 1.1238,
"step": 67500
},
{
"epoch": 0.78,
"learning_rate": 3.6992616396954746e-05,
"loss": 1.1184,
"step": 68000
},
{
"epoch": 0.79,
"learning_rate": 3.6896973870461766e-05,
"loss": 1.1025,
"step": 68500
},
{
"epoch": 0.79,
"learning_rate": 3.6801331343968785e-05,
"loss": 1.1305,
"step": 69000
},
{
"epoch": 0.8,
"learning_rate": 3.6705688817475804e-05,
"loss": 1.1119,
"step": 69500
},
{
"epoch": 0.8,
"learning_rate": 3.6610046290982823e-05,
"loss": 1.1053,
"step": 70000
},
{
"epoch": 0.81,
"learning_rate": 3.651440376448985e-05,
"loss": 1.1206,
"step": 70500
},
{
"epoch": 0.81,
"learning_rate": 3.641876123799686e-05,
"loss": 1.098,
"step": 71000
},
{
"epoch": 0.82,
"learning_rate": 3.632311871150388e-05,
"loss": 1.105,
"step": 71500
},
{
"epoch": 0.83,
"learning_rate": 3.622747618501091e-05,
"loss": 1.103,
"step": 72000
},
{
"epoch": 0.83,
"learning_rate": 3.613183365851793e-05,
"loss": 1.1052,
"step": 72500
},
{
"epoch": 0.84,
"learning_rate": 3.6036191132024946e-05,
"loss": 1.1197,
"step": 73000
},
{
"epoch": 0.84,
"learning_rate": 3.5940548605531966e-05,
"loss": 1.0951,
"step": 73500
},
{
"epoch": 0.85,
"learning_rate": 3.5844906079038985e-05,
"loss": 1.0932,
"step": 74000
},
{
"epoch": 0.86,
"learning_rate": 3.574926355254601e-05,
"loss": 1.09,
"step": 74500
},
{
"epoch": 0.86,
"learning_rate": 3.5653621026053024e-05,
"loss": 1.0815,
"step": 75000
},
{
"epoch": 0.87,
"learning_rate": 3.555797849956004e-05,
"loss": 1.1079,
"step": 75500
},
{
"epoch": 0.87,
"learning_rate": 3.546233597306707e-05,
"loss": 1.0864,
"step": 76000
},
{
"epoch": 0.88,
"learning_rate": 3.536669344657408e-05,
"loss": 1.1141,
"step": 76500
},
{
"epoch": 0.88,
"learning_rate": 3.527105092008111e-05,
"loss": 1.0849,
"step": 77000
},
{
"epoch": 0.89,
"learning_rate": 3.517540839358813e-05,
"loss": 1.0931,
"step": 77500
},
{
"epoch": 0.9,
"learning_rate": 3.5079765867095146e-05,
"loss": 1.1056,
"step": 78000
},
{
"epoch": 0.9,
"learning_rate": 3.4984123340602166e-05,
"loss": 1.0786,
"step": 78500
},
{
"epoch": 0.91,
"learning_rate": 3.4888480814109185e-05,
"loss": 1.0934,
"step": 79000
},
{
"epoch": 0.91,
"learning_rate": 3.479283828761621e-05,
"loss": 1.1029,
"step": 79500
},
{
"epoch": 0.92,
"learning_rate": 3.469719576112323e-05,
"loss": 1.084,
"step": 80000
},
{
"epoch": 0.92,
"learning_rate": 3.460155323463024e-05,
"loss": 1.1172,
"step": 80500
},
{
"epoch": 0.93,
"learning_rate": 3.450591070813727e-05,
"loss": 1.0789,
"step": 81000
},
{
"epoch": 0.94,
"learning_rate": 3.441026818164429e-05,
"loss": 1.0873,
"step": 81500
},
{
"epoch": 0.94,
"learning_rate": 3.431462565515131e-05,
"loss": 1.0903,
"step": 82000
},
{
"epoch": 0.95,
"learning_rate": 3.421898312865833e-05,
"loss": 1.0811,
"step": 82500
},
{
"epoch": 0.95,
"learning_rate": 3.4123340602165346e-05,
"loss": 1.0691,
"step": 83000
},
{
"epoch": 0.96,
"learning_rate": 3.402769807567237e-05,
"loss": 1.0954,
"step": 83500
},
{
"epoch": 0.96,
"learning_rate": 3.3932055549179385e-05,
"loss": 1.0743,
"step": 84000
},
{
"epoch": 0.97,
"learning_rate": 3.383641302268641e-05,
"loss": 1.0826,
"step": 84500
},
{
"epoch": 0.98,
"learning_rate": 3.374077049619343e-05,
"loss": 1.0678,
"step": 85000
},
{
"epoch": 0.98,
"learning_rate": 3.364512796970045e-05,
"loss": 1.0914,
"step": 85500
},
{
"epoch": 0.99,
"learning_rate": 3.354948544320747e-05,
"loss": 1.0577,
"step": 86000
},
{
"epoch": 0.99,
"learning_rate": 3.345384291671449e-05,
"loss": 1.0885,
"step": 86500
},
{
"epoch": 1.0,
"learning_rate": 3.3358200390221515e-05,
"loss": 1.0647,
"step": 87000
},
{
"epoch": 1.0,
"learning_rate": 3.326255786372853e-05,
"loss": 0.9806,
"step": 87500
},
{
"epoch": 1.01,
"learning_rate": 3.3166915337235546e-05,
"loss": 0.9697,
"step": 88000
},
{
"epoch": 1.02,
"learning_rate": 3.307127281074257e-05,
"loss": 0.9644,
"step": 88500
},
{
"epoch": 1.02,
"learning_rate": 3.297563028424959e-05,
"loss": 0.9666,
"step": 89000
},
{
"epoch": 1.03,
"learning_rate": 3.287998775775661e-05,
"loss": 0.9604,
"step": 89500
},
{
"epoch": 1.03,
"learning_rate": 3.278434523126363e-05,
"loss": 0.9707,
"step": 90000
},
{
"epoch": 1.04,
"learning_rate": 3.268870270477065e-05,
"loss": 0.9643,
"step": 90500
},
{
"epoch": 1.04,
"learning_rate": 3.2593060178277676e-05,
"loss": 0.9713,
"step": 91000
},
{
"epoch": 1.05,
"learning_rate": 3.249741765178469e-05,
"loss": 0.9726,
"step": 91500
},
{
"epoch": 1.06,
"learning_rate": 3.2401775125291715e-05,
"loss": 0.9616,
"step": 92000
},
{
