| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 185568, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9865278496292465e-05, | |
| "loss": 3.1195, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.973055699258493e-05, | |
| "loss": 2.3468, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.95958354888774e-05, | |
| "loss": 2.1174, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.946111398516986e-05, | |
| "loss": 1.9833, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9326392481462324e-05, | |
| "loss": 1.8805, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9191670977754786e-05, | |
| "loss": 1.803, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9056949474047256e-05, | |
| "loss": 1.741, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.892222797033972e-05, | |
| "loss": 1.7125, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.878750646663218e-05, | |
| "loss": 1.659, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8652784962924645e-05, | |
| "loss": 1.6171, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.8518063459217114e-05, | |
| "loss": 1.5846, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.838334195550957e-05, | |
| "loss": 1.57, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.824862045180203e-05, | |
| "loss": 1.5535, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8113898948094496e-05, | |
| "loss": 1.5426, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.7979177444386966e-05, | |
| "loss": 1.5053, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.784445594067943e-05, | |
| "loss": 1.4924, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.770973443697189e-05, | |
| "loss": 1.4624, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.7575012933264354e-05, | |
| "loss": 1.465, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.7440291429556824e-05, | |
| "loss": 1.4493, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.730556992584929e-05, | |
| "loss": 1.4349, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.717084842214175e-05, | |
| "loss": 1.4282, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.703612691843421e-05, | |
| "loss": 1.397, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.690140541472668e-05, | |
| "loss": 1.4094, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6766683911019145e-05, | |
| "loss": 1.3788, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.663196240731161e-05, | |
| "loss": 1.3869, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.649724090360407e-05, | |
| "loss": 1.3738, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.6362519399896534e-05, | |
| "loss": 1.3786, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6227797896189e-05, | |
| "loss": 1.3482, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6093076392481466e-05, | |
| "loss": 1.339, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.595835488877393e-05, | |
| "loss": 1.3486, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.582363338506639e-05, | |
| "loss": 1.32, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.568891188135886e-05, | |
| "loss": 1.3085, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5554190377651324e-05, | |
| "loss": 1.318, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.541946887394379e-05, | |
| "loss": 1.2964, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.528474737023625e-05, | |
| "loss": 1.3008, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.515002586652871e-05, | |
| "loss": 1.2903, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.5015304362821176e-05, | |
| "loss": 1.2753, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.488058285911364e-05, | |
| "loss": 1.2971, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.47458613554061e-05, | |
| "loss": 1.2741, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.461113985169857e-05, | |
| "loss": 1.2801, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.4476418347991034e-05, | |
| "loss": 1.2865, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.43416968442835e-05, | |
| "loss": 1.2644, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.420697534057596e-05, | |
| "loss": 1.2686, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.407225383686843e-05, | |
| "loss": 1.2518, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.393753233316089e-05, | |
| "loss": 1.2541, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.3802810829453355e-05, | |
| "loss": 1.2372, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.366808932574582e-05, | |
| "loss": 1.2392, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.353336782203829e-05, | |
| "loss": 1.2406, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.339864631833075e-05, | |
| "loss": 1.243, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.3263924814623213e-05, | |
| "loss": 1.2331, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.3129203310915676e-05, | |
| "loss": 1.2337, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.299448180720814e-05, | |
| "loss": 1.2269, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.285976030350061e-05, | |
| "loss": 1.2175, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.272503879979307e-05, | |
| "loss": 1.2061, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.2590317296085535e-05, | |
| "loss": 1.2246, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.2455595792378e-05, | |
| "loss": 1.2102, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.232087428867047e-05, | |
| "loss": 1.2054, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.218615278496293e-05, | |
| "loss": 1.2038, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.205143128125539e-05, | |
| "loss": 1.207, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.1916709777547856e-05, | |
| "loss": 1.1969, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.178198827384032e-05, | |
| "loss": 1.1912, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.164726677013278e-05, | |
| "loss": 1.1835, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.1512545266425244e-05, | |