"epoch": 1.06,
"learning_rate": 3.2306132598798734e-05,
"loss": 0.9632,
"step": 92500
},
{
"epoch": 1.07,
"learning_rate": 3.2210490072305746e-05,
"loss": 0.9728,
"step": 93000
},
{
"epoch": 1.07,
"learning_rate": 3.211484754581277e-05,
"loss": 0.9656,
"step": 93500
},
{
"epoch": 1.08,
"learning_rate": 3.201920501931979e-05,
"loss": 0.9649,
"step": 94000
},
{
"epoch": 1.08,
"learning_rate": 3.192356249282681e-05,
"loss": 0.9788,
"step": 94500
},
{
"epoch": 1.09,
"learning_rate": 3.182791996633383e-05,
"loss": 0.9396,
"step": 95000
},
{
"epoch": 1.1,
"learning_rate": 3.173227743984085e-05,
"loss": 0.9528,
"step": 95500
},
{
"epoch": 1.1,
"learning_rate": 3.1636634913347876e-05,
"loss": 0.9708,
"step": 96000
},
{
"epoch": 1.11,
"learning_rate": 3.1540992386854895e-05,
"loss": 0.9705,
"step": 96500
},
{
"epoch": 1.11,
"learning_rate": 3.144534986036191e-05,
"loss": 0.9715,
"step": 97000
},
{
"epoch": 1.12,
"learning_rate": 3.1349707333868934e-05,
"loss": 0.964,
"step": 97500
},
{
"epoch": 1.12,
"learning_rate": 3.125406480737595e-05,
"loss": 0.9567,
"step": 98000
},
{
"epoch": 1.13,
"learning_rate": 3.115842228088297e-05,
"loss": 0.9558,
"step": 98500
},
{
"epoch": 1.14,
"learning_rate": 3.106277975438999e-05,
"loss": 0.9592,
"step": 99000
},
{
"epoch": 1.14,
"learning_rate": 3.096713722789701e-05,
"loss": 0.9711,
"step": 99500
},
{
"epoch": 1.15,
"learning_rate": 3.087149470140404e-05,
"loss": 0.9692,
"step": 100000
},
{
"epoch": 1.15,
"learning_rate": 3.077585217491105e-05,
"loss": 0.9587,
"step": 100500
},
{
"epoch": 1.16,
"learning_rate": 3.0680209648418076e-05,
"loss": 0.9644,
"step": 101000
},
{
"epoch": 1.16,
"learning_rate": 3.0584567121925095e-05,
"loss": 0.9365,
"step": 101500
},
{
"epoch": 1.17,
"learning_rate": 3.0488924595432118e-05,
"loss": 0.9649,
"step": 102000
},
{
"epoch": 1.18,
"learning_rate": 3.0393282068939134e-05,
"loss": 0.9637,
"step": 102500
},
{
"epoch": 1.18,
"learning_rate": 3.0297639542446153e-05,
"loss": 0.9692,
"step": 103000
},
{
"epoch": 1.19,
"learning_rate": 3.0201997015953176e-05,
"loss": 0.9625,
"step": 103500
},
{
"epoch": 1.19,
"learning_rate": 3.0106354489460192e-05,
"loss": 0.9725,
"step": 104000
},
{
"epoch": 1.2,
"learning_rate": 3.0010711962967215e-05,
"loss": 0.9371,
"step": 104500
},
{
"epoch": 1.21,
"learning_rate": 2.9915069436474234e-05,
"loss": 0.9431,
"step": 105000
},
{
"epoch": 1.21,
"learning_rate": 2.9819426909981257e-05,
"loss": 0.9675,
"step": 105500
},
{
"epoch": 1.22,
"learning_rate": 2.9723784383488273e-05,
"loss": 0.9565,
"step": 106000
},
{
"epoch": 1.22,
"learning_rate": 2.9628141856995295e-05,
"loss": 0.9564,
"step": 106500
},
{
"epoch": 1.23,
"learning_rate": 2.9532499330502318e-05,
"loss": 0.9595,
"step": 107000
},
{
"epoch": 1.23,
"learning_rate": 2.9436856804009337e-05,
"loss": 0.9611,
"step": 107500
},
{
"epoch": 1.24,
"learning_rate": 2.9341214277516353e-05,
"loss": 0.9397,
"step": 108000
},
{
"epoch": 1.25,
"learning_rate": 2.9245571751023376e-05,
"loss": 0.9855,
"step": 108500
},
{
"epoch": 1.25,
"learning_rate": 2.91499292245304e-05,
"loss": 0.9613,
"step": 109000
},
{
"epoch": 1.26,
"learning_rate": 2.9054286698037415e-05,
"loss": 0.978,
"step": 109500
},
{
"epoch": 1.26,
"learning_rate": 2.8958644171544434e-05,
"loss": 0.9694,
"step": 110000
},
{
"epoch": 1.27,
"learning_rate": 2.8863001645051457e-05,
"loss": 0.9564,
"step": 110500
},
{
"epoch": 1.27,
"learning_rate": 2.876735911855848e-05,
"loss": 0.9546,
"step": 111000
},
{
"epoch": 1.28,
"learning_rate": 2.8671716592065495e-05,
"loss": 0.9345,
"step": 111500
},
{
"epoch": 1.29,
"learning_rate": 2.8576074065572518e-05,
"loss": 0.9413,
"step": 112000
},
{
"epoch": 1.29,
"learning_rate": 2.8480431539079538e-05,
"loss": 0.9674,
"step": 112500
},
{
"epoch": 1.3,
"learning_rate": 2.838478901258656e-05,
"loss": 0.9611,
"step": 113000
},
{
"epoch": 1.3,
"learning_rate": 2.8289146486093576e-05,
"loss": 0.944,
"step": 113500
},
{
"epoch": 1.31,
"learning_rate": 2.81935039596006e-05,
"loss": 0.9532,
"step": 114000
},
{
"epoch": 1.31,
"learning_rate": 2.8097861433107618e-05,
"loss": 0.9556,
"step": 114500
},
{
"epoch": 1.32,
"learning_rate": 2.800221890661464e-05,
"loss": 0.9539,
"step": 115000
},
{
"epoch": 1.33,
"learning_rate": 2.7906576380121657e-05,
"loss": 0.9525,
"step": 115500
},
{
"epoch": 1.33,
"learning_rate": 2.781093385362868e-05,
"loss": 0.9495,
"step": 116000
},
{
"epoch": 1.34,
"learning_rate": 2.7715291327135702e-05,
"loss": 0.9585,
"step": 116500