| "loss": 1.1881, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.137782376271771e-05, | |
| "loss": 1.1942, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.124310225901018e-05, | |
| "loss": 1.1888, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.110838075530264e-05, | |
| "loss": 1.1791, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.09736592515951e-05, | |
| "loss": 1.167, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.0838937747887565e-05, | |
| "loss": 1.1637, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.0704216244180035e-05, | |
| "loss": 1.1595, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.05694947404725e-05, | |
| "loss": 1.1715, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.043477323676496e-05, | |
| "loss": 1.1551, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.0300051733057424e-05, | |
| "loss": 1.1677, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.016533022934989e-05, | |
| "loss": 1.1316, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.0030608725642356e-05, | |
| "loss": 1.1413, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.989588722193482e-05, | |
| "loss": 1.1546, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.976116571822728e-05, | |
| "loss": 1.1584, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.9626444214519745e-05, | |
| "loss": 1.1517, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.9491722710812214e-05, | |
| "loss": 1.1307, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.935700120710468e-05, | |
| "loss": 1.1478, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.922227970339714e-05, | |
| "loss": 1.1346, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.90875581996896e-05, | |
| "loss": 1.1425, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.895283669598207e-05, | |
| "loss": 1.1365, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.8818115192274536e-05, | |
| "loss": 1.1413, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.868339368856699e-05, | |
| "loss": 1.1331, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.8548672184859454e-05, | |
| "loss": 1.1312, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.8413950681151924e-05, | |
| "loss": 1.1317, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.827922917744439e-05, | |
| "loss": 1.1213, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.814450767373685e-05, | |
| "loss": 1.1305, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.800978617002931e-05, | |
| "loss": 1.1299, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.787506466632178e-05, | |
| "loss": 1.1151, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.7740343162614245e-05, | |
| "loss": 1.1168, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.760562165890671e-05, | |
| "loss": 1.1149, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.747090015519917e-05, | |
| "loss": 1.1119, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.733617865149164e-05, | |
| "loss": 1.1265, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.7201457147784103e-05, | |
| "loss": 1.1231, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.7066735644076566e-05, | |
| "loss": 1.1267, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.693201414036903e-05, | |
| "loss": 1.1143, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.67972926366615e-05, | |
| "loss": 1.0906, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.666257113295396e-05, | |
| "loss": 1.1066, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.6527849629246425e-05, | |
| "loss": 1.1026, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.639312812553889e-05, | |
| "loss": 1.0802, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.625840662183135e-05, | |
| "loss": 1.1051, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.612368511812382e-05, | |
| "loss": 1.0943, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.598896361441628e-05, | |
| "loss": 1.0868, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.5854242110708746e-05, | |
| "loss": 1.0907, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.571952060700121e-05, | |
| "loss": 1.0774, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.558479910329368e-05, | |
| "loss": 1.0752, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.5450077599586134e-05, | |
| "loss": 1.088, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.53153560958786e-05, | |
| "loss": 1.0812, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.518063459217106e-05, | |
| "loss": 1.0844, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.504591308846353e-05, | |
| "loss": 1.0718, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.491119158475599e-05, | |
| "loss": 1.0805, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.4776470081048455e-05, | |
| "loss": 1.0649, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.464174857734092e-05, | |
| "loss": 1.0796, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.450702707363339e-05, | |
| "loss": 1.0731, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.437230556992585e-05, | |
| "loss": 1.0601, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.4237584066218314e-05, | |
| "loss": 1.0622, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.4102862562510777e-05, | |
| "loss": 1.0632, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.3968141058803246e-05, | |
| "loss": 1.0682, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.383341955509571e-05, | |
| "loss": 1.0501, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.369869805138817e-05, | |
| "loss": 1.0816, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.3563976547680635e-05, | |
| "loss": 1.0739, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.3429255043973104e-05, | |
| "loss": 1.0651, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.329453354026557e-05, | |
| "loss": 1.0508, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.315981203655803e-05, | |