},
{
"epoch": 1.34,
"learning_rate": 2.7619648800642718e-05,
"loss": 0.959,
"step": 117000
},
{
"epoch": 1.35,
"learning_rate": 2.7524006274149738e-05,
"loss": 0.9503,
"step": 117500
},
{
"epoch": 1.35,
"learning_rate": 2.742836374765676e-05,
"loss": 0.9444,
"step": 118000
},
{
"epoch": 1.36,
"learning_rate": 2.7332721221163783e-05,
"loss": 0.9495,
"step": 118500
},
{
"epoch": 1.37,
"learning_rate": 2.72370786946708e-05,
"loss": 0.9273,
"step": 119000
},
{
"epoch": 1.37,
"learning_rate": 2.7141436168177818e-05,
"loss": 0.952,
"step": 119500
},
{
"epoch": 1.38,
"learning_rate": 2.704579364168484e-05,
"loss": 0.9455,
"step": 120000
},
{
"epoch": 1.38,
"learning_rate": 2.6950151115191864e-05,
"loss": 0.9475,
"step": 120500
},
{
"epoch": 1.39,
"learning_rate": 2.685450858869888e-05,
"loss": 0.9348,
"step": 121000
},
{
"epoch": 1.39,
"learning_rate": 2.67588660622059e-05,
"loss": 0.9479,
"step": 121500
},
{
"epoch": 1.4,
"learning_rate": 2.666322353571292e-05,
"loss": 0.9603,
"step": 122000
},
{
"epoch": 1.41,
"learning_rate": 2.6567581009219938e-05,
"loss": 0.9276,
"step": 122500
},
{
"epoch": 1.41,
"learning_rate": 2.647193848272696e-05,
"loss": 0.9558,
"step": 123000
},
{
"epoch": 1.42,
"learning_rate": 2.6376295956233983e-05,
"loss": 0.9321,
"step": 123500
},
{
"epoch": 1.42,
"learning_rate": 2.6280653429741002e-05,
"loss": 0.9342,
"step": 124000
},
{
"epoch": 1.43,
"learning_rate": 2.618501090324802e-05,
"loss": 0.9395,
"step": 124500
},
{
"epoch": 1.43,
"learning_rate": 2.608936837675504e-05,
"loss": 0.9337,
"step": 125000
},
{
"epoch": 1.44,
"learning_rate": 2.5993725850262064e-05,
"loss": 0.9475,
"step": 125500
},
{
"epoch": 1.45,
"learning_rate": 2.5898083323769083e-05,
"loss": 0.9379,
"step": 126000
},
{
"epoch": 1.45,
"learning_rate": 2.58024407972761e-05,
"loss": 0.929,
"step": 126500
},
{
"epoch": 1.46,
"learning_rate": 2.5706798270783122e-05,
"loss": 0.9316,
"step": 127000
},
{
"epoch": 1.46,
"learning_rate": 2.5611155744290144e-05,
"loss": 0.9415,
"step": 127500
},
{
"epoch": 1.47,
"learning_rate": 2.551551321779716e-05,
"loss": 0.933,
"step": 128000
},
{
"epoch": 1.47,
"learning_rate": 2.5419870691304183e-05,
"loss": 0.9477,
"step": 128500
},
{
"epoch": 1.48,
"learning_rate": 2.5324228164811202e-05,
"loss": 0.9096,
"step": 129000
},
{
"epoch": 1.49,
"learning_rate": 2.5228585638318225e-05,
"loss": 0.9161,
"step": 129500
},
{
"epoch": 1.49,
"learning_rate": 2.513294311182524e-05,
"loss": 0.9535,
"step": 130000
},
{
"epoch": 1.5,
"learning_rate": 2.5037300585332264e-05,
"loss": 0.9386,
"step": 130500
},
{
"epoch": 1.5,
"learning_rate": 2.4941658058839283e-05,
"loss": 0.944,
"step": 131000
},
{
"epoch": 1.51,
"learning_rate": 2.4846015532346302e-05,
"loss": 0.934,
"step": 131500
},
{
"epoch": 1.51,
"learning_rate": 2.4750373005853325e-05,
"loss": 0.9243,
"step": 132000
},
{
"epoch": 1.52,
"learning_rate": 2.4654730479360344e-05,
"loss": 0.9215,
"step": 132500
},
{
"epoch": 1.53,
"learning_rate": 2.4559087952867364e-05,
"loss": 0.9271,
"step": 133000
},
{
"epoch": 1.53,
"learning_rate": 2.4463445426374383e-05,
"loss": 0.9442,
"step": 133500
},
{
"epoch": 1.54,
"learning_rate": 2.4367802899881402e-05,
"loss": 0.936,
"step": 134000
},
{
"epoch": 1.54,
"learning_rate": 2.4272160373388425e-05,
"loss": 0.9393,
"step": 134500
},
{
"epoch": 1.55,
"learning_rate": 2.4176517846895445e-05,
"loss": 0.9434,
"step": 135000
},
{
"epoch": 1.56,
"learning_rate": 2.4080875320402467e-05,
"loss": 0.9383,
"step": 135500
},
{
"epoch": 1.56,
"learning_rate": 2.3985232793909483e-05,
"loss": 0.9275,
"step": 136000
},
{
"epoch": 1.57,
"learning_rate": 2.3889590267416506e-05,
"loss": 0.9129,
"step": 136500
},
{
"epoch": 1.57,
"learning_rate": 2.3793947740923525e-05,
"loss": 0.937,
"step": 137000
},
{
"epoch": 1.58,
"learning_rate": 2.3698305214430548e-05,
"loss": 0.9158,
"step": 137500
},
{
"epoch": 1.58,
"learning_rate": 2.3602662687937567e-05,
"loss": 0.9184,
"step": 138000
},
{
"epoch": 1.59,
"learning_rate": 2.3507020161444583e-05,
"loss": 0.9214,
"step": 138500
},
{
"epoch": 1.6,
"learning_rate": 2.3411377634951606e-05,
"loss": 0.9188,
"step": 139000
},
{
"epoch": 1.6,
"learning_rate": 2.3315735108458625e-05,
"loss": 0.9209,
"step": 139500
},
{
"epoch": 1.61,
"learning_rate": 2.3220092581965648e-05,
"loss": 0.9111,
"step": 140000
},
{
"epoch": 1.61,
"learning_rate": 2.3124450055472667e-05,
"loss": 0.9279,
"step": 140500
},
{
"epoch": 1.62,
"learning_rate": 2.3028807528979687e-05,
"loss": 0.919,
"step": 141000
},
{
"epoch": 1.62,
"learning_rate": 2.2933165002486706e-05,
"loss": 0.9202,
"step": 141500
},
{
"epoch": 1.63,
"learning_rate": 2.283752247599373e-05,
"loss": 0.9126,
"step": 142000
},
{
"epoch": 1.64,
"learning_rate": 2.2741879949500748e-05,
"loss": 0.9277,
"step": 142500
},
{
"epoch": 1.64,
"learning_rate": 2.2646237423007767e-05,
"loss": 0.9171,
"step": 143000
},
{
"epoch": 1.65,
"learning_rate": 2.2550594896514787e-05,
"loss": 0.9017,
"step": 143500
},
{
"epoch": 1.65,
"learning_rate": 2.245495237002181e-05,
"loss": 0.9363,
"step": 144000
},
{
"epoch": 1.66,
"learning_rate": 2.235930984352883e-05,
"loss": 0.9139,
"step": 144500
},
{
"epoch": 1.66,
"learning_rate": 2.2263667317035848e-05,
"loss": 0.9055,
"step": 145000
},
{
"epoch": 1.67,
"learning_rate": 2.2168024790542867e-05,
"loss": 0.9028,
"step": 145500
},
{
"epoch": 1.68,
"learning_rate": 2.2072382264049887e-05,
"loss": 0.9026,
"step": 146000
},
{
"epoch": 1.68,
"learning_rate": 2.197673973755691e-05,
"loss": 0.9143,
"step": 146500
},
{
"epoch": 1.69,
"learning_rate": 2.188109721106393e-05,
"loss": 0.9231,
"step": 147000
},
{
"epoch": 1.69,
"learning_rate": 2.1785454684570948e-05,
"loss": 0.902,
"step": 147500
},
{
"epoch": 1.7,
"learning_rate": 2.1689812158077967e-05,
"loss": 0.9027,
"step": 148000
},
{
"epoch": 1.7,
"learning_rate": 2.159416963158499e-05,
"loss": 0.9204,
"step": 148500
},
{
"epoch": 1.71,
"learning_rate": 2.149852710509201e-05,
"loss": 0.9197,
"step": 149000
},
{
"epoch": 1.72,
"learning_rate": 2.1402884578599032e-05,
"loss": 0.9038,
"step": 149500
},
{
"epoch": 1.72,
"learning_rate": 2.1307242052106048e-05,
"loss": 0.8939,
"step": 150000
},
{
"epoch": 1.73,
"learning_rate": 2.1211599525613067e-05,
"loss": 0.9135,
"step": 150500
},
{
"epoch": 1.73,
"learning_rate": 2.111595699912009e-05,
"loss": 0.9099,
"step": 151000
},
{
"epoch": 1.74,
"learning_rate": 2.102031447262711e-05,
"loss": 0.9195,
"step": 151500
},
{
"epoch": 1.74,
"learning_rate": 2.0924671946134132e-05,
"loss": 0.9004,
"step": 152000
},
{
"epoch": 1.75,
"learning_rate": 2.0829029419641148e-05,
"loss": 0.9123,
"step": 152500
},
{
"epoch": 1.76,
"learning_rate": 2.073338689314817e-05,
"loss": 0.9008,
"step": 153000
},
{
"epoch": 1.76,
"learning_rate": 2.063774436665519e-05,
"loss": 0.9081,
"step": 153500
},
{
"epoch": 1.77,
"learning_rate": 2.0542101840162213e-05,
"loss": 0.9003,
"step": 154000
},
{
"epoch": 1.77,
"learning_rate": 2.0446459313669232e-05,
"loss": 0.9173,
"step": 154500
},
{
"epoch": 1.78,
"learning_rate": 2.035081678717625e-05,
"loss": 0.9174,
"step": 155000
},
{
"epoch": 1.78,
"learning_rate": 2.025517426068327e-05,
"loss": 0.8964,
"step": 155500
},
{
"epoch": 1.79,
"learning_rate": 2.015953173419029e-05,
"loss": 0.8951,
"step": 156000
},
{
"epoch": 1.8,
"learning_rate": 2.0063889207697313e-05,
"loss": 0.8963,
"step": 156500
},
{
"epoch": 1.8,
"learning_rate": 1.9968246681204332e-05,
"loss": 0.9004,
"step": 157000
},
{
"epoch": 1.81,
"learning_rate": 1.987260415471135e-05,
"loss": 0.9143,
"step": 157500
},
{
"epoch": 1.81,
"learning_rate": 1.977696162821837e-05,
"loss": 0.9057,
"step": 158000
},
{
"epoch": 1.82,
"learning_rate": 1.9681319101725394e-05,
"loss": 0.9035,
"step": 158500
},
{
"epoch": 1.82,
"learning_rate": 1.9585676575232413e-05,
"loss": 0.8944,
"step": 159000
},
{
"epoch": 1.83,
"learning_rate": 1.9490034048739432e-05,
"loss": 0.8942,
"step": 159500
},
{
"epoch": 1.84,
"learning_rate": 1.939439152224645e-05,
"loss": 0.8988,
"step": 160000
},
{
"epoch": 1.84,
"learning_rate": 1.9298748995753474e-05,
"loss": 0.8968,
"step": 160500
},
{
"epoch": 1.85,
"learning_rate": 1.9203106469260494e-05,
"loss": 0.9141,
"step": 161000
},
{
"epoch": 1.85,
"learning_rate": 1.9107463942767513e-05,
"loss": 0.8917,
"step": 161500
},
{
"epoch": 1.86,
"learning_rate": 1.9011821416274532e-05,
"loss": 0.8946,
"step": 162000
},
{
"epoch": 1.87,
"learning_rate": 1.891617888978155e-05,
"loss": 0.9182,
"step": 162500
},
{
"epoch": 1.87,
"learning_rate": 1.8820536363288574e-05,
"loss": 0.895,
"step": 163000
},
{
"epoch": 1.88,
"learning_rate": 1.8724893836795594e-05,
"loss": 0.8787,
"step": 163500
},
{
"epoch": 1.88,
"learning_rate": 1.8629251310302613e-05,
"loss": 0.9011,
"step": 164000
},
{
"epoch": 1.89,
"learning_rate": 1.8533608783809632e-05,
"loss": 0.8866,
"step": 164500
},
{
"epoch": 1.89,
"learning_rate": 1.8437966257316655e-05,
"loss": 0.903,
"step": 165000
},
{
"epoch": 1.9,
"learning_rate": 1.8342323730823674e-05,
"loss": 0.9039,
"step": 165500
},
{
"epoch": 1.91,
"learning_rate": 1.8246681204330697e-05,
"loss": 0.901,
"step": 166000
},
{
"epoch": 1.91,
"learning_rate": 1.8151038677837713e-05,
"loss": 0.906,
"step": 166500
},
{
"epoch": 1.92,
"learning_rate": 1.8055396151344736e-05,
"loss": 0.8961,
"step": 167000
},
{
"epoch": 1.92,
"learning_rate": 1.7959753624851755e-05,
"loss": 0.8867,
"step": 167500
},
{
"epoch": 1.93,
"learning_rate": 1.7864111098358774e-05,
"loss": 0.888,
"step": 168000
},
{
"epoch": 1.93,
"learning_rate": 1.7768468571865797e-05,
"loss": 0.908,
"step": 168500
},
{
"epoch": 1.94,
"learning_rate": 1.7672826045372813e-05,
"loss": 0.8863,
"step": 169000
},
{
"epoch": 1.95,
"learning_rate": 1.7577183518879836e-05,
"loss": 0.8975,
"step": 169500
},
{
"epoch": 1.95,
"learning_rate": 1.7481540992386855e-05,
"loss": 0.902,
"step": 170000
},
{
"epoch": 1.96,
"learning_rate": 1.7385898465893878e-05,
"loss": 0.8892,
"step": 170500
},
{
"epoch": 1.96,
"learning_rate": 1.7290255939400897e-05,
"loss": 0.8917,
"step": 171000
},
{
"epoch": 1.97,
"learning_rate": 1.7194613412907916e-05,
"loss": 0.8875,
"step": 171500
},
{
"epoch": 1.97,
"learning_rate": 1.7098970886414936e-05,
"loss": 0.8865,
"step": 172000
},
{
"epoch": 1.98,
"learning_rate": 1.700332835992196e-05,
"loss": 0.876,
"step": 172500
},
{
"epoch": 1.99,
"learning_rate": 1.6907685833428978e-05,
"loss": 0.8971,
"step": 173000
},
{
"epoch": 1.99,
"learning_rate": 1.6812043306935997e-05,
"loss": 0.8861,
"step": 173500
},
{
"epoch": 2.0,
"learning_rate": 1.6716400780443016e-05,
"loss": 0.9066,
"step": 174000
},
{
"epoch": 2.0,
"learning_rate": 1.6620758253950036e-05,
"loss": 0.8281,
"step": 174500
},
{
"epoch": 2.01,
"learning_rate": 1.652511572745706e-05,
"loss": 0.7694,
"step": 175000
},
{
"epoch": 2.01,
"learning_rate": 1.6429473200964078e-05,
"loss": 0.7686,
"step": 175500
},
{
"epoch": 2.02,
"learning_rate": 1.6333830674471097e-05,
"loss": 0.7825,
"step": 176000
},
{
"epoch": 2.03,
"learning_rate": 1.6238188147978116e-05,
"loss": 0.7796,
"step": 176500
},
{
"epoch": 2.03,
"learning_rate": 1.614254562148514e-05,
"loss": 0.7767,
"step": 177000
},
{
"epoch": 2.04,
"learning_rate": 1.604690309499216e-05,
"loss": 0.7784,
"step": 177500
},
{
"epoch": 2.04,
"learning_rate": 1.5951260568499178e-05,
"loss": 0.7735,
"step": 178000
},
{
"epoch": 2.05,
"learning_rate": 1.5855618042006197e-05,
"loss": 0.7804,
"step": 178500
},
{
"epoch": 2.05,
"learning_rate": 1.5759975515513217e-05,
"loss": 0.7778,
"step": 179000
},
{
"epoch": 2.06,
"learning_rate": 1.566433298902024e-05,
"loss": 0.7776,
"step": 179500
},
{
"epoch": 2.07,
"learning_rate": 1.556869046252726e-05,
"loss": 0.7612,
"step": 180000
},
{
"epoch": 2.07,
"learning_rate": 1.5473047936034278e-05,
"loss": 0.7675,
"step": 180500
},
{
"epoch": 2.08,
"learning_rate": 1.5377405409541297e-05,
"loss": 0.7906,
"step": 181000
},
{
"epoch": 2.08,
"learning_rate": 1.528176288304832e-05,
"loss": 0.7899,
"step": 181500
},
{
"epoch": 2.09,
"learning_rate": 1.518612035655534e-05,
"loss": 0.7917,
"step": 182000
},
{
"epoch": 2.09,
"learning_rate": 1.509047783006236e-05,
"loss": 0.7844,
"step": 182500
},
{
"epoch": 2.1,
"learning_rate": 1.499483530356938e-05,
"loss": 0.7891,
"step": 183000
},
{
"epoch": 2.11,
"learning_rate": 1.48991927770764e-05,
"loss": 0.7685,
"step": 183500
},
{
"epoch": 2.11,
"learning_rate": 1.480355025058342e-05,
"loss": 0.7708,
"step": 184000
},
{
"epoch": 2.12,
"learning_rate": 1.470790772409044e-05,
"loss": 0.7762,
"step": 184500
},
{
"epoch": 2.12,
"learning_rate": 1.461226519759746e-05,
"loss": 0.7842,
"step": 185000
},
{
"epoch": 2.13,
"learning_rate": 1.451662267110448e-05,
"loss": 0.7729,
"step": 185500
},
{
"epoch": 2.13,
"learning_rate": 1.44209801446115e-05,
"loss": 0.7683,
"step": 186000
},
{
"epoch": 2.14,
"learning_rate": 1.432533761811852e-05,
"loss": 0.779,
"step": 186500
},
{
"epoch": 2.15,
"learning_rate": 1.4229695091625541e-05,
"loss": 0.7759,
"step": 187000
},
{
"epoch": 2.15,
"learning_rate": 1.413405256513256e-05,
"loss": 0.7716,
"step": 187500
},
{
"epoch": 2.16,
"learning_rate": 1.4038410038639583e-05,
"loss": 0.7918,
"step": 188000
},
{
"epoch": 2.16,
"learning_rate": 1.39427675121466e-05,
"loss": 0.7656,
"step": 188500
},
{
"epoch": 2.17,
"learning_rate": 1.3847124985653623e-05,
"loss": 0.787,
"step": 189000
},
{
"epoch": 2.17,
"learning_rate": 1.3751482459160641e-05,
"loss": 0.7724,
"step": 189500
},
{