| "loss": 0.9982, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.302509053285049e-05, | |
| "loss": 0.9883, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.2890369029142956e-05, | |
| "loss": 0.9885, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.2755647525435426e-05, | |
| "loss": 0.987, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.262092602172789e-05, | |
| "loss": 0.981, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.248620451802035e-05, | |
| "loss": 0.9875, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.2351483014312814e-05, | |
| "loss": 0.9876, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.221676151060528e-05, | |
| "loss": 0.9906, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.208204000689774e-05, | |
| "loss": 0.996, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.19473185031902e-05, | |
| "loss": 0.9807, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.1812596999482666e-05, | |
| "loss": 0.9892, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.1677875495775135e-05, | |
| "loss": 0.9615, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.15431539920676e-05, | |
| "loss": 0.9805, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.140843248836006e-05, | |
| "loss": 0.9958, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.1273710984652524e-05, | |
| "loss": 0.9971, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.1138989480944993e-05, | |
| "loss": 0.971, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.1004267977237456e-05, | |
| "loss": 0.9822, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.086954647352992e-05, | |
| "loss": 0.9825, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.073482496982238e-05, | |
| "loss": 0.9795, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.060010346611485e-05, | |
| "loss": 0.9778, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.0465381962407315e-05, | |
| "loss": 0.9758, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.0330660458699777e-05, | |
| "loss": 0.9662, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.0195938954992244e-05, | |
| "loss": 0.966, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.0061217451284707e-05, | |
| "loss": 0.9717, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 2.992649594757717e-05, | |
| "loss": 0.9594, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.9791774443869636e-05, | |
| "loss": 0.9808, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.96570529401621e-05, | |
| "loss": 0.9743, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.9522331436454565e-05, | |
| "loss": 0.9745, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.9387609932747028e-05, | |
| "loss": 0.9836, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.9252888429039494e-05, | |
| "loss": 0.98, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.9118166925331957e-05, | |
| "loss": 0.9708, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.8983445421624416e-05, | |
| "loss": 0.9793, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.8848723917916883e-05, | |
| "loss": 0.9525, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.8714002414209345e-05, | |
| "loss": 0.9679, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.857928091050181e-05, | |
| "loss": 0.9749, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.8444559406794275e-05, | |
| "loss": 0.9701, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.8309837903086737e-05, | |
| "loss": 0.9681, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.8175116399379204e-05, | |
| "loss": 0.9745, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.8040394895671667e-05, | |
| "loss": 0.968, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.7905673391964133e-05, | |
| "loss": 0.9673, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.7770951888256596e-05, | |
| "loss": 0.9543, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.7636230384549062e-05, | |
| "loss": 0.9582, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.7501508880841525e-05, | |
| "loss": 0.9659, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.736678737713399e-05, | |
| "loss": 0.9692, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.7232065873426454e-05, | |
| "loss": 0.9556, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.709734436971892e-05, | |
| "loss": 0.9622, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.6962622866011383e-05, | |
| "loss": 0.9626, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.682790136230385e-05, | |
| "loss": 0.9571, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.6693179858596312e-05, | |
| "loss": 0.9588, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.6558458354888775e-05, | |
| "loss": 0.9544, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.642373685118124e-05, | |
| "loss": 0.9631, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.6289015347473704e-05, | |
| "loss": 0.9521, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.615429384376617e-05, | |
| "loss": 0.9605, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.6019572340058633e-05, | |
| "loss": 0.9685, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.58848508363511e-05, | |
| "loss": 0.9613, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.575012933264356e-05, | |
| "loss": 0.9532, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.5615407828936022e-05, | |
| "loss": 0.9633, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.5480686325228488e-05, | |
| "loss": 0.9481, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.534596482152095e-05, | |
| "loss": 0.9613, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.5211243317813417e-05, | |
| "loss": 0.9468, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.507652181410588e-05, | |
| "loss": 0.9594, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.4941800310398343e-05, | |
| "loss": 0.9495, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.480707880669081e-05, | |
| "loss": 0.9541, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.4672357302983272e-05, | |