"epoch": 2.18,
"learning_rate": 1.3655839932667664e-05,
"loss": 0.7769,
"step": 190000
},
{
"epoch": 2.19,
"learning_rate": 1.3560197406174683e-05,
"loss": 0.7756,
"step": 190500
},
{
"epoch": 2.19,
"learning_rate": 1.34645548796817e-05,
"loss": 0.7903,
"step": 191000
},
{
"epoch": 2.2,
"learning_rate": 1.3368912353188723e-05,
"loss": 0.768,
"step": 191500
},
{
"epoch": 2.2,
"learning_rate": 1.3273269826695741e-05,
"loss": 0.7644,
"step": 192000
},
{
"epoch": 2.21,
"learning_rate": 1.3177627300202764e-05,
"loss": 0.786,
"step": 192500
},
{
"epoch": 2.22,
"learning_rate": 1.3081984773709781e-05,
"loss": 0.7771,
"step": 193000
},
{
"epoch": 2.22,
"learning_rate": 1.2986342247216804e-05,
"loss": 0.7729,
"step": 193500
},
{
"epoch": 2.23,
"learning_rate": 1.2890699720723823e-05,
"loss": 0.7497,
"step": 194000
},
{
"epoch": 2.23,
"learning_rate": 1.2795057194230844e-05,
"loss": 0.7685,
"step": 194500
},
{
"epoch": 2.24,
"learning_rate": 1.2699414667737864e-05,
"loss": 0.7723,
"step": 195000
},
{
"epoch": 2.24,
"learning_rate": 1.2603772141244885e-05,
"loss": 0.7732,
"step": 195500
},
{
"epoch": 2.25,
"learning_rate": 1.2508129614751904e-05,
"loss": 0.7737,
"step": 196000
},
{
"epoch": 2.26,
"learning_rate": 1.2412487088258923e-05,
"loss": 0.7818,
"step": 196500
},
{
"epoch": 2.26,
"learning_rate": 1.2316844561765945e-05,
"loss": 0.7829,
"step": 197000
},
{
"epoch": 2.27,
"learning_rate": 1.2221202035272966e-05,
"loss": 0.785,
"step": 197500
},
{
"epoch": 2.27,
"learning_rate": 1.2125559508779985e-05,
"loss": 0.7778,
"step": 198000
},
{
"epoch": 2.28,
"learning_rate": 1.2029916982287006e-05,
"loss": 0.7865,
"step": 198500
},
{
"epoch": 2.28,
"learning_rate": 1.1934274455794024e-05,
"loss": 0.7675,
"step": 199000
},
{
"epoch": 2.29,
"learning_rate": 1.1838631929301045e-05,
"loss": 0.7751,
"step": 199500
},
{
"epoch": 2.3,
"learning_rate": 1.1742989402808066e-05,
"loss": 0.7961,
"step": 200000
},
{
"epoch": 2.3,
"learning_rate": 1.1647346876315085e-05,
"loss": 0.7809,
"step": 200500
},
{
"epoch": 2.31,
"learning_rate": 1.1551704349822106e-05,
"loss": 0.7711,
"step": 201000
},
{
"epoch": 2.31,
"learning_rate": 1.1456061823329125e-05,
"loss": 0.7639,
"step": 201500
},
{
"epoch": 2.32,
"learning_rate": 1.1360419296836146e-05,
"loss": 0.7762,
"step": 202000
},
{
"epoch": 2.32,
"learning_rate": 1.1264776770343166e-05,
"loss": 0.769,
"step": 202500
},
{
"epoch": 2.33,
"learning_rate": 1.1169134243850187e-05,
"loss": 0.7566,
"step": 203000
},
{
"epoch": 2.34,
"learning_rate": 1.1073491717357206e-05,
"loss": 0.7649,
"step": 203500
},
{
"epoch": 2.34,
"learning_rate": 1.0977849190864227e-05,
"loss": 0.7637,
"step": 204000
},
{
"epoch": 2.35,
"learning_rate": 1.0882206664371246e-05,
"loss": 0.7628,
"step": 204500
},
{
"epoch": 2.35,
"learning_rate": 1.0786564137878266e-05,
"loss": 0.7608,
"step": 205000
},
{
"epoch": 2.36,
"learning_rate": 1.0690921611385287e-05,
"loss": 0.7568,
"step": 205500
},
{
"epoch": 2.36,
"learning_rate": 1.0595279084892306e-05,
"loss": 0.7596,
"step": 206000
},
{
"epoch": 2.37,
"learning_rate": 1.0499636558399327e-05,
"loss": 0.7685,
"step": 206500
},
{
"epoch": 2.38,
"learning_rate": 1.0403994031906348e-05,
"loss": 0.7724,
"step": 207000
},
{
"epoch": 2.38,
"learning_rate": 1.0308351505413367e-05,
"loss": 0.7586,
"step": 207500
},
{
"epoch": 2.39,
"learning_rate": 1.0212708978920388e-05,
"loss": 0.7926,
"step": 208000
},
{
"epoch": 2.39,
"learning_rate": 1.0117066452427408e-05,
"loss": 0.7744,
"step": 208500
},
{
"epoch": 2.4,
"learning_rate": 1.0021423925934429e-05,
"loss": 0.767,
"step": 209000
},
{
"epoch": 2.4,
"learning_rate": 9.925781399441448e-06,
"loss": 0.7737,
"step": 209500
},
{
"epoch": 2.41,
"learning_rate": 9.830138872948469e-06,
"loss": 0.7559,
"step": 210000
},
{
"epoch": 2.42,
"learning_rate": 9.734496346455488e-06,
"loss": 0.7792,
"step": 210500
},
{
"epoch": 2.42,
"learning_rate": 9.638853819962508e-06,
"loss": 0.7495,
"step": 211000
},
{
"epoch": 2.43,
"learning_rate": 9.543211293469529e-06,
"loss": 0.7755,
"step": 211500
},
{
"epoch": 2.43,
"learning_rate": 9.447568766976548e-06,
"loss": 0.7536,
"step": 212000
},
{
"epoch": 2.44,
"learning_rate": 9.351926240483569e-06,
"loss": 0.7668,
"step": 212500
},
{
"epoch": 2.44,
"learning_rate": 9.256283713990588e-06,
"loss": 0.7739,
"step": 213000
},
{
"epoch": 2.45,
"learning_rate": 9.16064118749761e-06,
"loss": 0.7482,
"step": 213500
},
{
"epoch": 2.46,
"learning_rate": 9.06499866100463e-06,
"loss": 0.76,
"step": 214000
},
{
"epoch": 2.46,
"learning_rate": 8.96935613451165e-06,
"loss": 0.7528,
"step": 214500
},
{
"epoch": 2.47,
"learning_rate": 8.87371360801867e-06,
"loss": 0.7576,
"step": 215000
},
{
"epoch": 2.47,
"learning_rate": 8.77807108152569e-06,
"loss": 0.76,
"step": 215500
},
{
"epoch": 2.48,
"learning_rate": 8.682428555032711e-06,
"loss": 0.7653,
"step": 216000
},
{
"epoch": 2.48,
"learning_rate": 8.58678602853973e-06,
"loss": 0.7608,
"step": 216500
},
{
"epoch": 2.49,
"learning_rate": 8.49114350204675e-06,
"loss": 0.7622,
"step": 217000
},
{
"epoch": 2.5,
"learning_rate": 8.39550097555377e-06,
"loss": 0.7623,
"step": 217500
},
{
"epoch": 2.5,
"learning_rate": 8.29985844906079e-06,
"loss": 0.7605,
"step": 218000
},
{
"epoch": 2.51,
"learning_rate": 8.204215922567811e-06,
"loss": 0.7544,
"step": 218500
},
{
"epoch": 2.51,
"learning_rate": 8.10857339607483e-06,
"loss": 0.7346,
"step": 219000
},
{
"epoch": 2.52,
"learning_rate": 8.012930869581852e-06,
"loss": 0.7602,
"step": 219500
},
{
"epoch": 2.52,
"learning_rate": 7.917288343088871e-06,
"loss": 0.7568,
"step": 220000
},
{
"epoch": 2.53,
"learning_rate": 7.821645816595892e-06,
"loss": 0.7711,
"step": 220500
},
{
"epoch": 2.54,
"learning_rate": 7.726003290102913e-06,
"loss": 0.7503,
"step": 221000
},
{
"epoch": 2.54,
"learning_rate": 7.630360763609932e-06,
"loss": 0.7546,
"step": 221500
},
{
"epoch": 2.55,
"learning_rate": 7.5347182371169516e-06,
"loss": 0.7692,
"step": 222000
},
{
"epoch": 2.55,
"learning_rate": 7.439075710623972e-06,
"loss": 0.7611,
"step": 222500
},
{
"epoch": 2.56,
"learning_rate": 7.343433184130992e-06,
"loss": 0.7646,
"step": 223000
},
{
"epoch": 2.57,
"learning_rate": 7.247790657638012e-06,
"loss": 0.7501,
"step": 223500
},
{
"epoch": 2.57,
"learning_rate": 7.152148131145032e-06,
"loss": 0.7527,
"step": 224000
},
{
"epoch": 2.58,
"learning_rate": 7.0565056046520524e-06,
"loss": 0.7455,
"step": 224500
},
{
"epoch": 2.58,
"learning_rate": 6.9608630781590735e-06,
"loss": 0.7604,
"step": 225000
},
{
"epoch": 2.59,
"learning_rate": 6.865220551666094e-06,
"loss": 0.7662,
"step": 225500
},
{
"epoch": 2.59,
"learning_rate": 6.769578025173114e-06,
"loss": 0.7655,
"step": 226000
},
{
"epoch": 2.6,
"learning_rate": 6.673935498680134e-06,
"loss": 0.7766,
"step": 226500
},
{
"epoch": 2.61,
"learning_rate": 6.578292972187154e-06,
"loss": 0.7608,
"step": 227000
},
{
"epoch": 2.61,
"learning_rate": 6.482650445694174e-06,
"loss": 0.7483,
"step": 227500
},
{
"epoch": 2.62,
"learning_rate": 6.387007919201194e-06,
"loss": 0.7557,
"step": 228000
},
{
"epoch": 2.62,
"learning_rate": 6.291365392708214e-06,
"loss": 0.7531,
"step": 228500
},
{
"epoch": 2.63,
"learning_rate": 6.195722866215234e-06,
"loss": 0.7524,
"step": 229000
},
{
"epoch": 2.63,
"learning_rate": 6.100080339722254e-06,
"loss": 0.7607,
"step": 229500
},
{
"epoch": 2.64,
"learning_rate": 6.004437813229274e-06,
"loss": 0.7506,
"step": 230000
},
{
"epoch": 2.65,
"learning_rate": 5.9087952867362945e-06,
"loss": 0.7462,
"step": 230500
},
{
"epoch": 2.65,
"learning_rate": 5.813152760243315e-06,
"loss": 0.7501,
"step": 231000
},
{
"epoch": 2.66,
"learning_rate": 5.717510233750335e-06,
"loss": 0.7492,
"step": 231500
},
{
"epoch": 2.66,
"learning_rate": 5.621867707257355e-06,
"loss": 0.7569,
"step": 232000
},
{
"epoch": 2.67,
"learning_rate": 5.526225180764375e-06,
"loss": 0.7493,
"step": 232500
},
{
"epoch": 2.67,
"learning_rate": 5.430582654271395e-06,
"loss": 0.7689,
"step": 233000
},
{
"epoch": 2.68,
"learning_rate": 5.334940127778416e-06,
"loss": 0.7448,
"step": 233500
},
{
"epoch": 2.69,
"learning_rate": 5.239297601285436e-06,
"loss": 0.7439,
"step": 234000
},
{
"epoch": 2.69,
"learning_rate": 5.143655074792456e-06,
"loss": 0.7449,
"step": 234500
},
{
"epoch": 2.7,
"learning_rate": 5.048012548299476e-06,
"loss": 0.7691,
"step": 235000
},
{
"epoch": 2.7,
"learning_rate": 4.952370021806496e-06,
"loss": 0.7518,
"step": 235500
},
{
"epoch": 2.71,
"learning_rate": 4.8567274953135165e-06,
"loss": 0.7381,
"step": 236000
},
{
"epoch": 2.71,
"learning_rate": 4.761084968820537e-06,
"loss": 0.7502,
"step": 236500
},
{
"epoch": 2.72,
"learning_rate": 4.665442442327557e-06,
"loss": 0.7556,
"step": 237000
},
{
"epoch": 2.73,
"learning_rate": 4.569799915834577e-06,
"loss": 0.7539,
"step": 237500
},
{
"epoch": 2.73,
"learning_rate": 4.474157389341597e-06,
"loss": 0.7576,
"step": 238000
},
{
"epoch": 2.74,
"learning_rate": 4.378514862848617e-06,
"loss": 0.748,
"step": 238500
},
{
"epoch": 2.74,
"learning_rate": 4.2828723363556375e-06,
"loss": 0.75,
"step": 239000
},
{
"epoch": 2.75,
"learning_rate": 4.187229809862658e-06,
"loss": 0.7615,
"step": 239500
},
{
"epoch": 2.75,
"learning_rate": 4.091587283369678e-06,
"loss": 0.7556,
"step": 240000
},
{
"epoch": 2.76,
"learning_rate": 3.995944756876697e-06,
"loss": 0.7377,
"step": 240500
},
{
"epoch": 2.77,
"learning_rate": 3.900302230383718e-06,
"loss": 0.753,
"step": 241000
},
{
"epoch": 2.77,
"learning_rate": 3.804659703890738e-06,
"loss": 0.7488,
"step": 241500
},
{
"epoch": 2.78,
"learning_rate": 3.709017177397758e-06,
"loss": 0.7529,
"step": 242000
},
{
"epoch": 2.78,
"learning_rate": 3.6133746509047787e-06,
"loss": 0.7569,
"step": 242500
},
{
"epoch": 2.79,
"learning_rate": 3.517732124411799e-06,
"loss": 0.7491,
"step": 243000
},
{
"epoch": 2.79,
"learning_rate": 3.4220895979188187e-06,
"loss": 0.7499,
"step": 243500
},
{
"epoch": 2.8,
"learning_rate": 3.326447071425839e-06,
"loss": 0.7421,
"step": 244000
},
{
"epoch": 2.81,
"learning_rate": 3.230804544932859e-06,
"loss": 0.7479,
"step": 244500
},
{
"epoch": 2.81,
"learning_rate": 3.135162018439879e-06,
"loss": 0.7477,
"step": 245000
},
{
"epoch": 2.82,
"learning_rate": 3.0395194919468998e-06,
"loss": 0.7455,
"step": 245500
},
{
"epoch": 2.82,
"learning_rate": 2.9438769654539195e-06,
"loss": 0.7472,
"step": 246000
},
{
"epoch": 2.83,
"learning_rate": 2.8482344389609397e-06,
"loss": 0.7456,
"step": 246500
},
{
"epoch": 2.83,
"learning_rate": 2.75259191246796e-06,
"loss": 0.7286,
"step": 247000
},
{
"epoch": 2.84,
"learning_rate": 2.65694938597498e-06,
"loss": 0.7441,
"step": 247500
},
{
"epoch": 2.85,
"learning_rate": 2.5613068594820002e-06,
"loss": 0.7449,
"step": 248000
},
{
"epoch": 2.85,
"learning_rate": 2.4656643329890204e-06,
"loss": 0.7436,
"step": 248500
},
{
"epoch": 2.86,
"learning_rate": 2.3700218064960406e-06,
"loss": 0.7604,
"step": 249000
},
{
"epoch": 2.86,
"learning_rate": 2.2743792800030608e-06,
"loss": 0.7406,
"step": 249500
},
{
"epoch": 2.87,
"learning_rate": 2.1787367535100805e-06,
"loss": 0.7466,
"step": 250000
},
{
"epoch": 2.88,
"learning_rate": 2.083094227017101e-06,
"loss": 0.7616,
"step": 250500
},
{
"epoch": 2.88,
"learning_rate": 1.9874517005241213e-06,
"loss": 0.7614,
"step": 251000
},
{
"epoch": 2.89,
"learning_rate": 1.8918091740311412e-06,
"loss": 0.7505,
"step": 251500
},
{
"epoch": 2.89,
"learning_rate": 1.7961666475381614e-06,
"loss": 0.7455,
"step": 252000
},
{
"epoch": 2.9,
"learning_rate": 1.7005241210451818e-06,
"loss": 0.7404,
"step": 252500
},
{
"epoch": 2.9,
"learning_rate": 1.6048815945522018e-06,
"loss": 0.7533,
"step": 253000
},
{
"epoch": 2.91,
"learning_rate": 1.509239068059222e-06,
"loss": 0.7517,
"step": 253500
},
{
"epoch": 2.92,
"learning_rate": 1.4135965415662421e-06,
"loss": 0.7398,
"step": 254000
},
{
"epoch": 2.92,
"learning_rate": 1.3179540150732623e-06,
"loss": 0.7283,
"step": 254500
},
{
"epoch": 2.93,
"learning_rate": 1.2223114885802825e-06,
"loss": 0.7251,
"step": 255000
},
{
"epoch": 2.93,
"learning_rate": 1.1266689620873024e-06,
"loss": 0.7506,
"step": 255500
},
{
"epoch": 2.94,
"learning_rate": 1.0310264355943226e-06,
"loss": 0.7402,
"step": 256000
},
{
"epoch": 2.94,
"learning_rate": 9.353839091013429e-07,
"loss": 0.7402,
"step": 256500
},
{
"epoch": 2.95,
"learning_rate": 8.397413826083631e-07,
"loss": 0.7381,
"step": 257000
},
{
"epoch": 2.96,
"learning_rate": 7.440988561153832e-07,
"loss": 0.7342,
"step": 257500
},
{
"epoch": 2.96,
"learning_rate": 6.484563296224033e-07,
"loss": 0.74,
"step": 258000
},
{
"epoch": 2.97,
"learning_rate": 5.528138031294235e-07,
"loss": 0.7527,
"step": 258500
},
{
"epoch": 2.97,
"learning_rate": 4.571712766364437e-07,
"loss": 0.7476,
"step": 259000
},
{
"epoch": 2.98,
"learning_rate": 3.6152875014346377e-07,
"loss": 0.741,
"step": 259500
},
{
"epoch": 2.98,
"learning_rate": 2.65886223650484e-07,
"loss": 0.748,
"step": 260000
},
{
"epoch": 2.99,
"learning_rate": 1.7024369715750412e-07,
"loss": 0.7441,
"step": 260500
},
{
"epoch": 3.0,
"learning_rate": 7.460117066452428e-08,
"loss": 0.7257,
"step": 261000
},
{
"epoch": 3.0,
"step": 261390,
"total_flos": 4.60844775936e+17,
"train_loss": 0.9857338986367055,
"train_runtime": 73080.7529,
"train_samples_per_second": 35.767,
"train_steps_per_second": 3.577
}
],
"max_steps": 261390,
"num_train_epochs": 3,
"total_flos": 4.60844775936e+17,
"trial_name": null,
"trial_params": null
}