| "loss": 0.9411, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.453763579927574e-05, | |
| "loss": 0.9481, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.44029142955682e-05, | |
| "loss": 0.9534, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.4268192791860667e-05, | |
| "loss": 0.9563, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.413347128815313e-05, | |
| "loss": 0.9487, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.3998749784445597e-05, | |
| "loss": 0.9463, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.386402828073806e-05, | |
| "loss": 0.9479, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.3729306777030526e-05, | |
| "loss": 0.9388, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.359458527332299e-05, | |
| "loss": 0.9305, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.3459863769615455e-05, | |
| "loss": 0.9405, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.3325142265907914e-05, | |
| "loss": 0.936, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.319042076220038e-05, | |
| "loss": 0.9516, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.3055699258492843e-05, | |
| "loss": 0.9472, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.292097775478531e-05, | |
| "loss": 0.9376, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.2786256251077773e-05, | |
| "loss": 0.9483, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.265153474737024e-05, | |
| "loss": 0.94, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.25168132436627e-05, | |
| "loss": 0.9361, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.2382091739955165e-05, | |
| "loss": 0.947, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.224737023624763e-05, | |
| "loss": 0.9327, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.2112648732540094e-05, | |
| "loss": 0.9388, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.197792722883256e-05, | |
| "loss": 0.9282, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.1843205725125023e-05, | |
| "loss": 0.9314, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.1708484221417486e-05, | |
| "loss": 0.9222, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.157376271770995e-05, | |
| "loss": 0.9232, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.1439041214002415e-05, | |
| "loss": 0.9311, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.1304319710294878e-05, | |
| "loss": 0.9316, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.1169598206587344e-05, | |
| "loss": 0.9242, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.1034876702879807e-05, | |
| "loss": 0.9236, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.0900155199172273e-05, | |
| "loss": 0.93, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.0765433695464736e-05, | |
| "loss": 0.9386, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.0630712191757202e-05, | |
| "loss": 0.9246, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.0495990688049665e-05, | |
| "loss": 0.93, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.036126918434213e-05, | |
| "loss": 0.9226, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0226547680634594e-05, | |
| "loss": 0.9188, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0091826176927057e-05, | |
| "loss": 0.9269, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.995710467321952e-05, | |
| "loss": 0.9303, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.9822383169511986e-05, | |
| "loss": 0.9347, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.968766166580445e-05, | |
| "loss": 0.9217, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.9552940162096915e-05, | |
| "loss": 0.9387, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.9418218658389378e-05, | |
| "loss": 0.9241, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.9283497154681844e-05, | |
| "loss": 0.9269, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.9148775650974307e-05, | |
| "loss": 0.9132, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.901405414726677e-05, | |
| "loss": 0.9005, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.8879332643559236e-05, | |
| "loss": 0.9172, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.87446111398517e-05, | |
| "loss": 0.9289, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.8609889636144165e-05, | |
| "loss": 0.9153, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.8475168132436625e-05, | |
| "loss": 0.9163, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.834044662872909e-05, | |
| "loss": 0.9115, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.8205725125021554e-05, | |
| "loss": 0.9271, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.807100362131402e-05, | |
| "loss": 0.9104, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.7936282117606483e-05, | |
| "loss": 0.9099, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.780156061389895e-05, | |
| "loss": 0.9189, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.7666839110191412e-05, | |
| "loss": 0.9156, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.753211760648388e-05, | |
| "loss": 0.9096, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.739739610277634e-05, | |
| "loss": 0.9118, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.7262674599068808e-05, | |
| "loss": 0.9193, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.712795309536127e-05, | |
| "loss": 0.9009, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.6993231591653737e-05, | |
| "loss": 0.9068, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.6858510087946196e-05, | |
| "loss": 0.9101, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.6723788584238663e-05, | |
| "loss": 0.9142, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.6589067080531125e-05, | |
| "loss": 0.8746, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.645434557682359e-05, | |
| "loss": 0.8355, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.6319624073116054e-05, | |
| "loss": 0.8432, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.618490256940852e-05, | |
| "loss": 0.8385, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.6050181065700984e-05, | |
| "loss": 0.8273, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.591545956199345e-05, | |
| "loss": 0.8178, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.5780738058285913e-05, | |
| "loss": 0.831, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.5646016554578376e-05, | |
| "loss": 0.8338, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.5511295050870842e-05, | |
| "loss": 0.8318, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.5376573547163305e-05, | |
| "loss": 0.8405, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5241852043455768e-05, | |
| "loss": 0.8325, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5107130539748232e-05, | |
| "loss": 0.8221, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.4972409036040697e-05, | |
| "loss": 0.844, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.4837687532333161e-05, | |
| "loss": 0.8285, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4702966028625626e-05, | |
| "loss": 0.8404, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.456824452491809e-05, | |
| "loss": 0.8232, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.4433523021210555e-05, | |
| "loss": 0.83, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.429880151750302e-05, | |
| "loss": 0.8246, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.4164080013795482e-05, | |
| "loss": 0.8369, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.4029358510087947e-05, | |
| "loss": 0.8344, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.3894637006380412e-05, | |
| "loss": 0.8266, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.3759915502672876e-05, | |
| "loss": 0.842, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.3625193998965339e-05, | |
| "loss": 0.8304, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.3490472495257803e-05, | |
| "loss": 0.8235, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.3355750991550266e-05, | |
| "loss": 0.8343, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.3221029487842731e-05, | |
| "loss": 0.8277, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.3086307984135195e-05, | |
| "loss": 0.8305, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.295158648042766e-05, | |
| "loss": 0.8429, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.2816864976720125e-05, | |
| "loss": 0.8168, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.268214347301259e-05, | |
| "loss": 0.83, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.2547421969305054e-05, | |
| "loss": 0.8351, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.2412700465597517e-05, | |
| "loss": 0.8232, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.2277978961889981e-05, | |
| "loss": 0.8317, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.2143257458182446e-05, | |
| "loss": 0.8188, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.200853595447491e-05, | |
| "loss": 0.8267, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.1873814450767375e-05, | |
| "loss": 0.8215, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.173909294705984e-05, | |
| "loss": 0.8201, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.1604371443352302e-05, | |
| "loss": 0.8155, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.1469649939644767e-05, | |
| "loss": 0.8119, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.1334928435937231e-05, | |
| "loss": 0.8207, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.1200206932229696e-05, | |
| "loss": 0.8314, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.106548542852216e-05, | |
| "loss": 0.8212, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.0930763924814625e-05, | |
| "loss": 0.8173, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.0796042421107088e-05, | |
| "loss": 0.8195, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.0661320917399552e-05, | |
| "loss": 0.8237, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.0526599413692017e-05, | |
| "loss": 0.8198, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.039187790998448e-05, | |
| "loss": 0.8151, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.0257156406276944e-05, | |
| "loss": 0.8398, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.0122434902569409e-05, | |
| "loss": 0.8272, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 9.987713398861872e-06, | |
| "loss": 0.8155, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.852991895154336e-06, | |
| "loss": 0.8135, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 9.718270391446801e-06, | |
| "loss": 0.8246, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 9.583548887739266e-06, | |
| "loss": 0.8085, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.44882738403173e-06, | |
| "loss": 0.8156, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.314105880324195e-06, | |
| "loss": 0.8135, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.179384376616658e-06, | |
| "loss": 0.8184, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 9.044662872909122e-06, | |
| "loss": 0.8198, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 8.909941369201587e-06, | |
| "loss": 0.8167, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 8.775219865494051e-06, | |
| "loss": 0.8109, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 8.640498361786516e-06, | |
| "loss": 0.7983, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.50577685807898e-06, | |
| "loss": 0.8107, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 8.371055354371443e-06, | |
| "loss": 0.8026, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 8.236333850663908e-06, | |
| "loss": 0.8, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 8.101612346956372e-06, | |
| "loss": 0.8089, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 7.966890843248837e-06, | |
| "loss": 0.8055, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.832169339541301e-06, | |
| "loss": 0.8187, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.697447835833766e-06, | |
| "loss": 0.8103, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.562726332126228e-06, | |
| "loss": 0.8111, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.428004828418693e-06, | |
| "loss": 0.8168, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.293283324711157e-06, | |
| "loss": 0.815, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 7.158561821003622e-06, | |
| "loss": 0.8166, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 7.023840317296086e-06, | |
| "loss": 0.8057, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 6.889118813588551e-06, | |
| "loss": 0.8001, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 6.754397309881014e-06, | |
| "loss": 0.806, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.619675806173478e-06, | |
| "loss": 0.8116, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 6.484954302465943e-06, | |
| "loss": 0.8093, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 6.3502327987584074e-06, | |
| "loss": 0.815, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.215511295050871e-06, | |
| "loss": 0.7953, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 6.080789791343335e-06, | |
| "loss": 0.801, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 5.946068287635799e-06, | |
| "loss": 0.8272, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.811346783928263e-06, | |
| "loss": 0.8055, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 5.676625280220728e-06, | |
| "loss": 0.7992, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.541903776513192e-06, | |
| "loss": 0.8054, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.407182272805656e-06, | |
| "loss": 0.8029, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.2724607690981205e-06, | |
| "loss": 0.804, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.137739265390585e-06, | |
| "loss": 0.8052, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 5.003017761683049e-06, | |
| "loss": 0.7955, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.868296257975513e-06, | |
| "loss": 0.7952, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.733574754267978e-06, | |
| "loss": 0.8058, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.598853250560442e-06, | |
| "loss": 0.8031, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.464131746852906e-06, | |
| "loss": 0.8181, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.32941024314537e-06, | |
| "loss": 0.8076, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.194688739437834e-06, | |
| "loss": 0.8172, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.059967235730298e-06, | |
| "loss": 0.7994, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 3.925245732022763e-06, | |
| "loss": 0.801, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.7905242283152265e-06, | |
| "loss": 0.806, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 3.655802724607691e-06, | |
| "loss": 0.8023, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.5210812209001556e-06, | |
| "loss": 0.8102, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.3863597171926193e-06, | |
| "loss": 0.8119, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.251638213485084e-06, | |
| "loss": 0.807, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.116916709777548e-06, | |
| "loss": 0.7956, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 2.982195206070012e-06, | |
| "loss": 0.8035, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.8474737023624763e-06, | |
| "loss": 0.8076, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.712752198654941e-06, | |
| "loss": 0.7995, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.578030694947405e-06, | |
| "loss": 0.7979, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.443309191239869e-06, | |
| "loss": 0.7933, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.3085876875323332e-06, | |
| "loss": 0.8008, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.1738661838247974e-06, | |
| "loss": 0.7969, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 2.0391446801172615e-06, | |
| "loss": 0.8077, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.904423176409726e-06, | |
| "loss": 0.7988, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.7697016727021902e-06, | |
| "loss": 0.8078, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.6349801689946541e-06, | |
| "loss": 0.7973, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.5002586652871185e-06, | |
| "loss": 0.7927, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.3655371615795828e-06, | |
| "loss": 0.8139, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.230815657872047e-06, | |
| "loss": 0.7866, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.0960941541645111e-06, | |
| "loss": 0.7991, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 9.613726504569755e-07, | |
| "loss": 0.8022, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 8.266511467494397e-07, | |
| "loss": 0.7863, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 6.919296430419037e-07, | |
| "loss": 0.7973, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 5.572081393343681e-07, | |
| "loss": 0.7922, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.224866356268322e-07, | |
| "loss": 0.7922, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 2.8776513191929646e-07, | |
| "loss": 0.7878, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 1.5304362821176065e-07, | |
| "loss": 0.8012, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 1.8322124504224867e-08, | |
| "loss": 0.798, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 185568, | |
| "total_flos": 2.7769538443832525e+17, | |
| "train_loss": 1.0130458773570052, | |
| "train_runtime": 63769.0674, | |
| "train_samples_per_second": 29.1, | |
| "train_steps_per_second": 2.91 | |
| } | |
| ], | |
| "max_steps": 185568, | |
| "num_train_epochs": 3, | |
| "total_flos": 2.7769538443832525e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |