diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,17572 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 19.999726592937897, + "global_step": 1463000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.01, + "learning_rate": 4.998291205861847e-05, + "loss": 2.1473, + "step": 500 + }, + { + "epoch": 0.01, + "learning_rate": 4.996582411723695e-05, + "loss": 2.08, + "step": 1000 + }, + { + "epoch": 0.02, + "learning_rate": 4.994873617585543e-05, + "loss": 2.0483, + "step": 1500 + }, + { + "epoch": 0.03, + "learning_rate": 4.99316482344739e-05, + "loss": 1.988, + "step": 2000 + }, + { + "epoch": 0.03, + "learning_rate": 4.991456029309237e-05, + "loss": 2.0191, + "step": 2500 + }, + { + "epoch": 0.04, + "learning_rate": 4.9897472351710846e-05, + "loss": 1.9991, + "step": 3000 + }, + { + "epoch": 0.05, + "learning_rate": 4.988038441032932e-05, + "loss": 1.9509, + "step": 3500 + }, + { + "epoch": 0.05, + "learning_rate": 4.9863296468947794e-05, + "loss": 1.953, + "step": 4000 + }, + { + "epoch": 0.06, + "learning_rate": 4.984620852756627e-05, + "loss": 1.9231, + "step": 4500 + }, + { + "epoch": 0.07, + "learning_rate": 4.982912058618474e-05, + "loss": 1.9283, + "step": 5000 + }, + { + "epoch": 0.08, + "learning_rate": 4.981203264480322e-05, + "loss": 1.9113, + "step": 5500 + }, + { + "epoch": 0.08, + "learning_rate": 4.979494470342169e-05, + "loss": 1.9051, + "step": 6000 + }, + { + "epoch": 0.09, + "learning_rate": 4.977785676204017e-05, + "loss": 1.8648, + "step": 6500 + }, + { + "epoch": 0.1, + "learning_rate": 4.976076882065864e-05, + "loss": 1.915, + "step": 7000 + }, + { + "epoch": 0.1, + "learning_rate": 4.9743680879277115e-05, + "loss": 1.8784, + "step": 7500 + }, + { + "epoch": 0.11, + "learning_rate": 4.972659293789559e-05, + "loss": 1.8939, + "step": 8000 + }, + { + "epoch": 0.12, + "learning_rate": 4.970950499651406e-05, + "loss": 1.918, + "step": 8500 + }, + { + "epoch": 0.12, + "learning_rate": 4.969241705513253e-05, + "loss": 1.8854, + "step": 9000 + }, + { + "epoch": 0.13, + "learning_rate": 4.967532911375101e-05, + "loss": 1.8941, + "step": 9500 + }, + { + "epoch": 0.14, + "learning_rate": 4.965824117236949e-05, + "loss": 1.8544, + "step": 10000 + }, + { + "epoch": 0.14, + "learning_rate": 4.964115323098796e-05, + "loss": 1.8561, + "step": 10500 + }, + { + "epoch": 0.15, + "learning_rate": 4.962406528960643e-05, + "loss": 1.8602, + "step": 11000 + }, + { + "epoch": 0.16, + "learning_rate": 4.9606977348224906e-05, + "loss": 1.8556, + "step": 11500 + }, + { + "epoch": 0.16, + "learning_rate": 4.958988940684338e-05, + "loss": 1.83, + "step": 12000 + }, + { + "epoch": 0.17, + "learning_rate": 4.9572801465461854e-05, + "loss": 1.8515, + "step": 12500 + }, + { + "epoch": 0.18, + "learning_rate": 4.9555713524080324e-05, + "loss": 1.8384, + "step": 13000 + }, + { + "epoch": 0.18, + "learning_rate": 4.953862558269881e-05, + "loss": 1.8534, + "step": 13500 + }, + { + "epoch": 0.19, + "learning_rate": 4.952153764131728e-05, + "loss": 1.8488, + "step": 14000 + }, + { + "epoch": 0.2, + "learning_rate": 4.950444969993575e-05, + "loss": 1.8426, + "step": 14500 + }, + { + "epoch": 0.21, + "learning_rate": 4.948736175855423e-05, + "loss": 1.8037, + "step": 15000 + }, + { + "epoch": 0.21, + "learning_rate": 4.9470273817172704e-05, + "loss": 1.8529, + "step": 15500 + }, + { + "epoch": 0.22, + "learning_rate": 4.9453185875791174e-05, + "loss": 1.837, + "step": 16000 + }, + { + "epoch": 0.23, + "learning_rate": 4.9436097934409645e-05, + "loss": 1.8277, + "step": 16500 + }, + { + "epoch": 0.23, + "learning_rate": 4.941900999302812e-05, + "loss": 1.8321, + "step": 17000 + }, + { + "epoch": 0.24, + "learning_rate": 4.94019220516466e-05, + "loss": 1.8249, + "step": 17500 + }, + { + "epoch": 0.25, + "learning_rate": 4.938483411026507e-05, + "loss": 1.8092, + "step": 18000 + }, + { + "epoch": 0.25, + "learning_rate": 4.936774616888355e-05, + "loss": 1.8347, + "step": 18500 + }, + { + "epoch": 0.26, + "learning_rate": 4.935065822750202e-05, + "loss": 1.8339, + "step": 19000 + }, + { + "epoch": 0.27, + "learning_rate": 4.9333570286120495e-05, + "loss": 1.7946, + "step": 19500 + }, + { + "epoch": 0.27, + "learning_rate": 4.9316482344738966e-05, + "loss": 1.8443, + "step": 20000 + }, + { + "epoch": 0.28, + "learning_rate": 4.929939440335744e-05, + "loss": 1.8018, + "step": 20500 + }, + { + "epoch": 0.29, + "learning_rate": 4.9282306461975914e-05, + "loss": 1.8282, + "step": 21000 + }, + { + "epoch": 0.29, + "learning_rate": 4.926521852059439e-05, + "loss": 1.7917, + "step": 21500 + }, + { + "epoch": 0.3, + "learning_rate": 4.924813057921286e-05, + "loss": 1.833, + "step": 22000 + }, + { + "epoch": 0.31, + "learning_rate": 4.923104263783134e-05, + "loss": 1.8189, + "step": 22500 + }, + { + "epoch": 0.31, + "learning_rate": 4.921395469644981e-05, + "loss": 1.7933, + "step": 23000 + }, + { + "epoch": 0.32, + "learning_rate": 4.9196866755068287e-05, + "loss": 1.7949, + "step": 23500 + }, + { + "epoch": 0.33, + "learning_rate": 4.9179778813686764e-05, + "loss": 1.7883, + "step": 24000 + }, + { + "epoch": 0.33, + "learning_rate": 4.9162690872305234e-05, + "loss": 1.8027, + "step": 24500 + }, + { + "epoch": 0.34, + "learning_rate": 4.9145602930923705e-05, + "loss": 1.8035, + "step": 25000 + }, + { + "epoch": 0.35, + "learning_rate": 4.912851498954218e-05, + "loss": 1.7769, + "step": 25500 + }, + { + "epoch": 0.36, + "learning_rate": 4.911142704816066e-05, + "loss": 1.7845, + "step": 26000 + }, + { + "epoch": 0.36, + "learning_rate": 4.909433910677913e-05, + "loss": 1.7819, + "step": 26500 + }, + { + "epoch": 0.37, + "learning_rate": 4.90772511653976e-05, + "loss": 1.7828, + "step": 27000 + }, + { + "epoch": 0.38, + "learning_rate": 4.9060163224016085e-05, + "loss": 1.7437, + "step": 27500 + }, + { + "epoch": 0.38, + "learning_rate": 4.9043075282634555e-05, + "loss": 1.7717, + "step": 28000 + }, + { + "epoch": 0.39, + "learning_rate": 4.9025987341253026e-05, + "loss": 1.8001, + "step": 28500 + }, + { + "epoch": 0.4, + "learning_rate": 4.9008899399871496e-05, + "loss": 1.7801, + "step": 29000 + }, + { + "epoch": 0.4, + "learning_rate": 4.899181145848998e-05, + "loss": 1.7857, + "step": 29500 + }, + { + "epoch": 0.41, + "learning_rate": 4.897472351710845e-05, + "loss": 1.7677, + "step": 30000 + }, + { + "epoch": 0.42, + "learning_rate": 4.895763557572692e-05, + "loss": 1.7643, + "step": 30500 + }, + { + "epoch": 0.42, + "learning_rate": 4.89405476343454e-05, + "loss": 1.7547, + "step": 31000 + }, + { + "epoch": 0.43, + "learning_rate": 4.892345969296387e-05, + "loss": 1.7471, + "step": 31500 + }, + { + "epoch": 0.44, + "learning_rate": 4.8906371751582347e-05, + "loss": 1.7691, + "step": 32000 + }, + { + "epoch": 0.44, + "learning_rate": 4.888928381020082e-05, + "loss": 1.8062, + "step": 32500 + }, + { + "epoch": 0.45, + "learning_rate": 4.8872195868819294e-05, + "loss": 1.7589, + "step": 33000 + }, + { + "epoch": 0.46, + "learning_rate": 4.8855107927437765e-05, + "loss": 1.7461, + "step": 33500 + }, + { + "epoch": 0.46, + "learning_rate": 4.883801998605624e-05, + "loss": 1.7445, + "step": 34000 + }, + { + "epoch": 0.47, + "learning_rate": 4.882093204467472e-05, + "loss": 1.7616, + "step": 34500 + }, + { + "epoch": 0.48, + "learning_rate": 4.880384410329319e-05, + "loss": 1.7383, + "step": 35000 + }, + { + "epoch": 0.49, + "learning_rate": 4.878675616191166e-05, + "loss": 1.7593, + "step": 35500 + }, + { + "epoch": 0.49, + "learning_rate": 4.876966822053014e-05, + "loss": 1.7859, + "step": 36000 + }, + { + "epoch": 0.5, + "learning_rate": 4.8752580279148615e-05, + "loss": 1.7573, + "step": 36500 + }, + { + "epoch": 0.51, + "learning_rate": 4.8735492337767086e-05, + "loss": 1.7459, + "step": 37000 + }, + { + "epoch": 0.51, + "learning_rate": 4.8718404396385556e-05, + "loss": 1.7901, + "step": 37500 + }, + { + "epoch": 0.52, + "learning_rate": 4.8701316455004034e-05, + "loss": 1.7776, + "step": 38000 + }, + { + "epoch": 0.53, + "learning_rate": 4.868422851362251e-05, + "loss": 1.7564, + "step": 38500 + }, + { + "epoch": 0.53, + "learning_rate": 4.866714057224098e-05, + "loss": 1.7636, + "step": 39000 + }, + { + "epoch": 0.54, + "learning_rate": 4.865005263085945e-05, + "loss": 1.7517, + "step": 39500 + }, + { + "epoch": 0.55, + "learning_rate": 4.8632964689477936e-05, + "loss": 1.7487, + "step": 40000 + }, + { + "epoch": 0.55, + "learning_rate": 4.8615876748096407e-05, + "loss": 1.7673, + "step": 40500 + }, + { + "epoch": 0.56, + "learning_rate": 4.859878880671488e-05, + "loss": 1.759, + "step": 41000 + }, + { + "epoch": 0.57, + "learning_rate": 4.8581700865333354e-05, + "loss": 1.7641, + "step": 41500 + }, + { + "epoch": 0.57, + "learning_rate": 4.856461292395183e-05, + "loss": 1.7476, + "step": 42000 + }, + { + "epoch": 0.58, + "learning_rate": 4.85475249825703e-05, + "loss": 1.7699, + "step": 42500 + }, + { + "epoch": 0.59, + "learning_rate": 4.853043704118877e-05, + "loss": 1.7292, + "step": 43000 + }, + { + "epoch": 0.59, + "learning_rate": 4.851334909980725e-05, + "loss": 1.7342, + "step": 43500 + }, + { + "epoch": 0.6, + "learning_rate": 4.849626115842573e-05, + "loss": 1.7661, + "step": 44000 + }, + { + "epoch": 0.61, + "learning_rate": 4.84791732170442e-05, + "loss": 1.7214, + "step": 44500 + }, + { + "epoch": 0.62, + "learning_rate": 4.8462085275662675e-05, + "loss": 1.7612, + "step": 45000 + }, + { + "epoch": 0.62, + "learning_rate": 4.8444997334281146e-05, + "loss": 1.7241, + "step": 45500 + }, + { + "epoch": 0.63, + "learning_rate": 4.842790939289962e-05, + "loss": 1.7714, + "step": 46000 + }, + { + "epoch": 0.64, + "learning_rate": 4.8410821451518094e-05, + "loss": 1.7599, + "step": 46500 + }, + { + "epoch": 0.64, + "learning_rate": 4.839373351013657e-05, + "loss": 1.773, + "step": 47000 + }, + { + "epoch": 0.65, + "learning_rate": 4.837664556875504e-05, + "loss": 1.7551, + "step": 47500 + }, + { + "epoch": 0.66, + "learning_rate": 4.835955762737352e-05, + "loss": 1.7354, + "step": 48000 + }, + { + "epoch": 0.66, + "learning_rate": 4.834246968599199e-05, + "loss": 1.7099, + "step": 48500 + }, + { + "epoch": 0.67, + "learning_rate": 4.8325381744610466e-05, + "loss": 1.7452, + "step": 49000 + }, + { + "epoch": 0.68, + "learning_rate": 4.830829380322894e-05, + "loss": 1.7411, + "step": 49500 + }, + { + "epoch": 0.68, + "learning_rate": 4.8291205861847414e-05, + "loss": 1.7069, + "step": 50000 + }, + { + "epoch": 0.69, + "learning_rate": 4.827411792046589e-05, + "loss": 1.7445, + "step": 50500 + }, + { + "epoch": 0.7, + "learning_rate": 4.825702997908436e-05, + "loss": 1.7155, + "step": 51000 + }, + { + "epoch": 0.7, + "learning_rate": 4.823994203770283e-05, + "loss": 1.7694, + "step": 51500 + }, + { + "epoch": 0.71, + "learning_rate": 4.822285409632131e-05, + "loss": 1.7349, + "step": 52000 + }, + { + "epoch": 0.72, + "learning_rate": 4.820576615493979e-05, + "loss": 1.7219, + "step": 52500 + }, + { + "epoch": 0.72, + "learning_rate": 4.818867821355826e-05, + "loss": 1.7541, + "step": 53000 + }, + { + "epoch": 0.73, + "learning_rate": 4.817159027217673e-05, + "loss": 1.7453, + "step": 53500 + }, + { + "epoch": 0.74, + "learning_rate": 4.815450233079521e-05, + "loss": 1.7162, + "step": 54000 + }, + { + "epoch": 0.75, + "learning_rate": 4.813741438941368e-05, + "loss": 1.7243, + "step": 54500 + }, + { + "epoch": 0.75, + "learning_rate": 4.8120326448032153e-05, + "loss": 1.7148, + "step": 55000 + }, + { + "epoch": 0.76, + "learning_rate": 4.810323850665063e-05, + "loss": 1.7188, + "step": 55500 + }, + { + "epoch": 0.77, + "learning_rate": 4.808615056526911e-05, + "loss": 1.746, + "step": 56000 + }, + { + "epoch": 0.77, + "learning_rate": 4.806906262388758e-05, + "loss": 1.7448, + "step": 56500 + }, + { + "epoch": 0.78, + "learning_rate": 4.805197468250605e-05, + "loss": 1.7226, + "step": 57000 + }, + { + "epoch": 0.79, + "learning_rate": 4.8034886741124526e-05, + "loss": 1.7131, + "step": 57500 + }, + { + "epoch": 0.79, + "learning_rate": 4.8017798799743004e-05, + "loss": 1.7533, + "step": 58000 + }, + { + "epoch": 0.8, + "learning_rate": 4.8000710858361474e-05, + "loss": 1.6981, + "step": 58500 + }, + { + "epoch": 0.81, + "learning_rate": 4.7983622916979945e-05, + "loss": 1.7206, + "step": 59000 + }, + { + "epoch": 0.81, + "learning_rate": 4.796653497559842e-05, + "loss": 1.6682, + "step": 59500 + }, + { + "epoch": 0.82, + "learning_rate": 4.79494470342169e-05, + "loss": 1.7702, + "step": 60000 + }, + { + "epoch": 0.83, + "learning_rate": 4.793235909283537e-05, + "loss": 1.6938, + "step": 60500 + }, + { + "epoch": 0.83, + "learning_rate": 4.791527115145385e-05, + "loss": 1.6923, + "step": 61000 + }, + { + "epoch": 0.84, + "learning_rate": 4.789818321007232e-05, + "loss": 1.7227, + "step": 61500 + }, + { + "epoch": 0.85, + "learning_rate": 4.788109526869079e-05, + "loss": 1.7288, + "step": 62000 + }, + { + "epoch": 0.85, + "learning_rate": 4.7864007327309266e-05, + "loss": 1.738, + "step": 62500 + }, + { + "epoch": 0.86, + "learning_rate": 4.784691938592774e-05, + "loss": 1.7445, + "step": 63000 + }, + { + "epoch": 0.87, + "learning_rate": 4.7829831444546213e-05, + "loss": 1.7262, + "step": 63500 + }, + { + "epoch": 0.87, + "learning_rate": 4.7812743503164684e-05, + "loss": 1.7333, + "step": 64000 + }, + { + "epoch": 0.88, + "learning_rate": 4.779565556178317e-05, + "loss": 1.7155, + "step": 64500 + }, + { + "epoch": 0.89, + "learning_rate": 4.777856762040164e-05, + "loss": 1.7099, + "step": 65000 + }, + { + "epoch": 0.9, + "learning_rate": 4.776147967902011e-05, + "loss": 1.7325, + "step": 65500 + }, + { + "epoch": 0.9, + "learning_rate": 4.774439173763858e-05, + "loss": 1.7243, + "step": 66000 + }, + { + "epoch": 0.91, + "learning_rate": 4.7727303796257064e-05, + "loss": 1.7023, + "step": 66500 + }, + { + "epoch": 0.92, + "learning_rate": 4.7710215854875534e-05, + "loss": 1.7563, + "step": 67000 + }, + { + "epoch": 0.92, + "learning_rate": 4.7693127913494005e-05, + "loss": 1.7127, + "step": 67500 + }, + { + "epoch": 0.93, + "learning_rate": 4.767603997211248e-05, + "loss": 1.7144, + "step": 68000 + }, + { + "epoch": 0.94, + "learning_rate": 4.765895203073096e-05, + "loss": 1.7052, + "step": 68500 + }, + { + "epoch": 0.94, + "learning_rate": 4.764186408934943e-05, + "loss": 1.7193, + "step": 69000 + }, + { + "epoch": 0.95, + "learning_rate": 4.76247761479679e-05, + "loss": 1.7261, + "step": 69500 + }, + { + "epoch": 0.96, + "learning_rate": 4.760768820658638e-05, + "loss": 1.7249, + "step": 70000 + }, + { + "epoch": 0.96, + "learning_rate": 4.7590600265204855e-05, + "loss": 1.7228, + "step": 70500 + }, + { + "epoch": 0.97, + "learning_rate": 4.7573512323823326e-05, + "loss": 1.7152, + "step": 71000 + }, + { + "epoch": 0.98, + "learning_rate": 4.75564243824418e-05, + "loss": 1.7198, + "step": 71500 + }, + { + "epoch": 0.98, + "learning_rate": 4.7539336441060273e-05, + "loss": 1.7021, + "step": 72000 + }, + { + "epoch": 0.99, + "learning_rate": 4.752224849967875e-05, + "loss": 1.6822, + "step": 72500 + }, + { + "epoch": 1.0, + "learning_rate": 4.750516055829722e-05, + "loss": 1.738, + "step": 73000 + }, + { + "epoch": 1.0, + "learning_rate": 4.74880726169157e-05, + "loss": 1.7199, + "step": 73500 + }, + { + "epoch": 1.01, + "learning_rate": 4.747098467553417e-05, + "loss": 1.6915, + "step": 74000 + }, + { + "epoch": 1.02, + "learning_rate": 4.7453896734152646e-05, + "loss": 1.6833, + "step": 74500 + }, + { + "epoch": 1.03, + "learning_rate": 4.743680879277112e-05, + "loss": 1.7051, + "step": 75000 + }, + { + "epoch": 1.03, + "learning_rate": 4.7419720851389594e-05, + "loss": 1.6859, + "step": 75500 + }, + { + "epoch": 1.04, + "learning_rate": 4.7402632910008065e-05, + "loss": 1.6844, + "step": 76000 + }, + { + "epoch": 1.05, + "learning_rate": 4.738554496862654e-05, + "loss": 1.6841, + "step": 76500 + }, + { + "epoch": 1.05, + "learning_rate": 4.736845702724502e-05, + "loss": 1.6736, + "step": 77000 + }, + { + "epoch": 1.06, + "learning_rate": 4.735136908586349e-05, + "loss": 1.7044, + "step": 77500 + }, + { + "epoch": 1.07, + "learning_rate": 4.733428114448196e-05, + "loss": 1.6701, + "step": 78000 + }, + { + "epoch": 1.07, + "learning_rate": 4.731719320310044e-05, + "loss": 1.668, + "step": 78500 + }, + { + "epoch": 1.08, + "learning_rate": 4.7300105261718915e-05, + "loss": 1.6833, + "step": 79000 + }, + { + "epoch": 1.09, + "learning_rate": 4.7283017320337386e-05, + "loss": 1.7029, + "step": 79500 + }, + { + "epoch": 1.09, + "learning_rate": 4.7265929378955856e-05, + "loss": 1.6519, + "step": 80000 + }, + { + "epoch": 1.1, + "learning_rate": 4.724884143757434e-05, + "loss": 1.6832, + "step": 80500 + }, + { + "epoch": 1.11, + "learning_rate": 4.723175349619281e-05, + "loss": 1.668, + "step": 81000 + }, + { + "epoch": 1.11, + "learning_rate": 4.721466555481128e-05, + "loss": 1.708, + "step": 81500 + }, + { + "epoch": 1.12, + "learning_rate": 4.719757761342976e-05, + "loss": 1.6628, + "step": 82000 + }, + { + "epoch": 1.13, + "learning_rate": 4.7180489672048236e-05, + "loss": 1.6698, + "step": 82500 + }, + { + "epoch": 1.13, + "learning_rate": 4.7163401730666706e-05, + "loss": 1.6706, + "step": 83000 + }, + { + "epoch": 1.14, + "learning_rate": 4.714631378928518e-05, + "loss": 1.6827, + "step": 83500 + }, + { + "epoch": 1.15, + "learning_rate": 4.7129225847903654e-05, + "loss": 1.6742, + "step": 84000 + }, + { + "epoch": 1.16, + "learning_rate": 4.711213790652213e-05, + "loss": 1.6913, + "step": 84500 + }, + { + "epoch": 1.16, + "learning_rate": 4.70950499651406e-05, + "loss": 1.6682, + "step": 85000 + }, + { + "epoch": 1.17, + "learning_rate": 4.707796202375907e-05, + "loss": 1.6664, + "step": 85500 + }, + { + "epoch": 1.18, + "learning_rate": 4.706087408237755e-05, + "loss": 1.6785, + "step": 86000 + }, + { + "epoch": 1.18, + "learning_rate": 4.704378614099603e-05, + "loss": 1.6958, + "step": 86500 + }, + { + "epoch": 1.19, + "learning_rate": 4.70266981996145e-05, + "loss": 1.6947, + "step": 87000 + }, + { + "epoch": 1.2, + "learning_rate": 4.7009610258232975e-05, + "loss": 1.6733, + "step": 87500 + }, + { + "epoch": 1.2, + "learning_rate": 4.6992522316851445e-05, + "loss": 1.6944, + "step": 88000 + }, + { + "epoch": 1.21, + "learning_rate": 4.697543437546992e-05, + "loss": 1.6591, + "step": 88500 + }, + { + "epoch": 1.22, + "learning_rate": 4.695834643408839e-05, + "loss": 1.6788, + "step": 89000 + }, + { + "epoch": 1.22, + "learning_rate": 4.694125849270687e-05, + "loss": 1.6934, + "step": 89500 + }, + { + "epoch": 1.23, + "learning_rate": 4.692417055132534e-05, + "loss": 1.7039, + "step": 90000 + }, + { + "epoch": 1.24, + "learning_rate": 4.690708260994382e-05, + "loss": 1.6924, + "step": 90500 + }, + { + "epoch": 1.24, + "learning_rate": 4.6889994668562296e-05, + "loss": 1.6871, + "step": 91000 + }, + { + "epoch": 1.25, + "learning_rate": 4.6872906727180766e-05, + "loss": 1.6847, + "step": 91500 + }, + { + "epoch": 1.26, + "learning_rate": 4.685581878579924e-05, + "loss": 1.6631, + "step": 92000 + }, + { + "epoch": 1.26, + "learning_rate": 4.6838730844417714e-05, + "loss": 1.661, + "step": 92500 + }, + { + "epoch": 1.27, + "learning_rate": 4.682164290303619e-05, + "loss": 1.6839, + "step": 93000 + }, + { + "epoch": 1.28, + "learning_rate": 4.680455496165466e-05, + "loss": 1.6881, + "step": 93500 + }, + { + "epoch": 1.29, + "learning_rate": 4.678746702027313e-05, + "loss": 1.7025, + "step": 94000 + }, + { + "epoch": 1.29, + "learning_rate": 4.677037907889161e-05, + "loss": 1.6903, + "step": 94500 + }, + { + "epoch": 1.3, + "learning_rate": 4.675329113751009e-05, + "loss": 1.6555, + "step": 95000 + }, + { + "epoch": 1.31, + "learning_rate": 4.673620319612856e-05, + "loss": 1.6668, + "step": 95500 + }, + { + "epoch": 1.31, + "learning_rate": 4.671911525474703e-05, + "loss": 1.6824, + "step": 96000 + }, + { + "epoch": 1.32, + "learning_rate": 4.6702027313365505e-05, + "loss": 1.6683, + "step": 96500 + }, + { + "epoch": 1.33, + "learning_rate": 4.668493937198398e-05, + "loss": 1.6705, + "step": 97000 + }, + { + "epoch": 1.33, + "learning_rate": 4.666785143060245e-05, + "loss": 1.6714, + "step": 97500 + }, + { + "epoch": 1.34, + "learning_rate": 4.665076348922093e-05, + "loss": 1.68, + "step": 98000 + }, + { + "epoch": 1.35, + "learning_rate": 4.66336755478394e-05, + "loss": 1.6724, + "step": 98500 + }, + { + "epoch": 1.35, + "learning_rate": 4.661658760645788e-05, + "loss": 1.6726, + "step": 99000 + }, + { + "epoch": 1.36, + "learning_rate": 4.659949966507635e-05, + "loss": 1.6796, + "step": 99500 + }, + { + "epoch": 1.37, + "learning_rate": 4.6582411723694826e-05, + "loss": 1.6673, + "step": 100000 + }, + { + "epoch": 1.37, + "learning_rate": 4.65653237823133e-05, + "loss": 1.6927, + "step": 100500 + }, + { + "epoch": 1.38, + "learning_rate": 4.6548235840931774e-05, + "loss": 1.6485, + "step": 101000 + }, + { + "epoch": 1.39, + "learning_rate": 4.653114789955025e-05, + "loss": 1.681, + "step": 101500 + }, + { + "epoch": 1.39, + "learning_rate": 4.651405995816872e-05, + "loss": 1.6627, + "step": 102000 + }, + { + "epoch": 1.4, + "learning_rate": 4.649697201678719e-05, + "loss": 1.6904, + "step": 102500 + }, + { + "epoch": 1.41, + "learning_rate": 4.647988407540567e-05, + "loss": 1.6642, + "step": 103000 + }, + { + "epoch": 1.41, + "learning_rate": 4.646279613402415e-05, + "loss": 1.665, + "step": 103500 + }, + { + "epoch": 1.42, + "learning_rate": 4.644570819264262e-05, + "loss": 1.6653, + "step": 104000 + }, + { + "epoch": 1.43, + "learning_rate": 4.642862025126109e-05, + "loss": 1.6636, + "step": 104500 + }, + { + "epoch": 1.44, + "learning_rate": 4.6411532309879565e-05, + "loss": 1.6795, + "step": 105000 + }, + { + "epoch": 1.44, + "learning_rate": 4.639444436849804e-05, + "loss": 1.6647, + "step": 105500 + }, + { + "epoch": 1.45, + "learning_rate": 4.637735642711651e-05, + "loss": 1.6841, + "step": 106000 + }, + { + "epoch": 1.46, + "learning_rate": 4.6360268485734984e-05, + "loss": 1.6517, + "step": 106500 + }, + { + "epoch": 1.46, + "learning_rate": 4.634318054435347e-05, + "loss": 1.673, + "step": 107000 + }, + { + "epoch": 1.47, + "learning_rate": 4.632609260297194e-05, + "loss": 1.6962, + "step": 107500 + }, + { + "epoch": 1.48, + "learning_rate": 4.630900466159041e-05, + "loss": 1.6813, + "step": 108000 + }, + { + "epoch": 1.48, + "learning_rate": 4.6291916720208886e-05, + "loss": 1.6676, + "step": 108500 + }, + { + "epoch": 1.49, + "learning_rate": 4.6274828778827363e-05, + "loss": 1.668, + "step": 109000 + }, + { + "epoch": 1.5, + "learning_rate": 4.6257740837445834e-05, + "loss": 1.6711, + "step": 109500 + }, + { + "epoch": 1.5, + "learning_rate": 4.6240652896064305e-05, + "loss": 1.7066, + "step": 110000 + }, + { + "epoch": 1.51, + "learning_rate": 4.622356495468278e-05, + "loss": 1.6677, + "step": 110500 + }, + { + "epoch": 1.52, + "learning_rate": 4.620647701330126e-05, + "loss": 1.6588, + "step": 111000 + }, + { + "epoch": 1.52, + "learning_rate": 4.618938907191973e-05, + "loss": 1.6926, + "step": 111500 + }, + { + "epoch": 1.53, + "learning_rate": 4.617230113053821e-05, + "loss": 1.6514, + "step": 112000 + }, + { + "epoch": 1.54, + "learning_rate": 4.615521318915668e-05, + "loss": 1.6663, + "step": 112500 + }, + { + "epoch": 1.54, + "learning_rate": 4.6138125247775155e-05, + "loss": 1.6323, + "step": 113000 + }, + { + "epoch": 1.55, + "learning_rate": 4.6121037306393625e-05, + "loss": 1.6711, + "step": 113500 + }, + { + "epoch": 1.56, + "learning_rate": 4.61039493650121e-05, + "loss": 1.6964, + "step": 114000 + }, + { + "epoch": 1.57, + "learning_rate": 4.608686142363057e-05, + "loss": 1.6824, + "step": 114500 + }, + { + "epoch": 1.57, + "learning_rate": 4.606977348224905e-05, + "loss": 1.6623, + "step": 115000 + }, + { + "epoch": 1.58, + "learning_rate": 4.605268554086752e-05, + "loss": 1.666, + "step": 115500 + }, + { + "epoch": 1.59, + "learning_rate": 4.6035597599486e-05, + "loss": 1.681, + "step": 116000 + }, + { + "epoch": 1.59, + "learning_rate": 4.601850965810447e-05, + "loss": 1.6639, + "step": 116500 + }, + { + "epoch": 1.6, + "learning_rate": 4.6001421716722946e-05, + "loss": 1.7127, + "step": 117000 + }, + { + "epoch": 1.61, + "learning_rate": 4.5984333775341423e-05, + "loss": 1.6531, + "step": 117500 + }, + { + "epoch": 1.61, + "learning_rate": 4.5967245833959894e-05, + "loss": 1.6766, + "step": 118000 + }, + { + "epoch": 1.62, + "learning_rate": 4.5950157892578365e-05, + "loss": 1.681, + "step": 118500 + }, + { + "epoch": 1.63, + "learning_rate": 4.593306995119684e-05, + "loss": 1.6817, + "step": 119000 + }, + { + "epoch": 1.63, + "learning_rate": 4.591598200981532e-05, + "loss": 1.6909, + "step": 119500 + }, + { + "epoch": 1.64, + "learning_rate": 4.589889406843379e-05, + "loss": 1.6613, + "step": 120000 + }, + { + "epoch": 1.65, + "learning_rate": 4.588180612705226e-05, + "loss": 1.6732, + "step": 120500 + }, + { + "epoch": 1.65, + "learning_rate": 4.586471818567074e-05, + "loss": 1.6633, + "step": 121000 + }, + { + "epoch": 1.66, + "learning_rate": 4.5847630244289215e-05, + "loss": 1.654, + "step": 121500 + }, + { + "epoch": 1.67, + "learning_rate": 4.5830542302907685e-05, + "loss": 1.6858, + "step": 122000 + }, + { + "epoch": 1.67, + "learning_rate": 4.5813454361526156e-05, + "loss": 1.7091, + "step": 122500 + }, + { + "epoch": 1.68, + "learning_rate": 4.579636642014463e-05, + "loss": 1.6517, + "step": 123000 + }, + { + "epoch": 1.69, + "learning_rate": 4.577927847876311e-05, + "loss": 1.6496, + "step": 123500 + }, + { + "epoch": 1.7, + "learning_rate": 4.576219053738158e-05, + "loss": 1.6469, + "step": 124000 + }, + { + "epoch": 1.7, + "learning_rate": 4.574510259600006e-05, + "loss": 1.6625, + "step": 124500 + }, + { + "epoch": 1.71, + "learning_rate": 4.572801465461853e-05, + "loss": 1.6845, + "step": 125000 + }, + { + "epoch": 1.72, + "learning_rate": 4.5710926713237006e-05, + "loss": 1.6496, + "step": 125500 + }, + { + "epoch": 1.72, + "learning_rate": 4.569383877185548e-05, + "loss": 1.6746, + "step": 126000 + }, + { + "epoch": 1.73, + "learning_rate": 4.5676750830473954e-05, + "loss": 1.6498, + "step": 126500 + }, + { + "epoch": 1.74, + "learning_rate": 4.5659662889092424e-05, + "loss": 1.6495, + "step": 127000 + }, + { + "epoch": 1.74, + "learning_rate": 4.56425749477109e-05, + "loss": 1.6495, + "step": 127500 + }, + { + "epoch": 1.75, + "learning_rate": 4.562548700632938e-05, + "loss": 1.6483, + "step": 128000 + }, + { + "epoch": 1.76, + "learning_rate": 4.560839906494785e-05, + "loss": 1.6658, + "step": 128500 + }, + { + "epoch": 1.76, + "learning_rate": 4.559131112356632e-05, + "loss": 1.6807, + "step": 129000 + }, + { + "epoch": 1.77, + "learning_rate": 4.55742231821848e-05, + "loss": 1.6903, + "step": 129500 + }, + { + "epoch": 1.78, + "learning_rate": 4.5557135240803275e-05, + "loss": 1.6527, + "step": 130000 + }, + { + "epoch": 1.78, + "learning_rate": 4.5540047299421745e-05, + "loss": 1.6478, + "step": 130500 + }, + { + "epoch": 1.79, + "learning_rate": 4.5522959358040216e-05, + "loss": 1.6666, + "step": 131000 + }, + { + "epoch": 1.8, + "learning_rate": 4.550587141665869e-05, + "loss": 1.6773, + "step": 131500 + }, + { + "epoch": 1.8, + "learning_rate": 4.548878347527717e-05, + "loss": 1.6471, + "step": 132000 + }, + { + "epoch": 1.81, + "learning_rate": 4.547169553389564e-05, + "loss": 1.6603, + "step": 132500 + }, + { + "epoch": 1.82, + "learning_rate": 4.545460759251411e-05, + "loss": 1.6881, + "step": 133000 + }, + { + "epoch": 1.82, + "learning_rate": 4.5437519651132596e-05, + "loss": 1.6795, + "step": 133500 + }, + { + "epoch": 1.83, + "learning_rate": 4.5420431709751066e-05, + "loss": 1.6393, + "step": 134000 + }, + { + "epoch": 1.84, + "learning_rate": 4.5403343768369537e-05, + "loss": 1.6754, + "step": 134500 + }, + { + "epoch": 1.85, + "learning_rate": 4.5386255826988014e-05, + "loss": 1.6431, + "step": 135000 + }, + { + "epoch": 1.85, + "learning_rate": 4.536916788560649e-05, + "loss": 1.6613, + "step": 135500 + }, + { + "epoch": 1.86, + "learning_rate": 4.535207994422496e-05, + "loss": 1.6676, + "step": 136000 + }, + { + "epoch": 1.87, + "learning_rate": 4.533499200284343e-05, + "loss": 1.6418, + "step": 136500 + }, + { + "epoch": 1.87, + "learning_rate": 4.531790406146191e-05, + "loss": 1.6866, + "step": 137000 + }, + { + "epoch": 1.88, + "learning_rate": 4.530081612008039e-05, + "loss": 1.6698, + "step": 137500 + }, + { + "epoch": 1.89, + "learning_rate": 4.528372817869886e-05, + "loss": 1.6441, + "step": 138000 + }, + { + "epoch": 1.89, + "learning_rate": 4.5266640237317335e-05, + "loss": 1.6772, + "step": 138500 + }, + { + "epoch": 1.9, + "learning_rate": 4.5249552295935805e-05, + "loss": 1.6721, + "step": 139000 + }, + { + "epoch": 1.91, + "learning_rate": 4.523246435455428e-05, + "loss": 1.6459, + "step": 139500 + }, + { + "epoch": 1.91, + "learning_rate": 4.521537641317275e-05, + "loss": 1.6839, + "step": 140000 + }, + { + "epoch": 1.92, + "learning_rate": 4.519828847179123e-05, + "loss": 1.7087, + "step": 140500 + }, + { + "epoch": 1.93, + "learning_rate": 4.51812005304097e-05, + "loss": 1.6571, + "step": 141000 + }, + { + "epoch": 1.93, + "learning_rate": 4.516411258902818e-05, + "loss": 1.6484, + "step": 141500 + }, + { + "epoch": 1.94, + "learning_rate": 4.514702464764665e-05, + "loss": 1.6738, + "step": 142000 + }, + { + "epoch": 1.95, + "learning_rate": 4.5129936706265126e-05, + "loss": 1.6725, + "step": 142500 + }, + { + "epoch": 1.95, + "learning_rate": 4.5112848764883597e-05, + "loss": 1.6569, + "step": 143000 + }, + { + "epoch": 1.96, + "learning_rate": 4.5095760823502074e-05, + "loss": 1.675, + "step": 143500 + }, + { + "epoch": 1.97, + "learning_rate": 4.507867288212055e-05, + "loss": 1.6397, + "step": 144000 + }, + { + "epoch": 1.98, + "learning_rate": 4.506158494073902e-05, + "loss": 1.6479, + "step": 144500 + }, + { + "epoch": 1.98, + "learning_rate": 4.504449699935749e-05, + "loss": 1.6534, + "step": 145000 + }, + { + "epoch": 1.99, + "learning_rate": 4.502740905797597e-05, + "loss": 1.6376, + "step": 145500 + }, + { + "epoch": 2.0, + "learning_rate": 4.501032111659445e-05, + "loss": 1.6537, + "step": 146000 + }, + { + "epoch": 2.0, + "learning_rate": 4.499323317521292e-05, + "loss": 1.6765, + "step": 146500 + }, + { + "epoch": 2.01, + "learning_rate": 4.497614523383139e-05, + "loss": 1.6457, + "step": 147000 + }, + { + "epoch": 2.02, + "learning_rate": 4.495905729244987e-05, + "loss": 1.6054, + "step": 147500 + }, + { + "epoch": 2.02, + "learning_rate": 4.494196935106834e-05, + "loss": 1.6346, + "step": 148000 + }, + { + "epoch": 2.03, + "learning_rate": 4.492488140968681e-05, + "loss": 1.6306, + "step": 148500 + }, + { + "epoch": 2.04, + "learning_rate": 4.490779346830529e-05, + "loss": 1.6247, + "step": 149000 + }, + { + "epoch": 2.04, + "learning_rate": 4.489070552692377e-05, + "loss": 1.6361, + "step": 149500 + }, + { + "epoch": 2.05, + "learning_rate": 4.487361758554224e-05, + "loss": 1.6056, + "step": 150000 + }, + { + "epoch": 2.06, + "learning_rate": 4.485652964416071e-05, + "loss": 1.636, + "step": 150500 + }, + { + "epoch": 2.06, + "learning_rate": 4.4839441702779186e-05, + "loss": 1.657, + "step": 151000 + }, + { + "epoch": 2.07, + "learning_rate": 4.4822353761397657e-05, + "loss": 1.6613, + "step": 151500 + }, + { + "epoch": 2.08, + "learning_rate": 4.4805265820016134e-05, + "loss": 1.6168, + "step": 152000 + }, + { + "epoch": 2.08, + "learning_rate": 4.4788177878634604e-05, + "loss": 1.6497, + "step": 152500 + }, + { + "epoch": 2.09, + "learning_rate": 4.477108993725308e-05, + "loss": 1.6149, + "step": 153000 + }, + { + "epoch": 2.1, + "learning_rate": 4.475400199587155e-05, + "loss": 1.6698, + "step": 153500 + }, + { + "epoch": 2.11, + "learning_rate": 4.473691405449003e-05, + "loss": 1.6537, + "step": 154000 + }, + { + "epoch": 2.11, + "learning_rate": 4.471982611310851e-05, + "loss": 1.6259, + "step": 154500 + }, + { + "epoch": 2.12, + "learning_rate": 4.470273817172698e-05, + "loss": 1.6526, + "step": 155000 + }, + { + "epoch": 2.13, + "learning_rate": 4.468565023034545e-05, + "loss": 1.6334, + "step": 155500 + }, + { + "epoch": 2.13, + "learning_rate": 4.4668562288963925e-05, + "loss": 1.6551, + "step": 156000 + }, + { + "epoch": 2.14, + "learning_rate": 4.46514743475824e-05, + "loss": 1.6581, + "step": 156500 + }, + { + "epoch": 2.15, + "learning_rate": 4.463438640620087e-05, + "loss": 1.6311, + "step": 157000 + }, + { + "epoch": 2.15, + "learning_rate": 4.4617298464819344e-05, + "loss": 1.6158, + "step": 157500 + }, + { + "epoch": 2.16, + "learning_rate": 4.460021052343783e-05, + "loss": 1.6821, + "step": 158000 + }, + { + "epoch": 2.17, + "learning_rate": 4.45831225820563e-05, + "loss": 1.6481, + "step": 158500 + }, + { + "epoch": 2.17, + "learning_rate": 4.456603464067477e-05, + "loss": 1.6342, + "step": 159000 + }, + { + "epoch": 2.18, + "learning_rate": 4.454894669929324e-05, + "loss": 1.6489, + "step": 159500 + }, + { + "epoch": 2.19, + "learning_rate": 4.453185875791172e-05, + "loss": 1.6429, + "step": 160000 + }, + { + "epoch": 2.19, + "learning_rate": 4.4514770816530194e-05, + "loss": 1.6273, + "step": 160500 + }, + { + "epoch": 2.2, + "learning_rate": 4.4497682875148664e-05, + "loss": 1.6558, + "step": 161000 + }, + { + "epoch": 2.21, + "learning_rate": 4.448059493376714e-05, + "loss": 1.643, + "step": 161500 + }, + { + "epoch": 2.21, + "learning_rate": 4.446350699238562e-05, + "loss": 1.6613, + "step": 162000 + }, + { + "epoch": 2.22, + "learning_rate": 4.444641905100409e-05, + "loss": 1.6647, + "step": 162500 + }, + { + "epoch": 2.23, + "learning_rate": 4.442933110962256e-05, + "loss": 1.6445, + "step": 163000 + }, + { + "epoch": 2.24, + "learning_rate": 4.441224316824104e-05, + "loss": 1.6504, + "step": 163500 + }, + { + "epoch": 2.24, + "learning_rate": 4.4395155226859515e-05, + "loss": 1.624, + "step": 164000 + }, + { + "epoch": 2.25, + "learning_rate": 4.4378067285477985e-05, + "loss": 1.6553, + "step": 164500 + }, + { + "epoch": 2.26, + "learning_rate": 4.436097934409646e-05, + "loss": 1.6723, + "step": 165000 + }, + { + "epoch": 2.26, + "learning_rate": 4.434389140271493e-05, + "loss": 1.6559, + "step": 165500 + }, + { + "epoch": 2.27, + "learning_rate": 4.432680346133341e-05, + "loss": 1.6013, + "step": 166000 + }, + { + "epoch": 2.28, + "learning_rate": 4.430971551995188e-05, + "loss": 1.6809, + "step": 166500 + }, + { + "epoch": 2.28, + "learning_rate": 4.429262757857036e-05, + "loss": 1.6469, + "step": 167000 + }, + { + "epoch": 2.29, + "learning_rate": 4.427553963718883e-05, + "loss": 1.6107, + "step": 167500 + }, + { + "epoch": 2.3, + "learning_rate": 4.4258451695807306e-05, + "loss": 1.626, + "step": 168000 + }, + { + "epoch": 2.3, + "learning_rate": 4.4241363754425776e-05, + "loss": 1.6325, + "step": 168500 + }, + { + "epoch": 2.31, + "learning_rate": 4.4224275813044254e-05, + "loss": 1.6425, + "step": 169000 + }, + { + "epoch": 2.32, + "learning_rate": 4.4207187871662724e-05, + "loss": 1.6636, + "step": 169500 + }, + { + "epoch": 2.32, + "learning_rate": 4.41900999302812e-05, + "loss": 1.6331, + "step": 170000 + }, + { + "epoch": 2.33, + "learning_rate": 4.417301198889968e-05, + "loss": 1.6451, + "step": 170500 + }, + { + "epoch": 2.34, + "learning_rate": 4.415592404751815e-05, + "loss": 1.657, + "step": 171000 + }, + { + "epoch": 2.34, + "learning_rate": 4.413883610613662e-05, + "loss": 1.6428, + "step": 171500 + }, + { + "epoch": 2.35, + "learning_rate": 4.41217481647551e-05, + "loss": 1.6836, + "step": 172000 + }, + { + "epoch": 2.36, + "learning_rate": 4.4104660223373575e-05, + "loss": 1.6531, + "step": 172500 + }, + { + "epoch": 2.36, + "learning_rate": 4.4087572281992045e-05, + "loss": 1.6323, + "step": 173000 + }, + { + "epoch": 2.37, + "learning_rate": 4.4070484340610516e-05, + "loss": 1.6358, + "step": 173500 + }, + { + "epoch": 2.38, + "learning_rate": 4.4053396399229e-05, + "loss": 1.6504, + "step": 174000 + }, + { + "epoch": 2.39, + "learning_rate": 4.403630845784747e-05, + "loss": 1.65, + "step": 174500 + }, + { + "epoch": 2.39, + "learning_rate": 4.401922051646594e-05, + "loss": 1.6332, + "step": 175000 + }, + { + "epoch": 2.4, + "learning_rate": 4.400213257508442e-05, + "loss": 1.6275, + "step": 175500 + }, + { + "epoch": 2.41, + "learning_rate": 4.3985044633702895e-05, + "loss": 1.6483, + "step": 176000 + }, + { + "epoch": 2.41, + "learning_rate": 4.3967956692321366e-05, + "loss": 1.6269, + "step": 176500 + }, + { + "epoch": 2.42, + "learning_rate": 4.3950868750939836e-05, + "loss": 1.6639, + "step": 177000 + }, + { + "epoch": 2.43, + "learning_rate": 4.3933780809558314e-05, + "loss": 1.64, + "step": 177500 + }, + { + "epoch": 2.43, + "learning_rate": 4.391669286817679e-05, + "loss": 1.6439, + "step": 178000 + }, + { + "epoch": 2.44, + "learning_rate": 4.389960492679526e-05, + "loss": 1.6652, + "step": 178500 + }, + { + "epoch": 2.45, + "learning_rate": 4.388251698541373e-05, + "loss": 1.6623, + "step": 179000 + }, + { + "epoch": 2.45, + "learning_rate": 4.386542904403221e-05, + "loss": 1.651, + "step": 179500 + }, + { + "epoch": 2.46, + "learning_rate": 4.384834110265069e-05, + "loss": 1.6284, + "step": 180000 + }, + { + "epoch": 2.47, + "learning_rate": 4.383125316126916e-05, + "loss": 1.6354, + "step": 180500 + }, + { + "epoch": 2.47, + "learning_rate": 4.3814165219887634e-05, + "loss": 1.6287, + "step": 181000 + }, + { + "epoch": 2.48, + "learning_rate": 4.3797077278506105e-05, + "loss": 1.6103, + "step": 181500 + }, + { + "epoch": 2.49, + "learning_rate": 4.3779989337124576e-05, + "loss": 1.6191, + "step": 182000 + }, + { + "epoch": 2.49, + "learning_rate": 4.376290139574305e-05, + "loss": 1.6611, + "step": 182500 + }, + { + "epoch": 2.5, + "learning_rate": 4.374581345436153e-05, + "loss": 1.6783, + "step": 183000 + }, + { + "epoch": 2.51, + "learning_rate": 4.372872551298e-05, + "loss": 1.6511, + "step": 183500 + }, + { + "epoch": 2.52, + "learning_rate": 4.371163757159847e-05, + "loss": 1.6421, + "step": 184000 + }, + { + "epoch": 2.52, + "learning_rate": 4.3694549630216955e-05, + "loss": 1.6397, + "step": 184500 + }, + { + "epoch": 2.53, + "learning_rate": 4.3677461688835426e-05, + "loss": 1.6442, + "step": 185000 + }, + { + "epoch": 2.54, + "learning_rate": 4.3660373747453896e-05, + "loss": 1.6403, + "step": 185500 + }, + { + "epoch": 2.54, + "learning_rate": 4.3643285806072374e-05, + "loss": 1.642, + "step": 186000 + }, + { + "epoch": 2.55, + "learning_rate": 4.362619786469085e-05, + "loss": 1.6418, + "step": 186500 + }, + { + "epoch": 2.56, + "learning_rate": 4.360910992330932e-05, + "loss": 1.6533, + "step": 187000 + }, + { + "epoch": 2.56, + "learning_rate": 4.359202198192779e-05, + "loss": 1.6242, + "step": 187500 + }, + { + "epoch": 2.57, + "learning_rate": 4.357493404054627e-05, + "loss": 1.625, + "step": 188000 + }, + { + "epoch": 2.58, + "learning_rate": 4.355784609916475e-05, + "loss": 1.6101, + "step": 188500 + }, + { + "epoch": 2.58, + "learning_rate": 4.354075815778322e-05, + "loss": 1.6438, + "step": 189000 + }, + { + "epoch": 2.59, + "learning_rate": 4.352367021640169e-05, + "loss": 1.6322, + "step": 189500 + }, + { + "epoch": 2.6, + "learning_rate": 4.3506582275020165e-05, + "loss": 1.6285, + "step": 190000 + }, + { + "epoch": 2.6, + "learning_rate": 4.348949433363864e-05, + "loss": 1.617, + "step": 190500 + }, + { + "epoch": 2.61, + "learning_rate": 4.347240639225711e-05, + "loss": 1.651, + "step": 191000 + }, + { + "epoch": 2.62, + "learning_rate": 4.345531845087559e-05, + "loss": 1.6354, + "step": 191500 + }, + { + "epoch": 2.62, + "learning_rate": 4.343823050949406e-05, + "loss": 1.6221, + "step": 192000 + }, + { + "epoch": 2.63, + "learning_rate": 4.342114256811254e-05, + "loss": 1.6417, + "step": 192500 + }, + { + "epoch": 2.64, + "learning_rate": 4.340405462673101e-05, + "loss": 1.6138, + "step": 193000 + }, + { + "epoch": 2.65, + "learning_rate": 4.3386966685349486e-05, + "loss": 1.68, + "step": 193500 + }, + { + "epoch": 2.65, + "learning_rate": 4.3369878743967956e-05, + "loss": 1.6203, + "step": 194000 + }, + { + "epoch": 2.66, + "learning_rate": 4.3352790802586434e-05, + "loss": 1.65, + "step": 194500 + }, + { + "epoch": 2.67, + "learning_rate": 4.333570286120491e-05, + "loss": 1.6408, + "step": 195000 + }, + { + "epoch": 2.67, + "learning_rate": 4.331861491982338e-05, + "loss": 1.6471, + "step": 195500 + }, + { + "epoch": 2.68, + "learning_rate": 4.330152697844185e-05, + "loss": 1.5934, + "step": 196000 + }, + { + "epoch": 2.69, + "learning_rate": 4.328443903706033e-05, + "loss": 1.6501, + "step": 196500 + }, + { + "epoch": 2.69, + "learning_rate": 4.3267351095678807e-05, + "loss": 1.6508, + "step": 197000 + }, + { + "epoch": 2.7, + "learning_rate": 4.325026315429728e-05, + "loss": 1.6344, + "step": 197500 + }, + { + "epoch": 2.71, + "learning_rate": 4.323317521291575e-05, + "loss": 1.6142, + "step": 198000 + }, + { + "epoch": 2.71, + "learning_rate": 4.3216087271534225e-05, + "loss": 1.643, + "step": 198500 + }, + { + "epoch": 2.72, + "learning_rate": 4.31989993301527e-05, + "loss": 1.6459, + "step": 199000 + }, + { + "epoch": 2.73, + "learning_rate": 4.318191138877117e-05, + "loss": 1.6283, + "step": 199500 + }, + { + "epoch": 2.73, + "learning_rate": 4.316482344738964e-05, + "loss": 1.6283, + "step": 200000 + }, + { + "epoch": 2.74, + "learning_rate": 4.314773550600813e-05, + "loss": 1.6556, + "step": 200500 + }, + { + "epoch": 2.75, + "learning_rate": 4.31306475646266e-05, + "loss": 1.6557, + "step": 201000 + }, + { + "epoch": 2.75, + "learning_rate": 4.311355962324507e-05, + "loss": 1.655, + "step": 201500 + }, + { + "epoch": 2.76, + "learning_rate": 4.3096471681863546e-05, + "loss": 1.6659, + "step": 202000 + }, + { + "epoch": 2.77, + "learning_rate": 4.307938374048202e-05, + "loss": 1.6317, + "step": 202500 + }, + { + "epoch": 2.78, + "learning_rate": 4.3062295799100494e-05, + "loss": 1.6662, + "step": 203000 + }, + { + "epoch": 2.78, + "learning_rate": 4.3045207857718964e-05, + "loss": 1.6066, + "step": 203500 + }, + { + "epoch": 2.79, + "learning_rate": 4.302811991633744e-05, + "loss": 1.631, + "step": 204000 + }, + { + "epoch": 2.8, + "learning_rate": 4.301103197495592e-05, + "loss": 1.6333, + "step": 204500 + }, + { + "epoch": 2.8, + "learning_rate": 4.299394403357439e-05, + "loss": 1.6518, + "step": 205000 + }, + { + "epoch": 2.81, + "learning_rate": 4.297685609219286e-05, + "loss": 1.6474, + "step": 205500 + }, + { + "epoch": 2.82, + "learning_rate": 4.295976815081134e-05, + "loss": 1.6295, + "step": 206000 + }, + { + "epoch": 2.82, + "learning_rate": 4.2942680209429814e-05, + "loss": 1.6484, + "step": 206500 + }, + { + "epoch": 2.83, + "learning_rate": 4.2925592268048285e-05, + "loss": 1.6671, + "step": 207000 + }, + { + "epoch": 2.84, + "learning_rate": 4.290850432666676e-05, + "loss": 1.6483, + "step": 207500 + }, + { + "epoch": 2.84, + "learning_rate": 4.289141638528523e-05, + "loss": 1.6328, + "step": 208000 + }, + { + "epoch": 2.85, + "learning_rate": 4.287432844390371e-05, + "loss": 1.6684, + "step": 208500 + }, + { + "epoch": 2.86, + "learning_rate": 4.285724050252218e-05, + "loss": 1.6511, + "step": 209000 + }, + { + "epoch": 2.86, + "learning_rate": 4.284015256114066e-05, + "loss": 1.6195, + "step": 209500 + }, + { + "epoch": 2.87, + "learning_rate": 4.282306461975913e-05, + "loss": 1.6244, + "step": 210000 + }, + { + "epoch": 2.88, + "learning_rate": 4.2805976678377606e-05, + "loss": 1.6204, + "step": 210500 + }, + { + "epoch": 2.88, + "learning_rate": 4.278888873699608e-05, + "loss": 1.6365, + "step": 211000 + }, + { + "epoch": 2.89, + "learning_rate": 4.2771800795614554e-05, + "loss": 1.6723, + "step": 211500 + }, + { + "epoch": 2.9, + "learning_rate": 4.2754712854233024e-05, + "loss": 1.6274, + "step": 212000 + }, + { + "epoch": 2.9, + "learning_rate": 4.27376249128515e-05, + "loss": 1.6266, + "step": 212500 + }, + { + "epoch": 2.91, + "learning_rate": 4.272053697146998e-05, + "loss": 1.6364, + "step": 213000 + }, + { + "epoch": 2.92, + "learning_rate": 4.270344903008845e-05, + "loss": 1.6471, + "step": 213500 + }, + { + "epoch": 2.93, + "learning_rate": 4.268636108870692e-05, + "loss": 1.6439, + "step": 214000 + }, + { + "epoch": 2.93, + "learning_rate": 4.26692731473254e-05, + "loss": 1.6134, + "step": 214500 + }, + { + "epoch": 2.94, + "learning_rate": 4.2652185205943874e-05, + "loss": 1.6552, + "step": 215000 + }, + { + "epoch": 2.95, + "learning_rate": 4.2635097264562345e-05, + "loss": 1.6148, + "step": 215500 + }, + { + "epoch": 2.95, + "learning_rate": 4.2618009323180815e-05, + "loss": 1.6188, + "step": 216000 + }, + { + "epoch": 2.96, + "learning_rate": 4.260092138179929e-05, + "loss": 1.6389, + "step": 216500 + }, + { + "epoch": 2.97, + "learning_rate": 4.258383344041777e-05, + "loss": 1.6314, + "step": 217000 + }, + { + "epoch": 2.97, + "learning_rate": 4.256674549903624e-05, + "loss": 1.6391, + "step": 217500 + }, + { + "epoch": 2.98, + "learning_rate": 4.254965755765472e-05, + "loss": 1.6391, + "step": 218000 + }, + { + "epoch": 2.99, + "learning_rate": 4.253256961627319e-05, + "loss": 1.6566, + "step": 218500 + }, + { + "epoch": 2.99, + "learning_rate": 4.2515481674891666e-05, + "loss": 1.6127, + "step": 219000 + }, + { + "epoch": 3.0, + "learning_rate": 4.2498393733510136e-05, + "loss": 1.6265, + "step": 219500 + }, + { + "epoch": 3.01, + "learning_rate": 4.2481305792128613e-05, + "loss": 1.6037, + "step": 220000 + }, + { + "epoch": 3.01, + "learning_rate": 4.2464217850747084e-05, + "loss": 1.5997, + "step": 220500 + }, + { + "epoch": 3.02, + "learning_rate": 4.244712990936556e-05, + "loss": 1.6265, + "step": 221000 + }, + { + "epoch": 3.03, + "learning_rate": 4.243004196798404e-05, + "loss": 1.5955, + "step": 221500 + }, + { + "epoch": 3.03, + "learning_rate": 4.241295402660251e-05, + "loss": 1.5959, + "step": 222000 + }, + { + "epoch": 3.04, + "learning_rate": 4.239586608522098e-05, + "loss": 1.6474, + "step": 222500 + }, + { + "epoch": 3.05, + "learning_rate": 4.237877814383946e-05, + "loss": 1.6208, + "step": 223000 + }, + { + "epoch": 3.06, + "learning_rate": 4.2361690202457934e-05, + "loss": 1.611, + "step": 223500 + }, + { + "epoch": 3.06, + "learning_rate": 4.2344602261076405e-05, + "loss": 1.5959, + "step": 224000 + }, + { + "epoch": 3.07, + "learning_rate": 4.2327514319694875e-05, + "loss": 1.6089, + "step": 224500 + }, + { + "epoch": 3.08, + "learning_rate": 4.231042637831335e-05, + "loss": 1.6147, + "step": 225000 + }, + { + "epoch": 3.08, + "learning_rate": 4.229333843693183e-05, + "loss": 1.6128, + "step": 225500 + }, + { + "epoch": 3.09, + "learning_rate": 4.22762504955503e-05, + "loss": 1.6092, + "step": 226000 + }, + { + "epoch": 3.1, + "learning_rate": 4.225916255416877e-05, + "loss": 1.6027, + "step": 226500 + }, + { + "epoch": 3.1, + "learning_rate": 4.2242074612787255e-05, + "loss": 1.6262, + "step": 227000 + }, + { + "epoch": 3.11, + "learning_rate": 4.2224986671405726e-05, + "loss": 1.6483, + "step": 227500 + }, + { + "epoch": 3.12, + "learning_rate": 4.2207898730024196e-05, + "loss": 1.593, + "step": 228000 + }, + { + "epoch": 3.12, + "learning_rate": 4.2190810788642673e-05, + "loss": 1.5999, + "step": 228500 + }, + { + "epoch": 3.13, + "learning_rate": 4.217372284726115e-05, + "loss": 1.6586, + "step": 229000 + }, + { + "epoch": 3.14, + "learning_rate": 4.215663490587962e-05, + "loss": 1.5912, + "step": 229500 + }, + { + "epoch": 3.14, + "learning_rate": 4.213954696449809e-05, + "loss": 1.5827, + "step": 230000 + }, + { + "epoch": 3.15, + "learning_rate": 4.212245902311657e-05, + "loss": 1.6208, + "step": 230500 + }, + { + "epoch": 3.16, + "learning_rate": 4.2105371081735046e-05, + "loss": 1.6439, + "step": 231000 + }, + { + "epoch": 3.16, + "learning_rate": 4.208828314035352e-05, + "loss": 1.6056, + "step": 231500 + }, + { + "epoch": 3.17, + "learning_rate": 4.2071195198971994e-05, + "loss": 1.6364, + "step": 232000 + }, + { + "epoch": 3.18, + "learning_rate": 4.2054107257590465e-05, + "loss": 1.6276, + "step": 232500 + }, + { + "epoch": 3.19, + "learning_rate": 4.203701931620894e-05, + "loss": 1.6209, + "step": 233000 + }, + { + "epoch": 3.19, + "learning_rate": 4.201993137482741e-05, + "loss": 1.6183, + "step": 233500 + }, + { + "epoch": 3.2, + "learning_rate": 4.200284343344589e-05, + "loss": 1.63, + "step": 234000 + }, + { + "epoch": 3.21, + "learning_rate": 4.198575549206436e-05, + "loss": 1.6221, + "step": 234500 + }, + { + "epoch": 3.21, + "learning_rate": 4.196866755068284e-05, + "loss": 1.6147, + "step": 235000 + }, + { + "epoch": 3.22, + "learning_rate": 4.195157960930131e-05, + "loss": 1.6428, + "step": 235500 + }, + { + "epoch": 3.23, + "learning_rate": 4.1934491667919786e-05, + "loss": 1.611, + "step": 236000 + }, + { + "epoch": 3.23, + "learning_rate": 4.1917403726538256e-05, + "loss": 1.6306, + "step": 236500 + }, + { + "epoch": 3.24, + "learning_rate": 4.1900315785156733e-05, + "loss": 1.6389, + "step": 237000 + }, + { + "epoch": 3.25, + "learning_rate": 4.188322784377521e-05, + "loss": 1.6247, + "step": 237500 + }, + { + "epoch": 3.25, + "learning_rate": 4.186613990239368e-05, + "loss": 1.6132, + "step": 238000 + }, + { + "epoch": 3.26, + "learning_rate": 4.184905196101215e-05, + "loss": 1.593, + "step": 238500 + }, + { + "epoch": 3.27, + "learning_rate": 4.183196401963063e-05, + "loss": 1.6341, + "step": 239000 + }, + { + "epoch": 3.27, + "learning_rate": 4.1814876078249106e-05, + "loss": 1.6031, + "step": 239500 + }, + { + "epoch": 3.28, + "learning_rate": 4.179778813686758e-05, + "loss": 1.649, + "step": 240000 + }, + { + "epoch": 3.29, + "learning_rate": 4.178070019548605e-05, + "loss": 1.5947, + "step": 240500 + }, + { + "epoch": 3.29, + "learning_rate": 4.176361225410453e-05, + "loss": 1.5928, + "step": 241000 + }, + { + "epoch": 3.3, + "learning_rate": 4.1746524312723e-05, + "loss": 1.6091, + "step": 241500 + }, + { + "epoch": 3.31, + "learning_rate": 4.172943637134147e-05, + "loss": 1.6234, + "step": 242000 + }, + { + "epoch": 3.32, + "learning_rate": 4.171234842995995e-05, + "loss": 1.6106, + "step": 242500 + }, + { + "epoch": 3.32, + "learning_rate": 4.169526048857842e-05, + "loss": 1.6586, + "step": 243000 + }, + { + "epoch": 3.33, + "learning_rate": 4.16781725471969e-05, + "loss": 1.6672, + "step": 243500 + }, + { + "epoch": 3.34, + "learning_rate": 4.166108460581537e-05, + "loss": 1.6096, + "step": 244000 + }, + { + "epoch": 3.34, + "learning_rate": 4.1643996664433846e-05, + "loss": 1.5749, + "step": 244500 + }, + { + "epoch": 3.35, + "learning_rate": 4.1626908723052316e-05, + "loss": 1.6267, + "step": 245000 + }, + { + "epoch": 3.36, + "learning_rate": 4.160982078167079e-05, + "loss": 1.6081, + "step": 245500 + }, + { + "epoch": 3.36, + "learning_rate": 4.1592732840289264e-05, + "loss": 1.5886, + "step": 246000 + }, + { + "epoch": 3.37, + "learning_rate": 4.157564489890774e-05, + "loss": 1.61, + "step": 246500 + }, + { + "epoch": 3.38, + "learning_rate": 4.155855695752621e-05, + "loss": 1.6004, + "step": 247000 + }, + { + "epoch": 3.38, + "learning_rate": 4.154146901614469e-05, + "loss": 1.6207, + "step": 247500 + }, + { + "epoch": 3.39, + "learning_rate": 4.1524381074763166e-05, + "loss": 1.608, + "step": 248000 + }, + { + "epoch": 3.4, + "learning_rate": 4.150729313338164e-05, + "loss": 1.6069, + "step": 248500 + }, + { + "epoch": 3.4, + "learning_rate": 4.149020519200011e-05, + "loss": 1.6223, + "step": 249000 + }, + { + "epoch": 3.41, + "learning_rate": 4.1473117250618585e-05, + "loss": 1.6078, + "step": 249500 + }, + { + "epoch": 3.42, + "learning_rate": 4.145602930923706e-05, + "loss": 1.6075, + "step": 250000 + }, + { + "epoch": 3.42, + "learning_rate": 4.143894136785553e-05, + "loss": 1.6055, + "step": 250500 + }, + { + "epoch": 3.43, + "learning_rate": 4.1421853426474e-05, + "loss": 1.6174, + "step": 251000 + }, + { + "epoch": 3.44, + "learning_rate": 4.140476548509249e-05, + "loss": 1.6481, + "step": 251500 + }, + { + "epoch": 3.44, + "learning_rate": 4.138767754371096e-05, + "loss": 1.6233, + "step": 252000 + }, + { + "epoch": 3.45, + "learning_rate": 4.137058960232943e-05, + "loss": 1.6177, + "step": 252500 + }, + { + "epoch": 3.46, + "learning_rate": 4.13535016609479e-05, + "loss": 1.6349, + "step": 253000 + }, + { + "epoch": 3.47, + "learning_rate": 4.133641371956638e-05, + "loss": 1.6187, + "step": 253500 + }, + { + "epoch": 3.47, + "learning_rate": 4.131932577818485e-05, + "loss": 1.6479, + "step": 254000 + }, + { + "epoch": 3.48, + "learning_rate": 4.1302237836803324e-05, + "loss": 1.6293, + "step": 254500 + }, + { + "epoch": 3.49, + "learning_rate": 4.12851498954218e-05, + "loss": 1.6324, + "step": 255000 + }, + { + "epoch": 3.49, + "learning_rate": 4.126806195404028e-05, + "loss": 1.6175, + "step": 255500 + }, + { + "epoch": 3.5, + "learning_rate": 4.125097401265875e-05, + "loss": 1.5886, + "step": 256000 + }, + { + "epoch": 3.51, + "learning_rate": 4.123388607127722e-05, + "loss": 1.634, + "step": 256500 + }, + { + "epoch": 3.51, + "learning_rate": 4.12167981298957e-05, + "loss": 1.6426, + "step": 257000 + }, + { + "epoch": 3.52, + "learning_rate": 4.1199710188514174e-05, + "loss": 1.6297, + "step": 257500 + }, + { + "epoch": 3.53, + "learning_rate": 4.1182622247132645e-05, + "loss": 1.6096, + "step": 258000 + }, + { + "epoch": 3.53, + "learning_rate": 4.116553430575112e-05, + "loss": 1.6186, + "step": 258500 + }, + { + "epoch": 3.54, + "learning_rate": 4.114844636436959e-05, + "loss": 1.654, + "step": 259000 + }, + { + "epoch": 3.55, + "learning_rate": 4.113135842298807e-05, + "loss": 1.5966, + "step": 259500 + }, + { + "epoch": 3.55, + "learning_rate": 4.111427048160654e-05, + "loss": 1.63, + "step": 260000 + }, + { + "epoch": 3.56, + "learning_rate": 4.109718254022502e-05, + "loss": 1.5918, + "step": 260500 + }, + { + "epoch": 3.57, + "learning_rate": 4.108009459884349e-05, + "loss": 1.6217, + "step": 261000 + }, + { + "epoch": 3.57, + "learning_rate": 4.1063006657461965e-05, + "loss": 1.6231, + "step": 261500 + }, + { + "epoch": 3.58, + "learning_rate": 4.1045918716080436e-05, + "loss": 1.6416, + "step": 262000 + }, + { + "epoch": 3.59, + "learning_rate": 4.102883077469891e-05, + "loss": 1.6579, + "step": 262500 + }, + { + "epoch": 3.6, + "learning_rate": 4.1011742833317384e-05, + "loss": 1.6043, + "step": 263000 + }, + { + "epoch": 3.6, + "learning_rate": 4.099465489193586e-05, + "loss": 1.6305, + "step": 263500 + }, + { + "epoch": 3.61, + "learning_rate": 4.097756695055434e-05, + "loss": 1.6005, + "step": 264000 + }, + { + "epoch": 3.62, + "learning_rate": 4.096047900917281e-05, + "loss": 1.6023, + "step": 264500 + }, + { + "epoch": 3.62, + "learning_rate": 4.094339106779128e-05, + "loss": 1.6359, + "step": 265000 + }, + { + "epoch": 3.63, + "learning_rate": 4.092630312640976e-05, + "loss": 1.5947, + "step": 265500 + }, + { + "epoch": 3.64, + "learning_rate": 4.0909215185028234e-05, + "loss": 1.5842, + "step": 266000 + }, + { + "epoch": 3.64, + "learning_rate": 4.0892127243646705e-05, + "loss": 1.6164, + "step": 266500 + }, + { + "epoch": 3.65, + "learning_rate": 4.0875039302265175e-05, + "loss": 1.6292, + "step": 267000 + }, + { + "epoch": 3.66, + "learning_rate": 4.085795136088366e-05, + "loss": 1.6071, + "step": 267500 + }, + { + "epoch": 3.66, + "learning_rate": 4.084086341950213e-05, + "loss": 1.6022, + "step": 268000 + }, + { + "epoch": 3.67, + "learning_rate": 4.08237754781206e-05, + "loss": 1.6218, + "step": 268500 + }, + { + "epoch": 3.68, + "learning_rate": 4.080668753673908e-05, + "loss": 1.6306, + "step": 269000 + }, + { + "epoch": 3.68, + "learning_rate": 4.0789599595357555e-05, + "loss": 1.5959, + "step": 269500 + }, + { + "epoch": 3.69, + "learning_rate": 4.0772511653976025e-05, + "loss": 1.6175, + "step": 270000 + }, + { + "epoch": 3.7, + "learning_rate": 4.0755423712594496e-05, + "loss": 1.6211, + "step": 270500 + }, + { + "epoch": 3.7, + "learning_rate": 4.073833577121297e-05, + "loss": 1.6387, + "step": 271000 + }, + { + "epoch": 3.71, + "learning_rate": 4.072124782983145e-05, + "loss": 1.6074, + "step": 271500 + }, + { + "epoch": 3.72, + "learning_rate": 4.070415988844992e-05, + "loss": 1.6317, + "step": 272000 + }, + { + "epoch": 3.73, + "learning_rate": 4.068707194706839e-05, + "loss": 1.612, + "step": 272500 + }, + { + "epoch": 3.73, + "learning_rate": 4.066998400568687e-05, + "loss": 1.6207, + "step": 273000 + }, + { + "epoch": 3.74, + "learning_rate": 4.065289606430534e-05, + "loss": 1.6387, + "step": 273500 + }, + { + "epoch": 3.75, + "learning_rate": 4.063580812292382e-05, + "loss": 1.6036, + "step": 274000 + }, + { + "epoch": 3.75, + "learning_rate": 4.0618720181542294e-05, + "loss": 1.6404, + "step": 274500 + }, + { + "epoch": 3.76, + "learning_rate": 4.0601632240160765e-05, + "loss": 1.6153, + "step": 275000 + }, + { + "epoch": 3.77, + "learning_rate": 4.0584544298779235e-05, + "loss": 1.6175, + "step": 275500 + }, + { + "epoch": 3.77, + "learning_rate": 4.056745635739771e-05, + "loss": 1.6391, + "step": 276000 + }, + { + "epoch": 3.78, + "learning_rate": 4.055036841601619e-05, + "loss": 1.6055, + "step": 276500 + }, + { + "epoch": 3.79, + "learning_rate": 4.053328047463466e-05, + "loss": 1.6236, + "step": 277000 + }, + { + "epoch": 3.79, + "learning_rate": 4.051619253325313e-05, + "loss": 1.6041, + "step": 277500 + }, + { + "epoch": 3.8, + "learning_rate": 4.0499104591871615e-05, + "loss": 1.6166, + "step": 278000 + }, + { + "epoch": 3.81, + "learning_rate": 4.0482016650490085e-05, + "loss": 1.5995, + "step": 278500 + }, + { + "epoch": 3.81, + "learning_rate": 4.0464928709108556e-05, + "loss": 1.6138, + "step": 279000 + }, + { + "epoch": 3.82, + "learning_rate": 4.044784076772703e-05, + "loss": 1.6171, + "step": 279500 + }, + { + "epoch": 3.83, + "learning_rate": 4.043075282634551e-05, + "loss": 1.6145, + "step": 280000 + }, + { + "epoch": 3.83, + "learning_rate": 4.041366488496398e-05, + "loss": 1.6406, + "step": 280500 + }, + { + "epoch": 3.84, + "learning_rate": 4.039657694358245e-05, + "loss": 1.6121, + "step": 281000 + }, + { + "epoch": 3.85, + "learning_rate": 4.037948900220093e-05, + "loss": 1.5876, + "step": 281500 + }, + { + "epoch": 3.86, + "learning_rate": 4.0362401060819406e-05, + "loss": 1.5735, + "step": 282000 + }, + { + "epoch": 3.86, + "learning_rate": 4.034531311943788e-05, + "loss": 1.6097, + "step": 282500 + }, + { + "epoch": 3.87, + "learning_rate": 4.032822517805635e-05, + "loss": 1.6108, + "step": 283000 + }, + { + "epoch": 3.88, + "learning_rate": 4.0311137236674825e-05, + "loss": 1.5955, + "step": 283500 + }, + { + "epoch": 3.88, + "learning_rate": 4.02940492952933e-05, + "loss": 1.616, + "step": 284000 + }, + { + "epoch": 3.89, + "learning_rate": 4.027696135391177e-05, + "loss": 1.6267, + "step": 284500 + }, + { + "epoch": 3.9, + "learning_rate": 4.025987341253025e-05, + "loss": 1.6002, + "step": 285000 + }, + { + "epoch": 3.9, + "learning_rate": 4.024278547114872e-05, + "loss": 1.5917, + "step": 285500 + }, + { + "epoch": 3.91, + "learning_rate": 4.02256975297672e-05, + "loss": 1.5946, + "step": 286000 + }, + { + "epoch": 3.92, + "learning_rate": 4.020860958838567e-05, + "loss": 1.6007, + "step": 286500 + }, + { + "epoch": 3.92, + "learning_rate": 4.0191521647004145e-05, + "loss": 1.6225, + "step": 287000 + }, + { + "epoch": 3.93, + "learning_rate": 4.0174433705622616e-05, + "loss": 1.6242, + "step": 287500 + }, + { + "epoch": 3.94, + "learning_rate": 4.015734576424109e-05, + "loss": 1.5949, + "step": 288000 + }, + { + "epoch": 3.94, + "learning_rate": 4.014025782285957e-05, + "loss": 1.6002, + "step": 288500 + }, + { + "epoch": 3.95, + "learning_rate": 4.012316988147804e-05, + "loss": 1.5821, + "step": 289000 + }, + { + "epoch": 3.96, + "learning_rate": 4.010608194009651e-05, + "loss": 1.5979, + "step": 289500 + }, + { + "epoch": 3.96, + "learning_rate": 4.008899399871499e-05, + "loss": 1.6001, + "step": 290000 + }, + { + "epoch": 3.97, + "learning_rate": 4.0071906057333466e-05, + "loss": 1.6534, + "step": 290500 + }, + { + "epoch": 3.98, + "learning_rate": 4.005481811595194e-05, + "loss": 1.6129, + "step": 291000 + }, + { + "epoch": 3.98, + "learning_rate": 4.003773017457041e-05, + "loss": 1.6017, + "step": 291500 + }, + { + "epoch": 3.99, + "learning_rate": 4.0020642233188884e-05, + "loss": 1.6089, + "step": 292000 + }, + { + "epoch": 4.0, + "learning_rate": 4.000355429180736e-05, + "loss": 1.6214, + "step": 292500 + }, + { + "epoch": 4.01, + "learning_rate": 3.998646635042583e-05, + "loss": 1.618, + "step": 293000 + }, + { + "epoch": 4.01, + "learning_rate": 3.99693784090443e-05, + "loss": 1.579, + "step": 293500 + }, + { + "epoch": 4.02, + "learning_rate": 3.995229046766279e-05, + "loss": 1.6004, + "step": 294000 + }, + { + "epoch": 4.03, + "learning_rate": 3.993520252628126e-05, + "loss": 1.5648, + "step": 294500 + }, + { + "epoch": 4.03, + "learning_rate": 3.991811458489973e-05, + "loss": 1.5925, + "step": 295000 + }, + { + "epoch": 4.04, + "learning_rate": 3.9901026643518205e-05, + "loss": 1.6254, + "step": 295500 + }, + { + "epoch": 4.05, + "learning_rate": 3.988393870213668e-05, + "loss": 1.5743, + "step": 296000 + }, + { + "epoch": 4.05, + "learning_rate": 3.986685076075515e-05, + "loss": 1.5924, + "step": 296500 + }, + { + "epoch": 4.06, + "learning_rate": 3.9849762819373624e-05, + "loss": 1.5964, + "step": 297000 + }, + { + "epoch": 4.07, + "learning_rate": 3.98326748779921e-05, + "loss": 1.5947, + "step": 297500 + }, + { + "epoch": 4.07, + "learning_rate": 3.981558693661058e-05, + "loss": 1.6073, + "step": 298000 + }, + { + "epoch": 4.08, + "learning_rate": 3.979849899522905e-05, + "loss": 1.5981, + "step": 298500 + }, + { + "epoch": 4.09, + "learning_rate": 3.978141105384752e-05, + "loss": 1.6167, + "step": 299000 + }, + { + "epoch": 4.09, + "learning_rate": 3.9764323112465997e-05, + "loss": 1.5974, + "step": 299500 + }, + { + "epoch": 4.1, + "learning_rate": 3.9747235171084474e-05, + "loss": 1.5459, + "step": 300000 + }, + { + "epoch": 4.11, + "learning_rate": 3.9730147229702944e-05, + "loss": 1.613, + "step": 300500 + }, + { + "epoch": 4.11, + "learning_rate": 3.971305928832142e-05, + "loss": 1.5837, + "step": 301000 + }, + { + "epoch": 4.12, + "learning_rate": 3.969597134693989e-05, + "loss": 1.5969, + "step": 301500 + }, + { + "epoch": 4.13, + "learning_rate": 3.967888340555837e-05, + "loss": 1.5872, + "step": 302000 + }, + { + "epoch": 4.14, + "learning_rate": 3.966179546417684e-05, + "loss": 1.6263, + "step": 302500 + }, + { + "epoch": 4.14, + "learning_rate": 3.964470752279532e-05, + "loss": 1.5826, + "step": 303000 + }, + { + "epoch": 4.15, + "learning_rate": 3.962761958141379e-05, + "loss": 1.6152, + "step": 303500 + }, + { + "epoch": 4.16, + "learning_rate": 3.961053164003226e-05, + "loss": 1.6308, + "step": 304000 + }, + { + "epoch": 4.16, + "learning_rate": 3.959344369865074e-05, + "loss": 1.6129, + "step": 304500 + }, + { + "epoch": 4.17, + "learning_rate": 3.957635575726921e-05, + "loss": 1.6065, + "step": 305000 + }, + { + "epoch": 4.18, + "learning_rate": 3.9559267815887684e-05, + "loss": 1.5926, + "step": 305500 + }, + { + "epoch": 4.18, + "learning_rate": 3.954217987450616e-05, + "loss": 1.6035, + "step": 306000 + }, + { + "epoch": 4.19, + "learning_rate": 3.952509193312464e-05, + "loss": 1.5738, + "step": 306500 + }, + { + "epoch": 4.2, + "learning_rate": 3.950800399174311e-05, + "loss": 1.6096, + "step": 307000 + }, + { + "epoch": 4.2, + "learning_rate": 3.949091605036158e-05, + "loss": 1.596, + "step": 307500 + }, + { + "epoch": 4.21, + "learning_rate": 3.9473828108980057e-05, + "loss": 1.6036, + "step": 308000 + }, + { + "epoch": 4.22, + "learning_rate": 3.9456740167598534e-05, + "loss": 1.5874, + "step": 308500 + }, + { + "epoch": 4.22, + "learning_rate": 3.9439652226217004e-05, + "loss": 1.6451, + "step": 309000 + }, + { + "epoch": 4.23, + "learning_rate": 3.9422564284835475e-05, + "loss": 1.5666, + "step": 309500 + }, + { + "epoch": 4.24, + "learning_rate": 3.940547634345395e-05, + "loss": 1.6328, + "step": 310000 + }, + { + "epoch": 4.24, + "learning_rate": 3.938838840207243e-05, + "loss": 1.6183, + "step": 310500 + }, + { + "epoch": 4.25, + "learning_rate": 3.93713004606909e-05, + "loss": 1.5775, + "step": 311000 + }, + { + "epoch": 4.26, + "learning_rate": 3.935421251930938e-05, + "loss": 1.6084, + "step": 311500 + }, + { + "epoch": 4.27, + "learning_rate": 3.933712457792785e-05, + "loss": 1.6017, + "step": 312000 + }, + { + "epoch": 4.27, + "learning_rate": 3.9320036636546325e-05, + "loss": 1.6076, + "step": 312500 + }, + { + "epoch": 4.28, + "learning_rate": 3.9302948695164796e-05, + "loss": 1.5713, + "step": 313000 + }, + { + "epoch": 4.29, + "learning_rate": 3.928586075378327e-05, + "loss": 1.5692, + "step": 313500 + }, + { + "epoch": 4.29, + "learning_rate": 3.9268772812401744e-05, + "loss": 1.5944, + "step": 314000 + }, + { + "epoch": 4.3, + "learning_rate": 3.925168487102022e-05, + "loss": 1.5782, + "step": 314500 + }, + { + "epoch": 4.31, + "learning_rate": 3.92345969296387e-05, + "loss": 1.5874, + "step": 315000 + }, + { + "epoch": 4.31, + "learning_rate": 3.921750898825717e-05, + "loss": 1.6355, + "step": 315500 + }, + { + "epoch": 4.32, + "learning_rate": 3.920042104687564e-05, + "loss": 1.5743, + "step": 316000 + }, + { + "epoch": 4.33, + "learning_rate": 3.9183333105494117e-05, + "loss": 1.5881, + "step": 316500 + }, + { + "epoch": 4.33, + "learning_rate": 3.9166245164112594e-05, + "loss": 1.5771, + "step": 317000 + }, + { + "epoch": 4.34, + "learning_rate": 3.9149157222731064e-05, + "loss": 1.5818, + "step": 317500 + }, + { + "epoch": 4.35, + "learning_rate": 3.9132069281349535e-05, + "loss": 1.5915, + "step": 318000 + }, + { + "epoch": 4.35, + "learning_rate": 3.911498133996801e-05, + "loss": 1.5754, + "step": 318500 + }, + { + "epoch": 4.36, + "learning_rate": 3.909789339858649e-05, + "loss": 1.6006, + "step": 319000 + }, + { + "epoch": 4.37, + "learning_rate": 3.908080545720496e-05, + "loss": 1.5969, + "step": 319500 + }, + { + "epoch": 4.37, + "learning_rate": 3.906371751582343e-05, + "loss": 1.5704, + "step": 320000 + }, + { + "epoch": 4.38, + "learning_rate": 3.9046629574441915e-05, + "loss": 1.5763, + "step": 320500 + }, + { + "epoch": 4.39, + "learning_rate": 3.9029541633060385e-05, + "loss": 1.619, + "step": 321000 + }, + { + "epoch": 4.4, + "learning_rate": 3.9012453691678856e-05, + "loss": 1.5908, + "step": 321500 + }, + { + "epoch": 4.4, + "learning_rate": 3.899536575029733e-05, + "loss": 1.6183, + "step": 322000 + }, + { + "epoch": 4.41, + "learning_rate": 3.897827780891581e-05, + "loss": 1.5845, + "step": 322500 + }, + { + "epoch": 4.42, + "learning_rate": 3.896118986753428e-05, + "loss": 1.6037, + "step": 323000 + }, + { + "epoch": 4.42, + "learning_rate": 3.894410192615275e-05, + "loss": 1.5983, + "step": 323500 + }, + { + "epoch": 4.43, + "learning_rate": 3.892701398477123e-05, + "loss": 1.6113, + "step": 324000 + }, + { + "epoch": 4.44, + "learning_rate": 3.8909926043389706e-05, + "loss": 1.6198, + "step": 324500 + }, + { + "epoch": 4.44, + "learning_rate": 3.8892838102008176e-05, + "loss": 1.6155, + "step": 325000 + }, + { + "epoch": 4.45, + "learning_rate": 3.8875750160626654e-05, + "loss": 1.5569, + "step": 325500 + }, + { + "epoch": 4.46, + "learning_rate": 3.8858662219245124e-05, + "loss": 1.5904, + "step": 326000 + }, + { + "epoch": 4.46, + "learning_rate": 3.88415742778636e-05, + "loss": 1.6035, + "step": 326500 + }, + { + "epoch": 4.47, + "learning_rate": 3.882448633648207e-05, + "loss": 1.5855, + "step": 327000 + }, + { + "epoch": 4.48, + "learning_rate": 3.880739839510055e-05, + "loss": 1.5965, + "step": 327500 + }, + { + "epoch": 4.48, + "learning_rate": 3.879031045371902e-05, + "loss": 1.6115, + "step": 328000 + }, + { + "epoch": 4.49, + "learning_rate": 3.87732225123375e-05, + "loss": 1.5711, + "step": 328500 + }, + { + "epoch": 4.5, + "learning_rate": 3.875613457095597e-05, + "loss": 1.6042, + "step": 329000 + }, + { + "epoch": 4.5, + "learning_rate": 3.8739046629574445e-05, + "loss": 1.619, + "step": 329500 + }, + { + "epoch": 4.51, + "learning_rate": 3.8721958688192916e-05, + "loss": 1.5871, + "step": 330000 + }, + { + "epoch": 4.52, + "learning_rate": 3.870487074681139e-05, + "loss": 1.5885, + "step": 330500 + }, + { + "epoch": 4.52, + "learning_rate": 3.868778280542987e-05, + "loss": 1.6278, + "step": 331000 + }, + { + "epoch": 4.53, + "learning_rate": 3.867069486404834e-05, + "loss": 1.619, + "step": 331500 + }, + { + "epoch": 4.54, + "learning_rate": 3.865360692266681e-05, + "loss": 1.5891, + "step": 332000 + }, + { + "epoch": 4.55, + "learning_rate": 3.863651898128529e-05, + "loss": 1.6248, + "step": 332500 + }, + { + "epoch": 4.55, + "learning_rate": 3.8619431039903766e-05, + "loss": 1.6307, + "step": 333000 + }, + { + "epoch": 4.56, + "learning_rate": 3.8602343098522236e-05, + "loss": 1.589, + "step": 333500 + }, + { + "epoch": 4.57, + "learning_rate": 3.858525515714071e-05, + "loss": 1.5573, + "step": 334000 + }, + { + "epoch": 4.57, + "learning_rate": 3.8568167215759184e-05, + "loss": 1.5783, + "step": 334500 + }, + { + "epoch": 4.58, + "learning_rate": 3.855107927437766e-05, + "loss": 1.5849, + "step": 335000 + }, + { + "epoch": 4.59, + "learning_rate": 3.853399133299613e-05, + "loss": 1.5743, + "step": 335500 + }, + { + "epoch": 4.59, + "learning_rate": 3.851690339161461e-05, + "loss": 1.5911, + "step": 336000 + }, + { + "epoch": 4.6, + "learning_rate": 3.849981545023308e-05, + "loss": 1.5687, + "step": 336500 + }, + { + "epoch": 4.61, + "learning_rate": 3.848272750885156e-05, + "loss": 1.5717, + "step": 337000 + }, + { + "epoch": 4.61, + "learning_rate": 3.846563956747003e-05, + "loss": 1.6138, + "step": 337500 + }, + { + "epoch": 4.62, + "learning_rate": 3.8448551626088505e-05, + "loss": 1.6171, + "step": 338000 + }, + { + "epoch": 4.63, + "learning_rate": 3.8431463684706976e-05, + "loss": 1.6187, + "step": 338500 + }, + { + "epoch": 4.63, + "learning_rate": 3.841437574332545e-05, + "loss": 1.6153, + "step": 339000 + }, + { + "epoch": 4.64, + "learning_rate": 3.8397287801943923e-05, + "loss": 1.6024, + "step": 339500 + }, + { + "epoch": 4.65, + "learning_rate": 3.83801998605624e-05, + "loss": 1.599, + "step": 340000 + }, + { + "epoch": 4.65, + "learning_rate": 3.836311191918087e-05, + "loss": 1.6227, + "step": 340500 + }, + { + "epoch": 4.66, + "learning_rate": 3.834602397779935e-05, + "loss": 1.5773, + "step": 341000 + }, + { + "epoch": 4.67, + "learning_rate": 3.8328936036417826e-05, + "loss": 1.6151, + "step": 341500 + }, + { + "epoch": 4.68, + "learning_rate": 3.8311848095036296e-05, + "loss": 1.6207, + "step": 342000 + }, + { + "epoch": 4.68, + "learning_rate": 3.829476015365477e-05, + "loss": 1.5843, + "step": 342500 + }, + { + "epoch": 4.69, + "learning_rate": 3.8277672212273244e-05, + "loss": 1.5973, + "step": 343000 + }, + { + "epoch": 4.7, + "learning_rate": 3.826058427089172e-05, + "loss": 1.5805, + "step": 343500 + }, + { + "epoch": 4.7, + "learning_rate": 3.824349632951019e-05, + "loss": 1.5929, + "step": 344000 + }, + { + "epoch": 4.71, + "learning_rate": 3.822640838812866e-05, + "loss": 1.5916, + "step": 344500 + }, + { + "epoch": 4.72, + "learning_rate": 3.820932044674715e-05, + "loss": 1.595, + "step": 345000 + }, + { + "epoch": 4.72, + "learning_rate": 3.819223250536562e-05, + "loss": 1.6017, + "step": 345500 + }, + { + "epoch": 4.73, + "learning_rate": 3.817514456398409e-05, + "loss": 1.5872, + "step": 346000 + }, + { + "epoch": 4.74, + "learning_rate": 3.815805662260256e-05, + "loss": 1.6056, + "step": 346500 + }, + { + "epoch": 4.74, + "learning_rate": 3.814096868122104e-05, + "loss": 1.5708, + "step": 347000 + }, + { + "epoch": 4.75, + "learning_rate": 3.812388073983951e-05, + "loss": 1.5636, + "step": 347500 + }, + { + "epoch": 4.76, + "learning_rate": 3.810679279845798e-05, + "loss": 1.6, + "step": 348000 + }, + { + "epoch": 4.76, + "learning_rate": 3.808970485707646e-05, + "loss": 1.5867, + "step": 348500 + }, + { + "epoch": 4.77, + "learning_rate": 3.807261691569494e-05, + "loss": 1.5958, + "step": 349000 + }, + { + "epoch": 4.78, + "learning_rate": 3.805552897431341e-05, + "loss": 1.5755, + "step": 349500 + }, + { + "epoch": 4.78, + "learning_rate": 3.803844103293188e-05, + "loss": 1.5944, + "step": 350000 + }, + { + "epoch": 4.79, + "learning_rate": 3.8021353091550356e-05, + "loss": 1.5997, + "step": 350500 + }, + { + "epoch": 4.8, + "learning_rate": 3.8004265150168834e-05, + "loss": 1.6124, + "step": 351000 + }, + { + "epoch": 4.81, + "learning_rate": 3.7987177208787304e-05, + "loss": 1.5706, + "step": 351500 + }, + { + "epoch": 4.81, + "learning_rate": 3.797008926740578e-05, + "loss": 1.592, + "step": 352000 + }, + { + "epoch": 4.82, + "learning_rate": 3.795300132602425e-05, + "loss": 1.5865, + "step": 352500 + }, + { + "epoch": 4.83, + "learning_rate": 3.793591338464273e-05, + "loss": 1.6096, + "step": 353000 + }, + { + "epoch": 4.83, + "learning_rate": 3.79188254432612e-05, + "loss": 1.5948, + "step": 353500 + }, + { + "epoch": 4.84, + "learning_rate": 3.790173750187968e-05, + "loss": 1.6141, + "step": 354000 + }, + { + "epoch": 4.85, + "learning_rate": 3.788464956049815e-05, + "loss": 1.5993, + "step": 354500 + }, + { + "epoch": 4.85, + "learning_rate": 3.7867561619116625e-05, + "loss": 1.6195, + "step": 355000 + }, + { + "epoch": 4.86, + "learning_rate": 3.7850473677735096e-05, + "loss": 1.5853, + "step": 355500 + }, + { + "epoch": 4.87, + "learning_rate": 3.783338573635357e-05, + "loss": 1.6184, + "step": 356000 + }, + { + "epoch": 4.87, + "learning_rate": 3.781629779497204e-05, + "loss": 1.5648, + "step": 356500 + }, + { + "epoch": 4.88, + "learning_rate": 3.779920985359052e-05, + "loss": 1.6053, + "step": 357000 + }, + { + "epoch": 4.89, + "learning_rate": 3.7782121912209e-05, + "loss": 1.5906, + "step": 357500 + }, + { + "epoch": 4.89, + "learning_rate": 3.776503397082747e-05, + "loss": 1.5872, + "step": 358000 + }, + { + "epoch": 4.9, + "learning_rate": 3.774794602944594e-05, + "loss": 1.5689, + "step": 358500 + }, + { + "epoch": 4.91, + "learning_rate": 3.7730858088064416e-05, + "loss": 1.5887, + "step": 359000 + }, + { + "epoch": 4.91, + "learning_rate": 3.7713770146682894e-05, + "loss": 1.5918, + "step": 359500 + }, + { + "epoch": 4.92, + "learning_rate": 3.7696682205301364e-05, + "loss": 1.5705, + "step": 360000 + }, + { + "epoch": 4.93, + "learning_rate": 3.7679594263919835e-05, + "loss": 1.6155, + "step": 360500 + }, + { + "epoch": 4.93, + "learning_rate": 3.766250632253832e-05, + "loss": 1.5963, + "step": 361000 + }, + { + "epoch": 4.94, + "learning_rate": 3.764541838115679e-05, + "loss": 1.5871, + "step": 361500 + }, + { + "epoch": 4.95, + "learning_rate": 3.762833043977526e-05, + "loss": 1.5833, + "step": 362000 + }, + { + "epoch": 4.96, + "learning_rate": 3.761124249839374e-05, + "loss": 1.5777, + "step": 362500 + }, + { + "epoch": 4.96, + "learning_rate": 3.759415455701221e-05, + "loss": 1.5941, + "step": 363000 + }, + { + "epoch": 4.97, + "learning_rate": 3.7577066615630685e-05, + "loss": 1.5958, + "step": 363500 + }, + { + "epoch": 4.98, + "learning_rate": 3.7559978674249155e-05, + "loss": 1.5917, + "step": 364000 + }, + { + "epoch": 4.98, + "learning_rate": 3.754289073286763e-05, + "loss": 1.587, + "step": 364500 + }, + { + "epoch": 4.99, + "learning_rate": 3.75258027914861e-05, + "loss": 1.613, + "step": 365000 + }, + { + "epoch": 5.0, + "learning_rate": 3.750871485010458e-05, + "loss": 1.5923, + "step": 365500 + }, + { + "epoch": 5.0, + "learning_rate": 3.749162690872305e-05, + "loss": 1.5995, + "step": 366000 + }, + { + "epoch": 5.01, + "learning_rate": 3.747453896734153e-05, + "loss": 1.5548, + "step": 366500 + }, + { + "epoch": 5.02, + "learning_rate": 3.745745102596e-05, + "loss": 1.5625, + "step": 367000 + }, + { + "epoch": 5.02, + "learning_rate": 3.7440363084578476e-05, + "loss": 1.568, + "step": 367500 + }, + { + "epoch": 5.03, + "learning_rate": 3.7423275143196954e-05, + "loss": 1.5625, + "step": 368000 + }, + { + "epoch": 5.04, + "learning_rate": 3.7406187201815424e-05, + "loss": 1.5883, + "step": 368500 + }, + { + "epoch": 5.04, + "learning_rate": 3.7389099260433895e-05, + "loss": 1.5676, + "step": 369000 + }, + { + "epoch": 5.05, + "learning_rate": 3.737201131905237e-05, + "loss": 1.591, + "step": 369500 + }, + { + "epoch": 5.06, + "learning_rate": 3.735492337767085e-05, + "loss": 1.586, + "step": 370000 + }, + { + "epoch": 5.06, + "learning_rate": 3.733783543628932e-05, + "loss": 1.5899, + "step": 370500 + }, + { + "epoch": 5.07, + "learning_rate": 3.732074749490779e-05, + "loss": 1.5998, + "step": 371000 + }, + { + "epoch": 5.08, + "learning_rate": 3.7303659553526274e-05, + "loss": 1.6132, + "step": 371500 + }, + { + "epoch": 5.09, + "learning_rate": 3.7286571612144745e-05, + "loss": 1.552, + "step": 372000 + }, + { + "epoch": 5.09, + "learning_rate": 3.7269483670763215e-05, + "loss": 1.5731, + "step": 372500 + }, + { + "epoch": 5.1, + "learning_rate": 3.725239572938169e-05, + "loss": 1.5758, + "step": 373000 + }, + { + "epoch": 5.11, + "learning_rate": 3.723530778800017e-05, + "loss": 1.556, + "step": 373500 + }, + { + "epoch": 5.11, + "learning_rate": 3.721821984661864e-05, + "loss": 1.5617, + "step": 374000 + }, + { + "epoch": 5.12, + "learning_rate": 3.720113190523711e-05, + "loss": 1.5551, + "step": 374500 + }, + { + "epoch": 5.13, + "learning_rate": 3.718404396385559e-05, + "loss": 1.5855, + "step": 375000 + }, + { + "epoch": 5.13, + "learning_rate": 3.7166956022474066e-05, + "loss": 1.5725, + "step": 375500 + }, + { + "epoch": 5.14, + "learning_rate": 3.7149868081092536e-05, + "loss": 1.5615, + "step": 376000 + }, + { + "epoch": 5.15, + "learning_rate": 3.713278013971101e-05, + "loss": 1.6011, + "step": 376500 + }, + { + "epoch": 5.15, + "learning_rate": 3.7115692198329484e-05, + "loss": 1.5749, + "step": 377000 + }, + { + "epoch": 5.16, + "learning_rate": 3.709860425694796e-05, + "loss": 1.5851, + "step": 377500 + }, + { + "epoch": 5.17, + "learning_rate": 3.708151631556643e-05, + "loss": 1.5903, + "step": 378000 + }, + { + "epoch": 5.17, + "learning_rate": 3.706442837418491e-05, + "loss": 1.5879, + "step": 378500 + }, + { + "epoch": 5.18, + "learning_rate": 3.704734043280338e-05, + "loss": 1.5684, + "step": 379000 + }, + { + "epoch": 5.19, + "learning_rate": 3.703025249142186e-05, + "loss": 1.5872, + "step": 379500 + }, + { + "epoch": 5.19, + "learning_rate": 3.701316455004033e-05, + "loss": 1.5778, + "step": 380000 + }, + { + "epoch": 5.2, + "learning_rate": 3.6996076608658805e-05, + "loss": 1.5595, + "step": 380500 + }, + { + "epoch": 5.21, + "learning_rate": 3.6978988667277275e-05, + "loss": 1.5883, + "step": 381000 + }, + { + "epoch": 5.22, + "learning_rate": 3.696190072589575e-05, + "loss": 1.5772, + "step": 381500 + }, + { + "epoch": 5.22, + "learning_rate": 3.694481278451423e-05, + "loss": 1.5617, + "step": 382000 + }, + { + "epoch": 5.23, + "learning_rate": 3.69277248431327e-05, + "loss": 1.5691, + "step": 382500 + }, + { + "epoch": 5.24, + "learning_rate": 3.691063690175117e-05, + "loss": 1.5522, + "step": 383000 + }, + { + "epoch": 5.24, + "learning_rate": 3.689354896036965e-05, + "loss": 1.5555, + "step": 383500 + }, + { + "epoch": 5.25, + "learning_rate": 3.6876461018988126e-05, + "loss": 1.5727, + "step": 384000 + }, + { + "epoch": 5.26, + "learning_rate": 3.6859373077606596e-05, + "loss": 1.5299, + "step": 384500 + }, + { + "epoch": 5.26, + "learning_rate": 3.684228513622507e-05, + "loss": 1.581, + "step": 385000 + }, + { + "epoch": 5.27, + "learning_rate": 3.6825197194843544e-05, + "loss": 1.5619, + "step": 385500 + }, + { + "epoch": 5.28, + "learning_rate": 3.680810925346202e-05, + "loss": 1.5872, + "step": 386000 + }, + { + "epoch": 5.28, + "learning_rate": 3.679102131208049e-05, + "loss": 1.5671, + "step": 386500 + }, + { + "epoch": 5.29, + "learning_rate": 3.677393337069896e-05, + "loss": 1.5693, + "step": 387000 + }, + { + "epoch": 5.3, + "learning_rate": 3.6756845429317446e-05, + "loss": 1.5209, + "step": 387500 + }, + { + "epoch": 5.3, + "learning_rate": 3.673975748793592e-05, + "loss": 1.619, + "step": 388000 + }, + { + "epoch": 5.31, + "learning_rate": 3.672266954655439e-05, + "loss": 1.5472, + "step": 388500 + }, + { + "epoch": 5.32, + "learning_rate": 3.6705581605172865e-05, + "loss": 1.581, + "step": 389000 + }, + { + "epoch": 5.32, + "learning_rate": 3.668849366379134e-05, + "loss": 1.5855, + "step": 389500 + }, + { + "epoch": 5.33, + "learning_rate": 3.667140572240981e-05, + "loss": 1.5615, + "step": 390000 + }, + { + "epoch": 5.34, + "learning_rate": 3.665431778102828e-05, + "loss": 1.5749, + "step": 390500 + }, + { + "epoch": 5.35, + "learning_rate": 3.663722983964676e-05, + "loss": 1.5529, + "step": 391000 + }, + { + "epoch": 5.35, + "learning_rate": 3.662014189826524e-05, + "loss": 1.5822, + "step": 391500 + }, + { + "epoch": 5.36, + "learning_rate": 3.660305395688371e-05, + "loss": 1.5501, + "step": 392000 + }, + { + "epoch": 5.37, + "learning_rate": 3.658596601550218e-05, + "loss": 1.5599, + "step": 392500 + }, + { + "epoch": 5.37, + "learning_rate": 3.6568878074120656e-05, + "loss": 1.545, + "step": 393000 + }, + { + "epoch": 5.38, + "learning_rate": 3.655179013273913e-05, + "loss": 1.5657, + "step": 393500 + }, + { + "epoch": 5.39, + "learning_rate": 3.6534702191357604e-05, + "loss": 1.5867, + "step": 394000 + }, + { + "epoch": 5.39, + "learning_rate": 3.651761424997608e-05, + "loss": 1.536, + "step": 394500 + }, + { + "epoch": 5.4, + "learning_rate": 3.650052630859455e-05, + "loss": 1.5577, + "step": 395000 + }, + { + "epoch": 5.41, + "learning_rate": 3.648343836721302e-05, + "loss": 1.6016, + "step": 395500 + }, + { + "epoch": 5.41, + "learning_rate": 3.64663504258315e-05, + "loss": 1.5597, + "step": 396000 + }, + { + "epoch": 5.42, + "learning_rate": 3.644926248444998e-05, + "loss": 1.5452, + "step": 396500 + }, + { + "epoch": 5.43, + "learning_rate": 3.643217454306845e-05, + "loss": 1.5789, + "step": 397000 + }, + { + "epoch": 5.43, + "learning_rate": 3.641508660168692e-05, + "loss": 1.6404, + "step": 397500 + }, + { + "epoch": 5.44, + "learning_rate": 3.63979986603054e-05, + "loss": 1.5734, + "step": 398000 + }, + { + "epoch": 5.45, + "learning_rate": 3.638091071892387e-05, + "loss": 1.5561, + "step": 398500 + }, + { + "epoch": 5.45, + "learning_rate": 3.636382277754234e-05, + "loss": 1.5831, + "step": 399000 + }, + { + "epoch": 5.46, + "learning_rate": 3.634673483616082e-05, + "loss": 1.5535, + "step": 399500 + }, + { + "epoch": 5.47, + "learning_rate": 3.63296468947793e-05, + "loss": 1.5823, + "step": 400000 + }, + { + "epoch": 5.47, + "learning_rate": 3.631255895339777e-05, + "loss": 1.5717, + "step": 400500 + }, + { + "epoch": 5.48, + "learning_rate": 3.629547101201624e-05, + "loss": 1.5421, + "step": 401000 + }, + { + "epoch": 5.49, + "learning_rate": 3.6278383070634716e-05, + "loss": 1.5571, + "step": 401500 + }, + { + "epoch": 5.5, + "learning_rate": 3.6261295129253193e-05, + "loss": 1.5518, + "step": 402000 + }, + { + "epoch": 5.5, + "learning_rate": 3.6244207187871664e-05, + "loss": 1.5777, + "step": 402500 + }, + { + "epoch": 5.51, + "learning_rate": 3.6227119246490134e-05, + "loss": 1.5573, + "step": 403000 + }, + { + "epoch": 5.52, + "learning_rate": 3.621003130510861e-05, + "loss": 1.5685, + "step": 403500 + }, + { + "epoch": 5.52, + "learning_rate": 3.619294336372709e-05, + "loss": 1.5923, + "step": 404000 + }, + { + "epoch": 5.53, + "learning_rate": 3.617585542234556e-05, + "loss": 1.5958, + "step": 404500 + }, + { + "epoch": 5.54, + "learning_rate": 3.615876748096404e-05, + "loss": 1.5845, + "step": 405000 + }, + { + "epoch": 5.54, + "learning_rate": 3.614167953958251e-05, + "loss": 1.5464, + "step": 405500 + }, + { + "epoch": 5.55, + "learning_rate": 3.6124591598200985e-05, + "loss": 1.5708, + "step": 406000 + }, + { + "epoch": 5.56, + "learning_rate": 3.6107503656819455e-05, + "loss": 1.6005, + "step": 406500 + }, + { + "epoch": 5.56, + "learning_rate": 3.609041571543793e-05, + "loss": 1.557, + "step": 407000 + }, + { + "epoch": 5.57, + "learning_rate": 3.60733277740564e-05, + "loss": 1.5721, + "step": 407500 + }, + { + "epoch": 5.58, + "learning_rate": 3.605623983267488e-05, + "loss": 1.5633, + "step": 408000 + }, + { + "epoch": 5.58, + "learning_rate": 3.603915189129336e-05, + "loss": 1.5784, + "step": 408500 + }, + { + "epoch": 5.59, + "learning_rate": 3.602206394991183e-05, + "loss": 1.5724, + "step": 409000 + }, + { + "epoch": 5.6, + "learning_rate": 3.60049760085303e-05, + "loss": 1.5809, + "step": 409500 + }, + { + "epoch": 5.6, + "learning_rate": 3.5987888067148776e-05, + "loss": 1.5684, + "step": 410000 + }, + { + "epoch": 5.61, + "learning_rate": 3.597080012576725e-05, + "loss": 1.5858, + "step": 410500 + }, + { + "epoch": 5.62, + "learning_rate": 3.5953712184385724e-05, + "loss": 1.5441, + "step": 411000 + }, + { + "epoch": 5.63, + "learning_rate": 3.5936624243004194e-05, + "loss": 1.5682, + "step": 411500 + }, + { + "epoch": 5.63, + "learning_rate": 3.591953630162267e-05, + "loss": 1.57, + "step": 412000 + }, + { + "epoch": 5.64, + "learning_rate": 3.590244836024115e-05, + "loss": 1.5787, + "step": 412500 + }, + { + "epoch": 5.65, + "learning_rate": 3.588536041885962e-05, + "loss": 1.5641, + "step": 413000 + }, + { + "epoch": 5.65, + "learning_rate": 3.586827247747809e-05, + "loss": 1.5808, + "step": 413500 + }, + { + "epoch": 5.66, + "learning_rate": 3.5851184536096574e-05, + "loss": 1.5509, + "step": 414000 + }, + { + "epoch": 5.67, + "learning_rate": 3.5834096594715045e-05, + "loss": 1.5994, + "step": 414500 + }, + { + "epoch": 5.67, + "learning_rate": 3.5817008653333515e-05, + "loss": 1.5797, + "step": 415000 + }, + { + "epoch": 5.68, + "learning_rate": 3.579992071195199e-05, + "loss": 1.5722, + "step": 415500 + }, + { + "epoch": 5.69, + "learning_rate": 3.578283277057047e-05, + "loss": 1.5512, + "step": 416000 + }, + { + "epoch": 5.69, + "learning_rate": 3.576574482918894e-05, + "loss": 1.5743, + "step": 416500 + }, + { + "epoch": 5.7, + "learning_rate": 3.574865688780741e-05, + "loss": 1.5466, + "step": 417000 + }, + { + "epoch": 5.71, + "learning_rate": 3.573156894642589e-05, + "loss": 1.553, + "step": 417500 + }, + { + "epoch": 5.71, + "learning_rate": 3.5714481005044365e-05, + "loss": 1.6152, + "step": 418000 + }, + { + "epoch": 5.72, + "learning_rate": 3.5697393063662836e-05, + "loss": 1.5472, + "step": 418500 + }, + { + "epoch": 5.73, + "learning_rate": 3.568030512228131e-05, + "loss": 1.5732, + "step": 419000 + }, + { + "epoch": 5.73, + "learning_rate": 3.5663217180899784e-05, + "loss": 1.583, + "step": 419500 + }, + { + "epoch": 5.74, + "learning_rate": 3.564612923951826e-05, + "loss": 1.5569, + "step": 420000 + }, + { + "epoch": 5.75, + "learning_rate": 3.562904129813673e-05, + "loss": 1.5936, + "step": 420500 + }, + { + "epoch": 5.76, + "learning_rate": 3.561195335675521e-05, + "loss": 1.5827, + "step": 421000 + }, + { + "epoch": 5.76, + "learning_rate": 3.559486541537368e-05, + "loss": 1.5717, + "step": 421500 + }, + { + "epoch": 5.77, + "learning_rate": 3.557777747399216e-05, + "loss": 1.5642, + "step": 422000 + }, + { + "epoch": 5.78, + "learning_rate": 3.556068953261063e-05, + "loss": 1.5645, + "step": 422500 + }, + { + "epoch": 5.78, + "learning_rate": 3.5543601591229105e-05, + "loss": 1.5589, + "step": 423000 + }, + { + "epoch": 5.79, + "learning_rate": 3.5526513649847575e-05, + "loss": 1.5533, + "step": 423500 + }, + { + "epoch": 5.8, + "learning_rate": 3.5509425708466046e-05, + "loss": 1.5399, + "step": 424000 + }, + { + "epoch": 5.8, + "learning_rate": 3.549233776708453e-05, + "loss": 1.5901, + "step": 424500 + }, + { + "epoch": 5.81, + "learning_rate": 3.5475249825703e-05, + "loss": 1.5619, + "step": 425000 + }, + { + "epoch": 5.82, + "learning_rate": 3.545816188432147e-05, + "loss": 1.5339, + "step": 425500 + }, + { + "epoch": 5.82, + "learning_rate": 3.544107394293995e-05, + "loss": 1.6218, + "step": 426000 + }, + { + "epoch": 5.83, + "learning_rate": 3.5423986001558425e-05, + "loss": 1.582, + "step": 426500 + }, + { + "epoch": 5.84, + "learning_rate": 3.5406898060176896e-05, + "loss": 1.5562, + "step": 427000 + }, + { + "epoch": 5.84, + "learning_rate": 3.5389810118795367e-05, + "loss": 1.5593, + "step": 427500 + }, + { + "epoch": 5.85, + "learning_rate": 3.5372722177413844e-05, + "loss": 1.5533, + "step": 428000 + }, + { + "epoch": 5.86, + "learning_rate": 3.535563423603232e-05, + "loss": 1.5567, + "step": 428500 + }, + { + "epoch": 5.86, + "learning_rate": 3.533854629465079e-05, + "loss": 1.5493, + "step": 429000 + }, + { + "epoch": 5.87, + "learning_rate": 3.532145835326926e-05, + "loss": 1.565, + "step": 429500 + }, + { + "epoch": 5.88, + "learning_rate": 3.530437041188774e-05, + "loss": 1.5725, + "step": 430000 + }, + { + "epoch": 5.89, + "learning_rate": 3.528728247050622e-05, + "loss": 1.5581, + "step": 430500 + }, + { + "epoch": 5.89, + "learning_rate": 3.527019452912469e-05, + "loss": 1.5693, + "step": 431000 + }, + { + "epoch": 5.9, + "learning_rate": 3.5253106587743165e-05, + "loss": 1.5644, + "step": 431500 + }, + { + "epoch": 5.91, + "learning_rate": 3.5236018646361635e-05, + "loss": 1.5665, + "step": 432000 + }, + { + "epoch": 5.91, + "learning_rate": 3.521893070498011e-05, + "loss": 1.5734, + "step": 432500 + }, + { + "epoch": 5.92, + "learning_rate": 3.520184276359858e-05, + "loss": 1.5784, + "step": 433000 + }, + { + "epoch": 5.93, + "learning_rate": 3.518475482221706e-05, + "loss": 1.5715, + "step": 433500 + }, + { + "epoch": 5.93, + "learning_rate": 3.516766688083553e-05, + "loss": 1.5456, + "step": 434000 + }, + { + "epoch": 5.94, + "learning_rate": 3.515057893945401e-05, + "loss": 1.5365, + "step": 434500 + }, + { + "epoch": 5.95, + "learning_rate": 3.5133490998072485e-05, + "loss": 1.5648, + "step": 435000 + }, + { + "epoch": 5.95, + "learning_rate": 3.5116403056690956e-05, + "loss": 1.5804, + "step": 435500 + }, + { + "epoch": 5.96, + "learning_rate": 3.5099315115309426e-05, + "loss": 1.566, + "step": 436000 + }, + { + "epoch": 5.97, + "learning_rate": 3.5082227173927904e-05, + "loss": 1.5702, + "step": 436500 + }, + { + "epoch": 5.97, + "learning_rate": 3.506513923254638e-05, + "loss": 1.569, + "step": 437000 + }, + { + "epoch": 5.98, + "learning_rate": 3.504805129116485e-05, + "loss": 1.566, + "step": 437500 + }, + { + "epoch": 5.99, + "learning_rate": 3.503096334978332e-05, + "loss": 1.5541, + "step": 438000 + }, + { + "epoch": 5.99, + "learning_rate": 3.50138754084018e-05, + "loss": 1.5844, + "step": 438500 + }, + { + "epoch": 6.0, + "learning_rate": 3.499678746702028e-05, + "loss": 1.5793, + "step": 439000 + }, + { + "epoch": 6.01, + "learning_rate": 3.497969952563875e-05, + "loss": 1.5578, + "step": 439500 + }, + { + "epoch": 6.01, + "learning_rate": 3.496261158425722e-05, + "loss": 1.575, + "step": 440000 + }, + { + "epoch": 6.02, + "learning_rate": 3.49455236428757e-05, + "loss": 1.5636, + "step": 440500 + }, + { + "epoch": 6.03, + "learning_rate": 3.492843570149417e-05, + "loss": 1.5482, + "step": 441000 + }, + { + "epoch": 6.04, + "learning_rate": 3.491134776011264e-05, + "loss": 1.5839, + "step": 441500 + }, + { + "epoch": 6.04, + "learning_rate": 3.489425981873112e-05, + "loss": 1.5534, + "step": 442000 + }, + { + "epoch": 6.05, + "learning_rate": 3.48771718773496e-05, + "loss": 1.5353, + "step": 442500 + }, + { + "epoch": 6.06, + "learning_rate": 3.486008393596807e-05, + "loss": 1.5359, + "step": 443000 + }, + { + "epoch": 6.06, + "learning_rate": 3.484299599458654e-05, + "loss": 1.5631, + "step": 443500 + }, + { + "epoch": 6.07, + "learning_rate": 3.4825908053205016e-05, + "loss": 1.5583, + "step": 444000 + }, + { + "epoch": 6.08, + "learning_rate": 3.480882011182349e-05, + "loss": 1.5408, + "step": 444500 + }, + { + "epoch": 6.08, + "learning_rate": 3.4791732170441964e-05, + "loss": 1.5633, + "step": 445000 + }, + { + "epoch": 6.09, + "learning_rate": 3.477464422906044e-05, + "loss": 1.5532, + "step": 445500 + }, + { + "epoch": 6.1, + "learning_rate": 3.475755628767891e-05, + "loss": 1.5852, + "step": 446000 + }, + { + "epoch": 6.1, + "learning_rate": 3.474046834629739e-05, + "loss": 1.5353, + "step": 446500 + }, + { + "epoch": 6.11, + "learning_rate": 3.472338040491586e-05, + "loss": 1.5434, + "step": 447000 + }, + { + "epoch": 6.12, + "learning_rate": 3.470629246353434e-05, + "loss": 1.5615, + "step": 447500 + }, + { + "epoch": 6.12, + "learning_rate": 3.468920452215281e-05, + "loss": 1.5043, + "step": 448000 + }, + { + "epoch": 6.13, + "learning_rate": 3.4672116580771285e-05, + "loss": 1.581, + "step": 448500 + }, + { + "epoch": 6.14, + "learning_rate": 3.4655028639389755e-05, + "loss": 1.5333, + "step": 449000 + }, + { + "epoch": 6.14, + "learning_rate": 3.463794069800823e-05, + "loss": 1.5749, + "step": 449500 + }, + { + "epoch": 6.15, + "learning_rate": 3.46208527566267e-05, + "loss": 1.5383, + "step": 450000 + }, + { + "epoch": 6.16, + "learning_rate": 3.460376481524518e-05, + "loss": 1.5492, + "step": 450500 + }, + { + "epoch": 6.17, + "learning_rate": 3.458667687386366e-05, + "loss": 1.5571, + "step": 451000 + }, + { + "epoch": 6.17, + "learning_rate": 3.456958893248213e-05, + "loss": 1.561, + "step": 451500 + }, + { + "epoch": 6.18, + "learning_rate": 3.45525009911006e-05, + "loss": 1.5557, + "step": 452000 + }, + { + "epoch": 6.19, + "learning_rate": 3.4535413049719076e-05, + "loss": 1.5327, + "step": 452500 + }, + { + "epoch": 6.19, + "learning_rate": 3.451832510833755e-05, + "loss": 1.53, + "step": 453000 + }, + { + "epoch": 6.2, + "learning_rate": 3.4501237166956024e-05, + "loss": 1.5791, + "step": 453500 + }, + { + "epoch": 6.21, + "learning_rate": 3.4484149225574494e-05, + "loss": 1.547, + "step": 454000 + }, + { + "epoch": 6.21, + "learning_rate": 3.446706128419297e-05, + "loss": 1.5452, + "step": 454500 + }, + { + "epoch": 6.22, + "learning_rate": 3.444997334281145e-05, + "loss": 1.5543, + "step": 455000 + }, + { + "epoch": 6.23, + "learning_rate": 3.443288540142992e-05, + "loss": 1.5395, + "step": 455500 + }, + { + "epoch": 6.23, + "learning_rate": 3.44157974600484e-05, + "loss": 1.5383, + "step": 456000 + }, + { + "epoch": 6.24, + "learning_rate": 3.439870951866687e-05, + "loss": 1.5357, + "step": 456500 + }, + { + "epoch": 6.25, + "learning_rate": 3.4381621577285344e-05, + "loss": 1.5481, + "step": 457000 + }, + { + "epoch": 6.25, + "learning_rate": 3.4364533635903815e-05, + "loss": 1.5646, + "step": 457500 + }, + { + "epoch": 6.26, + "learning_rate": 3.434744569452229e-05, + "loss": 1.5372, + "step": 458000 + }, + { + "epoch": 6.27, + "learning_rate": 3.433035775314076e-05, + "loss": 1.5167, + "step": 458500 + }, + { + "epoch": 6.27, + "learning_rate": 3.431326981175924e-05, + "loss": 1.557, + "step": 459000 + }, + { + "epoch": 6.28, + "learning_rate": 3.429618187037771e-05, + "loss": 1.5306, + "step": 459500 + }, + { + "epoch": 6.29, + "learning_rate": 3.427909392899619e-05, + "loss": 1.5453, + "step": 460000 + }, + { + "epoch": 6.3, + "learning_rate": 3.426200598761466e-05, + "loss": 1.5515, + "step": 460500 + }, + { + "epoch": 6.3, + "learning_rate": 3.4244918046233136e-05, + "loss": 1.5457, + "step": 461000 + }, + { + "epoch": 6.31, + "learning_rate": 3.422783010485161e-05, + "loss": 1.5619, + "step": 461500 + }, + { + "epoch": 6.32, + "learning_rate": 3.4210742163470084e-05, + "loss": 1.5175, + "step": 462000 + }, + { + "epoch": 6.32, + "learning_rate": 3.4193654222088554e-05, + "loss": 1.5552, + "step": 462500 + }, + { + "epoch": 6.33, + "learning_rate": 3.417656628070703e-05, + "loss": 1.5615, + "step": 463000 + }, + { + "epoch": 6.34, + "learning_rate": 3.415947833932551e-05, + "loss": 1.5753, + "step": 463500 + }, + { + "epoch": 6.34, + "learning_rate": 3.414239039794398e-05, + "loss": 1.5117, + "step": 464000 + }, + { + "epoch": 6.35, + "learning_rate": 3.412530245656245e-05, + "loss": 1.5577, + "step": 464500 + }, + { + "epoch": 6.36, + "learning_rate": 3.4108214515180934e-05, + "loss": 1.5283, + "step": 465000 + }, + { + "epoch": 6.36, + "learning_rate": 3.4091126573799404e-05, + "loss": 1.5372, + "step": 465500 + }, + { + "epoch": 6.37, + "learning_rate": 3.4074038632417875e-05, + "loss": 1.5184, + "step": 466000 + }, + { + "epoch": 6.38, + "learning_rate": 3.405695069103635e-05, + "loss": 1.5096, + "step": 466500 + }, + { + "epoch": 6.38, + "learning_rate": 3.403986274965483e-05, + "loss": 1.5571, + "step": 467000 + }, + { + "epoch": 6.39, + "learning_rate": 3.40227748082733e-05, + "loss": 1.5488, + "step": 467500 + }, + { + "epoch": 6.4, + "learning_rate": 3.400568686689177e-05, + "loss": 1.5647, + "step": 468000 + }, + { + "epoch": 6.4, + "learning_rate": 3.398859892551025e-05, + "loss": 1.5332, + "step": 468500 + }, + { + "epoch": 6.41, + "learning_rate": 3.3971510984128725e-05, + "loss": 1.5426, + "step": 469000 + }, + { + "epoch": 6.42, + "learning_rate": 3.3954423042747196e-05, + "loss": 1.5484, + "step": 469500 + }, + { + "epoch": 6.43, + "learning_rate": 3.3937335101365666e-05, + "loss": 1.5454, + "step": 470000 + }, + { + "epoch": 6.43, + "learning_rate": 3.3920247159984144e-05, + "loss": 1.5499, + "step": 470500 + }, + { + "epoch": 6.44, + "learning_rate": 3.390315921860262e-05, + "loss": 1.5406, + "step": 471000 + }, + { + "epoch": 6.45, + "learning_rate": 3.388607127722109e-05, + "loss": 1.5283, + "step": 471500 + }, + { + "epoch": 6.45, + "learning_rate": 3.386898333583957e-05, + "loss": 1.5301, + "step": 472000 + }, + { + "epoch": 6.46, + "learning_rate": 3.385189539445804e-05, + "loss": 1.5239, + "step": 472500 + }, + { + "epoch": 6.47, + "learning_rate": 3.3834807453076517e-05, + "loss": 1.5675, + "step": 473000 + }, + { + "epoch": 6.47, + "learning_rate": 3.381771951169499e-05, + "loss": 1.5308, + "step": 473500 + }, + { + "epoch": 6.48, + "learning_rate": 3.3800631570313464e-05, + "loss": 1.5381, + "step": 474000 + }, + { + "epoch": 6.49, + "learning_rate": 3.3783543628931935e-05, + "loss": 1.5742, + "step": 474500 + }, + { + "epoch": 6.49, + "learning_rate": 3.376645568755041e-05, + "loss": 1.5408, + "step": 475000 + }, + { + "epoch": 6.5, + "learning_rate": 3.374936774616889e-05, + "loss": 1.5415, + "step": 475500 + }, + { + "epoch": 6.51, + "learning_rate": 3.373227980478736e-05, + "loss": 1.5456, + "step": 476000 + }, + { + "epoch": 6.51, + "learning_rate": 3.371519186340583e-05, + "loss": 1.5237, + "step": 476500 + }, + { + "epoch": 6.52, + "learning_rate": 3.369810392202431e-05, + "loss": 1.5356, + "step": 477000 + }, + { + "epoch": 6.53, + "learning_rate": 3.3681015980642785e-05, + "loss": 1.5491, + "step": 477500 + }, + { + "epoch": 6.53, + "learning_rate": 3.3663928039261256e-05, + "loss": 1.5266, + "step": 478000 + }, + { + "epoch": 6.54, + "learning_rate": 3.3646840097879726e-05, + "loss": 1.5224, + "step": 478500 + }, + { + "epoch": 6.55, + "learning_rate": 3.3629752156498204e-05, + "loss": 1.5241, + "step": 479000 + }, + { + "epoch": 6.55, + "learning_rate": 3.361266421511668e-05, + "loss": 1.5153, + "step": 479500 + }, + { + "epoch": 6.56, + "learning_rate": 3.359557627373515e-05, + "loss": 1.4995, + "step": 480000 + }, + { + "epoch": 6.57, + "learning_rate": 3.357848833235362e-05, + "loss": 1.5614, + "step": 480500 + }, + { + "epoch": 6.58, + "learning_rate": 3.3561400390972106e-05, + "loss": 1.5745, + "step": 481000 + }, + { + "epoch": 6.58, + "learning_rate": 3.3544312449590577e-05, + "loss": 1.5262, + "step": 481500 + }, + { + "epoch": 6.59, + "learning_rate": 3.352722450820905e-05, + "loss": 1.5751, + "step": 482000 + }, + { + "epoch": 6.6, + "learning_rate": 3.3510136566827524e-05, + "loss": 1.5416, + "step": 482500 + }, + { + "epoch": 6.6, + "learning_rate": 3.3493048625446e-05, + "loss": 1.5391, + "step": 483000 + }, + { + "epoch": 6.61, + "learning_rate": 3.347596068406447e-05, + "loss": 1.5446, + "step": 483500 + }, + { + "epoch": 6.62, + "learning_rate": 3.345887274268294e-05, + "loss": 1.5555, + "step": 484000 + }, + { + "epoch": 6.62, + "learning_rate": 3.344178480130142e-05, + "loss": 1.5565, + "step": 484500 + }, + { + "epoch": 6.63, + "learning_rate": 3.342469685991989e-05, + "loss": 1.5761, + "step": 485000 + }, + { + "epoch": 6.64, + "learning_rate": 3.340760891853837e-05, + "loss": 1.5509, + "step": 485500 + }, + { + "epoch": 6.64, + "learning_rate": 3.339052097715684e-05, + "loss": 1.5478, + "step": 486000 + }, + { + "epoch": 6.65, + "learning_rate": 3.3373433035775316e-05, + "loss": 1.5152, + "step": 486500 + }, + { + "epoch": 6.66, + "learning_rate": 3.3356345094393786e-05, + "loss": 1.5622, + "step": 487000 + }, + { + "epoch": 6.66, + "learning_rate": 3.3339257153012264e-05, + "loss": 1.5564, + "step": 487500 + }, + { + "epoch": 6.67, + "learning_rate": 3.332216921163074e-05, + "loss": 1.5427, + "step": 488000 + }, + { + "epoch": 6.68, + "learning_rate": 3.330508127024921e-05, + "loss": 1.5522, + "step": 488500 + }, + { + "epoch": 6.68, + "learning_rate": 3.328799332886768e-05, + "loss": 1.5525, + "step": 489000 + }, + { + "epoch": 6.69, + "learning_rate": 3.327090538748616e-05, + "loss": 1.572, + "step": 489500 + }, + { + "epoch": 6.7, + "learning_rate": 3.3253817446104636e-05, + "loss": 1.5274, + "step": 490000 + }, + { + "epoch": 6.71, + "learning_rate": 3.323672950472311e-05, + "loss": 1.5633, + "step": 490500 + }, + { + "epoch": 6.71, + "learning_rate": 3.321964156334158e-05, + "loss": 1.5416, + "step": 491000 + }, + { + "epoch": 6.72, + "learning_rate": 3.320255362196006e-05, + "loss": 1.5238, + "step": 491500 + }, + { + "epoch": 6.73, + "learning_rate": 3.318546568057853e-05, + "loss": 1.5249, + "step": 492000 + }, + { + "epoch": 6.73, + "learning_rate": 3.3168377739197e-05, + "loss": 1.5568, + "step": 492500 + }, + { + "epoch": 6.74, + "learning_rate": 3.315128979781548e-05, + "loss": 1.5222, + "step": 493000 + }, + { + "epoch": 6.75, + "learning_rate": 3.313420185643396e-05, + "loss": 1.537, + "step": 493500 + }, + { + "epoch": 6.75, + "learning_rate": 3.311711391505243e-05, + "loss": 1.5748, + "step": 494000 + }, + { + "epoch": 6.76, + "learning_rate": 3.31000259736709e-05, + "loss": 1.5538, + "step": 494500 + }, + { + "epoch": 6.77, + "learning_rate": 3.3082938032289376e-05, + "loss": 1.5065, + "step": 495000 + }, + { + "epoch": 6.77, + "learning_rate": 3.306585009090785e-05, + "loss": 1.544, + "step": 495500 + }, + { + "epoch": 6.78, + "learning_rate": 3.3048762149526323e-05, + "loss": 1.5751, + "step": 496000 + }, + { + "epoch": 6.79, + "learning_rate": 3.3031674208144794e-05, + "loss": 1.54, + "step": 496500 + }, + { + "epoch": 6.79, + "learning_rate": 3.301458626676327e-05, + "loss": 1.5562, + "step": 497000 + }, + { + "epoch": 6.8, + "learning_rate": 3.299749832538175e-05, + "loss": 1.5621, + "step": 497500 + }, + { + "epoch": 6.81, + "learning_rate": 3.298041038400022e-05, + "loss": 1.55, + "step": 498000 + }, + { + "epoch": 6.81, + "learning_rate": 3.2963322442618696e-05, + "loss": 1.5528, + "step": 498500 + }, + { + "epoch": 6.82, + "learning_rate": 3.294623450123717e-05, + "loss": 1.5238, + "step": 499000 + }, + { + "epoch": 6.83, + "learning_rate": 3.2929146559855644e-05, + "loss": 1.5371, + "step": 499500 + }, + { + "epoch": 6.84, + "learning_rate": 3.2912058618474115e-05, + "loss": 1.544, + "step": 500000 + }, + { + "epoch": 6.84, + "learning_rate": 3.289497067709259e-05, + "loss": 1.5542, + "step": 500500 + }, + { + "epoch": 6.85, + "learning_rate": 3.287788273571106e-05, + "loss": 1.5367, + "step": 501000 + }, + { + "epoch": 6.86, + "learning_rate": 3.286079479432954e-05, + "loss": 1.5233, + "step": 501500 + }, + { + "epoch": 6.86, + "learning_rate": 3.284370685294802e-05, + "loss": 1.5313, + "step": 502000 + }, + { + "epoch": 6.87, + "learning_rate": 3.282661891156649e-05, + "loss": 1.5319, + "step": 502500 + }, + { + "epoch": 6.88, + "learning_rate": 3.280953097018496e-05, + "loss": 1.523, + "step": 503000 + }, + { + "epoch": 6.88, + "learning_rate": 3.2792443028803436e-05, + "loss": 1.5313, + "step": 503500 + }, + { + "epoch": 6.89, + "learning_rate": 3.277535508742191e-05, + "loss": 1.5532, + "step": 504000 + }, + { + "epoch": 6.9, + "learning_rate": 3.2758267146040383e-05, + "loss": 1.5591, + "step": 504500 + }, + { + "epoch": 6.9, + "learning_rate": 3.2741179204658854e-05, + "loss": 1.5637, + "step": 505000 + }, + { + "epoch": 6.91, + "learning_rate": 3.272409126327733e-05, + "loss": 1.5745, + "step": 505500 + }, + { + "epoch": 6.92, + "learning_rate": 3.270700332189581e-05, + "loss": 1.5701, + "step": 506000 + }, + { + "epoch": 6.92, + "learning_rate": 3.268991538051428e-05, + "loss": 1.5145, + "step": 506500 + }, + { + "epoch": 6.93, + "learning_rate": 3.267282743913275e-05, + "loss": 1.5454, + "step": 507000 + }, + { + "epoch": 6.94, + "learning_rate": 3.2655739497751234e-05, + "loss": 1.5264, + "step": 507500 + }, + { + "epoch": 6.94, + "learning_rate": 3.2638651556369704e-05, + "loss": 1.5382, + "step": 508000 + }, + { + "epoch": 6.95, + "learning_rate": 3.2621563614988175e-05, + "loss": 1.5547, + "step": 508500 + }, + { + "epoch": 6.96, + "learning_rate": 3.260447567360665e-05, + "loss": 1.5306, + "step": 509000 + }, + { + "epoch": 6.97, + "learning_rate": 3.258738773222513e-05, + "loss": 1.5549, + "step": 509500 + }, + { + "epoch": 6.97, + "learning_rate": 3.25702997908436e-05, + "loss": 1.5407, + "step": 510000 + }, + { + "epoch": 6.98, + "learning_rate": 3.255321184946207e-05, + "loss": 1.5194, + "step": 510500 + }, + { + "epoch": 6.99, + "learning_rate": 3.253612390808055e-05, + "loss": 1.5389, + "step": 511000 + }, + { + "epoch": 6.99, + "learning_rate": 3.2519035966699025e-05, + "loss": 1.5692, + "step": 511500 + }, + { + "epoch": 7.0, + "learning_rate": 3.2501948025317496e-05, + "loss": 1.5397, + "step": 512000 + }, + { + "epoch": 7.01, + "learning_rate": 3.248486008393597e-05, + "loss": 1.5248, + "step": 512500 + }, + { + "epoch": 7.01, + "learning_rate": 3.2467772142554443e-05, + "loss": 1.5311, + "step": 513000 + }, + { + "epoch": 7.02, + "learning_rate": 3.2450684201172914e-05, + "loss": 1.5084, + "step": 513500 + }, + { + "epoch": 7.03, + "learning_rate": 3.243359625979139e-05, + "loss": 1.5212, + "step": 514000 + }, + { + "epoch": 7.03, + "learning_rate": 3.241650831840987e-05, + "loss": 1.5148, + "step": 514500 + }, + { + "epoch": 7.04, + "learning_rate": 3.239942037702834e-05, + "loss": 1.5318, + "step": 515000 + }, + { + "epoch": 7.05, + "learning_rate": 3.238233243564681e-05, + "loss": 1.5638, + "step": 515500 + }, + { + "epoch": 7.05, + "learning_rate": 3.236524449426529e-05, + "loss": 1.5184, + "step": 516000 + }, + { + "epoch": 7.06, + "learning_rate": 3.2348156552883764e-05, + "loss": 1.5239, + "step": 516500 + }, + { + "epoch": 7.07, + "learning_rate": 3.2331068611502235e-05, + "loss": 1.5684, + "step": 517000 + }, + { + "epoch": 7.07, + "learning_rate": 3.2313980670120705e-05, + "loss": 1.528, + "step": 517500 + }, + { + "epoch": 7.08, + "learning_rate": 3.229689272873919e-05, + "loss": 1.5339, + "step": 518000 + }, + { + "epoch": 7.09, + "learning_rate": 3.227980478735766e-05, + "loss": 1.5115, + "step": 518500 + }, + { + "epoch": 7.09, + "learning_rate": 3.226271684597613e-05, + "loss": 1.4908, + "step": 519000 + }, + { + "epoch": 7.1, + "learning_rate": 3.224562890459461e-05, + "loss": 1.5348, + "step": 519500 + }, + { + "epoch": 7.11, + "learning_rate": 3.2228540963213085e-05, + "loss": 1.5192, + "step": 520000 + }, + { + "epoch": 7.12, + "learning_rate": 3.2211453021831556e-05, + "loss": 1.5409, + "step": 520500 + }, + { + "epoch": 7.12, + "learning_rate": 3.2194365080450026e-05, + "loss": 1.5328, + "step": 521000 + }, + { + "epoch": 7.13, + "learning_rate": 3.21772771390685e-05, + "loss": 1.5296, + "step": 521500 + }, + { + "epoch": 7.14, + "learning_rate": 3.216018919768698e-05, + "loss": 1.5292, + "step": 522000 + }, + { + "epoch": 7.14, + "learning_rate": 3.214310125630545e-05, + "loss": 1.5025, + "step": 522500 + }, + { + "epoch": 7.15, + "learning_rate": 3.212601331492392e-05, + "loss": 1.5158, + "step": 523000 + }, + { + "epoch": 7.16, + "learning_rate": 3.21089253735424e-05, + "loss": 1.5034, + "step": 523500 + }, + { + "epoch": 7.16, + "learning_rate": 3.2091837432160876e-05, + "loss": 1.5372, + "step": 524000 + }, + { + "epoch": 7.17, + "learning_rate": 3.207474949077935e-05, + "loss": 1.5339, + "step": 524500 + }, + { + "epoch": 7.18, + "learning_rate": 3.2057661549397824e-05, + "loss": 1.5341, + "step": 525000 + }, + { + "epoch": 7.18, + "learning_rate": 3.2040573608016295e-05, + "loss": 1.5184, + "step": 525500 + }, + { + "epoch": 7.19, + "learning_rate": 3.202348566663477e-05, + "loss": 1.513, + "step": 526000 + }, + { + "epoch": 7.2, + "learning_rate": 3.200639772525324e-05, + "loss": 1.5276, + "step": 526500 + }, + { + "epoch": 7.2, + "learning_rate": 3.198930978387172e-05, + "loss": 1.5359, + "step": 527000 + }, + { + "epoch": 7.21, + "learning_rate": 3.197222184249019e-05, + "loss": 1.5318, + "step": 527500 + }, + { + "epoch": 7.22, + "learning_rate": 3.195513390110867e-05, + "loss": 1.513, + "step": 528000 + }, + { + "epoch": 7.22, + "learning_rate": 3.1938045959727145e-05, + "loss": 1.504, + "step": 528500 + }, + { + "epoch": 7.23, + "learning_rate": 3.1920958018345615e-05, + "loss": 1.5088, + "step": 529000 + }, + { + "epoch": 7.24, + "learning_rate": 3.1903870076964086e-05, + "loss": 1.5236, + "step": 529500 + }, + { + "epoch": 7.25, + "learning_rate": 3.188678213558256e-05, + "loss": 1.5217, + "step": 530000 + }, + { + "epoch": 7.25, + "learning_rate": 3.186969419420104e-05, + "loss": 1.5241, + "step": 530500 + }, + { + "epoch": 7.26, + "learning_rate": 3.185260625281951e-05, + "loss": 1.5056, + "step": 531000 + }, + { + "epoch": 7.27, + "learning_rate": 3.183551831143798e-05, + "loss": 1.5195, + "step": 531500 + }, + { + "epoch": 7.27, + "learning_rate": 3.181843037005646e-05, + "loss": 1.5289, + "step": 532000 + }, + { + "epoch": 7.28, + "learning_rate": 3.1801342428674936e-05, + "loss": 1.5018, + "step": 532500 + }, + { + "epoch": 7.29, + "learning_rate": 3.178425448729341e-05, + "loss": 1.4901, + "step": 533000 + }, + { + "epoch": 7.29, + "learning_rate": 3.176716654591188e-05, + "loss": 1.5055, + "step": 533500 + }, + { + "epoch": 7.3, + "learning_rate": 3.175007860453036e-05, + "loss": 1.5241, + "step": 534000 + }, + { + "epoch": 7.31, + "learning_rate": 3.173299066314883e-05, + "loss": 1.5217, + "step": 534500 + }, + { + "epoch": 7.31, + "learning_rate": 3.17159027217673e-05, + "loss": 1.5158, + "step": 535000 + }, + { + "epoch": 7.32, + "learning_rate": 3.169881478038578e-05, + "loss": 1.549, + "step": 535500 + }, + { + "epoch": 7.33, + "learning_rate": 3.168172683900426e-05, + "loss": 1.5058, + "step": 536000 + }, + { + "epoch": 7.33, + "learning_rate": 3.166463889762273e-05, + "loss": 1.4947, + "step": 536500 + }, + { + "epoch": 7.34, + "learning_rate": 3.16475509562412e-05, + "loss": 1.5366, + "step": 537000 + }, + { + "epoch": 7.35, + "learning_rate": 3.1630463014859675e-05, + "loss": 1.5475, + "step": 537500 + }, + { + "epoch": 7.35, + "learning_rate": 3.161337507347815e-05, + "loss": 1.4898, + "step": 538000 + }, + { + "epoch": 7.36, + "learning_rate": 3.159628713209662e-05, + "loss": 1.5315, + "step": 538500 + }, + { + "epoch": 7.37, + "learning_rate": 3.15791991907151e-05, + "loss": 1.5053, + "step": 539000 + }, + { + "epoch": 7.38, + "learning_rate": 3.156211124933357e-05, + "loss": 1.5072, + "step": 539500 + }, + { + "epoch": 7.38, + "learning_rate": 3.154502330795205e-05, + "loss": 1.5366, + "step": 540000 + }, + { + "epoch": 7.39, + "learning_rate": 3.152793536657052e-05, + "loss": 1.527, + "step": 540500 + }, + { + "epoch": 7.4, + "learning_rate": 3.1510847425188996e-05, + "loss": 1.5053, + "step": 541000 + }, + { + "epoch": 7.4, + "learning_rate": 3.149375948380747e-05, + "loss": 1.5502, + "step": 541500 + }, + { + "epoch": 7.41, + "learning_rate": 3.1476671542425944e-05, + "loss": 1.5429, + "step": 542000 + }, + { + "epoch": 7.42, + "learning_rate": 3.1459583601044415e-05, + "loss": 1.517, + "step": 542500 + }, + { + "epoch": 7.42, + "learning_rate": 3.144249565966289e-05, + "loss": 1.4951, + "step": 543000 + }, + { + "epoch": 7.43, + "learning_rate": 3.142540771828136e-05, + "loss": 1.5219, + "step": 543500 + }, + { + "epoch": 7.44, + "learning_rate": 3.140831977689983e-05, + "loss": 1.5131, + "step": 544000 + }, + { + "epoch": 7.44, + "learning_rate": 3.139123183551832e-05, + "loss": 1.5097, + "step": 544500 + }, + { + "epoch": 7.45, + "learning_rate": 3.137414389413679e-05, + "loss": 1.5225, + "step": 545000 + }, + { + "epoch": 7.46, + "learning_rate": 3.135705595275526e-05, + "loss": 1.5209, + "step": 545500 + }, + { + "epoch": 7.46, + "learning_rate": 3.1339968011373735e-05, + "loss": 1.5213, + "step": 546000 + }, + { + "epoch": 7.47, + "learning_rate": 3.132288006999221e-05, + "loss": 1.496, + "step": 546500 + }, + { + "epoch": 7.48, + "learning_rate": 3.130579212861068e-05, + "loss": 1.486, + "step": 547000 + }, + { + "epoch": 7.48, + "learning_rate": 3.1288704187229154e-05, + "loss": 1.5273, + "step": 547500 + }, + { + "epoch": 7.49, + "learning_rate": 3.127161624584763e-05, + "loss": 1.4992, + "step": 548000 + }, + { + "epoch": 7.5, + "learning_rate": 3.125452830446611e-05, + "loss": 1.5354, + "step": 548500 + }, + { + "epoch": 7.51, + "learning_rate": 3.123744036308458e-05, + "loss": 1.5287, + "step": 549000 + }, + { + "epoch": 7.51, + "learning_rate": 3.1220352421703056e-05, + "loss": 1.5567, + "step": 549500 + }, + { + "epoch": 7.52, + "learning_rate": 3.120326448032153e-05, + "loss": 1.5254, + "step": 550000 + }, + { + "epoch": 7.53, + "learning_rate": 3.1186176538940004e-05, + "loss": 1.5291, + "step": 550500 + }, + { + "epoch": 7.53, + "learning_rate": 3.1169088597558475e-05, + "loss": 1.528, + "step": 551000 + }, + { + "epoch": 7.54, + "learning_rate": 3.115200065617695e-05, + "loss": 1.5015, + "step": 551500 + }, + { + "epoch": 7.55, + "learning_rate": 3.113491271479542e-05, + "loss": 1.5086, + "step": 552000 + }, + { + "epoch": 7.55, + "learning_rate": 3.11178247734139e-05, + "loss": 1.5506, + "step": 552500 + }, + { + "epoch": 7.56, + "learning_rate": 3.110073683203237e-05, + "loss": 1.5202, + "step": 553000 + }, + { + "epoch": 7.57, + "learning_rate": 3.108364889065085e-05, + "loss": 1.5225, + "step": 553500 + }, + { + "epoch": 7.57, + "learning_rate": 3.106656094926932e-05, + "loss": 1.5213, + "step": 554000 + }, + { + "epoch": 7.58, + "learning_rate": 3.1049473007887795e-05, + "loss": 1.5335, + "step": 554500 + }, + { + "epoch": 7.59, + "learning_rate": 3.103238506650627e-05, + "loss": 1.5152, + "step": 555000 + }, + { + "epoch": 7.59, + "learning_rate": 3.101529712512474e-05, + "loss": 1.4791, + "step": 555500 + }, + { + "epoch": 7.6, + "learning_rate": 3.0998209183743214e-05, + "loss": 1.5322, + "step": 556000 + }, + { + "epoch": 7.61, + "learning_rate": 3.098112124236169e-05, + "loss": 1.4964, + "step": 556500 + }, + { + "epoch": 7.61, + "learning_rate": 3.096403330098017e-05, + "loss": 1.5139, + "step": 557000 + }, + { + "epoch": 7.62, + "learning_rate": 3.094694535959864e-05, + "loss": 1.4994, + "step": 557500 + }, + { + "epoch": 7.63, + "learning_rate": 3.092985741821711e-05, + "loss": 1.5465, + "step": 558000 + }, + { + "epoch": 7.63, + "learning_rate": 3.0912769476835593e-05, + "loss": 1.5179, + "step": 558500 + }, + { + "epoch": 7.64, + "learning_rate": 3.0895681535454064e-05, + "loss": 1.5304, + "step": 559000 + }, + { + "epoch": 7.65, + "learning_rate": 3.0878593594072535e-05, + "loss": 1.5579, + "step": 559500 + }, + { + "epoch": 7.66, + "learning_rate": 3.086150565269101e-05, + "loss": 1.5394, + "step": 560000 + }, + { + "epoch": 7.66, + "learning_rate": 3.084441771130949e-05, + "loss": 1.5217, + "step": 560500 + }, + { + "epoch": 7.67, + "learning_rate": 3.082732976992796e-05, + "loss": 1.5014, + "step": 561000 + }, + { + "epoch": 7.68, + "learning_rate": 3.081024182854643e-05, + "loss": 1.5444, + "step": 561500 + }, + { + "epoch": 7.68, + "learning_rate": 3.079315388716491e-05, + "loss": 1.5268, + "step": 562000 + }, + { + "epoch": 7.69, + "learning_rate": 3.0776065945783385e-05, + "loss": 1.5246, + "step": 562500 + }, + { + "epoch": 7.7, + "learning_rate": 3.0758978004401855e-05, + "loss": 1.4976, + "step": 563000 + }, + { + "epoch": 7.7, + "learning_rate": 3.0741890063020326e-05, + "loss": 1.5283, + "step": 563500 + }, + { + "epoch": 7.71, + "learning_rate": 3.07248021216388e-05, + "loss": 1.5341, + "step": 564000 + }, + { + "epoch": 7.72, + "learning_rate": 3.070771418025728e-05, + "loss": 1.5313, + "step": 564500 + }, + { + "epoch": 7.72, + "learning_rate": 3.069062623887575e-05, + "loss": 1.5199, + "step": 565000 + }, + { + "epoch": 7.73, + "learning_rate": 3.067353829749423e-05, + "loss": 1.5199, + "step": 565500 + }, + { + "epoch": 7.74, + "learning_rate": 3.06564503561127e-05, + "loss": 1.5456, + "step": 566000 + }, + { + "epoch": 7.74, + "learning_rate": 3.0639362414731176e-05, + "loss": 1.5267, + "step": 566500 + }, + { + "epoch": 7.75, + "learning_rate": 3.062227447334965e-05, + "loss": 1.5248, + "step": 567000 + }, + { + "epoch": 7.76, + "learning_rate": 3.0605186531968124e-05, + "loss": 1.5146, + "step": 567500 + }, + { + "epoch": 7.76, + "learning_rate": 3.0588098590586594e-05, + "loss": 1.5425, + "step": 568000 + }, + { + "epoch": 7.77, + "learning_rate": 3.057101064920507e-05, + "loss": 1.5498, + "step": 568500 + }, + { + "epoch": 7.78, + "learning_rate": 3.055392270782355e-05, + "loss": 1.4988, + "step": 569000 + }, + { + "epoch": 7.79, + "learning_rate": 3.053683476644202e-05, + "loss": 1.533, + "step": 569500 + }, + { + "epoch": 7.79, + "learning_rate": 3.051974682506049e-05, + "loss": 1.5044, + "step": 570000 + }, + { + "epoch": 7.8, + "learning_rate": 3.050265888367897e-05, + "loss": 1.5205, + "step": 570500 + }, + { + "epoch": 7.81, + "learning_rate": 3.048557094229744e-05, + "loss": 1.5155, + "step": 571000 + }, + { + "epoch": 7.81, + "learning_rate": 3.0468483000915915e-05, + "loss": 1.5177, + "step": 571500 + }, + { + "epoch": 7.82, + "learning_rate": 3.045139505953439e-05, + "loss": 1.521, + "step": 572000 + }, + { + "epoch": 7.83, + "learning_rate": 3.0434307118152866e-05, + "loss": 1.5155, + "step": 572500 + }, + { + "epoch": 7.83, + "learning_rate": 3.041721917677134e-05, + "loss": 1.4831, + "step": 573000 + }, + { + "epoch": 7.84, + "learning_rate": 3.040013123538981e-05, + "loss": 1.5296, + "step": 573500 + }, + { + "epoch": 7.85, + "learning_rate": 3.0383043294008285e-05, + "loss": 1.491, + "step": 574000 + }, + { + "epoch": 7.85, + "learning_rate": 3.0365955352626755e-05, + "loss": 1.525, + "step": 574500 + }, + { + "epoch": 7.86, + "learning_rate": 3.0348867411245236e-05, + "loss": 1.514, + "step": 575000 + }, + { + "epoch": 7.87, + "learning_rate": 3.0331779469863707e-05, + "loss": 1.5506, + "step": 575500 + }, + { + "epoch": 7.87, + "learning_rate": 3.031469152848218e-05, + "loss": 1.4966, + "step": 576000 + }, + { + "epoch": 7.88, + "learning_rate": 3.0297603587100654e-05, + "loss": 1.5237, + "step": 576500 + }, + { + "epoch": 7.89, + "learning_rate": 3.0280515645719132e-05, + "loss": 1.5381, + "step": 577000 + }, + { + "epoch": 7.89, + "learning_rate": 3.0263427704337606e-05, + "loss": 1.5263, + "step": 577500 + }, + { + "epoch": 7.9, + "learning_rate": 3.0246339762956076e-05, + "loss": 1.5327, + "step": 578000 + }, + { + "epoch": 7.91, + "learning_rate": 3.022925182157455e-05, + "loss": 1.5247, + "step": 578500 + }, + { + "epoch": 7.92, + "learning_rate": 3.0212163880193027e-05, + "loss": 1.5136, + "step": 579000 + }, + { + "epoch": 7.92, + "learning_rate": 3.01950759388115e-05, + "loss": 1.5123, + "step": 579500 + }, + { + "epoch": 7.93, + "learning_rate": 3.0177987997429975e-05, + "loss": 1.4809, + "step": 580000 + }, + { + "epoch": 7.94, + "learning_rate": 3.0160900056048446e-05, + "loss": 1.5199, + "step": 580500 + }, + { + "epoch": 7.94, + "learning_rate": 3.0143812114666926e-05, + "loss": 1.5053, + "step": 581000 + }, + { + "epoch": 7.95, + "learning_rate": 3.0126724173285397e-05, + "loss": 1.4964, + "step": 581500 + }, + { + "epoch": 7.96, + "learning_rate": 3.010963623190387e-05, + "loss": 1.5175, + "step": 582000 + }, + { + "epoch": 7.96, + "learning_rate": 3.0092548290522345e-05, + "loss": 1.5182, + "step": 582500 + }, + { + "epoch": 7.97, + "learning_rate": 3.0075460349140822e-05, + "loss": 1.4836, + "step": 583000 + }, + { + "epoch": 7.98, + "learning_rate": 3.0058372407759293e-05, + "loss": 1.5693, + "step": 583500 + }, + { + "epoch": 7.98, + "learning_rate": 3.0041284466377767e-05, + "loss": 1.5146, + "step": 584000 + }, + { + "epoch": 7.99, + "learning_rate": 3.002419652499624e-05, + "loss": 1.5016, + "step": 584500 + }, + { + "epoch": 8.0, + "learning_rate": 3.0007108583614718e-05, + "loss": 1.5213, + "step": 585000 + }, + { + "epoch": 8.0, + "learning_rate": 2.9990020642233192e-05, + "loss": 1.4901, + "step": 585500 + }, + { + "epoch": 8.01, + "learning_rate": 2.9972932700851662e-05, + "loss": 1.4989, + "step": 586000 + }, + { + "epoch": 8.02, + "learning_rate": 2.9955844759470136e-05, + "loss": 1.4987, + "step": 586500 + }, + { + "epoch": 8.02, + "learning_rate": 2.9938756818088613e-05, + "loss": 1.5039, + "step": 587000 + }, + { + "epoch": 8.03, + "learning_rate": 2.9921668876707087e-05, + "loss": 1.5082, + "step": 587500 + }, + { + "epoch": 8.04, + "learning_rate": 2.990458093532556e-05, + "loss": 1.4895, + "step": 588000 + }, + { + "epoch": 8.05, + "learning_rate": 2.9887492993944032e-05, + "loss": 1.4526, + "step": 588500 + }, + { + "epoch": 8.05, + "learning_rate": 2.9870405052562512e-05, + "loss": 1.5076, + "step": 589000 + }, + { + "epoch": 8.06, + "learning_rate": 2.9853317111180983e-05, + "loss": 1.4863, + "step": 589500 + }, + { + "epoch": 8.07, + "learning_rate": 2.9836229169799457e-05, + "loss": 1.5033, + "step": 590000 + }, + { + "epoch": 8.07, + "learning_rate": 2.981914122841793e-05, + "loss": 1.5025, + "step": 590500 + }, + { + "epoch": 8.08, + "learning_rate": 2.9802053287036408e-05, + "loss": 1.4894, + "step": 591000 + }, + { + "epoch": 8.09, + "learning_rate": 2.9784965345654882e-05, + "loss": 1.4785, + "step": 591500 + }, + { + "epoch": 8.09, + "learning_rate": 2.9767877404273353e-05, + "loss": 1.5096, + "step": 592000 + }, + { + "epoch": 8.1, + "learning_rate": 2.9750789462891827e-05, + "loss": 1.5039, + "step": 592500 + }, + { + "epoch": 8.11, + "learning_rate": 2.9733701521510304e-05, + "loss": 1.5387, + "step": 593000 + }, + { + "epoch": 8.11, + "learning_rate": 2.9716613580128778e-05, + "loss": 1.5217, + "step": 593500 + }, + { + "epoch": 8.12, + "learning_rate": 2.9699525638747248e-05, + "loss": 1.4908, + "step": 594000 + }, + { + "epoch": 8.13, + "learning_rate": 2.9682437697365722e-05, + "loss": 1.5035, + "step": 594500 + }, + { + "epoch": 8.13, + "learning_rate": 2.96653497559842e-05, + "loss": 1.5094, + "step": 595000 + }, + { + "epoch": 8.14, + "learning_rate": 2.9648261814602673e-05, + "loss": 1.4902, + "step": 595500 + }, + { + "epoch": 8.15, + "learning_rate": 2.9631173873221147e-05, + "loss": 1.5087, + "step": 596000 + }, + { + "epoch": 8.15, + "learning_rate": 2.9614085931839618e-05, + "loss": 1.4734, + "step": 596500 + }, + { + "epoch": 8.16, + "learning_rate": 2.95969979904581e-05, + "loss": 1.5039, + "step": 597000 + }, + { + "epoch": 8.17, + "learning_rate": 2.957991004907657e-05, + "loss": 1.4942, + "step": 597500 + }, + { + "epoch": 8.17, + "learning_rate": 2.9562822107695043e-05, + "loss": 1.5051, + "step": 598000 + }, + { + "epoch": 8.18, + "learning_rate": 2.9545734166313517e-05, + "loss": 1.4932, + "step": 598500 + }, + { + "epoch": 8.19, + "learning_rate": 2.9528646224931994e-05, + "loss": 1.4889, + "step": 599000 + }, + { + "epoch": 8.2, + "learning_rate": 2.9511558283550468e-05, + "loss": 1.4516, + "step": 599500 + }, + { + "epoch": 8.2, + "learning_rate": 2.949447034216894e-05, + "loss": 1.5125, + "step": 600000 + }, + { + "epoch": 8.21, + "learning_rate": 2.9477382400787413e-05, + "loss": 1.486, + "step": 600500 + }, + { + "epoch": 8.22, + "learning_rate": 2.946029445940589e-05, + "loss": 1.4924, + "step": 601000 + }, + { + "epoch": 8.22, + "learning_rate": 2.9443206518024364e-05, + "loss": 1.4962, + "step": 601500 + }, + { + "epoch": 8.23, + "learning_rate": 2.9426118576642834e-05, + "loss": 1.4966, + "step": 602000 + }, + { + "epoch": 8.24, + "learning_rate": 2.9409030635261308e-05, + "loss": 1.5036, + "step": 602500 + }, + { + "epoch": 8.24, + "learning_rate": 2.9391942693879786e-05, + "loss": 1.5087, + "step": 603000 + }, + { + "epoch": 8.25, + "learning_rate": 2.937485475249826e-05, + "loss": 1.5141, + "step": 603500 + }, + { + "epoch": 8.26, + "learning_rate": 2.9357766811116733e-05, + "loss": 1.4993, + "step": 604000 + }, + { + "epoch": 8.26, + "learning_rate": 2.9340678869735204e-05, + "loss": 1.4949, + "step": 604500 + }, + { + "epoch": 8.27, + "learning_rate": 2.9323590928353678e-05, + "loss": 1.5242, + "step": 605000 + }, + { + "epoch": 8.28, + "learning_rate": 2.9306502986972155e-05, + "loss": 1.4843, + "step": 605500 + }, + { + "epoch": 8.28, + "learning_rate": 2.928941504559063e-05, + "loss": 1.5068, + "step": 606000 + }, + { + "epoch": 8.29, + "learning_rate": 2.9272327104209103e-05, + "loss": 1.5028, + "step": 606500 + }, + { + "epoch": 8.3, + "learning_rate": 2.9255239162827573e-05, + "loss": 1.4802, + "step": 607000 + }, + { + "epoch": 8.3, + "learning_rate": 2.9238151221446054e-05, + "loss": 1.4709, + "step": 607500 + }, + { + "epoch": 8.31, + "learning_rate": 2.9221063280064525e-05, + "loss": 1.4903, + "step": 608000 + }, + { + "epoch": 8.32, + "learning_rate": 2.9203975338683e-05, + "loss": 1.4965, + "step": 608500 + }, + { + "epoch": 8.33, + "learning_rate": 2.9186887397301473e-05, + "loss": 1.5031, + "step": 609000 + }, + { + "epoch": 8.33, + "learning_rate": 2.916979945591995e-05, + "loss": 1.4853, + "step": 609500 + }, + { + "epoch": 8.34, + "learning_rate": 2.9152711514538424e-05, + "loss": 1.4961, + "step": 610000 + }, + { + "epoch": 8.35, + "learning_rate": 2.9135623573156894e-05, + "loss": 1.4854, + "step": 610500 + }, + { + "epoch": 8.35, + "learning_rate": 2.9118535631775368e-05, + "loss": 1.4725, + "step": 611000 + }, + { + "epoch": 8.36, + "learning_rate": 2.9101447690393845e-05, + "loss": 1.5003, + "step": 611500 + }, + { + "epoch": 8.37, + "learning_rate": 2.908435974901232e-05, + "loss": 1.5201, + "step": 612000 + }, + { + "epoch": 8.37, + "learning_rate": 2.906727180763079e-05, + "loss": 1.4893, + "step": 612500 + }, + { + "epoch": 8.38, + "learning_rate": 2.9050183866249264e-05, + "loss": 1.4777, + "step": 613000 + }, + { + "epoch": 8.39, + "learning_rate": 2.903309592486774e-05, + "loss": 1.4732, + "step": 613500 + }, + { + "epoch": 8.39, + "learning_rate": 2.9016007983486215e-05, + "loss": 1.5074, + "step": 614000 + }, + { + "epoch": 8.4, + "learning_rate": 2.899892004210469e-05, + "loss": 1.4759, + "step": 614500 + }, + { + "epoch": 8.41, + "learning_rate": 2.898183210072316e-05, + "loss": 1.5268, + "step": 615000 + }, + { + "epoch": 8.41, + "learning_rate": 2.896474415934164e-05, + "loss": 1.4667, + "step": 615500 + }, + { + "epoch": 8.42, + "learning_rate": 2.894765621796011e-05, + "loss": 1.4969, + "step": 616000 + }, + { + "epoch": 8.43, + "learning_rate": 2.8930568276578585e-05, + "loss": 1.5058, + "step": 616500 + }, + { + "epoch": 8.43, + "learning_rate": 2.891348033519706e-05, + "loss": 1.4504, + "step": 617000 + }, + { + "epoch": 8.44, + "learning_rate": 2.8896392393815536e-05, + "loss": 1.4885, + "step": 617500 + }, + { + "epoch": 8.45, + "learning_rate": 2.887930445243401e-05, + "loss": 1.4745, + "step": 618000 + }, + { + "epoch": 8.46, + "learning_rate": 2.886221651105248e-05, + "loss": 1.5071, + "step": 618500 + }, + { + "epoch": 8.46, + "learning_rate": 2.8845128569670954e-05, + "loss": 1.4908, + "step": 619000 + }, + { + "epoch": 8.47, + "learning_rate": 2.882804062828943e-05, + "loss": 1.4772, + "step": 619500 + }, + { + "epoch": 8.48, + "learning_rate": 2.8810952686907905e-05, + "loss": 1.4896, + "step": 620000 + }, + { + "epoch": 8.48, + "learning_rate": 2.879386474552638e-05, + "loss": 1.4895, + "step": 620500 + }, + { + "epoch": 8.49, + "learning_rate": 2.877677680414485e-05, + "loss": 1.4765, + "step": 621000 + }, + { + "epoch": 8.5, + "learning_rate": 2.8759688862763327e-05, + "loss": 1.5223, + "step": 621500 + }, + { + "epoch": 8.5, + "learning_rate": 2.87426009213818e-05, + "loss": 1.4831, + "step": 622000 + }, + { + "epoch": 8.51, + "learning_rate": 2.8725512980000275e-05, + "loss": 1.5148, + "step": 622500 + }, + { + "epoch": 8.52, + "learning_rate": 2.8708425038618746e-05, + "loss": 1.4784, + "step": 623000 + }, + { + "epoch": 8.52, + "learning_rate": 2.8691337097237226e-05, + "loss": 1.4985, + "step": 623500 + }, + { + "epoch": 8.53, + "learning_rate": 2.8674249155855697e-05, + "loss": 1.5064, + "step": 624000 + }, + { + "epoch": 8.54, + "learning_rate": 2.865716121447417e-05, + "loss": 1.4928, + "step": 624500 + }, + { + "epoch": 8.54, + "learning_rate": 2.8640073273092645e-05, + "loss": 1.4865, + "step": 625000 + }, + { + "epoch": 8.55, + "learning_rate": 2.8622985331711122e-05, + "loss": 1.4454, + "step": 625500 + }, + { + "epoch": 8.56, + "learning_rate": 2.8605897390329596e-05, + "loss": 1.4874, + "step": 626000 + }, + { + "epoch": 8.56, + "learning_rate": 2.8588809448948066e-05, + "loss": 1.5061, + "step": 626500 + }, + { + "epoch": 8.57, + "learning_rate": 2.857172150756654e-05, + "loss": 1.4841, + "step": 627000 + }, + { + "epoch": 8.58, + "learning_rate": 2.8554633566185018e-05, + "loss": 1.5026, + "step": 627500 + }, + { + "epoch": 8.58, + "learning_rate": 2.853754562480349e-05, + "loss": 1.5028, + "step": 628000 + }, + { + "epoch": 8.59, + "learning_rate": 2.8520457683421965e-05, + "loss": 1.476, + "step": 628500 + }, + { + "epoch": 8.6, + "learning_rate": 2.8503369742040436e-05, + "loss": 1.4771, + "step": 629000 + }, + { + "epoch": 8.61, + "learning_rate": 2.8486281800658917e-05, + "loss": 1.4767, + "step": 629500 + }, + { + "epoch": 8.61, + "learning_rate": 2.8469193859277387e-05, + "loss": 1.5119, + "step": 630000 + }, + { + "epoch": 8.62, + "learning_rate": 2.845210591789586e-05, + "loss": 1.5151, + "step": 630500 + }, + { + "epoch": 8.63, + "learning_rate": 2.843501797651433e-05, + "loss": 1.5125, + "step": 631000 + }, + { + "epoch": 8.63, + "learning_rate": 2.8417930035132812e-05, + "loss": 1.4958, + "step": 631500 + }, + { + "epoch": 8.64, + "learning_rate": 2.8400842093751283e-05, + "loss": 1.5199, + "step": 632000 + }, + { + "epoch": 8.65, + "learning_rate": 2.8383754152369757e-05, + "loss": 1.4864, + "step": 632500 + }, + { + "epoch": 8.65, + "learning_rate": 2.836666621098823e-05, + "loss": 1.4854, + "step": 633000 + }, + { + "epoch": 8.66, + "learning_rate": 2.8349578269606708e-05, + "loss": 1.4798, + "step": 633500 + }, + { + "epoch": 8.67, + "learning_rate": 2.8332490328225182e-05, + "loss": 1.5015, + "step": 634000 + }, + { + "epoch": 8.67, + "learning_rate": 2.8315402386843652e-05, + "loss": 1.4907, + "step": 634500 + }, + { + "epoch": 8.68, + "learning_rate": 2.8298314445462126e-05, + "loss": 1.4988, + "step": 635000 + }, + { + "epoch": 8.69, + "learning_rate": 2.82812265040806e-05, + "loss": 1.4921, + "step": 635500 + }, + { + "epoch": 8.69, + "learning_rate": 2.8264138562699078e-05, + "loss": 1.486, + "step": 636000 + }, + { + "epoch": 8.7, + "learning_rate": 2.824705062131755e-05, + "loss": 1.481, + "step": 636500 + }, + { + "epoch": 8.71, + "learning_rate": 2.8229962679936022e-05, + "loss": 1.4972, + "step": 637000 + }, + { + "epoch": 8.71, + "learning_rate": 2.8212874738554496e-05, + "loss": 1.4883, + "step": 637500 + }, + { + "epoch": 8.72, + "learning_rate": 2.8195786797172973e-05, + "loss": 1.5092, + "step": 638000 + }, + { + "epoch": 8.73, + "learning_rate": 2.8178698855791447e-05, + "loss": 1.4981, + "step": 638500 + }, + { + "epoch": 8.74, + "learning_rate": 2.816161091440992e-05, + "loss": 1.5086, + "step": 639000 + }, + { + "epoch": 8.74, + "learning_rate": 2.814452297302839e-05, + "loss": 1.4738, + "step": 639500 + }, + { + "epoch": 8.75, + "learning_rate": 2.812743503164687e-05, + "loss": 1.4877, + "step": 640000 + }, + { + "epoch": 8.76, + "learning_rate": 2.8110347090265343e-05, + "loss": 1.5222, + "step": 640500 + }, + { + "epoch": 8.76, + "learning_rate": 2.8093259148883817e-05, + "loss": 1.4875, + "step": 641000 + }, + { + "epoch": 8.77, + "learning_rate": 2.8076171207502287e-05, + "loss": 1.4824, + "step": 641500 + }, + { + "epoch": 8.78, + "learning_rate": 2.8059083266120768e-05, + "loss": 1.497, + "step": 642000 + }, + { + "epoch": 8.78, + "learning_rate": 2.804199532473924e-05, + "loss": 1.4865, + "step": 642500 + }, + { + "epoch": 8.79, + "learning_rate": 2.8024907383357712e-05, + "loss": 1.5044, + "step": 643000 + }, + { + "epoch": 8.8, + "learning_rate": 2.8007819441976186e-05, + "loss": 1.4704, + "step": 643500 + }, + { + "epoch": 8.8, + "learning_rate": 2.7990731500594664e-05, + "loss": 1.4811, + "step": 644000 + }, + { + "epoch": 8.81, + "learning_rate": 2.7973643559213137e-05, + "loss": 1.4861, + "step": 644500 + }, + { + "epoch": 8.82, + "learning_rate": 2.7956555617831608e-05, + "loss": 1.4989, + "step": 645000 + }, + { + "epoch": 8.82, + "learning_rate": 2.7939467676450082e-05, + "loss": 1.4845, + "step": 645500 + }, + { + "epoch": 8.83, + "learning_rate": 2.792237973506856e-05, + "loss": 1.5079, + "step": 646000 + }, + { + "epoch": 8.84, + "learning_rate": 2.7905291793687033e-05, + "loss": 1.4834, + "step": 646500 + }, + { + "epoch": 8.84, + "learning_rate": 2.7888203852305507e-05, + "loss": 1.499, + "step": 647000 + }, + { + "epoch": 8.85, + "learning_rate": 2.7871115910923978e-05, + "loss": 1.5081, + "step": 647500 + }, + { + "epoch": 8.86, + "learning_rate": 2.7854027969542458e-05, + "loss": 1.5199, + "step": 648000 + }, + { + "epoch": 8.87, + "learning_rate": 2.783694002816093e-05, + "loss": 1.5139, + "step": 648500 + }, + { + "epoch": 8.87, + "learning_rate": 2.7819852086779403e-05, + "loss": 1.4496, + "step": 649000 + }, + { + "epoch": 8.88, + "learning_rate": 2.7802764145397873e-05, + "loss": 1.4594, + "step": 649500 + }, + { + "epoch": 8.89, + "learning_rate": 2.7785676204016354e-05, + "loss": 1.5356, + "step": 650000 + }, + { + "epoch": 8.89, + "learning_rate": 2.7768588262634824e-05, + "loss": 1.4932, + "step": 650500 + }, + { + "epoch": 8.9, + "learning_rate": 2.77515003212533e-05, + "loss": 1.491, + "step": 651000 + }, + { + "epoch": 8.91, + "learning_rate": 2.7734412379871772e-05, + "loss": 1.4971, + "step": 651500 + }, + { + "epoch": 8.91, + "learning_rate": 2.771732443849025e-05, + "loss": 1.4776, + "step": 652000 + }, + { + "epoch": 8.92, + "learning_rate": 2.7700236497108724e-05, + "loss": 1.5026, + "step": 652500 + }, + { + "epoch": 8.93, + "learning_rate": 2.7683148555727194e-05, + "loss": 1.4995, + "step": 653000 + }, + { + "epoch": 8.93, + "learning_rate": 2.7666060614345668e-05, + "loss": 1.4945, + "step": 653500 + }, + { + "epoch": 8.94, + "learning_rate": 2.7648972672964145e-05, + "loss": 1.536, + "step": 654000 + }, + { + "epoch": 8.95, + "learning_rate": 2.763188473158262e-05, + "loss": 1.5035, + "step": 654500 + }, + { + "epoch": 8.95, + "learning_rate": 2.7614796790201093e-05, + "loss": 1.4934, + "step": 655000 + }, + { + "epoch": 8.96, + "learning_rate": 2.7597708848819564e-05, + "loss": 1.5146, + "step": 655500 + }, + { + "epoch": 8.97, + "learning_rate": 2.7580620907438044e-05, + "loss": 1.4982, + "step": 656000 + }, + { + "epoch": 8.97, + "learning_rate": 2.7563532966056515e-05, + "loss": 1.4556, + "step": 656500 + }, + { + "epoch": 8.98, + "learning_rate": 2.754644502467499e-05, + "loss": 1.4727, + "step": 657000 + }, + { + "epoch": 8.99, + "learning_rate": 2.7529357083293463e-05, + "loss": 1.478, + "step": 657500 + }, + { + "epoch": 9.0, + "learning_rate": 2.751226914191194e-05, + "loss": 1.4869, + "step": 658000 + }, + { + "epoch": 9.0, + "learning_rate": 2.749518120053041e-05, + "loss": 1.5087, + "step": 658500 + }, + { + "epoch": 9.01, + "learning_rate": 2.7478093259148884e-05, + "loss": 1.4459, + "step": 659000 + }, + { + "epoch": 9.02, + "learning_rate": 2.746100531776736e-05, + "loss": 1.4671, + "step": 659500 + }, + { + "epoch": 9.02, + "learning_rate": 2.7443917376385836e-05, + "loss": 1.4956, + "step": 660000 + }, + { + "epoch": 9.03, + "learning_rate": 2.742682943500431e-05, + "loss": 1.4879, + "step": 660500 + }, + { + "epoch": 9.04, + "learning_rate": 2.740974149362278e-05, + "loss": 1.482, + "step": 661000 + }, + { + "epoch": 9.04, + "learning_rate": 2.7392653552241254e-05, + "loss": 1.4711, + "step": 661500 + }, + { + "epoch": 9.05, + "learning_rate": 2.737556561085973e-05, + "loss": 1.4622, + "step": 662000 + }, + { + "epoch": 9.06, + "learning_rate": 2.7358477669478205e-05, + "loss": 1.4777, + "step": 662500 + }, + { + "epoch": 9.06, + "learning_rate": 2.734138972809668e-05, + "loss": 1.4635, + "step": 663000 + }, + { + "epoch": 9.07, + "learning_rate": 2.732430178671515e-05, + "loss": 1.478, + "step": 663500 + }, + { + "epoch": 9.08, + "learning_rate": 2.730721384533363e-05, + "loss": 1.4862, + "step": 664000 + }, + { + "epoch": 9.08, + "learning_rate": 2.72901259039521e-05, + "loss": 1.4822, + "step": 664500 + }, + { + "epoch": 9.09, + "learning_rate": 2.7273037962570575e-05, + "loss": 1.4666, + "step": 665000 + }, + { + "epoch": 9.1, + "learning_rate": 2.725595002118905e-05, + "loss": 1.4881, + "step": 665500 + }, + { + "epoch": 9.1, + "learning_rate": 2.723886207980752e-05, + "loss": 1.4811, + "step": 666000 + }, + { + "epoch": 9.11, + "learning_rate": 2.7221774138426e-05, + "loss": 1.4577, + "step": 666500 + }, + { + "epoch": 9.12, + "learning_rate": 2.720468619704447e-05, + "loss": 1.4929, + "step": 667000 + }, + { + "epoch": 9.12, + "learning_rate": 2.7187598255662944e-05, + "loss": 1.4764, + "step": 667500 + }, + { + "epoch": 9.13, + "learning_rate": 2.7170510314281415e-05, + "loss": 1.4601, + "step": 668000 + }, + { + "epoch": 9.14, + "learning_rate": 2.7153422372899896e-05, + "loss": 1.4539, + "step": 668500 + }, + { + "epoch": 9.15, + "learning_rate": 2.7136334431518366e-05, + "loss": 1.5083, + "step": 669000 + }, + { + "epoch": 9.15, + "learning_rate": 2.711924649013684e-05, + "loss": 1.4844, + "step": 669500 + }, + { + "epoch": 9.16, + "learning_rate": 2.7102158548755314e-05, + "loss": 1.4702, + "step": 670000 + }, + { + "epoch": 9.17, + "learning_rate": 2.708507060737379e-05, + "loss": 1.476, + "step": 670500 + }, + { + "epoch": 9.17, + "learning_rate": 2.7067982665992265e-05, + "loss": 1.4659, + "step": 671000 + }, + { + "epoch": 9.18, + "learning_rate": 2.7050894724610736e-05, + "loss": 1.4655, + "step": 671500 + }, + { + "epoch": 9.19, + "learning_rate": 2.703380678322921e-05, + "loss": 1.4903, + "step": 672000 + }, + { + "epoch": 9.19, + "learning_rate": 2.7016718841847687e-05, + "loss": 1.4695, + "step": 672500 + }, + { + "epoch": 9.2, + "learning_rate": 2.699963090046616e-05, + "loss": 1.4484, + "step": 673000 + }, + { + "epoch": 9.21, + "learning_rate": 2.6982542959084635e-05, + "loss": 1.4799, + "step": 673500 + }, + { + "epoch": 9.21, + "learning_rate": 2.6965455017703105e-05, + "loss": 1.4826, + "step": 674000 + }, + { + "epoch": 9.22, + "learning_rate": 2.6948367076321586e-05, + "loss": 1.4598, + "step": 674500 + }, + { + "epoch": 9.23, + "learning_rate": 2.6931279134940057e-05, + "loss": 1.4494, + "step": 675000 + }, + { + "epoch": 9.23, + "learning_rate": 2.691419119355853e-05, + "loss": 1.4694, + "step": 675500 + }, + { + "epoch": 9.24, + "learning_rate": 2.6897103252177004e-05, + "loss": 1.4724, + "step": 676000 + }, + { + "epoch": 9.25, + "learning_rate": 2.688001531079548e-05, + "loss": 1.4878, + "step": 676500 + }, + { + "epoch": 9.25, + "learning_rate": 2.6862927369413952e-05, + "loss": 1.4612, + "step": 677000 + }, + { + "epoch": 9.26, + "learning_rate": 2.6845839428032426e-05, + "loss": 1.4534, + "step": 677500 + }, + { + "epoch": 9.27, + "learning_rate": 2.68287514866509e-05, + "loss": 1.4807, + "step": 678000 + }, + { + "epoch": 9.28, + "learning_rate": 2.6811663545269377e-05, + "loss": 1.48, + "step": 678500 + }, + { + "epoch": 9.28, + "learning_rate": 2.679457560388785e-05, + "loss": 1.4708, + "step": 679000 + }, + { + "epoch": 9.29, + "learning_rate": 2.6777487662506322e-05, + "loss": 1.4739, + "step": 679500 + }, + { + "epoch": 9.3, + "learning_rate": 2.6760399721124796e-05, + "loss": 1.4664, + "step": 680000 + }, + { + "epoch": 9.3, + "learning_rate": 2.6743311779743273e-05, + "loss": 1.4306, + "step": 680500 + }, + { + "epoch": 9.31, + "learning_rate": 2.6726223838361747e-05, + "loss": 1.4716, + "step": 681000 + }, + { + "epoch": 9.32, + "learning_rate": 2.670913589698022e-05, + "loss": 1.4611, + "step": 681500 + }, + { + "epoch": 9.32, + "learning_rate": 2.669204795559869e-05, + "loss": 1.4693, + "step": 682000 + }, + { + "epoch": 9.33, + "learning_rate": 2.6674960014217172e-05, + "loss": 1.4863, + "step": 682500 + }, + { + "epoch": 9.34, + "learning_rate": 2.6657872072835643e-05, + "loss": 1.4744, + "step": 683000 + }, + { + "epoch": 9.34, + "learning_rate": 2.6640784131454116e-05, + "loss": 1.4623, + "step": 683500 + }, + { + "epoch": 9.35, + "learning_rate": 2.662369619007259e-05, + "loss": 1.4557, + "step": 684000 + }, + { + "epoch": 9.36, + "learning_rate": 2.6606608248691068e-05, + "loss": 1.4559, + "step": 684500 + }, + { + "epoch": 9.36, + "learning_rate": 2.658952030730954e-05, + "loss": 1.4421, + "step": 685000 + }, + { + "epoch": 9.37, + "learning_rate": 2.6572432365928012e-05, + "loss": 1.4704, + "step": 685500 + }, + { + "epoch": 9.38, + "learning_rate": 2.6555344424546486e-05, + "loss": 1.455, + "step": 686000 + }, + { + "epoch": 9.38, + "learning_rate": 2.6538256483164963e-05, + "loss": 1.4648, + "step": 686500 + }, + { + "epoch": 9.39, + "learning_rate": 2.6521168541783437e-05, + "loss": 1.48, + "step": 687000 + }, + { + "epoch": 9.4, + "learning_rate": 2.6504080600401908e-05, + "loss": 1.4584, + "step": 687500 + }, + { + "epoch": 9.41, + "learning_rate": 2.6486992659020382e-05, + "loss": 1.4861, + "step": 688000 + }, + { + "epoch": 9.41, + "learning_rate": 2.646990471763886e-05, + "loss": 1.4904, + "step": 688500 + }, + { + "epoch": 9.42, + "learning_rate": 2.6452816776257333e-05, + "loss": 1.4836, + "step": 689000 + }, + { + "epoch": 9.43, + "learning_rate": 2.6435728834875807e-05, + "loss": 1.4917, + "step": 689500 + }, + { + "epoch": 9.43, + "learning_rate": 2.6418640893494277e-05, + "loss": 1.4744, + "step": 690000 + }, + { + "epoch": 9.44, + "learning_rate": 2.6401552952112758e-05, + "loss": 1.46, + "step": 690500 + }, + { + "epoch": 9.45, + "learning_rate": 2.638446501073123e-05, + "loss": 1.4962, + "step": 691000 + }, + { + "epoch": 9.45, + "learning_rate": 2.6367377069349703e-05, + "loss": 1.4929, + "step": 691500 + }, + { + "epoch": 9.46, + "learning_rate": 2.6350289127968176e-05, + "loss": 1.4657, + "step": 692000 + }, + { + "epoch": 9.47, + "learning_rate": 2.6333201186586654e-05, + "loss": 1.4484, + "step": 692500 + }, + { + "epoch": 9.47, + "learning_rate": 2.6316113245205128e-05, + "loss": 1.4549, + "step": 693000 + }, + { + "epoch": 9.48, + "learning_rate": 2.6299025303823598e-05, + "loss": 1.4841, + "step": 693500 + }, + { + "epoch": 9.49, + "learning_rate": 2.6281937362442072e-05, + "loss": 1.4464, + "step": 694000 + }, + { + "epoch": 9.49, + "learning_rate": 2.626484942106055e-05, + "loss": 1.4748, + "step": 694500 + }, + { + "epoch": 9.5, + "learning_rate": 2.6247761479679023e-05, + "loss": 1.4784, + "step": 695000 + }, + { + "epoch": 9.51, + "learning_rate": 2.6230673538297494e-05, + "loss": 1.4859, + "step": 695500 + }, + { + "epoch": 9.51, + "learning_rate": 2.6213585596915968e-05, + "loss": 1.443, + "step": 696000 + }, + { + "epoch": 9.52, + "learning_rate": 2.619649765553444e-05, + "loss": 1.4743, + "step": 696500 + }, + { + "epoch": 9.53, + "learning_rate": 2.617940971415292e-05, + "loss": 1.4812, + "step": 697000 + }, + { + "epoch": 9.54, + "learning_rate": 2.6162321772771393e-05, + "loss": 1.4608, + "step": 697500 + }, + { + "epoch": 9.54, + "learning_rate": 2.6145233831389863e-05, + "loss": 1.4727, + "step": 698000 + }, + { + "epoch": 9.55, + "learning_rate": 2.6128145890008337e-05, + "loss": 1.4619, + "step": 698500 + }, + { + "epoch": 9.56, + "learning_rate": 2.6111057948626815e-05, + "loss": 1.4781, + "step": 699000 + }, + { + "epoch": 9.56, + "learning_rate": 2.609397000724529e-05, + "loss": 1.4766, + "step": 699500 + }, + { + "epoch": 9.57, + "learning_rate": 2.6076882065863762e-05, + "loss": 1.4792, + "step": 700000 + }, + { + "epoch": 9.58, + "learning_rate": 2.6059794124482233e-05, + "loss": 1.4828, + "step": 700500 + }, + { + "epoch": 9.58, + "learning_rate": 2.6042706183100714e-05, + "loss": 1.4681, + "step": 701000 + }, + { + "epoch": 9.59, + "learning_rate": 2.6025618241719184e-05, + "loss": 1.4731, + "step": 701500 + }, + { + "epoch": 9.6, + "learning_rate": 2.6008530300337658e-05, + "loss": 1.4552, + "step": 702000 + }, + { + "epoch": 9.6, + "learning_rate": 2.5991442358956132e-05, + "loss": 1.4954, + "step": 702500 + }, + { + "epoch": 9.61, + "learning_rate": 2.597435441757461e-05, + "loss": 1.4919, + "step": 703000 + }, + { + "epoch": 9.62, + "learning_rate": 2.5957266476193083e-05, + "loss": 1.4645, + "step": 703500 + }, + { + "epoch": 9.62, + "learning_rate": 2.5940178534811554e-05, + "loss": 1.4499, + "step": 704000 + }, + { + "epoch": 9.63, + "learning_rate": 2.5923090593430028e-05, + "loss": 1.4748, + "step": 704500 + }, + { + "epoch": 9.64, + "learning_rate": 2.5906002652048505e-05, + "loss": 1.459, + "step": 705000 + }, + { + "epoch": 9.64, + "learning_rate": 2.588891471066698e-05, + "loss": 1.4623, + "step": 705500 + }, + { + "epoch": 9.65, + "learning_rate": 2.587182676928545e-05, + "loss": 1.4547, + "step": 706000 + }, + { + "epoch": 9.66, + "learning_rate": 2.5854738827903923e-05, + "loss": 1.4786, + "step": 706500 + }, + { + "epoch": 9.66, + "learning_rate": 2.58376508865224e-05, + "loss": 1.4739, + "step": 707000 + }, + { + "epoch": 9.67, + "learning_rate": 2.5820562945140875e-05, + "loss": 1.4506, + "step": 707500 + }, + { + "epoch": 9.68, + "learning_rate": 2.580347500375935e-05, + "loss": 1.4542, + "step": 708000 + }, + { + "epoch": 9.69, + "learning_rate": 2.578638706237782e-05, + "loss": 1.4389, + "step": 708500 + }, + { + "epoch": 9.69, + "learning_rate": 2.57692991209963e-05, + "loss": 1.4588, + "step": 709000 + }, + { + "epoch": 9.7, + "learning_rate": 2.575221117961477e-05, + "loss": 1.4706, + "step": 709500 + }, + { + "epoch": 9.71, + "learning_rate": 2.5735123238233244e-05, + "loss": 1.471, + "step": 710000 + }, + { + "epoch": 9.71, + "learning_rate": 2.5718035296851718e-05, + "loss": 1.4677, + "step": 710500 + }, + { + "epoch": 9.72, + "learning_rate": 2.5700947355470195e-05, + "loss": 1.4727, + "step": 711000 + }, + { + "epoch": 9.73, + "learning_rate": 2.568385941408867e-05, + "loss": 1.4379, + "step": 711500 + }, + { + "epoch": 9.73, + "learning_rate": 2.566677147270714e-05, + "loss": 1.4929, + "step": 712000 + }, + { + "epoch": 9.74, + "learning_rate": 2.5649683531325614e-05, + "loss": 1.4682, + "step": 712500 + }, + { + "epoch": 9.75, + "learning_rate": 2.563259558994409e-05, + "loss": 1.4787, + "step": 713000 + }, + { + "epoch": 9.75, + "learning_rate": 2.5615507648562565e-05, + "loss": 1.4591, + "step": 713500 + }, + { + "epoch": 9.76, + "learning_rate": 2.559841970718104e-05, + "loss": 1.5071, + "step": 714000 + }, + { + "epoch": 9.77, + "learning_rate": 2.558133176579951e-05, + "loss": 1.4586, + "step": 714500 + }, + { + "epoch": 9.77, + "learning_rate": 2.5564243824417987e-05, + "loss": 1.4619, + "step": 715000 + }, + { + "epoch": 9.78, + "learning_rate": 2.554715588303646e-05, + "loss": 1.4858, + "step": 715500 + }, + { + "epoch": 9.79, + "learning_rate": 2.5530067941654935e-05, + "loss": 1.4724, + "step": 716000 + }, + { + "epoch": 9.79, + "learning_rate": 2.5512980000273405e-05, + "loss": 1.4345, + "step": 716500 + }, + { + "epoch": 9.8, + "learning_rate": 2.5495892058891886e-05, + "loss": 1.4674, + "step": 717000 + }, + { + "epoch": 9.81, + "learning_rate": 2.5478804117510356e-05, + "loss": 1.4863, + "step": 717500 + }, + { + "epoch": 9.82, + "learning_rate": 2.546171617612883e-05, + "loss": 1.4946, + "step": 718000 + }, + { + "epoch": 9.82, + "learning_rate": 2.5444628234747304e-05, + "loss": 1.4588, + "step": 718500 + }, + { + "epoch": 9.83, + "learning_rate": 2.542754029336578e-05, + "loss": 1.4579, + "step": 719000 + }, + { + "epoch": 9.84, + "learning_rate": 2.5410452351984255e-05, + "loss": 1.4757, + "step": 719500 + }, + { + "epoch": 9.84, + "learning_rate": 2.5393364410602726e-05, + "loss": 1.4671, + "step": 720000 + }, + { + "epoch": 9.85, + "learning_rate": 2.53762764692212e-05, + "loss": 1.4597, + "step": 720500 + }, + { + "epoch": 9.86, + "learning_rate": 2.5359188527839677e-05, + "loss": 1.4758, + "step": 721000 + }, + { + "epoch": 9.86, + "learning_rate": 2.534210058645815e-05, + "loss": 1.4617, + "step": 721500 + }, + { + "epoch": 9.87, + "learning_rate": 2.5325012645076625e-05, + "loss": 1.4998, + "step": 722000 + }, + { + "epoch": 9.88, + "learning_rate": 2.5307924703695095e-05, + "loss": 1.4656, + "step": 722500 + }, + { + "epoch": 9.88, + "learning_rate": 2.5290836762313573e-05, + "loss": 1.4736, + "step": 723000 + }, + { + "epoch": 9.89, + "learning_rate": 2.5273748820932047e-05, + "loss": 1.4629, + "step": 723500 + }, + { + "epoch": 9.9, + "learning_rate": 2.525666087955052e-05, + "loss": 1.4405, + "step": 724000 + }, + { + "epoch": 9.9, + "learning_rate": 2.523957293816899e-05, + "loss": 1.4663, + "step": 724500 + }, + { + "epoch": 9.91, + "learning_rate": 2.5222484996787465e-05, + "loss": 1.4757, + "step": 725000 + }, + { + "epoch": 9.92, + "learning_rate": 2.5205397055405942e-05, + "loss": 1.5058, + "step": 725500 + }, + { + "epoch": 9.92, + "learning_rate": 2.5188309114024416e-05, + "loss": 1.4642, + "step": 726000 + }, + { + "epoch": 9.93, + "learning_rate": 2.517122117264289e-05, + "loss": 1.4744, + "step": 726500 + }, + { + "epoch": 9.94, + "learning_rate": 2.515413323126136e-05, + "loss": 1.4967, + "step": 727000 + }, + { + "epoch": 9.95, + "learning_rate": 2.513704528987984e-05, + "loss": 1.4522, + "step": 727500 + }, + { + "epoch": 9.95, + "learning_rate": 2.5119957348498312e-05, + "loss": 1.4532, + "step": 728000 + }, + { + "epoch": 9.96, + "learning_rate": 2.5102869407116786e-05, + "loss": 1.4628, + "step": 728500 + }, + { + "epoch": 9.97, + "learning_rate": 2.508578146573526e-05, + "loss": 1.4559, + "step": 729000 + }, + { + "epoch": 9.97, + "learning_rate": 2.5068693524353737e-05, + "loss": 1.4486, + "step": 729500 + }, + { + "epoch": 9.98, + "learning_rate": 2.505160558297221e-05, + "loss": 1.4469, + "step": 730000 + }, + { + "epoch": 9.99, + "learning_rate": 2.503451764159068e-05, + "loss": 1.4191, + "step": 730500 + }, + { + "epoch": 9.99, + "learning_rate": 2.5017429700209155e-05, + "loss": 1.4698, + "step": 731000 + }, + { + "epoch": 10.0, + "learning_rate": 2.5000341758827633e-05, + "loss": 1.4583, + "step": 731500 + }, + { + "epoch": 10.01, + "learning_rate": 2.4983253817446107e-05, + "loss": 1.4458, + "step": 732000 + }, + { + "epoch": 10.01, + "learning_rate": 2.496616587606458e-05, + "loss": 1.4466, + "step": 732500 + }, + { + "epoch": 10.02, + "learning_rate": 2.4949077934683054e-05, + "loss": 1.4315, + "step": 733000 + }, + { + "epoch": 10.03, + "learning_rate": 2.493198999330153e-05, + "loss": 1.4482, + "step": 733500 + }, + { + "epoch": 10.03, + "learning_rate": 2.4914902051920002e-05, + "loss": 1.4337, + "step": 734000 + }, + { + "epoch": 10.04, + "learning_rate": 2.4897814110538476e-05, + "loss": 1.4641, + "step": 734500 + }, + { + "epoch": 10.05, + "learning_rate": 2.488072616915695e-05, + "loss": 1.4523, + "step": 735000 + }, + { + "epoch": 10.05, + "learning_rate": 2.4863638227775424e-05, + "loss": 1.4392, + "step": 735500 + }, + { + "epoch": 10.06, + "learning_rate": 2.4846550286393898e-05, + "loss": 1.4516, + "step": 736000 + }, + { + "epoch": 10.07, + "learning_rate": 2.4829462345012372e-05, + "loss": 1.4297, + "step": 736500 + }, + { + "epoch": 10.08, + "learning_rate": 2.481237440363085e-05, + "loss": 1.4208, + "step": 737000 + }, + { + "epoch": 10.08, + "learning_rate": 2.479528646224932e-05, + "loss": 1.4294, + "step": 737500 + }, + { + "epoch": 10.09, + "learning_rate": 2.4778198520867797e-05, + "loss": 1.4208, + "step": 738000 + }, + { + "epoch": 10.1, + "learning_rate": 2.4761110579486268e-05, + "loss": 1.4434, + "step": 738500 + }, + { + "epoch": 10.1, + "learning_rate": 2.4744022638104745e-05, + "loss": 1.4261, + "step": 739000 + }, + { + "epoch": 10.11, + "learning_rate": 2.4726934696723215e-05, + "loss": 1.4279, + "step": 739500 + }, + { + "epoch": 10.12, + "learning_rate": 2.4709846755341693e-05, + "loss": 1.455, + "step": 740000 + }, + { + "epoch": 10.12, + "learning_rate": 2.4692758813960167e-05, + "loss": 1.4626, + "step": 740500 + }, + { + "epoch": 10.13, + "learning_rate": 2.467567087257864e-05, + "loss": 1.4422, + "step": 741000 + }, + { + "epoch": 10.14, + "learning_rate": 2.4658582931197114e-05, + "loss": 1.4399, + "step": 741500 + }, + { + "epoch": 10.14, + "learning_rate": 2.464149498981559e-05, + "loss": 1.4176, + "step": 742000 + }, + { + "epoch": 10.15, + "learning_rate": 2.4624407048434062e-05, + "loss": 1.4448, + "step": 742500 + }, + { + "epoch": 10.16, + "learning_rate": 2.4607319107052536e-05, + "loss": 1.4325, + "step": 743000 + }, + { + "epoch": 10.16, + "learning_rate": 2.459023116567101e-05, + "loss": 1.4607, + "step": 743500 + }, + { + "epoch": 10.17, + "learning_rate": 2.4573143224289484e-05, + "loss": 1.4387, + "step": 744000 + }, + { + "epoch": 10.18, + "learning_rate": 2.4556055282907958e-05, + "loss": 1.4423, + "step": 744500 + }, + { + "epoch": 10.18, + "learning_rate": 2.4538967341526435e-05, + "loss": 1.4206, + "step": 745000 + }, + { + "epoch": 10.19, + "learning_rate": 2.4521879400144906e-05, + "loss": 1.4713, + "step": 745500 + }, + { + "epoch": 10.2, + "learning_rate": 2.4504791458763383e-05, + "loss": 1.4698, + "step": 746000 + }, + { + "epoch": 10.2, + "learning_rate": 2.4487703517381854e-05, + "loss": 1.4352, + "step": 746500 + }, + { + "epoch": 10.21, + "learning_rate": 2.447061557600033e-05, + "loss": 1.443, + "step": 747000 + }, + { + "epoch": 10.22, + "learning_rate": 2.44535276346188e-05, + "loss": 1.4518, + "step": 747500 + }, + { + "epoch": 10.23, + "learning_rate": 2.443643969323728e-05, + "loss": 1.4619, + "step": 748000 + }, + { + "epoch": 10.23, + "learning_rate": 2.4419351751855753e-05, + "loss": 1.406, + "step": 748500 + }, + { + "epoch": 10.24, + "learning_rate": 2.4402263810474227e-05, + "loss": 1.4557, + "step": 749000 + }, + { + "epoch": 10.25, + "learning_rate": 2.43851758690927e-05, + "loss": 1.477, + "step": 749500 + }, + { + "epoch": 10.25, + "learning_rate": 2.4368087927711174e-05, + "loss": 1.4519, + "step": 750000 + }, + { + "epoch": 10.26, + "learning_rate": 2.435099998632965e-05, + "loss": 1.4586, + "step": 750500 + }, + { + "epoch": 10.27, + "learning_rate": 2.4333912044948122e-05, + "loss": 1.4505, + "step": 751000 + }, + { + "epoch": 10.27, + "learning_rate": 2.4316824103566596e-05, + "loss": 1.4307, + "step": 751500 + }, + { + "epoch": 10.28, + "learning_rate": 2.429973616218507e-05, + "loss": 1.4761, + "step": 752000 + }, + { + "epoch": 10.29, + "learning_rate": 2.4282648220803544e-05, + "loss": 1.4639, + "step": 752500 + }, + { + "epoch": 10.29, + "learning_rate": 2.426556027942202e-05, + "loss": 1.4433, + "step": 753000 + }, + { + "epoch": 10.3, + "learning_rate": 2.4248472338040492e-05, + "loss": 1.4655, + "step": 753500 + }, + { + "epoch": 10.31, + "learning_rate": 2.423138439665897e-05, + "loss": 1.4583, + "step": 754000 + }, + { + "epoch": 10.31, + "learning_rate": 2.421429645527744e-05, + "loss": 1.4314, + "step": 754500 + }, + { + "epoch": 10.32, + "learning_rate": 2.4197208513895917e-05, + "loss": 1.4385, + "step": 755000 + }, + { + "epoch": 10.33, + "learning_rate": 2.418012057251439e-05, + "loss": 1.443, + "step": 755500 + }, + { + "epoch": 10.33, + "learning_rate": 2.416303263113286e-05, + "loss": 1.4263, + "step": 756000 + }, + { + "epoch": 10.34, + "learning_rate": 2.414594468975134e-05, + "loss": 1.4422, + "step": 756500 + }, + { + "epoch": 10.35, + "learning_rate": 2.412885674836981e-05, + "loss": 1.4477, + "step": 757000 + }, + { + "epoch": 10.36, + "learning_rate": 2.4111768806988287e-05, + "loss": 1.4592, + "step": 757500 + }, + { + "epoch": 10.36, + "learning_rate": 2.4094680865606757e-05, + "loss": 1.4432, + "step": 758000 + }, + { + "epoch": 10.37, + "learning_rate": 2.4077592924225234e-05, + "loss": 1.4244, + "step": 758500 + }, + { + "epoch": 10.38, + "learning_rate": 2.4060504982843708e-05, + "loss": 1.4251, + "step": 759000 + }, + { + "epoch": 10.38, + "learning_rate": 2.4043417041462182e-05, + "loss": 1.4401, + "step": 759500 + }, + { + "epoch": 10.39, + "learning_rate": 2.4026329100080656e-05, + "loss": 1.4411, + "step": 760000 + }, + { + "epoch": 10.4, + "learning_rate": 2.400924115869913e-05, + "loss": 1.395, + "step": 760500 + }, + { + "epoch": 10.4, + "learning_rate": 2.3992153217317604e-05, + "loss": 1.4492, + "step": 761000 + }, + { + "epoch": 10.41, + "learning_rate": 2.3975065275936078e-05, + "loss": 1.4324, + "step": 761500 + }, + { + "epoch": 10.42, + "learning_rate": 2.3957977334554552e-05, + "loss": 1.434, + "step": 762000 + }, + { + "epoch": 10.42, + "learning_rate": 2.3940889393173026e-05, + "loss": 1.4603, + "step": 762500 + }, + { + "epoch": 10.43, + "learning_rate": 2.39238014517915e-05, + "loss": 1.4694, + "step": 763000 + }, + { + "epoch": 10.44, + "learning_rate": 2.3906713510409977e-05, + "loss": 1.4796, + "step": 763500 + }, + { + "epoch": 10.44, + "learning_rate": 2.3889625569028447e-05, + "loss": 1.4439, + "step": 764000 + }, + { + "epoch": 10.45, + "learning_rate": 2.3872537627646925e-05, + "loss": 1.4581, + "step": 764500 + }, + { + "epoch": 10.46, + "learning_rate": 2.3855449686265395e-05, + "loss": 1.419, + "step": 765000 + }, + { + "epoch": 10.46, + "learning_rate": 2.3838361744883873e-05, + "loss": 1.468, + "step": 765500 + }, + { + "epoch": 10.47, + "learning_rate": 2.3821273803502343e-05, + "loss": 1.4358, + "step": 766000 + }, + { + "epoch": 10.48, + "learning_rate": 2.380418586212082e-05, + "loss": 1.4254, + "step": 766500 + }, + { + "epoch": 10.49, + "learning_rate": 2.3787097920739294e-05, + "loss": 1.4328, + "step": 767000 + }, + { + "epoch": 10.49, + "learning_rate": 2.3770009979357768e-05, + "loss": 1.4519, + "step": 767500 + }, + { + "epoch": 10.5, + "learning_rate": 2.3752922037976242e-05, + "loss": 1.4651, + "step": 768000 + }, + { + "epoch": 10.51, + "learning_rate": 2.3735834096594716e-05, + "loss": 1.476, + "step": 768500 + }, + { + "epoch": 10.51, + "learning_rate": 2.371874615521319e-05, + "loss": 1.4275, + "step": 769000 + }, + { + "epoch": 10.52, + "learning_rate": 2.3701658213831664e-05, + "loss": 1.4703, + "step": 769500 + }, + { + "epoch": 10.53, + "learning_rate": 2.3684570272450138e-05, + "loss": 1.4432, + "step": 770000 + }, + { + "epoch": 10.53, + "learning_rate": 2.3667482331068612e-05, + "loss": 1.4483, + "step": 770500 + }, + { + "epoch": 10.54, + "learning_rate": 2.3650394389687086e-05, + "loss": 1.4012, + "step": 771000 + }, + { + "epoch": 10.55, + "learning_rate": 2.3633306448305563e-05, + "loss": 1.4423, + "step": 771500 + }, + { + "epoch": 10.55, + "learning_rate": 2.3616218506924033e-05, + "loss": 1.4772, + "step": 772000 + }, + { + "epoch": 10.56, + "learning_rate": 2.359913056554251e-05, + "loss": 1.4243, + "step": 772500 + }, + { + "epoch": 10.57, + "learning_rate": 2.358204262416098e-05, + "loss": 1.4345, + "step": 773000 + }, + { + "epoch": 10.57, + "learning_rate": 2.356495468277946e-05, + "loss": 1.4637, + "step": 773500 + }, + { + "epoch": 10.58, + "learning_rate": 2.3547866741397933e-05, + "loss": 1.4623, + "step": 774000 + }, + { + "epoch": 10.59, + "learning_rate": 2.3530778800016406e-05, + "loss": 1.4311, + "step": 774500 + }, + { + "epoch": 10.59, + "learning_rate": 2.351369085863488e-05, + "loss": 1.4576, + "step": 775000 + }, + { + "epoch": 10.6, + "learning_rate": 2.3496602917253354e-05, + "loss": 1.4253, + "step": 775500 + }, + { + "epoch": 10.61, + "learning_rate": 2.3479514975871828e-05, + "loss": 1.4507, + "step": 776000 + }, + { + "epoch": 10.62, + "learning_rate": 2.3462427034490302e-05, + "loss": 1.4421, + "step": 776500 + }, + { + "epoch": 10.62, + "learning_rate": 2.3445339093108776e-05, + "loss": 1.4613, + "step": 777000 + }, + { + "epoch": 10.63, + "learning_rate": 2.342825115172725e-05, + "loss": 1.4451, + "step": 777500 + }, + { + "epoch": 10.64, + "learning_rate": 2.3411163210345724e-05, + "loss": 1.4364, + "step": 778000 + }, + { + "epoch": 10.64, + "learning_rate": 2.33940752689642e-05, + "loss": 1.4348, + "step": 778500 + }, + { + "epoch": 10.65, + "learning_rate": 2.337698732758267e-05, + "loss": 1.4115, + "step": 779000 + }, + { + "epoch": 10.66, + "learning_rate": 2.335989938620115e-05, + "loss": 1.4101, + "step": 779500 + }, + { + "epoch": 10.66, + "learning_rate": 2.334281144481962e-05, + "loss": 1.4391, + "step": 780000 + }, + { + "epoch": 10.67, + "learning_rate": 2.3325723503438097e-05, + "loss": 1.4519, + "step": 780500 + }, + { + "epoch": 10.68, + "learning_rate": 2.3308635562056567e-05, + "loss": 1.4354, + "step": 781000 + }, + { + "epoch": 10.68, + "learning_rate": 2.3291547620675045e-05, + "loss": 1.4378, + "step": 781500 + }, + { + "epoch": 10.69, + "learning_rate": 2.327445967929352e-05, + "loss": 1.4608, + "step": 782000 + }, + { + "epoch": 10.7, + "learning_rate": 2.3257371737911992e-05, + "loss": 1.4468, + "step": 782500 + }, + { + "epoch": 10.7, + "learning_rate": 2.3240283796530466e-05, + "loss": 1.4298, + "step": 783000 + }, + { + "epoch": 10.71, + "learning_rate": 2.322319585514894e-05, + "loss": 1.4259, + "step": 783500 + }, + { + "epoch": 10.72, + "learning_rate": 2.3206107913767414e-05, + "loss": 1.4224, + "step": 784000 + }, + { + "epoch": 10.72, + "learning_rate": 2.3189019972385888e-05, + "loss": 1.4356, + "step": 784500 + }, + { + "epoch": 10.73, + "learning_rate": 2.3171932031004362e-05, + "loss": 1.4321, + "step": 785000 + }, + { + "epoch": 10.74, + "learning_rate": 2.3154844089622836e-05, + "loss": 1.4348, + "step": 785500 + }, + { + "epoch": 10.74, + "learning_rate": 2.313775614824131e-05, + "loss": 1.4787, + "step": 786000 + }, + { + "epoch": 10.75, + "learning_rate": 2.3120668206859784e-05, + "loss": 1.4289, + "step": 786500 + }, + { + "epoch": 10.76, + "learning_rate": 2.3103580265478258e-05, + "loss": 1.4303, + "step": 787000 + }, + { + "epoch": 10.77, + "learning_rate": 2.308649232409673e-05, + "loss": 1.4241, + "step": 787500 + }, + { + "epoch": 10.77, + "learning_rate": 2.3069404382715206e-05, + "loss": 1.4187, + "step": 788000 + }, + { + "epoch": 10.78, + "learning_rate": 2.305231644133368e-05, + "loss": 1.4354, + "step": 788500 + }, + { + "epoch": 10.79, + "learning_rate": 2.3035228499952153e-05, + "loss": 1.4168, + "step": 789000 + }, + { + "epoch": 10.79, + "learning_rate": 2.3018140558570627e-05, + "loss": 1.4418, + "step": 789500 + }, + { + "epoch": 10.8, + "learning_rate": 2.3001052617189105e-05, + "loss": 1.4588, + "step": 790000 + }, + { + "epoch": 10.81, + "learning_rate": 2.2983964675807575e-05, + "loss": 1.445, + "step": 790500 + }, + { + "epoch": 10.81, + "learning_rate": 2.2966876734426052e-05, + "loss": 1.4398, + "step": 791000 + }, + { + "epoch": 10.82, + "learning_rate": 2.2949788793044523e-05, + "loss": 1.4615, + "step": 791500 + }, + { + "epoch": 10.83, + "learning_rate": 2.2932700851663e-05, + "loss": 1.4624, + "step": 792000 + }, + { + "epoch": 10.83, + "learning_rate": 2.2915612910281474e-05, + "loss": 1.4255, + "step": 792500 + }, + { + "epoch": 10.84, + "learning_rate": 2.2898524968899948e-05, + "loss": 1.4596, + "step": 793000 + }, + { + "epoch": 10.85, + "learning_rate": 2.2881437027518422e-05, + "loss": 1.4006, + "step": 793500 + }, + { + "epoch": 10.85, + "learning_rate": 2.2864349086136896e-05, + "loss": 1.4482, + "step": 794000 + }, + { + "epoch": 10.86, + "learning_rate": 2.284726114475537e-05, + "loss": 1.4319, + "step": 794500 + }, + { + "epoch": 10.87, + "learning_rate": 2.2830173203373844e-05, + "loss": 1.4492, + "step": 795000 + }, + { + "epoch": 10.87, + "learning_rate": 2.2813085261992318e-05, + "loss": 1.459, + "step": 795500 + }, + { + "epoch": 10.88, + "learning_rate": 2.279599732061079e-05, + "loss": 1.4415, + "step": 796000 + }, + { + "epoch": 10.89, + "learning_rate": 2.2778909379229266e-05, + "loss": 1.4218, + "step": 796500 + }, + { + "epoch": 10.9, + "learning_rate": 2.2761821437847743e-05, + "loss": 1.4431, + "step": 797000 + }, + { + "epoch": 10.9, + "learning_rate": 2.2744733496466213e-05, + "loss": 1.441, + "step": 797500 + }, + { + "epoch": 10.91, + "learning_rate": 2.272764555508469e-05, + "loss": 1.4859, + "step": 798000 + }, + { + "epoch": 10.92, + "learning_rate": 2.271055761370316e-05, + "loss": 1.407, + "step": 798500 + }, + { + "epoch": 10.92, + "learning_rate": 2.269346967232164e-05, + "loss": 1.4517, + "step": 799000 + }, + { + "epoch": 10.93, + "learning_rate": 2.267638173094011e-05, + "loss": 1.4544, + "step": 799500 + }, + { + "epoch": 10.94, + "learning_rate": 2.2659293789558586e-05, + "loss": 1.4431, + "step": 800000 + }, + { + "epoch": 10.94, + "learning_rate": 2.264220584817706e-05, + "loss": 1.4242, + "step": 800500 + }, + { + "epoch": 10.95, + "learning_rate": 2.2625117906795534e-05, + "loss": 1.4466, + "step": 801000 + }, + { + "epoch": 10.96, + "learning_rate": 2.2608029965414008e-05, + "loss": 1.4097, + "step": 801500 + }, + { + "epoch": 10.96, + "learning_rate": 2.2590942024032482e-05, + "loss": 1.4563, + "step": 802000 + }, + { + "epoch": 10.97, + "learning_rate": 2.2573854082650956e-05, + "loss": 1.4416, + "step": 802500 + }, + { + "epoch": 10.98, + "learning_rate": 2.255676614126943e-05, + "loss": 1.4184, + "step": 803000 + }, + { + "epoch": 10.98, + "learning_rate": 2.2539678199887904e-05, + "loss": 1.4515, + "step": 803500 + }, + { + "epoch": 10.99, + "learning_rate": 2.2522590258506378e-05, + "loss": 1.4244, + "step": 804000 + }, + { + "epoch": 11.0, + "learning_rate": 2.250550231712485e-05, + "loss": 1.4086, + "step": 804500 + }, + { + "epoch": 11.0, + "learning_rate": 2.248841437574333e-05, + "loss": 1.4123, + "step": 805000 + }, + { + "epoch": 11.01, + "learning_rate": 2.24713264343618e-05, + "loss": 1.422, + "step": 805500 + }, + { + "epoch": 11.02, + "learning_rate": 2.2454238492980277e-05, + "loss": 1.4184, + "step": 806000 + }, + { + "epoch": 11.03, + "learning_rate": 2.2437150551598747e-05, + "loss": 1.405, + "step": 806500 + }, + { + "epoch": 11.03, + "learning_rate": 2.2420062610217225e-05, + "loss": 1.4136, + "step": 807000 + }, + { + "epoch": 11.04, + "learning_rate": 2.2402974668835695e-05, + "loss": 1.441, + "step": 807500 + }, + { + "epoch": 11.05, + "learning_rate": 2.2385886727454172e-05, + "loss": 1.4193, + "step": 808000 + }, + { + "epoch": 11.05, + "learning_rate": 2.2368798786072646e-05, + "loss": 1.4295, + "step": 808500 + }, + { + "epoch": 11.06, + "learning_rate": 2.235171084469112e-05, + "loss": 1.3932, + "step": 809000 + }, + { + "epoch": 11.07, + "learning_rate": 2.2334622903309594e-05, + "loss": 1.4045, + "step": 809500 + }, + { + "epoch": 11.07, + "learning_rate": 2.2317534961928068e-05, + "loss": 1.4187, + "step": 810000 + }, + { + "epoch": 11.08, + "learning_rate": 2.2300447020546542e-05, + "loss": 1.4372, + "step": 810500 + }, + { + "epoch": 11.09, + "learning_rate": 2.2283359079165016e-05, + "loss": 1.4185, + "step": 811000 + }, + { + "epoch": 11.09, + "learning_rate": 2.226627113778349e-05, + "loss": 1.4155, + "step": 811500 + }, + { + "epoch": 11.1, + "learning_rate": 2.2249183196401964e-05, + "loss": 1.4103, + "step": 812000 + }, + { + "epoch": 11.11, + "learning_rate": 2.2232095255020438e-05, + "loss": 1.4021, + "step": 812500 + }, + { + "epoch": 11.11, + "learning_rate": 2.2215007313638915e-05, + "loss": 1.404, + "step": 813000 + }, + { + "epoch": 11.12, + "learning_rate": 2.2197919372257385e-05, + "loss": 1.4458, + "step": 813500 + }, + { + "epoch": 11.13, + "learning_rate": 2.2180831430875863e-05, + "loss": 1.427, + "step": 814000 + }, + { + "epoch": 11.13, + "learning_rate": 2.2163743489494333e-05, + "loss": 1.441, + "step": 814500 + }, + { + "epoch": 11.14, + "learning_rate": 2.214665554811281e-05, + "loss": 1.4101, + "step": 815000 + }, + { + "epoch": 11.15, + "learning_rate": 2.2129567606731284e-05, + "loss": 1.4242, + "step": 815500 + }, + { + "epoch": 11.16, + "learning_rate": 2.2112479665349755e-05, + "loss": 1.4189, + "step": 816000 + }, + { + "epoch": 11.16, + "learning_rate": 2.2095391723968232e-05, + "loss": 1.4189, + "step": 816500 + }, + { + "epoch": 11.17, + "learning_rate": 2.2078303782586703e-05, + "loss": 1.4285, + "step": 817000 + }, + { + "epoch": 11.18, + "learning_rate": 2.206121584120518e-05, + "loss": 1.4388, + "step": 817500 + }, + { + "epoch": 11.18, + "learning_rate": 2.204412789982365e-05, + "loss": 1.43, + "step": 818000 + }, + { + "epoch": 11.19, + "learning_rate": 2.2027039958442128e-05, + "loss": 1.398, + "step": 818500 + }, + { + "epoch": 11.2, + "learning_rate": 2.2009952017060602e-05, + "loss": 1.4299, + "step": 819000 + }, + { + "epoch": 11.2, + "learning_rate": 2.1992864075679076e-05, + "loss": 1.4006, + "step": 819500 + }, + { + "epoch": 11.21, + "learning_rate": 2.197577613429755e-05, + "loss": 1.423, + "step": 820000 + }, + { + "epoch": 11.22, + "learning_rate": 2.1958688192916024e-05, + "loss": 1.4344, + "step": 820500 + }, + { + "epoch": 11.22, + "learning_rate": 2.1941600251534498e-05, + "loss": 1.3991, + "step": 821000 + }, + { + "epoch": 11.23, + "learning_rate": 2.192451231015297e-05, + "loss": 1.4235, + "step": 821500 + }, + { + "epoch": 11.24, + "learning_rate": 2.1907424368771445e-05, + "loss": 1.4361, + "step": 822000 + }, + { + "epoch": 11.24, + "learning_rate": 2.189033642738992e-05, + "loss": 1.41, + "step": 822500 + }, + { + "epoch": 11.25, + "learning_rate": 2.1873248486008393e-05, + "loss": 1.4346, + "step": 823000 + }, + { + "epoch": 11.26, + "learning_rate": 2.185616054462687e-05, + "loss": 1.3958, + "step": 823500 + }, + { + "epoch": 11.26, + "learning_rate": 2.183907260324534e-05, + "loss": 1.4485, + "step": 824000 + }, + { + "epoch": 11.27, + "learning_rate": 2.182198466186382e-05, + "loss": 1.4246, + "step": 824500 + }, + { + "epoch": 11.28, + "learning_rate": 2.180489672048229e-05, + "loss": 1.4338, + "step": 825000 + }, + { + "epoch": 11.28, + "learning_rate": 2.1787808779100766e-05, + "loss": 1.4244, + "step": 825500 + }, + { + "epoch": 11.29, + "learning_rate": 2.177072083771924e-05, + "loss": 1.426, + "step": 826000 + }, + { + "epoch": 11.3, + "learning_rate": 2.1753632896337714e-05, + "loss": 1.4114, + "step": 826500 + }, + { + "epoch": 11.31, + "learning_rate": 2.1736544954956188e-05, + "loss": 1.4018, + "step": 827000 + }, + { + "epoch": 11.31, + "learning_rate": 2.1719457013574662e-05, + "loss": 1.4223, + "step": 827500 + }, + { + "epoch": 11.32, + "learning_rate": 2.1702369072193136e-05, + "loss": 1.4516, + "step": 828000 + }, + { + "epoch": 11.33, + "learning_rate": 2.168528113081161e-05, + "loss": 1.4289, + "step": 828500 + }, + { + "epoch": 11.33, + "learning_rate": 2.1668193189430084e-05, + "loss": 1.4107, + "step": 829000 + }, + { + "epoch": 11.34, + "learning_rate": 2.1651105248048558e-05, + "loss": 1.4264, + "step": 829500 + }, + { + "epoch": 11.35, + "learning_rate": 2.163401730666703e-05, + "loss": 1.4082, + "step": 830000 + }, + { + "epoch": 11.35, + "learning_rate": 2.161692936528551e-05, + "loss": 1.4112, + "step": 830500 + }, + { + "epoch": 11.36, + "learning_rate": 2.159984142390398e-05, + "loss": 1.3898, + "step": 831000 + }, + { + "epoch": 11.37, + "learning_rate": 2.1582753482522457e-05, + "loss": 1.4113, + "step": 831500 + }, + { + "epoch": 11.37, + "learning_rate": 2.1565665541140927e-05, + "loss": 1.4247, + "step": 832000 + }, + { + "epoch": 11.38, + "learning_rate": 2.1548577599759404e-05, + "loss": 1.4019, + "step": 832500 + }, + { + "epoch": 11.39, + "learning_rate": 2.1531489658377875e-05, + "loss": 1.3879, + "step": 833000 + }, + { + "epoch": 11.39, + "learning_rate": 2.1514401716996352e-05, + "loss": 1.4048, + "step": 833500 + }, + { + "epoch": 11.4, + "learning_rate": 2.1497313775614826e-05, + "loss": 1.4085, + "step": 834000 + }, + { + "epoch": 11.41, + "learning_rate": 2.14802258342333e-05, + "loss": 1.4049, + "step": 834500 + }, + { + "epoch": 11.41, + "learning_rate": 2.1463137892851774e-05, + "loss": 1.4117, + "step": 835000 + }, + { + "epoch": 11.42, + "learning_rate": 2.1446049951470248e-05, + "loss": 1.3849, + "step": 835500 + }, + { + "epoch": 11.43, + "learning_rate": 2.1428962010088722e-05, + "loss": 1.4048, + "step": 836000 + }, + { + "epoch": 11.44, + "learning_rate": 2.1411874068707196e-05, + "loss": 1.4391, + "step": 836500 + }, + { + "epoch": 11.44, + "learning_rate": 2.139478612732567e-05, + "loss": 1.3813, + "step": 837000 + }, + { + "epoch": 11.45, + "learning_rate": 2.1377698185944144e-05, + "loss": 1.42, + "step": 837500 + }, + { + "epoch": 11.46, + "learning_rate": 2.1360610244562617e-05, + "loss": 1.4215, + "step": 838000 + }, + { + "epoch": 11.46, + "learning_rate": 2.1343522303181095e-05, + "loss": 1.4286, + "step": 838500 + }, + { + "epoch": 11.47, + "learning_rate": 2.1326434361799565e-05, + "loss": 1.4019, + "step": 839000 + }, + { + "epoch": 11.48, + "learning_rate": 2.1309346420418043e-05, + "loss": 1.4373, + "step": 839500 + }, + { + "epoch": 11.48, + "learning_rate": 2.1292258479036513e-05, + "loss": 1.3991, + "step": 840000 + }, + { + "epoch": 11.49, + "learning_rate": 2.127517053765499e-05, + "loss": 1.3881, + "step": 840500 + }, + { + "epoch": 11.5, + "learning_rate": 2.125808259627346e-05, + "loss": 1.4237, + "step": 841000 + }, + { + "epoch": 11.5, + "learning_rate": 2.1240994654891938e-05, + "loss": 1.4341, + "step": 841500 + }, + { + "epoch": 11.51, + "learning_rate": 2.1223906713510412e-05, + "loss": 1.425, + "step": 842000 + }, + { + "epoch": 11.52, + "learning_rate": 2.1206818772128886e-05, + "loss": 1.4366, + "step": 842500 + }, + { + "epoch": 11.52, + "learning_rate": 2.118973083074736e-05, + "loss": 1.4187, + "step": 843000 + }, + { + "epoch": 11.53, + "learning_rate": 2.1172642889365834e-05, + "loss": 1.4109, + "step": 843500 + }, + { + "epoch": 11.54, + "learning_rate": 2.1155554947984308e-05, + "loss": 1.4124, + "step": 844000 + }, + { + "epoch": 11.54, + "learning_rate": 2.1138467006602782e-05, + "loss": 1.4174, + "step": 844500 + }, + { + "epoch": 11.55, + "learning_rate": 2.1121379065221256e-05, + "loss": 1.4214, + "step": 845000 + }, + { + "epoch": 11.56, + "learning_rate": 2.110429112383973e-05, + "loss": 1.4134, + "step": 845500 + }, + { + "epoch": 11.57, + "learning_rate": 2.1087203182458204e-05, + "loss": 1.4008, + "step": 846000 + }, + { + "epoch": 11.57, + "learning_rate": 2.1070115241076677e-05, + "loss": 1.4148, + "step": 846500 + }, + { + "epoch": 11.58, + "learning_rate": 2.105302729969515e-05, + "loss": 1.3788, + "step": 847000 + }, + { + "epoch": 11.59, + "learning_rate": 2.1035939358313625e-05, + "loss": 1.417, + "step": 847500 + }, + { + "epoch": 11.59, + "learning_rate": 2.10188514169321e-05, + "loss": 1.4134, + "step": 848000 + }, + { + "epoch": 11.6, + "learning_rate": 2.1001763475550573e-05, + "loss": 1.4178, + "step": 848500 + }, + { + "epoch": 11.61, + "learning_rate": 2.098467553416905e-05, + "loss": 1.4193, + "step": 849000 + }, + { + "epoch": 11.61, + "learning_rate": 2.096758759278752e-05, + "loss": 1.4005, + "step": 849500 + }, + { + "epoch": 11.62, + "learning_rate": 2.0950499651405998e-05, + "loss": 1.399, + "step": 850000 + }, + { + "epoch": 11.63, + "learning_rate": 2.093341171002447e-05, + "loss": 1.4221, + "step": 850500 + }, + { + "epoch": 11.63, + "learning_rate": 2.0916323768642946e-05, + "loss": 1.4193, + "step": 851000 + }, + { + "epoch": 11.64, + "learning_rate": 2.0899235827261417e-05, + "loss": 1.4244, + "step": 851500 + }, + { + "epoch": 11.65, + "learning_rate": 2.0882147885879894e-05, + "loss": 1.4075, + "step": 852000 + }, + { + "epoch": 11.65, + "learning_rate": 2.0865059944498368e-05, + "loss": 1.3956, + "step": 852500 + }, + { + "epoch": 11.66, + "learning_rate": 2.0847972003116842e-05, + "loss": 1.43, + "step": 853000 + }, + { + "epoch": 11.67, + "learning_rate": 2.0830884061735316e-05, + "loss": 1.4412, + "step": 853500 + }, + { + "epoch": 11.67, + "learning_rate": 2.081379612035379e-05, + "loss": 1.4081, + "step": 854000 + }, + { + "epoch": 11.68, + "learning_rate": 2.0796708178972263e-05, + "loss": 1.4048, + "step": 854500 + }, + { + "epoch": 11.69, + "learning_rate": 2.0779620237590737e-05, + "loss": 1.4088, + "step": 855000 + }, + { + "epoch": 11.69, + "learning_rate": 2.076253229620921e-05, + "loss": 1.4177, + "step": 855500 + }, + { + "epoch": 11.7, + "learning_rate": 2.0745444354827685e-05, + "loss": 1.393, + "step": 856000 + }, + { + "epoch": 11.71, + "learning_rate": 2.072835641344616e-05, + "loss": 1.4212, + "step": 856500 + }, + { + "epoch": 11.72, + "learning_rate": 2.0711268472064636e-05, + "loss": 1.4128, + "step": 857000 + }, + { + "epoch": 11.72, + "learning_rate": 2.0694180530683107e-05, + "loss": 1.4216, + "step": 857500 + }, + { + "epoch": 11.73, + "learning_rate": 2.0677092589301584e-05, + "loss": 1.4018, + "step": 858000 + }, + { + "epoch": 11.74, + "learning_rate": 2.0660004647920055e-05, + "loss": 1.4162, + "step": 858500 + }, + { + "epoch": 11.74, + "learning_rate": 2.0642916706538532e-05, + "loss": 1.4239, + "step": 859000 + }, + { + "epoch": 11.75, + "learning_rate": 2.0625828765157003e-05, + "loss": 1.3925, + "step": 859500 + }, + { + "epoch": 11.76, + "learning_rate": 2.060874082377548e-05, + "loss": 1.4089, + "step": 860000 + }, + { + "epoch": 11.76, + "learning_rate": 2.0591652882393954e-05, + "loss": 1.4182, + "step": 860500 + }, + { + "epoch": 11.77, + "learning_rate": 2.0574564941012428e-05, + "loss": 1.4075, + "step": 861000 + }, + { + "epoch": 11.78, + "learning_rate": 2.05574769996309e-05, + "loss": 1.427, + "step": 861500 + }, + { + "epoch": 11.78, + "learning_rate": 2.0540389058249376e-05, + "loss": 1.41, + "step": 862000 + }, + { + "epoch": 11.79, + "learning_rate": 2.052330111686785e-05, + "loss": 1.4211, + "step": 862500 + }, + { + "epoch": 11.8, + "learning_rate": 2.0506213175486323e-05, + "loss": 1.4378, + "step": 863000 + }, + { + "epoch": 11.8, + "learning_rate": 2.0489125234104797e-05, + "loss": 1.4301, + "step": 863500 + }, + { + "epoch": 11.81, + "learning_rate": 2.047203729272327e-05, + "loss": 1.405, + "step": 864000 + }, + { + "epoch": 11.82, + "learning_rate": 2.0454949351341745e-05, + "loss": 1.43, + "step": 864500 + }, + { + "epoch": 11.82, + "learning_rate": 2.0437861409960222e-05, + "loss": 1.4084, + "step": 865000 + }, + { + "epoch": 11.83, + "learning_rate": 2.0420773468578693e-05, + "loss": 1.423, + "step": 865500 + }, + { + "epoch": 11.84, + "learning_rate": 2.040368552719717e-05, + "loss": 1.4406, + "step": 866000 + }, + { + "epoch": 11.85, + "learning_rate": 2.038659758581564e-05, + "loss": 1.3974, + "step": 866500 + }, + { + "epoch": 11.85, + "learning_rate": 2.0369509644434118e-05, + "loss": 1.3956, + "step": 867000 + }, + { + "epoch": 11.86, + "learning_rate": 2.0352421703052592e-05, + "loss": 1.4038, + "step": 867500 + }, + { + "epoch": 11.87, + "learning_rate": 2.0335333761671066e-05, + "loss": 1.4004, + "step": 868000 + }, + { + "epoch": 11.87, + "learning_rate": 2.031824582028954e-05, + "loss": 1.4231, + "step": 868500 + }, + { + "epoch": 11.88, + "learning_rate": 2.0301157878908014e-05, + "loss": 1.4, + "step": 869000 + }, + { + "epoch": 11.89, + "learning_rate": 2.0284069937526488e-05, + "loss": 1.4317, + "step": 869500 + }, + { + "epoch": 11.89, + "learning_rate": 2.026698199614496e-05, + "loss": 1.4097, + "step": 870000 + }, + { + "epoch": 11.9, + "learning_rate": 2.0249894054763436e-05, + "loss": 1.4207, + "step": 870500 + }, + { + "epoch": 11.91, + "learning_rate": 2.023280611338191e-05, + "loss": 1.4129, + "step": 871000 + }, + { + "epoch": 11.91, + "learning_rate": 2.0215718172000383e-05, + "loss": 1.4042, + "step": 871500 + }, + { + "epoch": 11.92, + "learning_rate": 2.019863023061886e-05, + "loss": 1.429, + "step": 872000 + }, + { + "epoch": 11.93, + "learning_rate": 2.018154228923733e-05, + "loss": 1.4133, + "step": 872500 + }, + { + "epoch": 11.93, + "learning_rate": 2.016445434785581e-05, + "loss": 1.408, + "step": 873000 + }, + { + "epoch": 11.94, + "learning_rate": 2.014736640647428e-05, + "loss": 1.4229, + "step": 873500 + }, + { + "epoch": 11.95, + "learning_rate": 2.0130278465092756e-05, + "loss": 1.3853, + "step": 874000 + }, + { + "epoch": 11.95, + "learning_rate": 2.0113190523711227e-05, + "loss": 1.4045, + "step": 874500 + }, + { + "epoch": 11.96, + "learning_rate": 2.0096102582329704e-05, + "loss": 1.418, + "step": 875000 + }, + { + "epoch": 11.97, + "learning_rate": 2.0079014640948178e-05, + "loss": 1.3965, + "step": 875500 + }, + { + "epoch": 11.98, + "learning_rate": 2.006192669956665e-05, + "loss": 1.4061, + "step": 876000 + }, + { + "epoch": 11.98, + "learning_rate": 2.0044838758185126e-05, + "loss": 1.4067, + "step": 876500 + }, + { + "epoch": 11.99, + "learning_rate": 2.0027750816803596e-05, + "loss": 1.4012, + "step": 877000 + }, + { + "epoch": 12.0, + "learning_rate": 2.0010662875422074e-05, + "loss": 1.403, + "step": 877500 + }, + { + "epoch": 12.0, + "learning_rate": 1.9993574934040544e-05, + "loss": 1.428, + "step": 878000 + }, + { + "epoch": 12.01, + "learning_rate": 1.997648699265902e-05, + "loss": 1.3597, + "step": 878500 + }, + { + "epoch": 12.02, + "learning_rate": 1.9959399051277496e-05, + "loss": 1.3953, + "step": 879000 + }, + { + "epoch": 12.02, + "learning_rate": 1.994231110989597e-05, + "loss": 1.3716, + "step": 879500 + }, + { + "epoch": 12.03, + "learning_rate": 1.9925223168514443e-05, + "loss": 1.3643, + "step": 880000 + }, + { + "epoch": 12.04, + "learning_rate": 1.9908135227132917e-05, + "loss": 1.3743, + "step": 880500 + }, + { + "epoch": 12.04, + "learning_rate": 1.989104728575139e-05, + "loss": 1.3838, + "step": 881000 + }, + { + "epoch": 12.05, + "learning_rate": 1.9873959344369865e-05, + "loss": 1.4069, + "step": 881500 + }, + { + "epoch": 12.06, + "learning_rate": 1.985687140298834e-05, + "loss": 1.402, + "step": 882000 + }, + { + "epoch": 12.06, + "learning_rate": 1.9839783461606813e-05, + "loss": 1.3624, + "step": 882500 + }, + { + "epoch": 12.07, + "learning_rate": 1.9822695520225287e-05, + "loss": 1.3822, + "step": 883000 + }, + { + "epoch": 12.08, + "learning_rate": 1.9805607578843764e-05, + "loss": 1.3759, + "step": 883500 + }, + { + "epoch": 12.08, + "learning_rate": 1.9788519637462235e-05, + "loss": 1.405, + "step": 884000 + }, + { + "epoch": 12.09, + "learning_rate": 1.9771431696080712e-05, + "loss": 1.3791, + "step": 884500 + }, + { + "epoch": 12.1, + "learning_rate": 1.9754343754699183e-05, + "loss": 1.402, + "step": 885000 + }, + { + "epoch": 12.11, + "learning_rate": 1.973725581331766e-05, + "loss": 1.4245, + "step": 885500 + }, + { + "epoch": 12.11, + "learning_rate": 1.9720167871936134e-05, + "loss": 1.415, + "step": 886000 + }, + { + "epoch": 12.12, + "learning_rate": 1.9703079930554608e-05, + "loss": 1.4018, + "step": 886500 + }, + { + "epoch": 12.13, + "learning_rate": 1.968599198917308e-05, + "loss": 1.4046, + "step": 887000 + }, + { + "epoch": 12.13, + "learning_rate": 1.9668904047791555e-05, + "loss": 1.3926, + "step": 887500 + }, + { + "epoch": 12.14, + "learning_rate": 1.965181610641003e-05, + "loss": 1.388, + "step": 888000 + }, + { + "epoch": 12.15, + "learning_rate": 1.9634728165028503e-05, + "loss": 1.4188, + "step": 888500 + }, + { + "epoch": 12.15, + "learning_rate": 1.9617640223646977e-05, + "loss": 1.3826, + "step": 889000 + }, + { + "epoch": 12.16, + "learning_rate": 1.960055228226545e-05, + "loss": 1.4191, + "step": 889500 + }, + { + "epoch": 12.17, + "learning_rate": 1.9583464340883925e-05, + "loss": 1.4033, + "step": 890000 + }, + { + "epoch": 12.17, + "learning_rate": 1.9566376399502402e-05, + "loss": 1.3934, + "step": 890500 + }, + { + "epoch": 12.18, + "learning_rate": 1.9549288458120873e-05, + "loss": 1.3983, + "step": 891000 + }, + { + "epoch": 12.19, + "learning_rate": 1.953220051673935e-05, + "loss": 1.391, + "step": 891500 + }, + { + "epoch": 12.19, + "learning_rate": 1.951511257535782e-05, + "loss": 1.4166, + "step": 892000 + }, + { + "epoch": 12.2, + "learning_rate": 1.9498024633976298e-05, + "loss": 1.4106, + "step": 892500 + }, + { + "epoch": 12.21, + "learning_rate": 1.948093669259477e-05, + "loss": 1.3885, + "step": 893000 + }, + { + "epoch": 12.21, + "learning_rate": 1.9463848751213246e-05, + "loss": 1.4151, + "step": 893500 + }, + { + "epoch": 12.22, + "learning_rate": 1.944676080983172e-05, + "loss": 1.4211, + "step": 894000 + }, + { + "epoch": 12.23, + "learning_rate": 1.9429672868450194e-05, + "loss": 1.413, + "step": 894500 + }, + { + "epoch": 12.23, + "learning_rate": 1.9412584927068668e-05, + "loss": 1.4164, + "step": 895000 + }, + { + "epoch": 12.24, + "learning_rate": 1.939549698568714e-05, + "loss": 1.3836, + "step": 895500 + }, + { + "epoch": 12.25, + "learning_rate": 1.9378409044305615e-05, + "loss": 1.4095, + "step": 896000 + }, + { + "epoch": 12.26, + "learning_rate": 1.936132110292409e-05, + "loss": 1.3841, + "step": 896500 + }, + { + "epoch": 12.26, + "learning_rate": 1.9344233161542563e-05, + "loss": 1.3932, + "step": 897000 + }, + { + "epoch": 12.27, + "learning_rate": 1.9327145220161037e-05, + "loss": 1.3845, + "step": 897500 + }, + { + "epoch": 12.28, + "learning_rate": 1.931005727877951e-05, + "loss": 1.4001, + "step": 898000 + }, + { + "epoch": 12.28, + "learning_rate": 1.929296933739799e-05, + "loss": 1.3725, + "step": 898500 + }, + { + "epoch": 12.29, + "learning_rate": 1.927588139601646e-05, + "loss": 1.3963, + "step": 899000 + }, + { + "epoch": 12.3, + "learning_rate": 1.9258793454634936e-05, + "loss": 1.4062, + "step": 899500 + }, + { + "epoch": 12.3, + "learning_rate": 1.9241705513253407e-05, + "loss": 1.4224, + "step": 900000 + }, + { + "epoch": 12.31, + "learning_rate": 1.9224617571871884e-05, + "loss": 1.3943, + "step": 900500 + }, + { + "epoch": 12.32, + "learning_rate": 1.9207529630490355e-05, + "loss": 1.3978, + "step": 901000 + }, + { + "epoch": 12.32, + "learning_rate": 1.9190441689108832e-05, + "loss": 1.408, + "step": 901500 + }, + { + "epoch": 12.33, + "learning_rate": 1.9173353747727306e-05, + "loss": 1.3731, + "step": 902000 + }, + { + "epoch": 12.34, + "learning_rate": 1.915626580634578e-05, + "loss": 1.4162, + "step": 902500 + }, + { + "epoch": 12.34, + "learning_rate": 1.9139177864964254e-05, + "loss": 1.3904, + "step": 903000 + }, + { + "epoch": 12.35, + "learning_rate": 1.9122089923582728e-05, + "loss": 1.4044, + "step": 903500 + }, + { + "epoch": 12.36, + "learning_rate": 1.91050019822012e-05, + "loss": 1.4133, + "step": 904000 + }, + { + "epoch": 12.36, + "learning_rate": 1.9087914040819675e-05, + "loss": 1.4031, + "step": 904500 + }, + { + "epoch": 12.37, + "learning_rate": 1.907082609943815e-05, + "loss": 1.409, + "step": 905000 + }, + { + "epoch": 12.38, + "learning_rate": 1.9053738158056623e-05, + "loss": 1.3877, + "step": 905500 + }, + { + "epoch": 12.39, + "learning_rate": 1.9036650216675097e-05, + "loss": 1.4091, + "step": 906000 + }, + { + "epoch": 12.39, + "learning_rate": 1.901956227529357e-05, + "loss": 1.3811, + "step": 906500 + }, + { + "epoch": 12.4, + "learning_rate": 1.9002474333912045e-05, + "loss": 1.3912, + "step": 907000 + }, + { + "epoch": 12.41, + "learning_rate": 1.898538639253052e-05, + "loss": 1.3741, + "step": 907500 + }, + { + "epoch": 12.41, + "learning_rate": 1.8968298451148993e-05, + "loss": 1.4079, + "step": 908000 + }, + { + "epoch": 12.42, + "learning_rate": 1.8951210509767467e-05, + "loss": 1.4285, + "step": 908500 + }, + { + "epoch": 12.43, + "learning_rate": 1.8934122568385944e-05, + "loss": 1.3757, + "step": 909000 + }, + { + "epoch": 12.43, + "learning_rate": 1.8917034627004415e-05, + "loss": 1.3783, + "step": 909500 + }, + { + "epoch": 12.44, + "learning_rate": 1.8899946685622892e-05, + "loss": 1.3952, + "step": 910000 + }, + { + "epoch": 12.45, + "learning_rate": 1.8882858744241362e-05, + "loss": 1.4172, + "step": 910500 + }, + { + "epoch": 12.45, + "learning_rate": 1.886577080285984e-05, + "loss": 1.4118, + "step": 911000 + }, + { + "epoch": 12.46, + "learning_rate": 1.884868286147831e-05, + "loss": 1.3954, + "step": 911500 + }, + { + "epoch": 12.47, + "learning_rate": 1.8831594920096788e-05, + "loss": 1.3783, + "step": 912000 + }, + { + "epoch": 12.47, + "learning_rate": 1.881450697871526e-05, + "loss": 1.3858, + "step": 912500 + }, + { + "epoch": 12.48, + "learning_rate": 1.8797419037333735e-05, + "loss": 1.3901, + "step": 913000 + }, + { + "epoch": 12.49, + "learning_rate": 1.878033109595221e-05, + "loss": 1.3933, + "step": 913500 + }, + { + "epoch": 12.49, + "learning_rate": 1.8763243154570683e-05, + "loss": 1.3804, + "step": 914000 + }, + { + "epoch": 12.5, + "learning_rate": 1.8746155213189157e-05, + "loss": 1.4148, + "step": 914500 + }, + { + "epoch": 12.51, + "learning_rate": 1.872906727180763e-05, + "loss": 1.3958, + "step": 915000 + }, + { + "epoch": 12.52, + "learning_rate": 1.8711979330426105e-05, + "loss": 1.4045, + "step": 915500 + }, + { + "epoch": 12.52, + "learning_rate": 1.869489138904458e-05, + "loss": 1.3988, + "step": 916000 + }, + { + "epoch": 12.53, + "learning_rate": 1.8677803447663053e-05, + "loss": 1.3655, + "step": 916500 + }, + { + "epoch": 12.54, + "learning_rate": 1.866071550628153e-05, + "loss": 1.3893, + "step": 917000 + }, + { + "epoch": 12.54, + "learning_rate": 1.86436275649e-05, + "loss": 1.3806, + "step": 917500 + }, + { + "epoch": 12.55, + "learning_rate": 1.8626539623518478e-05, + "loss": 1.4144, + "step": 918000 + }, + { + "epoch": 12.56, + "learning_rate": 1.860945168213695e-05, + "loss": 1.3692, + "step": 918500 + }, + { + "epoch": 12.56, + "learning_rate": 1.8592363740755426e-05, + "loss": 1.3915, + "step": 919000 + }, + { + "epoch": 12.57, + "learning_rate": 1.8575275799373896e-05, + "loss": 1.352, + "step": 919500 + }, + { + "epoch": 12.58, + "learning_rate": 1.8558187857992374e-05, + "loss": 1.4019, + "step": 920000 + }, + { + "epoch": 12.58, + "learning_rate": 1.8541099916610847e-05, + "loss": 1.3796, + "step": 920500 + }, + { + "epoch": 12.59, + "learning_rate": 1.852401197522932e-05, + "loss": 1.3965, + "step": 921000 + }, + { + "epoch": 12.6, + "learning_rate": 1.8506924033847795e-05, + "loss": 1.3854, + "step": 921500 + }, + { + "epoch": 12.6, + "learning_rate": 1.848983609246627e-05, + "loss": 1.3723, + "step": 922000 + }, + { + "epoch": 12.61, + "learning_rate": 1.8472748151084743e-05, + "loss": 1.3959, + "step": 922500 + }, + { + "epoch": 12.62, + "learning_rate": 1.8455660209703217e-05, + "loss": 1.3698, + "step": 923000 + }, + { + "epoch": 12.62, + "learning_rate": 1.843857226832169e-05, + "loss": 1.3788, + "step": 923500 + }, + { + "epoch": 12.63, + "learning_rate": 1.8421484326940165e-05, + "loss": 1.3824, + "step": 924000 + }, + { + "epoch": 12.64, + "learning_rate": 1.840439638555864e-05, + "loss": 1.3952, + "step": 924500 + }, + { + "epoch": 12.65, + "learning_rate": 1.8387308444177116e-05, + "loss": 1.3675, + "step": 925000 + }, + { + "epoch": 12.65, + "learning_rate": 1.8370220502795587e-05, + "loss": 1.4015, + "step": 925500 + }, + { + "epoch": 12.66, + "learning_rate": 1.8353132561414064e-05, + "loss": 1.3895, + "step": 926000 + }, + { + "epoch": 12.67, + "learning_rate": 1.8336044620032534e-05, + "loss": 1.3744, + "step": 926500 + }, + { + "epoch": 12.67, + "learning_rate": 1.8318956678651012e-05, + "loss": 1.3954, + "step": 927000 + }, + { + "epoch": 12.68, + "learning_rate": 1.8301868737269486e-05, + "loss": 1.4018, + "step": 927500 + }, + { + "epoch": 12.69, + "learning_rate": 1.828478079588796e-05, + "loss": 1.4192, + "step": 928000 + }, + { + "epoch": 12.69, + "learning_rate": 1.8267692854506434e-05, + "loss": 1.4196, + "step": 928500 + }, + { + "epoch": 12.7, + "learning_rate": 1.8250604913124907e-05, + "loss": 1.3688, + "step": 929000 + }, + { + "epoch": 12.71, + "learning_rate": 1.823351697174338e-05, + "loss": 1.3732, + "step": 929500 + }, + { + "epoch": 12.71, + "learning_rate": 1.8216429030361855e-05, + "loss": 1.3905, + "step": 930000 + }, + { + "epoch": 12.72, + "learning_rate": 1.819934108898033e-05, + "loss": 1.3767, + "step": 930500 + }, + { + "epoch": 12.73, + "learning_rate": 1.8182253147598803e-05, + "loss": 1.4093, + "step": 931000 + }, + { + "epoch": 12.73, + "learning_rate": 1.8165165206217277e-05, + "loss": 1.3879, + "step": 931500 + }, + { + "epoch": 12.74, + "learning_rate": 1.8148077264835754e-05, + "loss": 1.3752, + "step": 932000 + }, + { + "epoch": 12.75, + "learning_rate": 1.8130989323454225e-05, + "loss": 1.3806, + "step": 932500 + }, + { + "epoch": 12.75, + "learning_rate": 1.8113901382072702e-05, + "loss": 1.3905, + "step": 933000 + }, + { + "epoch": 12.76, + "learning_rate": 1.8096813440691173e-05, + "loss": 1.3774, + "step": 933500 + }, + { + "epoch": 12.77, + "learning_rate": 1.807972549930965e-05, + "loss": 1.3644, + "step": 934000 + }, + { + "epoch": 12.77, + "learning_rate": 1.806263755792812e-05, + "loss": 1.3946, + "step": 934500 + }, + { + "epoch": 12.78, + "learning_rate": 1.8045549616546598e-05, + "loss": 1.4163, + "step": 935000 + }, + { + "epoch": 12.79, + "learning_rate": 1.8028461675165072e-05, + "loss": 1.3738, + "step": 935500 + }, + { + "epoch": 12.8, + "learning_rate": 1.8011373733783546e-05, + "loss": 1.3904, + "step": 936000 + }, + { + "epoch": 12.8, + "learning_rate": 1.799428579240202e-05, + "loss": 1.3906, + "step": 936500 + }, + { + "epoch": 12.81, + "learning_rate": 1.797719785102049e-05, + "loss": 1.3971, + "step": 937000 + }, + { + "epoch": 12.82, + "learning_rate": 1.7960109909638967e-05, + "loss": 1.3868, + "step": 937500 + }, + { + "epoch": 12.82, + "learning_rate": 1.794302196825744e-05, + "loss": 1.401, + "step": 938000 + }, + { + "epoch": 12.83, + "learning_rate": 1.7925934026875915e-05, + "loss": 1.3937, + "step": 938500 + }, + { + "epoch": 12.84, + "learning_rate": 1.790884608549439e-05, + "loss": 1.3763, + "step": 939000 + }, + { + "epoch": 12.84, + "learning_rate": 1.7891758144112863e-05, + "loss": 1.3791, + "step": 939500 + }, + { + "epoch": 12.85, + "learning_rate": 1.7874670202731337e-05, + "loss": 1.3596, + "step": 940000 + }, + { + "epoch": 12.86, + "learning_rate": 1.785758226134981e-05, + "loss": 1.3728, + "step": 940500 + }, + { + "epoch": 12.86, + "learning_rate": 1.7840494319968285e-05, + "loss": 1.4034, + "step": 941000 + }, + { + "epoch": 12.87, + "learning_rate": 1.782340637858676e-05, + "loss": 1.3815, + "step": 941500 + }, + { + "epoch": 12.88, + "learning_rate": 1.7806318437205233e-05, + "loss": 1.4149, + "step": 942000 + }, + { + "epoch": 12.88, + "learning_rate": 1.778923049582371e-05, + "loss": 1.3907, + "step": 942500 + }, + { + "epoch": 12.89, + "learning_rate": 1.777214255444218e-05, + "loss": 1.3789, + "step": 943000 + }, + { + "epoch": 12.9, + "learning_rate": 1.7755054613060658e-05, + "loss": 1.39, + "step": 943500 + }, + { + "epoch": 12.9, + "learning_rate": 1.773796667167913e-05, + "loss": 1.3708, + "step": 944000 + }, + { + "epoch": 12.91, + "learning_rate": 1.7720878730297606e-05, + "loss": 1.3846, + "step": 944500 + }, + { + "epoch": 12.92, + "learning_rate": 1.7703790788916076e-05, + "loss": 1.3868, + "step": 945000 + }, + { + "epoch": 12.93, + "learning_rate": 1.7686702847534553e-05, + "loss": 1.4117, + "step": 945500 + }, + { + "epoch": 12.93, + "learning_rate": 1.7669614906153027e-05, + "loss": 1.3479, + "step": 946000 + }, + { + "epoch": 12.94, + "learning_rate": 1.76525269647715e-05, + "loss": 1.3717, + "step": 946500 + }, + { + "epoch": 12.95, + "learning_rate": 1.7635439023389975e-05, + "loss": 1.4105, + "step": 947000 + }, + { + "epoch": 12.95, + "learning_rate": 1.761835108200845e-05, + "loss": 1.38, + "step": 947500 + }, + { + "epoch": 12.96, + "learning_rate": 1.7601263140626923e-05, + "loss": 1.3589, + "step": 948000 + }, + { + "epoch": 12.97, + "learning_rate": 1.7584175199245397e-05, + "loss": 1.3673, + "step": 948500 + }, + { + "epoch": 12.97, + "learning_rate": 1.756708725786387e-05, + "loss": 1.3584, + "step": 949000 + }, + { + "epoch": 12.98, + "learning_rate": 1.7549999316482345e-05, + "loss": 1.3775, + "step": 949500 + }, + { + "epoch": 12.99, + "learning_rate": 1.753291137510082e-05, + "loss": 1.384, + "step": 950000 + }, + { + "epoch": 12.99, + "learning_rate": 1.7515823433719296e-05, + "loss": 1.3769, + "step": 950500 + }, + { + "epoch": 13.0, + "learning_rate": 1.7498735492337767e-05, + "loss": 1.4147, + "step": 951000 + }, + { + "epoch": 13.01, + "learning_rate": 1.7481647550956244e-05, + "loss": 1.3622, + "step": 951500 + }, + { + "epoch": 13.01, + "learning_rate": 1.7464559609574714e-05, + "loss": 1.3679, + "step": 952000 + }, + { + "epoch": 13.02, + "learning_rate": 1.744747166819319e-05, + "loss": 1.3701, + "step": 952500 + }, + { + "epoch": 13.03, + "learning_rate": 1.7430383726811662e-05, + "loss": 1.3679, + "step": 953000 + }, + { + "epoch": 13.03, + "learning_rate": 1.741329578543014e-05, + "loss": 1.373, + "step": 953500 + }, + { + "epoch": 13.04, + "learning_rate": 1.7396207844048613e-05, + "loss": 1.3628, + "step": 954000 + }, + { + "epoch": 13.05, + "learning_rate": 1.7379119902667087e-05, + "loss": 1.353, + "step": 954500 + }, + { + "epoch": 13.06, + "learning_rate": 1.736203196128556e-05, + "loss": 1.387, + "step": 955000 + }, + { + "epoch": 13.06, + "learning_rate": 1.7344944019904035e-05, + "loss": 1.4016, + "step": 955500 + }, + { + "epoch": 13.07, + "learning_rate": 1.732785607852251e-05, + "loss": 1.3651, + "step": 956000 + }, + { + "epoch": 13.08, + "learning_rate": 1.7310768137140983e-05, + "loss": 1.3322, + "step": 956500 + }, + { + "epoch": 13.08, + "learning_rate": 1.7293680195759457e-05, + "loss": 1.3638, + "step": 957000 + }, + { + "epoch": 13.09, + "learning_rate": 1.727659225437793e-05, + "loss": 1.3552, + "step": 957500 + }, + { + "epoch": 13.1, + "learning_rate": 1.7259504312996405e-05, + "loss": 1.399, + "step": 958000 + }, + { + "epoch": 13.1, + "learning_rate": 1.7242416371614882e-05, + "loss": 1.3942, + "step": 958500 + }, + { + "epoch": 13.11, + "learning_rate": 1.7225328430233353e-05, + "loss": 1.3473, + "step": 959000 + }, + { + "epoch": 13.12, + "learning_rate": 1.720824048885183e-05, + "loss": 1.3562, + "step": 959500 + }, + { + "epoch": 13.12, + "learning_rate": 1.71911525474703e-05, + "loss": 1.3407, + "step": 960000 + }, + { + "epoch": 13.13, + "learning_rate": 1.7174064606088778e-05, + "loss": 1.3607, + "step": 960500 + }, + { + "epoch": 13.14, + "learning_rate": 1.715697666470725e-05, + "loss": 1.3389, + "step": 961000 + }, + { + "epoch": 13.14, + "learning_rate": 1.7139888723325726e-05, + "loss": 1.3505, + "step": 961500 + }, + { + "epoch": 13.15, + "learning_rate": 1.71228007819442e-05, + "loss": 1.3551, + "step": 962000 + }, + { + "epoch": 13.16, + "learning_rate": 1.7105712840562673e-05, + "loss": 1.3778, + "step": 962500 + }, + { + "epoch": 13.16, + "learning_rate": 1.7088624899181147e-05, + "loss": 1.3597, + "step": 963000 + }, + { + "epoch": 13.17, + "learning_rate": 1.707153695779962e-05, + "loss": 1.3645, + "step": 963500 + }, + { + "epoch": 13.18, + "learning_rate": 1.7054449016418095e-05, + "loss": 1.3837, + "step": 964000 + }, + { + "epoch": 13.19, + "learning_rate": 1.703736107503657e-05, + "loss": 1.3554, + "step": 964500 + }, + { + "epoch": 13.19, + "learning_rate": 1.7020273133655043e-05, + "loss": 1.3255, + "step": 965000 + }, + { + "epoch": 13.2, + "learning_rate": 1.700318519227352e-05, + "loss": 1.3741, + "step": 965500 + }, + { + "epoch": 13.21, + "learning_rate": 1.698609725089199e-05, + "loss": 1.3695, + "step": 966000 + }, + { + "epoch": 13.21, + "learning_rate": 1.6969009309510465e-05, + "loss": 1.3542, + "step": 966500 + }, + { + "epoch": 13.22, + "learning_rate": 1.695192136812894e-05, + "loss": 1.3621, + "step": 967000 + }, + { + "epoch": 13.23, + "learning_rate": 1.6934833426747413e-05, + "loss": 1.3594, + "step": 967500 + }, + { + "epoch": 13.23, + "learning_rate": 1.6917745485365886e-05, + "loss": 1.3493, + "step": 968000 + }, + { + "epoch": 13.24, + "learning_rate": 1.690065754398436e-05, + "loss": 1.376, + "step": 968500 + }, + { + "epoch": 13.25, + "learning_rate": 1.6883569602602838e-05, + "loss": 1.3929, + "step": 969000 + }, + { + "epoch": 13.25, + "learning_rate": 1.6866481661221308e-05, + "loss": 1.3766, + "step": 969500 + }, + { + "epoch": 13.26, + "learning_rate": 1.6849393719839785e-05, + "loss": 1.3325, + "step": 970000 + }, + { + "epoch": 13.27, + "learning_rate": 1.6832305778458256e-05, + "loss": 1.3892, + "step": 970500 + }, + { + "epoch": 13.27, + "learning_rate": 1.6815217837076733e-05, + "loss": 1.3362, + "step": 971000 + }, + { + "epoch": 13.28, + "learning_rate": 1.6798129895695204e-05, + "loss": 1.3753, + "step": 971500 + }, + { + "epoch": 13.29, + "learning_rate": 1.678104195431368e-05, + "loss": 1.3372, + "step": 972000 + }, + { + "epoch": 13.29, + "learning_rate": 1.6763954012932155e-05, + "loss": 1.359, + "step": 972500 + }, + { + "epoch": 13.3, + "learning_rate": 1.674686607155063e-05, + "loss": 1.3863, + "step": 973000 + }, + { + "epoch": 13.31, + "learning_rate": 1.6729778130169103e-05, + "loss": 1.3622, + "step": 973500 + }, + { + "epoch": 13.31, + "learning_rate": 1.6712690188787577e-05, + "loss": 1.3516, + "step": 974000 + }, + { + "epoch": 13.32, + "learning_rate": 1.669560224740605e-05, + "loss": 1.3595, + "step": 974500 + }, + { + "epoch": 13.33, + "learning_rate": 1.6678514306024525e-05, + "loss": 1.3733, + "step": 975000 + }, + { + "epoch": 13.34, + "learning_rate": 1.6661426364643e-05, + "loss": 1.374, + "step": 975500 + }, + { + "epoch": 13.34, + "learning_rate": 1.6644338423261472e-05, + "loss": 1.3662, + "step": 976000 + }, + { + "epoch": 13.35, + "learning_rate": 1.6627250481879946e-05, + "loss": 1.3703, + "step": 976500 + }, + { + "epoch": 13.36, + "learning_rate": 1.6610162540498424e-05, + "loss": 1.3864, + "step": 977000 + }, + { + "epoch": 13.36, + "learning_rate": 1.6593074599116894e-05, + "loss": 1.3872, + "step": 977500 + }, + { + "epoch": 13.37, + "learning_rate": 1.657598665773537e-05, + "loss": 1.357, + "step": 978000 + }, + { + "epoch": 13.38, + "learning_rate": 1.6558898716353842e-05, + "loss": 1.3753, + "step": 978500 + }, + { + "epoch": 13.38, + "learning_rate": 1.654181077497232e-05, + "loss": 1.3798, + "step": 979000 + }, + { + "epoch": 13.39, + "learning_rate": 1.6524722833590793e-05, + "loss": 1.3528, + "step": 979500 + }, + { + "epoch": 13.4, + "learning_rate": 1.6507634892209267e-05, + "loss": 1.3743, + "step": 980000 + }, + { + "epoch": 13.4, + "learning_rate": 1.649054695082774e-05, + "loss": 1.371, + "step": 980500 + }, + { + "epoch": 13.41, + "learning_rate": 1.6473459009446215e-05, + "loss": 1.3787, + "step": 981000 + }, + { + "epoch": 13.42, + "learning_rate": 1.645637106806469e-05, + "loss": 1.3793, + "step": 981500 + }, + { + "epoch": 13.42, + "learning_rate": 1.6439283126683163e-05, + "loss": 1.3659, + "step": 982000 + }, + { + "epoch": 13.43, + "learning_rate": 1.6422195185301637e-05, + "loss": 1.3892, + "step": 982500 + }, + { + "epoch": 13.44, + "learning_rate": 1.640510724392011e-05, + "loss": 1.3661, + "step": 983000 + }, + { + "epoch": 13.44, + "learning_rate": 1.6388019302538585e-05, + "loss": 1.3855, + "step": 983500 + }, + { + "epoch": 13.45, + "learning_rate": 1.6370931361157062e-05, + "loss": 1.3553, + "step": 984000 + }, + { + "epoch": 13.46, + "learning_rate": 1.6353843419775532e-05, + "loss": 1.3809, + "step": 984500 + }, + { + "epoch": 13.47, + "learning_rate": 1.633675547839401e-05, + "loss": 1.3344, + "step": 985000 + }, + { + "epoch": 13.47, + "learning_rate": 1.631966753701248e-05, + "loss": 1.3644, + "step": 985500 + }, + { + "epoch": 13.48, + "learning_rate": 1.6302579595630958e-05, + "loss": 1.3406, + "step": 986000 + }, + { + "epoch": 13.49, + "learning_rate": 1.6285491654249428e-05, + "loss": 1.3613, + "step": 986500 + }, + { + "epoch": 13.49, + "learning_rate": 1.6268403712867905e-05, + "loss": 1.3719, + "step": 987000 + }, + { + "epoch": 13.5, + "learning_rate": 1.625131577148638e-05, + "loss": 1.3555, + "step": 987500 + }, + { + "epoch": 13.51, + "learning_rate": 1.6234227830104853e-05, + "loss": 1.3953, + "step": 988000 + }, + { + "epoch": 13.51, + "learning_rate": 1.6217139888723327e-05, + "loss": 1.3573, + "step": 988500 + }, + { + "epoch": 13.52, + "learning_rate": 1.62000519473418e-05, + "loss": 1.3458, + "step": 989000 + }, + { + "epoch": 13.53, + "learning_rate": 1.6182964005960275e-05, + "loss": 1.391, + "step": 989500 + }, + { + "epoch": 13.53, + "learning_rate": 1.616587606457875e-05, + "loss": 1.3588, + "step": 990000 + }, + { + "epoch": 13.54, + "learning_rate": 1.6148788123197223e-05, + "loss": 1.3458, + "step": 990500 + }, + { + "epoch": 13.55, + "learning_rate": 1.6131700181815697e-05, + "loss": 1.3823, + "step": 991000 + }, + { + "epoch": 13.55, + "learning_rate": 1.611461224043417e-05, + "loss": 1.3595, + "step": 991500 + }, + { + "epoch": 13.56, + "learning_rate": 1.6097524299052648e-05, + "loss": 1.3355, + "step": 992000 + }, + { + "epoch": 13.57, + "learning_rate": 1.608043635767112e-05, + "loss": 1.376, + "step": 992500 + }, + { + "epoch": 13.57, + "learning_rate": 1.6063348416289596e-05, + "loss": 1.3431, + "step": 993000 + }, + { + "epoch": 13.58, + "learning_rate": 1.6046260474908066e-05, + "loss": 1.3749, + "step": 993500 + }, + { + "epoch": 13.59, + "learning_rate": 1.6029172533526544e-05, + "loss": 1.3722, + "step": 994000 + }, + { + "epoch": 13.6, + "learning_rate": 1.6012084592145014e-05, + "loss": 1.3432, + "step": 994500 + }, + { + "epoch": 13.6, + "learning_rate": 1.599499665076349e-05, + "loss": 1.368, + "step": 995000 + }, + { + "epoch": 13.61, + "learning_rate": 1.5977908709381965e-05, + "loss": 1.3707, + "step": 995500 + }, + { + "epoch": 13.62, + "learning_rate": 1.596082076800044e-05, + "loss": 1.3722, + "step": 996000 + }, + { + "epoch": 13.62, + "learning_rate": 1.5943732826618913e-05, + "loss": 1.3521, + "step": 996500 + }, + { + "epoch": 13.63, + "learning_rate": 1.5926644885237384e-05, + "loss": 1.3837, + "step": 997000 + }, + { + "epoch": 13.64, + "learning_rate": 1.590955694385586e-05, + "loss": 1.3793, + "step": 997500 + }, + { + "epoch": 13.64, + "learning_rate": 1.5892469002474335e-05, + "loss": 1.366, + "step": 998000 + }, + { + "epoch": 13.65, + "learning_rate": 1.587538106109281e-05, + "loss": 1.3425, + "step": 998500 + }, + { + "epoch": 13.66, + "learning_rate": 1.5858293119711283e-05, + "loss": 1.3244, + "step": 999000 + }, + { + "epoch": 13.66, + "learning_rate": 1.5841205178329757e-05, + "loss": 1.3586, + "step": 999500 + }, + { + "epoch": 13.67, + "learning_rate": 1.582411723694823e-05, + "loss": 1.3511, + "step": 1000000 + }, + { + "epoch": 13.68, + "learning_rate": 1.5807029295566705e-05, + "loss": 1.3535, + "step": 1000500 + }, + { + "epoch": 13.68, + "learning_rate": 1.578994135418518e-05, + "loss": 1.3618, + "step": 1001000 + }, + { + "epoch": 13.69, + "learning_rate": 1.5772853412803652e-05, + "loss": 1.3403, + "step": 1001500 + }, + { + "epoch": 13.7, + "learning_rate": 1.5755765471422126e-05, + "loss": 1.3577, + "step": 1002000 + }, + { + "epoch": 13.7, + "learning_rate": 1.5738677530040604e-05, + "loss": 1.3535, + "step": 1002500 + }, + { + "epoch": 13.71, + "learning_rate": 1.5721589588659074e-05, + "loss": 1.3699, + "step": 1003000 + }, + { + "epoch": 13.72, + "learning_rate": 1.570450164727755e-05, + "loss": 1.3682, + "step": 1003500 + }, + { + "epoch": 13.73, + "learning_rate": 1.5687413705896022e-05, + "loss": 1.3987, + "step": 1004000 + }, + { + "epoch": 13.73, + "learning_rate": 1.56703257645145e-05, + "loss": 1.3976, + "step": 1004500 + }, + { + "epoch": 13.74, + "learning_rate": 1.565323782313297e-05, + "loss": 1.3283, + "step": 1005000 + }, + { + "epoch": 13.75, + "learning_rate": 1.5636149881751447e-05, + "loss": 1.3329, + "step": 1005500 + }, + { + "epoch": 13.75, + "learning_rate": 1.561906194036992e-05, + "loss": 1.3589, + "step": 1006000 + }, + { + "epoch": 13.76, + "learning_rate": 1.5601973998988395e-05, + "loss": 1.3988, + "step": 1006500 + }, + { + "epoch": 13.77, + "learning_rate": 1.558488605760687e-05, + "loss": 1.3511, + "step": 1007000 + }, + { + "epoch": 13.77, + "learning_rate": 1.5567798116225343e-05, + "loss": 1.3441, + "step": 1007500 + }, + { + "epoch": 13.78, + "learning_rate": 1.5550710174843817e-05, + "loss": 1.3989, + "step": 1008000 + }, + { + "epoch": 13.79, + "learning_rate": 1.553362223346229e-05, + "loss": 1.3804, + "step": 1008500 + }, + { + "epoch": 13.79, + "learning_rate": 1.5516534292080764e-05, + "loss": 1.3514, + "step": 1009000 + }, + { + "epoch": 13.8, + "learning_rate": 1.549944635069924e-05, + "loss": 1.3387, + "step": 1009500 + }, + { + "epoch": 13.81, + "learning_rate": 1.5482358409317712e-05, + "loss": 1.3521, + "step": 1010000 + }, + { + "epoch": 13.81, + "learning_rate": 1.546527046793619e-05, + "loss": 1.3713, + "step": 1010500 + }, + { + "epoch": 13.82, + "learning_rate": 1.544818252655466e-05, + "loss": 1.3696, + "step": 1011000 + }, + { + "epoch": 13.83, + "learning_rate": 1.5431094585173137e-05, + "loss": 1.3577, + "step": 1011500 + }, + { + "epoch": 13.83, + "learning_rate": 1.5414006643791608e-05, + "loss": 1.3621, + "step": 1012000 + }, + { + "epoch": 13.84, + "learning_rate": 1.5396918702410085e-05, + "loss": 1.3609, + "step": 1012500 + }, + { + "epoch": 13.85, + "learning_rate": 1.5379830761028556e-05, + "loss": 1.3188, + "step": 1013000 + }, + { + "epoch": 13.85, + "learning_rate": 1.5362742819647033e-05, + "loss": 1.3481, + "step": 1013500 + }, + { + "epoch": 13.86, + "learning_rate": 1.5345654878265507e-05, + "loss": 1.3366, + "step": 1014000 + }, + { + "epoch": 13.87, + "learning_rate": 1.532856693688398e-05, + "loss": 1.3774, + "step": 1014500 + }, + { + "epoch": 13.88, + "learning_rate": 1.5311478995502455e-05, + "loss": 1.3712, + "step": 1015000 + }, + { + "epoch": 13.88, + "learning_rate": 1.529439105412093e-05, + "loss": 1.3696, + "step": 1015500 + }, + { + "epoch": 13.89, + "learning_rate": 1.5277303112739403e-05, + "loss": 1.3542, + "step": 1016000 + }, + { + "epoch": 13.9, + "learning_rate": 1.5260215171357877e-05, + "loss": 1.3566, + "step": 1016500 + }, + { + "epoch": 13.9, + "learning_rate": 1.524312722997635e-05, + "loss": 1.3777, + "step": 1017000 + }, + { + "epoch": 13.91, + "learning_rate": 1.5226039288594826e-05, + "loss": 1.3957, + "step": 1017500 + }, + { + "epoch": 13.92, + "learning_rate": 1.5208951347213298e-05, + "loss": 1.3622, + "step": 1018000 + }, + { + "epoch": 13.92, + "learning_rate": 1.5191863405831774e-05, + "loss": 1.3886, + "step": 1018500 + }, + { + "epoch": 13.93, + "learning_rate": 1.5174775464450248e-05, + "loss": 1.3571, + "step": 1019000 + }, + { + "epoch": 13.94, + "learning_rate": 1.5157687523068722e-05, + "loss": 1.3806, + "step": 1019500 + }, + { + "epoch": 13.94, + "learning_rate": 1.5140599581687196e-05, + "loss": 1.3694, + "step": 1020000 + }, + { + "epoch": 13.95, + "learning_rate": 1.5123511640305671e-05, + "loss": 1.3567, + "step": 1020500 + }, + { + "epoch": 13.96, + "learning_rate": 1.5106423698924144e-05, + "loss": 1.3582, + "step": 1021000 + }, + { + "epoch": 13.96, + "learning_rate": 1.508933575754262e-05, + "loss": 1.3565, + "step": 1021500 + }, + { + "epoch": 13.97, + "learning_rate": 1.5072247816161091e-05, + "loss": 1.3746, + "step": 1022000 + }, + { + "epoch": 13.98, + "learning_rate": 1.5055159874779567e-05, + "loss": 1.3304, + "step": 1022500 + }, + { + "epoch": 13.98, + "learning_rate": 1.5038071933398041e-05, + "loss": 1.35, + "step": 1023000 + }, + { + "epoch": 13.99, + "learning_rate": 1.5020983992016517e-05, + "loss": 1.372, + "step": 1023500 + }, + { + "epoch": 14.0, + "learning_rate": 1.5003896050634989e-05, + "loss": 1.3305, + "step": 1024000 + }, + { + "epoch": 14.01, + "learning_rate": 1.4986808109253464e-05, + "loss": 1.3304, + "step": 1024500 + }, + { + "epoch": 14.01, + "learning_rate": 1.4969720167871937e-05, + "loss": 1.3343, + "step": 1025000 + }, + { + "epoch": 14.02, + "learning_rate": 1.4952632226490412e-05, + "loss": 1.3397, + "step": 1025500 + }, + { + "epoch": 14.03, + "learning_rate": 1.4935544285108884e-05, + "loss": 1.381, + "step": 1026000 + }, + { + "epoch": 14.03, + "learning_rate": 1.491845634372736e-05, + "loss": 1.3438, + "step": 1026500 + }, + { + "epoch": 14.04, + "learning_rate": 1.4901368402345834e-05, + "loss": 1.3526, + "step": 1027000 + }, + { + "epoch": 14.05, + "learning_rate": 1.4884280460964306e-05, + "loss": 1.3409, + "step": 1027500 + }, + { + "epoch": 14.05, + "learning_rate": 1.4867192519582782e-05, + "loss": 1.3264, + "step": 1028000 + }, + { + "epoch": 14.06, + "learning_rate": 1.4850104578201254e-05, + "loss": 1.3425, + "step": 1028500 + }, + { + "epoch": 14.07, + "learning_rate": 1.483301663681973e-05, + "loss": 1.3343, + "step": 1029000 + }, + { + "epoch": 14.07, + "learning_rate": 1.4815928695438202e-05, + "loss": 1.3246, + "step": 1029500 + }, + { + "epoch": 14.08, + "learning_rate": 1.4798840754056677e-05, + "loss": 1.3322, + "step": 1030000 + }, + { + "epoch": 14.09, + "learning_rate": 1.4781752812675151e-05, + "loss": 1.3389, + "step": 1030500 + }, + { + "epoch": 14.09, + "learning_rate": 1.4764664871293627e-05, + "loss": 1.3543, + "step": 1031000 + }, + { + "epoch": 14.1, + "learning_rate": 1.47475769299121e-05, + "loss": 1.3546, + "step": 1031500 + }, + { + "epoch": 14.11, + "learning_rate": 1.4730488988530575e-05, + "loss": 1.3311, + "step": 1032000 + }, + { + "epoch": 14.11, + "learning_rate": 1.4713401047149047e-05, + "loss": 1.3343, + "step": 1032500 + }, + { + "epoch": 14.12, + "learning_rate": 1.4696313105767523e-05, + "loss": 1.3312, + "step": 1033000 + }, + { + "epoch": 14.13, + "learning_rate": 1.4679225164385997e-05, + "loss": 1.319, + "step": 1033500 + }, + { + "epoch": 14.14, + "learning_rate": 1.466213722300447e-05, + "loss": 1.3334, + "step": 1034000 + }, + { + "epoch": 14.14, + "learning_rate": 1.4645049281622944e-05, + "loss": 1.3493, + "step": 1034500 + }, + { + "epoch": 14.15, + "learning_rate": 1.462796134024142e-05, + "loss": 1.3591, + "step": 1035000 + }, + { + "epoch": 14.16, + "learning_rate": 1.4610873398859892e-05, + "loss": 1.3498, + "step": 1035500 + }, + { + "epoch": 14.16, + "learning_rate": 1.4593785457478368e-05, + "loss": 1.3336, + "step": 1036000 + }, + { + "epoch": 14.17, + "learning_rate": 1.457669751609684e-05, + "loss": 1.3635, + "step": 1036500 + }, + { + "epoch": 14.18, + "learning_rate": 1.4559609574715316e-05, + "loss": 1.3639, + "step": 1037000 + }, + { + "epoch": 14.18, + "learning_rate": 1.454252163333379e-05, + "loss": 1.3534, + "step": 1037500 + }, + { + "epoch": 14.19, + "learning_rate": 1.4525433691952265e-05, + "loss": 1.3413, + "step": 1038000 + }, + { + "epoch": 14.2, + "learning_rate": 1.4508345750570737e-05, + "loss": 1.3151, + "step": 1038500 + }, + { + "epoch": 14.2, + "learning_rate": 1.4491257809189213e-05, + "loss": 1.3359, + "step": 1039000 + }, + { + "epoch": 14.21, + "learning_rate": 1.4474169867807685e-05, + "loss": 1.3671, + "step": 1039500 + }, + { + "epoch": 14.22, + "learning_rate": 1.445708192642616e-05, + "loss": 1.345, + "step": 1040000 + }, + { + "epoch": 14.22, + "learning_rate": 1.4439993985044633e-05, + "loss": 1.3677, + "step": 1040500 + }, + { + "epoch": 14.23, + "learning_rate": 1.4422906043663109e-05, + "loss": 1.3704, + "step": 1041000 + }, + { + "epoch": 14.24, + "learning_rate": 1.4405818102281583e-05, + "loss": 1.339, + "step": 1041500 + }, + { + "epoch": 14.24, + "learning_rate": 1.4388730160900058e-05, + "loss": 1.3626, + "step": 1042000 + }, + { + "epoch": 14.25, + "learning_rate": 1.437164221951853e-05, + "loss": 1.3643, + "step": 1042500 + }, + { + "epoch": 14.26, + "learning_rate": 1.4354554278137006e-05, + "loss": 1.3393, + "step": 1043000 + }, + { + "epoch": 14.27, + "learning_rate": 1.4337466336755478e-05, + "loss": 1.3555, + "step": 1043500 + }, + { + "epoch": 14.27, + "learning_rate": 1.4320378395373954e-05, + "loss": 1.3408, + "step": 1044000 + }, + { + "epoch": 14.28, + "learning_rate": 1.4303290453992426e-05, + "loss": 1.3094, + "step": 1044500 + }, + { + "epoch": 14.29, + "learning_rate": 1.4286202512610902e-05, + "loss": 1.3187, + "step": 1045000 + }, + { + "epoch": 14.29, + "learning_rate": 1.4269114571229376e-05, + "loss": 1.3538, + "step": 1045500 + }, + { + "epoch": 14.3, + "learning_rate": 1.4252026629847851e-05, + "loss": 1.3429, + "step": 1046000 + }, + { + "epoch": 14.31, + "learning_rate": 1.4234938688466323e-05, + "loss": 1.3312, + "step": 1046500 + }, + { + "epoch": 14.31, + "learning_rate": 1.4217850747084799e-05, + "loss": 1.3277, + "step": 1047000 + }, + { + "epoch": 14.32, + "learning_rate": 1.4200762805703271e-05, + "loss": 1.3738, + "step": 1047500 + }, + { + "epoch": 14.33, + "learning_rate": 1.4183674864321747e-05, + "loss": 1.3166, + "step": 1048000 + }, + { + "epoch": 14.33, + "learning_rate": 1.4166586922940219e-05, + "loss": 1.3557, + "step": 1048500 + }, + { + "epoch": 14.34, + "learning_rate": 1.4149498981558695e-05, + "loss": 1.3458, + "step": 1049000 + }, + { + "epoch": 14.35, + "learning_rate": 1.4132411040177169e-05, + "loss": 1.3338, + "step": 1049500 + }, + { + "epoch": 14.35, + "learning_rate": 1.4115323098795644e-05, + "loss": 1.3457, + "step": 1050000 + }, + { + "epoch": 14.36, + "learning_rate": 1.4098235157414116e-05, + "loss": 1.3149, + "step": 1050500 + }, + { + "epoch": 14.37, + "learning_rate": 1.4081147216032592e-05, + "loss": 1.3305, + "step": 1051000 + }, + { + "epoch": 14.37, + "learning_rate": 1.4064059274651064e-05, + "loss": 1.3481, + "step": 1051500 + }, + { + "epoch": 14.38, + "learning_rate": 1.404697133326954e-05, + "loss": 1.3322, + "step": 1052000 + }, + { + "epoch": 14.39, + "learning_rate": 1.4029883391888012e-05, + "loss": 1.339, + "step": 1052500 + }, + { + "epoch": 14.39, + "learning_rate": 1.4012795450506488e-05, + "loss": 1.3306, + "step": 1053000 + }, + { + "epoch": 14.4, + "learning_rate": 1.3995707509124962e-05, + "loss": 1.3378, + "step": 1053500 + }, + { + "epoch": 14.41, + "learning_rate": 1.3978619567743437e-05, + "loss": 1.3072, + "step": 1054000 + }, + { + "epoch": 14.42, + "learning_rate": 1.396153162636191e-05, + "loss": 1.3332, + "step": 1054500 + }, + { + "epoch": 14.42, + "learning_rate": 1.3944443684980385e-05, + "loss": 1.3282, + "step": 1055000 + }, + { + "epoch": 14.43, + "learning_rate": 1.3927355743598857e-05, + "loss": 1.3362, + "step": 1055500 + }, + { + "epoch": 14.44, + "learning_rate": 1.3910267802217333e-05, + "loss": 1.3478, + "step": 1056000 + }, + { + "epoch": 14.44, + "learning_rate": 1.3893179860835807e-05, + "loss": 1.3631, + "step": 1056500 + }, + { + "epoch": 14.45, + "learning_rate": 1.387609191945428e-05, + "loss": 1.3183, + "step": 1057000 + }, + { + "epoch": 14.46, + "learning_rate": 1.3859003978072755e-05, + "loss": 1.3491, + "step": 1057500 + }, + { + "epoch": 14.46, + "learning_rate": 1.3841916036691227e-05, + "loss": 1.3294, + "step": 1058000 + }, + { + "epoch": 14.47, + "learning_rate": 1.3824828095309702e-05, + "loss": 1.325, + "step": 1058500 + }, + { + "epoch": 14.48, + "learning_rate": 1.3807740153928175e-05, + "loss": 1.3314, + "step": 1059000 + }, + { + "epoch": 14.48, + "learning_rate": 1.379065221254665e-05, + "loss": 1.3549, + "step": 1059500 + }, + { + "epoch": 14.49, + "learning_rate": 1.3773564271165124e-05, + "loss": 1.3305, + "step": 1060000 + }, + { + "epoch": 14.5, + "learning_rate": 1.37564763297836e-05, + "loss": 1.326, + "step": 1060500 + }, + { + "epoch": 14.5, + "learning_rate": 1.3739388388402072e-05, + "loss": 1.3348, + "step": 1061000 + }, + { + "epoch": 14.51, + "learning_rate": 1.3722300447020548e-05, + "loss": 1.3542, + "step": 1061500 + }, + { + "epoch": 14.52, + "learning_rate": 1.370521250563902e-05, + "loss": 1.3369, + "step": 1062000 + }, + { + "epoch": 14.52, + "learning_rate": 1.3688124564257496e-05, + "loss": 1.3393, + "step": 1062500 + }, + { + "epoch": 14.53, + "learning_rate": 1.3671036622875968e-05, + "loss": 1.3433, + "step": 1063000 + }, + { + "epoch": 14.54, + "learning_rate": 1.3653948681494443e-05, + "loss": 1.3562, + "step": 1063500 + }, + { + "epoch": 14.55, + "learning_rate": 1.3636860740112917e-05, + "loss": 1.3464, + "step": 1064000 + }, + { + "epoch": 14.55, + "learning_rate": 1.3619772798731393e-05, + "loss": 1.3524, + "step": 1064500 + }, + { + "epoch": 14.56, + "learning_rate": 1.3602684857349865e-05, + "loss": 1.3418, + "step": 1065000 + }, + { + "epoch": 14.57, + "learning_rate": 1.358559691596834e-05, + "loss": 1.3541, + "step": 1065500 + }, + { + "epoch": 14.57, + "learning_rate": 1.3568508974586813e-05, + "loss": 1.3335, + "step": 1066000 + }, + { + "epoch": 14.58, + "learning_rate": 1.3551421033205289e-05, + "loss": 1.3153, + "step": 1066500 + }, + { + "epoch": 14.59, + "learning_rate": 1.353433309182376e-05, + "loss": 1.3246, + "step": 1067000 + }, + { + "epoch": 14.59, + "learning_rate": 1.3517245150442236e-05, + "loss": 1.3291, + "step": 1067500 + }, + { + "epoch": 14.6, + "learning_rate": 1.350015720906071e-05, + "loss": 1.3468, + "step": 1068000 + }, + { + "epoch": 14.61, + "learning_rate": 1.3483069267679186e-05, + "loss": 1.3316, + "step": 1068500 + }, + { + "epoch": 14.61, + "learning_rate": 1.3465981326297658e-05, + "loss": 1.3487, + "step": 1069000 + }, + { + "epoch": 14.62, + "learning_rate": 1.3448893384916134e-05, + "loss": 1.3681, + "step": 1069500 + }, + { + "epoch": 14.63, + "learning_rate": 1.3431805443534606e-05, + "loss": 1.3353, + "step": 1070000 + }, + { + "epoch": 14.63, + "learning_rate": 1.3414717502153082e-05, + "loss": 1.3127, + "step": 1070500 + }, + { + "epoch": 14.64, + "learning_rate": 1.3397629560771555e-05, + "loss": 1.3469, + "step": 1071000 + }, + { + "epoch": 14.65, + "learning_rate": 1.338054161939003e-05, + "loss": 1.3277, + "step": 1071500 + }, + { + "epoch": 14.65, + "learning_rate": 1.3363453678008503e-05, + "loss": 1.3034, + "step": 1072000 + }, + { + "epoch": 14.66, + "learning_rate": 1.3346365736626979e-05, + "loss": 1.3366, + "step": 1072500 + }, + { + "epoch": 14.67, + "learning_rate": 1.3329277795245451e-05, + "loss": 1.336, + "step": 1073000 + }, + { + "epoch": 14.68, + "learning_rate": 1.3312189853863927e-05, + "loss": 1.329, + "step": 1073500 + }, + { + "epoch": 14.68, + "learning_rate": 1.3295101912482399e-05, + "loss": 1.3769, + "step": 1074000 + }, + { + "epoch": 14.69, + "learning_rate": 1.3278013971100875e-05, + "loss": 1.3676, + "step": 1074500 + }, + { + "epoch": 14.7, + "learning_rate": 1.3260926029719348e-05, + "loss": 1.339, + "step": 1075000 + }, + { + "epoch": 14.7, + "learning_rate": 1.3243838088337822e-05, + "loss": 1.34, + "step": 1075500 + }, + { + "epoch": 14.71, + "learning_rate": 1.3226750146956296e-05, + "loss": 1.342, + "step": 1076000 + }, + { + "epoch": 14.72, + "learning_rate": 1.3209662205574772e-05, + "loss": 1.3576, + "step": 1076500 + }, + { + "epoch": 14.72, + "learning_rate": 1.3192574264193244e-05, + "loss": 1.3546, + "step": 1077000 + }, + { + "epoch": 14.73, + "learning_rate": 1.317548632281172e-05, + "loss": 1.3241, + "step": 1077500 + }, + { + "epoch": 14.74, + "learning_rate": 1.3158398381430192e-05, + "loss": 1.3238, + "step": 1078000 + }, + { + "epoch": 14.74, + "learning_rate": 1.3141310440048668e-05, + "loss": 1.3205, + "step": 1078500 + }, + { + "epoch": 14.75, + "learning_rate": 1.3124222498667142e-05, + "loss": 1.2948, + "step": 1079000 + }, + { + "epoch": 14.76, + "learning_rate": 1.3107134557285617e-05, + "loss": 1.3632, + "step": 1079500 + }, + { + "epoch": 14.76, + "learning_rate": 1.309004661590409e-05, + "loss": 1.3123, + "step": 1080000 + }, + { + "epoch": 14.77, + "learning_rate": 1.3072958674522565e-05, + "loss": 1.3436, + "step": 1080500 + }, + { + "epoch": 14.78, + "learning_rate": 1.3055870733141037e-05, + "loss": 1.3213, + "step": 1081000 + }, + { + "epoch": 14.78, + "learning_rate": 1.3038782791759513e-05, + "loss": 1.3258, + "step": 1081500 + }, + { + "epoch": 14.79, + "learning_rate": 1.3021694850377985e-05, + "loss": 1.3596, + "step": 1082000 + }, + { + "epoch": 14.8, + "learning_rate": 1.300460690899646e-05, + "loss": 1.3233, + "step": 1082500 + }, + { + "epoch": 14.8, + "learning_rate": 1.2987518967614935e-05, + "loss": 1.3134, + "step": 1083000 + }, + { + "epoch": 14.81, + "learning_rate": 1.297043102623341e-05, + "loss": 1.3519, + "step": 1083500 + }, + { + "epoch": 14.82, + "learning_rate": 1.2953343084851882e-05, + "loss": 1.3361, + "step": 1084000 + }, + { + "epoch": 14.83, + "learning_rate": 1.2936255143470358e-05, + "loss": 1.3459, + "step": 1084500 + }, + { + "epoch": 14.83, + "learning_rate": 1.291916720208883e-05, + "loss": 1.3532, + "step": 1085000 + }, + { + "epoch": 14.84, + "learning_rate": 1.2902079260707306e-05, + "loss": 1.3098, + "step": 1085500 + }, + { + "epoch": 14.85, + "learning_rate": 1.2884991319325778e-05, + "loss": 1.3458, + "step": 1086000 + }, + { + "epoch": 14.85, + "learning_rate": 1.2867903377944254e-05, + "loss": 1.3599, + "step": 1086500 + }, + { + "epoch": 14.86, + "learning_rate": 1.2850815436562728e-05, + "loss": 1.3445, + "step": 1087000 + }, + { + "epoch": 14.87, + "learning_rate": 1.28337274951812e-05, + "loss": 1.3484, + "step": 1087500 + }, + { + "epoch": 14.87, + "learning_rate": 1.2816639553799675e-05, + "loss": 1.3129, + "step": 1088000 + }, + { + "epoch": 14.88, + "learning_rate": 1.2799551612418148e-05, + "loss": 1.3485, + "step": 1088500 + }, + { + "epoch": 14.89, + "learning_rate": 1.2782463671036623e-05, + "loss": 1.3508, + "step": 1089000 + }, + { + "epoch": 14.89, + "learning_rate": 1.2765375729655097e-05, + "loss": 1.3161, + "step": 1089500 + }, + { + "epoch": 14.9, + "learning_rate": 1.2748287788273571e-05, + "loss": 1.3333, + "step": 1090000 + }, + { + "epoch": 14.91, + "learning_rate": 1.2731199846892045e-05, + "loss": 1.3421, + "step": 1090500 + }, + { + "epoch": 14.91, + "learning_rate": 1.271411190551052e-05, + "loss": 1.3349, + "step": 1091000 + }, + { + "epoch": 14.92, + "learning_rate": 1.2697023964128993e-05, + "loss": 1.2992, + "step": 1091500 + }, + { + "epoch": 14.93, + "learning_rate": 1.2679936022747468e-05, + "loss": 1.347, + "step": 1092000 + }, + { + "epoch": 14.93, + "learning_rate": 1.266284808136594e-05, + "loss": 1.3528, + "step": 1092500 + }, + { + "epoch": 14.94, + "learning_rate": 1.2645760139984416e-05, + "loss": 1.3496, + "step": 1093000 + }, + { + "epoch": 14.95, + "learning_rate": 1.262867219860289e-05, + "loss": 1.3188, + "step": 1093500 + }, + { + "epoch": 14.96, + "learning_rate": 1.2611584257221366e-05, + "loss": 1.321, + "step": 1094000 + }, + { + "epoch": 14.96, + "learning_rate": 1.2594496315839838e-05, + "loss": 1.289, + "step": 1094500 + }, + { + "epoch": 14.97, + "learning_rate": 1.2577408374458314e-05, + "loss": 1.3408, + "step": 1095000 + }, + { + "epoch": 14.98, + "learning_rate": 1.2560320433076786e-05, + "loss": 1.3391, + "step": 1095500 + }, + { + "epoch": 14.98, + "learning_rate": 1.2543232491695261e-05, + "loss": 1.3206, + "step": 1096000 + }, + { + "epoch": 14.99, + "learning_rate": 1.2526144550313734e-05, + "loss": 1.33, + "step": 1096500 + }, + { + "epoch": 15.0, + "learning_rate": 1.250905660893221e-05, + "loss": 1.3239, + "step": 1097000 + }, + { + "epoch": 15.0, + "learning_rate": 1.2491968667550683e-05, + "loss": 1.346, + "step": 1097500 + }, + { + "epoch": 15.01, + "learning_rate": 1.2474880726169159e-05, + "loss": 1.3238, + "step": 1098000 + }, + { + "epoch": 15.02, + "learning_rate": 1.2457792784787633e-05, + "loss": 1.2923, + "step": 1098500 + }, + { + "epoch": 15.02, + "learning_rate": 1.2440704843406107e-05, + "loss": 1.3567, + "step": 1099000 + }, + { + "epoch": 15.03, + "learning_rate": 1.242361690202458e-05, + "loss": 1.3256, + "step": 1099500 + }, + { + "epoch": 15.04, + "learning_rate": 1.2406528960643054e-05, + "loss": 1.3291, + "step": 1100000 + }, + { + "epoch": 15.04, + "learning_rate": 1.2389441019261528e-05, + "loss": 1.3163, + "step": 1100500 + }, + { + "epoch": 15.05, + "learning_rate": 1.2372353077880002e-05, + "loss": 1.3098, + "step": 1101000 + }, + { + "epoch": 15.06, + "learning_rate": 1.2355265136498476e-05, + "loss": 1.3252, + "step": 1101500 + }, + { + "epoch": 15.06, + "learning_rate": 1.2338177195116952e-05, + "loss": 1.3114, + "step": 1102000 + }, + { + "epoch": 15.07, + "learning_rate": 1.2321089253735424e-05, + "loss": 1.3328, + "step": 1102500 + }, + { + "epoch": 15.08, + "learning_rate": 1.2304001312353898e-05, + "loss": 1.3283, + "step": 1103000 + }, + { + "epoch": 15.09, + "learning_rate": 1.2286913370972372e-05, + "loss": 1.3359, + "step": 1103500 + }, + { + "epoch": 15.09, + "learning_rate": 1.2269825429590846e-05, + "loss": 1.3094, + "step": 1104000 + }, + { + "epoch": 15.1, + "learning_rate": 1.225273748820932e-05, + "loss": 1.3108, + "step": 1104500 + }, + { + "epoch": 15.11, + "learning_rate": 1.2235649546827795e-05, + "loss": 1.3197, + "step": 1105000 + }, + { + "epoch": 15.11, + "learning_rate": 1.221856160544627e-05, + "loss": 1.3096, + "step": 1105500 + }, + { + "epoch": 15.12, + "learning_rate": 1.2201473664064743e-05, + "loss": 1.2786, + "step": 1106000 + }, + { + "epoch": 15.13, + "learning_rate": 1.2184385722683217e-05, + "loss": 1.2826, + "step": 1106500 + }, + { + "epoch": 15.13, + "learning_rate": 1.2167297781301691e-05, + "loss": 1.3181, + "step": 1107000 + }, + { + "epoch": 15.14, + "learning_rate": 1.2150209839920165e-05, + "loss": 1.3088, + "step": 1107500 + }, + { + "epoch": 15.15, + "learning_rate": 1.2133121898538639e-05, + "loss": 1.3264, + "step": 1108000 + }, + { + "epoch": 15.15, + "learning_rate": 1.2116033957157113e-05, + "loss": 1.3152, + "step": 1108500 + }, + { + "epoch": 15.16, + "learning_rate": 1.2098946015775588e-05, + "loss": 1.3242, + "step": 1109000 + }, + { + "epoch": 15.17, + "learning_rate": 1.2081858074394062e-05, + "loss": 1.3499, + "step": 1109500 + }, + { + "epoch": 15.17, + "learning_rate": 1.2064770133012536e-05, + "loss": 1.3522, + "step": 1110000 + }, + { + "epoch": 15.18, + "learning_rate": 1.204768219163101e-05, + "loss": 1.3184, + "step": 1110500 + }, + { + "epoch": 15.19, + "learning_rate": 1.2030594250249484e-05, + "loss": 1.3014, + "step": 1111000 + }, + { + "epoch": 15.19, + "learning_rate": 1.2013506308867958e-05, + "loss": 1.2929, + "step": 1111500 + }, + { + "epoch": 15.2, + "learning_rate": 1.1996418367486432e-05, + "loss": 1.3226, + "step": 1112000 + }, + { + "epoch": 15.21, + "learning_rate": 1.1979330426104907e-05, + "loss": 1.3005, + "step": 1112500 + }, + { + "epoch": 15.22, + "learning_rate": 1.1962242484723381e-05, + "loss": 1.3346, + "step": 1113000 + }, + { + "epoch": 15.22, + "learning_rate": 1.1945154543341855e-05, + "loss": 1.3301, + "step": 1113500 + }, + { + "epoch": 15.23, + "learning_rate": 1.192806660196033e-05, + "loss": 1.3127, + "step": 1114000 + }, + { + "epoch": 15.24, + "learning_rate": 1.1910978660578803e-05, + "loss": 1.2922, + "step": 1114500 + }, + { + "epoch": 15.24, + "learning_rate": 1.1893890719197277e-05, + "loss": 1.3358, + "step": 1115000 + }, + { + "epoch": 15.25, + "learning_rate": 1.1876802777815751e-05, + "loss": 1.3227, + "step": 1115500 + }, + { + "epoch": 15.26, + "learning_rate": 1.1859714836434225e-05, + "loss": 1.3193, + "step": 1116000 + }, + { + "epoch": 15.26, + "learning_rate": 1.18426268950527e-05, + "loss": 1.3217, + "step": 1116500 + }, + { + "epoch": 15.27, + "learning_rate": 1.1825538953671174e-05, + "loss": 1.3095, + "step": 1117000 + }, + { + "epoch": 15.28, + "learning_rate": 1.1808451012289648e-05, + "loss": 1.3144, + "step": 1117500 + }, + { + "epoch": 15.28, + "learning_rate": 1.1791363070908122e-05, + "loss": 1.313, + "step": 1118000 + }, + { + "epoch": 15.29, + "learning_rate": 1.1774275129526596e-05, + "loss": 1.3129, + "step": 1118500 + }, + { + "epoch": 15.3, + "learning_rate": 1.175718718814507e-05, + "loss": 1.2818, + "step": 1119000 + }, + { + "epoch": 15.3, + "learning_rate": 1.1740099246763544e-05, + "loss": 1.3111, + "step": 1119500 + }, + { + "epoch": 15.31, + "learning_rate": 1.1723011305382018e-05, + "loss": 1.3078, + "step": 1120000 + }, + { + "epoch": 15.32, + "learning_rate": 1.1705923364000493e-05, + "loss": 1.3398, + "step": 1120500 + }, + { + "epoch": 15.32, + "learning_rate": 1.1688835422618967e-05, + "loss": 1.3084, + "step": 1121000 + }, + { + "epoch": 15.33, + "learning_rate": 1.1671747481237441e-05, + "loss": 1.3127, + "step": 1121500 + }, + { + "epoch": 15.34, + "learning_rate": 1.1654659539855915e-05, + "loss": 1.3423, + "step": 1122000 + }, + { + "epoch": 15.34, + "learning_rate": 1.1637571598474389e-05, + "loss": 1.3088, + "step": 1122500 + }, + { + "epoch": 15.35, + "learning_rate": 1.1620483657092863e-05, + "loss": 1.3047, + "step": 1123000 + }, + { + "epoch": 15.36, + "learning_rate": 1.1603395715711337e-05, + "loss": 1.3182, + "step": 1123500 + }, + { + "epoch": 15.37, + "learning_rate": 1.1586307774329813e-05, + "loss": 1.3164, + "step": 1124000 + }, + { + "epoch": 15.37, + "learning_rate": 1.1569219832948286e-05, + "loss": 1.3345, + "step": 1124500 + }, + { + "epoch": 15.38, + "learning_rate": 1.155213189156676e-05, + "loss": 1.2972, + "step": 1125000 + }, + { + "epoch": 15.39, + "learning_rate": 1.1535043950185234e-05, + "loss": 1.3078, + "step": 1125500 + }, + { + "epoch": 15.39, + "learning_rate": 1.1517956008803708e-05, + "loss": 1.3504, + "step": 1126000 + }, + { + "epoch": 15.4, + "learning_rate": 1.1500868067422182e-05, + "loss": 1.2945, + "step": 1126500 + }, + { + "epoch": 15.41, + "learning_rate": 1.1483780126040656e-05, + "loss": 1.3067, + "step": 1127000 + }, + { + "epoch": 15.41, + "learning_rate": 1.146669218465913e-05, + "loss": 1.333, + "step": 1127500 + }, + { + "epoch": 15.42, + "learning_rate": 1.1449604243277606e-05, + "loss": 1.3218, + "step": 1128000 + }, + { + "epoch": 15.43, + "learning_rate": 1.143251630189608e-05, + "loss": 1.32, + "step": 1128500 + }, + { + "epoch": 15.43, + "learning_rate": 1.1415428360514553e-05, + "loss": 1.2754, + "step": 1129000 + }, + { + "epoch": 15.44, + "learning_rate": 1.1398340419133027e-05, + "loss": 1.3235, + "step": 1129500 + }, + { + "epoch": 15.45, + "learning_rate": 1.1381252477751501e-05, + "loss": 1.2847, + "step": 1130000 + }, + { + "epoch": 15.45, + "learning_rate": 1.1364164536369975e-05, + "loss": 1.3307, + "step": 1130500 + }, + { + "epoch": 15.46, + "learning_rate": 1.1347076594988449e-05, + "loss": 1.3093, + "step": 1131000 + }, + { + "epoch": 15.47, + "learning_rate": 1.1329988653606925e-05, + "loss": 1.3203, + "step": 1131500 + }, + { + "epoch": 15.47, + "learning_rate": 1.1312900712225399e-05, + "loss": 1.3006, + "step": 1132000 + }, + { + "epoch": 15.48, + "learning_rate": 1.1295812770843873e-05, + "loss": 1.3029, + "step": 1132500 + }, + { + "epoch": 15.49, + "learning_rate": 1.1278724829462345e-05, + "loss": 1.2848, + "step": 1133000 + }, + { + "epoch": 15.5, + "learning_rate": 1.1261636888080819e-05, + "loss": 1.3441, + "step": 1133500 + }, + { + "epoch": 15.5, + "learning_rate": 1.1244548946699293e-05, + "loss": 1.316, + "step": 1134000 + }, + { + "epoch": 15.51, + "learning_rate": 1.1227461005317766e-05, + "loss": 1.3179, + "step": 1134500 + }, + { + "epoch": 15.52, + "learning_rate": 1.1210373063936242e-05, + "loss": 1.33, + "step": 1135000 + }, + { + "epoch": 15.52, + "learning_rate": 1.1193285122554716e-05, + "loss": 1.3019, + "step": 1135500 + }, + { + "epoch": 15.53, + "learning_rate": 1.117619718117319e-05, + "loss": 1.303, + "step": 1136000 + }, + { + "epoch": 15.54, + "learning_rate": 1.1159109239791664e-05, + "loss": 1.3238, + "step": 1136500 + }, + { + "epoch": 15.54, + "learning_rate": 1.1142021298410138e-05, + "loss": 1.3403, + "step": 1137000 + }, + { + "epoch": 15.55, + "learning_rate": 1.1124933357028612e-05, + "loss": 1.3096, + "step": 1137500 + }, + { + "epoch": 15.56, + "learning_rate": 1.1107845415647086e-05, + "loss": 1.3184, + "step": 1138000 + }, + { + "epoch": 15.56, + "learning_rate": 1.1090757474265561e-05, + "loss": 1.3158, + "step": 1138500 + }, + { + "epoch": 15.57, + "learning_rate": 1.1073669532884035e-05, + "loss": 1.3144, + "step": 1139000 + }, + { + "epoch": 15.58, + "learning_rate": 1.1056581591502509e-05, + "loss": 1.313, + "step": 1139500 + }, + { + "epoch": 15.58, + "learning_rate": 1.1039493650120983e-05, + "loss": 1.3238, + "step": 1140000 + }, + { + "epoch": 15.59, + "learning_rate": 1.1022405708739457e-05, + "loss": 1.3135, + "step": 1140500 + }, + { + "epoch": 15.6, + "learning_rate": 1.100531776735793e-05, + "loss": 1.306, + "step": 1141000 + }, + { + "epoch": 15.6, + "learning_rate": 1.0988229825976405e-05, + "loss": 1.3378, + "step": 1141500 + }, + { + "epoch": 15.61, + "learning_rate": 1.0971141884594879e-05, + "loss": 1.3357, + "step": 1142000 + }, + { + "epoch": 15.62, + "learning_rate": 1.0954053943213354e-05, + "loss": 1.2928, + "step": 1142500 + }, + { + "epoch": 15.63, + "learning_rate": 1.0936966001831828e-05, + "loss": 1.3265, + "step": 1143000 + }, + { + "epoch": 15.63, + "learning_rate": 1.0919878060450302e-05, + "loss": 1.3176, + "step": 1143500 + }, + { + "epoch": 15.64, + "learning_rate": 1.0902790119068776e-05, + "loss": 1.3029, + "step": 1144000 + }, + { + "epoch": 15.65, + "learning_rate": 1.088570217768725e-05, + "loss": 1.3221, + "step": 1144500 + }, + { + "epoch": 15.65, + "learning_rate": 1.0868614236305724e-05, + "loss": 1.3083, + "step": 1145000 + }, + { + "epoch": 15.66, + "learning_rate": 1.0851526294924198e-05, + "loss": 1.3009, + "step": 1145500 + }, + { + "epoch": 15.67, + "learning_rate": 1.0834438353542672e-05, + "loss": 1.3268, + "step": 1146000 + }, + { + "epoch": 15.67, + "learning_rate": 1.0817350412161147e-05, + "loss": 1.3132, + "step": 1146500 + }, + { + "epoch": 15.68, + "learning_rate": 1.0800262470779621e-05, + "loss": 1.3197, + "step": 1147000 + }, + { + "epoch": 15.69, + "learning_rate": 1.0783174529398095e-05, + "loss": 1.2956, + "step": 1147500 + }, + { + "epoch": 15.69, + "learning_rate": 1.0766086588016569e-05, + "loss": 1.288, + "step": 1148000 + }, + { + "epoch": 15.7, + "learning_rate": 1.0748998646635043e-05, + "loss": 1.3244, + "step": 1148500 + }, + { + "epoch": 15.71, + "learning_rate": 1.0731910705253517e-05, + "loss": 1.3092, + "step": 1149000 + }, + { + "epoch": 15.71, + "learning_rate": 1.071482276387199e-05, + "loss": 1.3009, + "step": 1149500 + }, + { + "epoch": 15.72, + "learning_rate": 1.0697734822490466e-05, + "loss": 1.3076, + "step": 1150000 + }, + { + "epoch": 15.73, + "learning_rate": 1.068064688110894e-05, + "loss": 1.3036, + "step": 1150500 + }, + { + "epoch": 15.73, + "learning_rate": 1.0663558939727414e-05, + "loss": 1.3148, + "step": 1151000 + }, + { + "epoch": 15.74, + "learning_rate": 1.0646470998345888e-05, + "loss": 1.3112, + "step": 1151500 + }, + { + "epoch": 15.75, + "learning_rate": 1.0629383056964362e-05, + "loss": 1.3055, + "step": 1152000 + }, + { + "epoch": 15.76, + "learning_rate": 1.0612295115582836e-05, + "loss": 1.33, + "step": 1152500 + }, + { + "epoch": 15.76, + "learning_rate": 1.059520717420131e-05, + "loss": 1.3226, + "step": 1153000 + }, + { + "epoch": 15.77, + "learning_rate": 1.0578119232819784e-05, + "loss": 1.3306, + "step": 1153500 + }, + { + "epoch": 15.78, + "learning_rate": 1.056103129143826e-05, + "loss": 1.3358, + "step": 1154000 + }, + { + "epoch": 15.78, + "learning_rate": 1.0543943350056733e-05, + "loss": 1.3089, + "step": 1154500 + }, + { + "epoch": 15.79, + "learning_rate": 1.0526855408675207e-05, + "loss": 1.3189, + "step": 1155000 + }, + { + "epoch": 15.8, + "learning_rate": 1.0509767467293681e-05, + "loss": 1.3419, + "step": 1155500 + }, + { + "epoch": 15.8, + "learning_rate": 1.0492679525912155e-05, + "loss": 1.3014, + "step": 1156000 + }, + { + "epoch": 15.81, + "learning_rate": 1.0475591584530629e-05, + "loss": 1.3235, + "step": 1156500 + }, + { + "epoch": 15.82, + "learning_rate": 1.0458503643149103e-05, + "loss": 1.3359, + "step": 1157000 + }, + { + "epoch": 15.82, + "learning_rate": 1.0441415701767577e-05, + "loss": 1.3445, + "step": 1157500 + }, + { + "epoch": 15.83, + "learning_rate": 1.0424327760386052e-05, + "loss": 1.2798, + "step": 1158000 + }, + { + "epoch": 15.84, + "learning_rate": 1.0407239819004526e-05, + "loss": 1.283, + "step": 1158500 + }, + { + "epoch": 15.84, + "learning_rate": 1.0390151877623e-05, + "loss": 1.2817, + "step": 1159000 + }, + { + "epoch": 15.85, + "learning_rate": 1.0373063936241474e-05, + "loss": 1.3172, + "step": 1159500 + }, + { + "epoch": 15.86, + "learning_rate": 1.0355975994859948e-05, + "loss": 1.307, + "step": 1160000 + }, + { + "epoch": 15.86, + "learning_rate": 1.0338888053478422e-05, + "loss": 1.2882, + "step": 1160500 + }, + { + "epoch": 15.87, + "learning_rate": 1.0321800112096896e-05, + "loss": 1.3118, + "step": 1161000 + }, + { + "epoch": 15.88, + "learning_rate": 1.0304712170715372e-05, + "loss": 1.2914, + "step": 1161500 + }, + { + "epoch": 15.88, + "learning_rate": 1.0287624229333845e-05, + "loss": 1.328, + "step": 1162000 + }, + { + "epoch": 15.89, + "learning_rate": 1.027053628795232e-05, + "loss": 1.3196, + "step": 1162500 + }, + { + "epoch": 15.9, + "learning_rate": 1.0253448346570792e-05, + "loss": 1.334, + "step": 1163000 + }, + { + "epoch": 15.91, + "learning_rate": 1.0236360405189265e-05, + "loss": 1.3133, + "step": 1163500 + }, + { + "epoch": 15.91, + "learning_rate": 1.021927246380774e-05, + "loss": 1.3193, + "step": 1164000 + }, + { + "epoch": 15.92, + "learning_rate": 1.0202184522426213e-05, + "loss": 1.3448, + "step": 1164500 + }, + { + "epoch": 15.93, + "learning_rate": 1.0185096581044689e-05, + "loss": 1.3235, + "step": 1165000 + }, + { + "epoch": 15.93, + "learning_rate": 1.0168008639663163e-05, + "loss": 1.2919, + "step": 1165500 + }, + { + "epoch": 15.94, + "learning_rate": 1.0150920698281637e-05, + "loss": 1.2999, + "step": 1166000 + }, + { + "epoch": 15.95, + "learning_rate": 1.013383275690011e-05, + "loss": 1.2879, + "step": 1166500 + }, + { + "epoch": 15.95, + "learning_rate": 1.0116744815518585e-05, + "loss": 1.3026, + "step": 1167000 + }, + { + "epoch": 15.96, + "learning_rate": 1.0099656874137059e-05, + "loss": 1.3319, + "step": 1167500 + }, + { + "epoch": 15.97, + "learning_rate": 1.0082568932755532e-05, + "loss": 1.3293, + "step": 1168000 + }, + { + "epoch": 15.97, + "learning_rate": 1.0065480991374008e-05, + "loss": 1.3274, + "step": 1168500 + }, + { + "epoch": 15.98, + "learning_rate": 1.0048393049992482e-05, + "loss": 1.328, + "step": 1169000 + }, + { + "epoch": 15.99, + "learning_rate": 1.0031305108610956e-05, + "loss": 1.3017, + "step": 1169500 + }, + { + "epoch": 15.99, + "learning_rate": 1.001421716722943e-05, + "loss": 1.3092, + "step": 1170000 + }, + { + "epoch": 16.0, + "learning_rate": 9.997129225847904e-06, + "loss": 1.3214, + "step": 1170500 + }, + { + "epoch": 16.01, + "learning_rate": 9.980041284466378e-06, + "loss": 1.3097, + "step": 1171000 + }, + { + "epoch": 16.01, + "learning_rate": 9.962953343084852e-06, + "loss": 1.3306, + "step": 1171500 + }, + { + "epoch": 16.02, + "learning_rate": 9.945865401703325e-06, + "loss": 1.305, + "step": 1172000 + }, + { + "epoch": 16.03, + "learning_rate": 9.928777460321801e-06, + "loss": 1.2962, + "step": 1172500 + }, + { + "epoch": 16.04, + "learning_rate": 9.911689518940275e-06, + "loss": 1.2928, + "step": 1173000 + }, + { + "epoch": 16.04, + "learning_rate": 9.894601577558749e-06, + "loss": 1.3125, + "step": 1173500 + }, + { + "epoch": 16.05, + "learning_rate": 9.877513636177223e-06, + "loss": 1.2934, + "step": 1174000 + }, + { + "epoch": 16.06, + "learning_rate": 9.860425694795697e-06, + "loss": 1.3043, + "step": 1174500 + }, + { + "epoch": 16.06, + "learning_rate": 9.84333775341417e-06, + "loss": 1.27, + "step": 1175000 + }, + { + "epoch": 16.07, + "learning_rate": 9.826249812032645e-06, + "loss": 1.2875, + "step": 1175500 + }, + { + "epoch": 16.08, + "learning_rate": 9.809161870651118e-06, + "loss": 1.2998, + "step": 1176000 + }, + { + "epoch": 16.08, + "learning_rate": 9.792073929269594e-06, + "loss": 1.3044, + "step": 1176500 + }, + { + "epoch": 16.09, + "learning_rate": 9.774985987888068e-06, + "loss": 1.3005, + "step": 1177000 + }, + { + "epoch": 16.1, + "learning_rate": 9.757898046506542e-06, + "loss": 1.289, + "step": 1177500 + }, + { + "epoch": 16.1, + "learning_rate": 9.740810105125016e-06, + "loss": 1.3049, + "step": 1178000 + }, + { + "epoch": 16.11, + "learning_rate": 9.72372216374349e-06, + "loss": 1.2788, + "step": 1178500 + }, + { + "epoch": 16.12, + "learning_rate": 9.706634222361964e-06, + "loss": 1.31, + "step": 1179000 + }, + { + "epoch": 16.12, + "learning_rate": 9.689546280980438e-06, + "loss": 1.2912, + "step": 1179500 + }, + { + "epoch": 16.13, + "learning_rate": 9.672458339598913e-06, + "loss": 1.2876, + "step": 1180000 + }, + { + "epoch": 16.14, + "learning_rate": 9.655370398217387e-06, + "loss": 1.2775, + "step": 1180500 + }, + { + "epoch": 16.14, + "learning_rate": 9.638282456835861e-06, + "loss": 1.2865, + "step": 1181000 + }, + { + "epoch": 16.15, + "learning_rate": 9.621194515454335e-06, + "loss": 1.2904, + "step": 1181500 + }, + { + "epoch": 16.16, + "learning_rate": 9.604106574072809e-06, + "loss": 1.3168, + "step": 1182000 + }, + { + "epoch": 16.17, + "learning_rate": 9.587018632691283e-06, + "loss": 1.2556, + "step": 1182500 + }, + { + "epoch": 16.17, + "learning_rate": 9.569930691309757e-06, + "loss": 1.3124, + "step": 1183000 + }, + { + "epoch": 16.18, + "learning_rate": 9.55284274992823e-06, + "loss": 1.3078, + "step": 1183500 + }, + { + "epoch": 16.19, + "learning_rate": 9.535754808546706e-06, + "loss": 1.2964, + "step": 1184000 + }, + { + "epoch": 16.19, + "learning_rate": 9.51866686716518e-06, + "loss": 1.2765, + "step": 1184500 + }, + { + "epoch": 16.2, + "learning_rate": 9.501578925783654e-06, + "loss": 1.3125, + "step": 1185000 + }, + { + "epoch": 16.21, + "learning_rate": 9.484490984402128e-06, + "loss": 1.3093, + "step": 1185500 + }, + { + "epoch": 16.21, + "learning_rate": 9.467403043020602e-06, + "loss": 1.2985, + "step": 1186000 + }, + { + "epoch": 16.22, + "learning_rate": 9.450315101639076e-06, + "loss": 1.2998, + "step": 1186500 + }, + { + "epoch": 16.23, + "learning_rate": 9.43322716025755e-06, + "loss": 1.3191, + "step": 1187000 + }, + { + "epoch": 16.23, + "learning_rate": 9.416139218876025e-06, + "loss": 1.2874, + "step": 1187500 + }, + { + "epoch": 16.24, + "learning_rate": 9.3990512774945e-06, + "loss": 1.2947, + "step": 1188000 + }, + { + "epoch": 16.25, + "learning_rate": 9.381963336112973e-06, + "loss": 1.2937, + "step": 1188500 + }, + { + "epoch": 16.25, + "learning_rate": 9.364875394731447e-06, + "loss": 1.3093, + "step": 1189000 + }, + { + "epoch": 16.26, + "learning_rate": 9.347787453349921e-06, + "loss": 1.2875, + "step": 1189500 + }, + { + "epoch": 16.27, + "learning_rate": 9.330699511968395e-06, + "loss": 1.2669, + "step": 1190000 + }, + { + "epoch": 16.27, + "learning_rate": 9.313611570586869e-06, + "loss": 1.2725, + "step": 1190500 + }, + { + "epoch": 16.28, + "learning_rate": 9.296523629205343e-06, + "loss": 1.3211, + "step": 1191000 + }, + { + "epoch": 16.29, + "learning_rate": 9.279435687823818e-06, + "loss": 1.2576, + "step": 1191500 + }, + { + "epoch": 16.3, + "learning_rate": 9.262347746442292e-06, + "loss": 1.2941, + "step": 1192000 + }, + { + "epoch": 16.3, + "learning_rate": 9.245259805060766e-06, + "loss": 1.291, + "step": 1192500 + }, + { + "epoch": 16.31, + "learning_rate": 9.22817186367924e-06, + "loss": 1.2876, + "step": 1193000 + }, + { + "epoch": 16.32, + "learning_rate": 9.211083922297712e-06, + "loss": 1.3025, + "step": 1193500 + }, + { + "epoch": 16.32, + "learning_rate": 9.193995980916186e-06, + "loss": 1.2883, + "step": 1194000 + }, + { + "epoch": 16.33, + "learning_rate": 9.176908039534662e-06, + "loss": 1.2828, + "step": 1194500 + }, + { + "epoch": 16.34, + "learning_rate": 9.159820098153136e-06, + "loss": 1.2884, + "step": 1195000 + }, + { + "epoch": 16.34, + "learning_rate": 9.14273215677161e-06, + "loss": 1.3054, + "step": 1195500 + }, + { + "epoch": 16.35, + "learning_rate": 9.125644215390084e-06, + "loss": 1.2922, + "step": 1196000 + }, + { + "epoch": 16.36, + "learning_rate": 9.108556274008557e-06, + "loss": 1.31, + "step": 1196500 + }, + { + "epoch": 16.36, + "learning_rate": 9.091468332627031e-06, + "loss": 1.3077, + "step": 1197000 + }, + { + "epoch": 16.37, + "learning_rate": 9.074380391245505e-06, + "loss": 1.2931, + "step": 1197500 + }, + { + "epoch": 16.38, + "learning_rate": 9.05729244986398e-06, + "loss": 1.3092, + "step": 1198000 + }, + { + "epoch": 16.38, + "learning_rate": 9.040204508482455e-06, + "loss": 1.2903, + "step": 1198500 + }, + { + "epoch": 16.39, + "learning_rate": 9.023116567100929e-06, + "loss": 1.2762, + "step": 1199000 + }, + { + "epoch": 16.4, + "learning_rate": 9.006028625719403e-06, + "loss": 1.2965, + "step": 1199500 + }, + { + "epoch": 16.4, + "learning_rate": 8.988940684337877e-06, + "loss": 1.2949, + "step": 1200000 + }, + { + "epoch": 16.41, + "learning_rate": 8.97185274295635e-06, + "loss": 1.3042, + "step": 1200500 + }, + { + "epoch": 16.42, + "learning_rate": 8.954764801574824e-06, + "loss": 1.261, + "step": 1201000 + }, + { + "epoch": 16.42, + "learning_rate": 8.937676860193298e-06, + "loss": 1.3111, + "step": 1201500 + }, + { + "epoch": 16.43, + "learning_rate": 8.920588918811772e-06, + "loss": 1.2922, + "step": 1202000 + }, + { + "epoch": 16.44, + "learning_rate": 8.903500977430248e-06, + "loss": 1.3004, + "step": 1202500 + }, + { + "epoch": 16.45, + "learning_rate": 8.886413036048722e-06, + "loss": 1.2634, + "step": 1203000 + }, + { + "epoch": 16.45, + "learning_rate": 8.869325094667196e-06, + "loss": 1.2704, + "step": 1203500 + }, + { + "epoch": 16.46, + "learning_rate": 8.85223715328567e-06, + "loss": 1.2686, + "step": 1204000 + }, + { + "epoch": 16.47, + "learning_rate": 8.835149211904144e-06, + "loss": 1.2916, + "step": 1204500 + }, + { + "epoch": 16.47, + "learning_rate": 8.818061270522617e-06, + "loss": 1.303, + "step": 1205000 + }, + { + "epoch": 16.48, + "learning_rate": 8.800973329141091e-06, + "loss": 1.3124, + "step": 1205500 + }, + { + "epoch": 16.49, + "learning_rate": 8.783885387759567e-06, + "loss": 1.2906, + "step": 1206000 + }, + { + "epoch": 16.49, + "learning_rate": 8.766797446378041e-06, + "loss": 1.3153, + "step": 1206500 + }, + { + "epoch": 16.5, + "learning_rate": 8.749709504996515e-06, + "loss": 1.3059, + "step": 1207000 + }, + { + "epoch": 16.51, + "learning_rate": 8.732621563614989e-06, + "loss": 1.2763, + "step": 1207500 + }, + { + "epoch": 16.51, + "learning_rate": 8.715533622233463e-06, + "loss": 1.3012, + "step": 1208000 + }, + { + "epoch": 16.52, + "learning_rate": 8.698445680851937e-06, + "loss": 1.2966, + "step": 1208500 + }, + { + "epoch": 16.53, + "learning_rate": 8.68135773947041e-06, + "loss": 1.2821, + "step": 1209000 + }, + { + "epoch": 16.53, + "learning_rate": 8.664269798088884e-06, + "loss": 1.3103, + "step": 1209500 + }, + { + "epoch": 16.54, + "learning_rate": 8.64718185670736e-06, + "loss": 1.3074, + "step": 1210000 + }, + { + "epoch": 16.55, + "learning_rate": 8.630093915325834e-06, + "loss": 1.2932, + "step": 1210500 + }, + { + "epoch": 16.55, + "learning_rate": 8.613005973944308e-06, + "loss": 1.2867, + "step": 1211000 + }, + { + "epoch": 16.56, + "learning_rate": 8.595918032562782e-06, + "loss": 1.297, + "step": 1211500 + }, + { + "epoch": 16.57, + "learning_rate": 8.578830091181256e-06, + "loss": 1.293, + "step": 1212000 + }, + { + "epoch": 16.58, + "learning_rate": 8.56174214979973e-06, + "loss": 1.2762, + "step": 1212500 + }, + { + "epoch": 16.58, + "learning_rate": 8.544654208418203e-06, + "loss": 1.2876, + "step": 1213000 + }, + { + "epoch": 16.59, + "learning_rate": 8.527566267036677e-06, + "loss": 1.3001, + "step": 1213500 + }, + { + "epoch": 16.6, + "learning_rate": 8.510478325655153e-06, + "loss": 1.2819, + "step": 1214000 + }, + { + "epoch": 16.6, + "learning_rate": 8.493390384273627e-06, + "loss": 1.3149, + "step": 1214500 + }, + { + "epoch": 16.61, + "learning_rate": 8.4763024428921e-06, + "loss": 1.2575, + "step": 1215000 + }, + { + "epoch": 16.62, + "learning_rate": 8.459214501510575e-06, + "loss": 1.2764, + "step": 1215500 + }, + { + "epoch": 16.62, + "learning_rate": 8.442126560129049e-06, + "loss": 1.2647, + "step": 1216000 + }, + { + "epoch": 16.63, + "learning_rate": 8.425038618747523e-06, + "loss": 1.2796, + "step": 1216500 + }, + { + "epoch": 16.64, + "learning_rate": 8.407950677365997e-06, + "loss": 1.3069, + "step": 1217000 + }, + { + "epoch": 16.64, + "learning_rate": 8.390862735984472e-06, + "loss": 1.2584, + "step": 1217500 + }, + { + "epoch": 16.65, + "learning_rate": 8.373774794602946e-06, + "loss": 1.3012, + "step": 1218000 + }, + { + "epoch": 16.66, + "learning_rate": 8.35668685322142e-06, + "loss": 1.301, + "step": 1218500 + }, + { + "epoch": 16.66, + "learning_rate": 8.339598911839894e-06, + "loss": 1.3091, + "step": 1219000 + }, + { + "epoch": 16.67, + "learning_rate": 8.322510970458368e-06, + "loss": 1.2601, + "step": 1219500 + }, + { + "epoch": 16.68, + "learning_rate": 8.305423029076842e-06, + "loss": 1.2828, + "step": 1220000 + }, + { + "epoch": 16.68, + "learning_rate": 8.288335087695316e-06, + "loss": 1.2796, + "step": 1220500 + }, + { + "epoch": 16.69, + "learning_rate": 8.27124714631379e-06, + "loss": 1.2677, + "step": 1221000 + }, + { + "epoch": 16.7, + "learning_rate": 8.254159204932265e-06, + "loss": 1.2861, + "step": 1221500 + }, + { + "epoch": 16.71, + "learning_rate": 8.237071263550739e-06, + "loss": 1.3139, + "step": 1222000 + }, + { + "epoch": 16.71, + "learning_rate": 8.219983322169213e-06, + "loss": 1.297, + "step": 1222500 + }, + { + "epoch": 16.72, + "learning_rate": 8.202895380787687e-06, + "loss": 1.2778, + "step": 1223000 + }, + { + "epoch": 16.73, + "learning_rate": 8.185807439406159e-06, + "loss": 1.2816, + "step": 1223500 + }, + { + "epoch": 16.73, + "learning_rate": 8.168719498024633e-06, + "loss": 1.3015, + "step": 1224000 + }, + { + "epoch": 16.74, + "learning_rate": 8.151631556643109e-06, + "loss": 1.2973, + "step": 1224500 + }, + { + "epoch": 16.75, + "learning_rate": 8.134543615261583e-06, + "loss": 1.2984, + "step": 1225000 + }, + { + "epoch": 16.75, + "learning_rate": 8.117455673880056e-06, + "loss": 1.2942, + "step": 1225500 + }, + { + "epoch": 16.76, + "learning_rate": 8.10036773249853e-06, + "loss": 1.2744, + "step": 1226000 + }, + { + "epoch": 16.77, + "learning_rate": 8.083279791117004e-06, + "loss": 1.2996, + "step": 1226500 + }, + { + "epoch": 16.77, + "learning_rate": 8.066191849735478e-06, + "loss": 1.2885, + "step": 1227000 + }, + { + "epoch": 16.78, + "learning_rate": 8.049103908353952e-06, + "loss": 1.3102, + "step": 1227500 + }, + { + "epoch": 16.79, + "learning_rate": 8.032015966972426e-06, + "loss": 1.2964, + "step": 1228000 + }, + { + "epoch": 16.79, + "learning_rate": 8.014928025590902e-06, + "loss": 1.3049, + "step": 1228500 + }, + { + "epoch": 16.8, + "learning_rate": 7.997840084209376e-06, + "loss": 1.2997, + "step": 1229000 + }, + { + "epoch": 16.81, + "learning_rate": 7.98075214282785e-06, + "loss": 1.2985, + "step": 1229500 + }, + { + "epoch": 16.81, + "learning_rate": 7.963664201446323e-06, + "loss": 1.2783, + "step": 1230000 + }, + { + "epoch": 16.82, + "learning_rate": 7.946576260064797e-06, + "loss": 1.2875, + "step": 1230500 + }, + { + "epoch": 16.83, + "learning_rate": 7.929488318683271e-06, + "loss": 1.278, + "step": 1231000 + }, + { + "epoch": 16.84, + "learning_rate": 7.912400377301745e-06, + "loss": 1.2893, + "step": 1231500 + }, + { + "epoch": 16.84, + "learning_rate": 7.895312435920219e-06, + "loss": 1.3008, + "step": 1232000 + }, + { + "epoch": 16.85, + "learning_rate": 7.878224494538695e-06, + "loss": 1.3221, + "step": 1232500 + }, + { + "epoch": 16.86, + "learning_rate": 7.861136553157169e-06, + "loss": 1.2799, + "step": 1233000 + }, + { + "epoch": 16.86, + "learning_rate": 7.844048611775643e-06, + "loss": 1.2686, + "step": 1233500 + }, + { + "epoch": 16.87, + "learning_rate": 7.826960670394116e-06, + "loss": 1.2944, + "step": 1234000 + }, + { + "epoch": 16.88, + "learning_rate": 7.80987272901259e-06, + "loss": 1.2919, + "step": 1234500 + }, + { + "epoch": 16.88, + "learning_rate": 7.792784787631064e-06, + "loss": 1.2702, + "step": 1235000 + }, + { + "epoch": 16.89, + "learning_rate": 7.775696846249538e-06, + "loss": 1.2889, + "step": 1235500 + }, + { + "epoch": 16.9, + "learning_rate": 7.758608904868014e-06, + "loss": 1.2831, + "step": 1236000 + }, + { + "epoch": 16.9, + "learning_rate": 7.741520963486488e-06, + "loss": 1.2552, + "step": 1236500 + }, + { + "epoch": 16.91, + "learning_rate": 7.724433022104962e-06, + "loss": 1.2891, + "step": 1237000 + }, + { + "epoch": 16.92, + "learning_rate": 7.707345080723436e-06, + "loss": 1.3071, + "step": 1237500 + }, + { + "epoch": 16.92, + "learning_rate": 7.69025713934191e-06, + "loss": 1.2983, + "step": 1238000 + }, + { + "epoch": 16.93, + "learning_rate": 7.673169197960383e-06, + "loss": 1.2558, + "step": 1238500 + }, + { + "epoch": 16.94, + "learning_rate": 7.656081256578857e-06, + "loss": 1.3088, + "step": 1239000 + }, + { + "epoch": 16.94, + "learning_rate": 7.638993315197331e-06, + "loss": 1.2833, + "step": 1239500 + }, + { + "epoch": 16.95, + "learning_rate": 7.621905373815806e-06, + "loss": 1.2668, + "step": 1240000 + }, + { + "epoch": 16.96, + "learning_rate": 7.60481743243428e-06, + "loss": 1.3023, + "step": 1240500 + }, + { + "epoch": 16.96, + "learning_rate": 7.587729491052755e-06, + "loss": 1.2959, + "step": 1241000 + }, + { + "epoch": 16.97, + "learning_rate": 7.5706415496712285e-06, + "loss": 1.2764, + "step": 1241500 + }, + { + "epoch": 16.98, + "learning_rate": 7.5535536082897025e-06, + "loss": 1.3203, + "step": 1242000 + }, + { + "epoch": 16.99, + "learning_rate": 7.536465666908176e-06, + "loss": 1.2776, + "step": 1242500 + }, + { + "epoch": 16.99, + "learning_rate": 7.519377725526651e-06, + "loss": 1.2784, + "step": 1243000 + }, + { + "epoch": 17.0, + "learning_rate": 7.502289784145125e-06, + "loss": 1.2786, + "step": 1243500 + }, + { + "epoch": 17.01, + "learning_rate": 7.485201842763599e-06, + "loss": 1.2937, + "step": 1244000 + }, + { + "epoch": 17.01, + "learning_rate": 7.468113901382074e-06, + "loss": 1.2661, + "step": 1244500 + }, + { + "epoch": 17.02, + "learning_rate": 7.451025960000548e-06, + "loss": 1.2704, + "step": 1245000 + }, + { + "epoch": 17.03, + "learning_rate": 7.4339380186190216e-06, + "loss": 1.2813, + "step": 1245500 + }, + { + "epoch": 17.03, + "learning_rate": 7.4168500772374955e-06, + "loss": 1.2902, + "step": 1246000 + }, + { + "epoch": 17.04, + "learning_rate": 7.39976213585597e-06, + "loss": 1.2595, + "step": 1246500 + }, + { + "epoch": 17.05, + "learning_rate": 7.382674194474444e-06, + "loss": 1.2583, + "step": 1247000 + }, + { + "epoch": 17.05, + "learning_rate": 7.365586253092918e-06, + "loss": 1.2809, + "step": 1247500 + }, + { + "epoch": 17.06, + "learning_rate": 7.348498311711392e-06, + "loss": 1.2713, + "step": 1248000 + }, + { + "epoch": 17.07, + "learning_rate": 7.331410370329867e-06, + "loss": 1.2763, + "step": 1248500 + }, + { + "epoch": 17.07, + "learning_rate": 7.314322428948341e-06, + "loss": 1.2767, + "step": 1249000 + }, + { + "epoch": 17.08, + "learning_rate": 7.297234487566815e-06, + "loss": 1.2754, + "step": 1249500 + }, + { + "epoch": 17.09, + "learning_rate": 7.2801465461852885e-06, + "loss": 1.2637, + "step": 1250000 + }, + { + "epoch": 17.09, + "learning_rate": 7.263058604803763e-06, + "loss": 1.2772, + "step": 1250500 + }, + { + "epoch": 17.1, + "learning_rate": 7.245970663422237e-06, + "loss": 1.2993, + "step": 1251000 + }, + { + "epoch": 17.11, + "learning_rate": 7.228882722040711e-06, + "loss": 1.2704, + "step": 1251500 + }, + { + "epoch": 17.12, + "learning_rate": 7.211794780659185e-06, + "loss": 1.2495, + "step": 1252000 + }, + { + "epoch": 17.12, + "learning_rate": 7.19470683927766e-06, + "loss": 1.2884, + "step": 1252500 + }, + { + "epoch": 17.13, + "learning_rate": 7.177618897896134e-06, + "loss": 1.2804, + "step": 1253000 + }, + { + "epoch": 17.14, + "learning_rate": 7.160530956514607e-06, + "loss": 1.2764, + "step": 1253500 + }, + { + "epoch": 17.14, + "learning_rate": 7.143443015133081e-06, + "loss": 1.2886, + "step": 1254000 + }, + { + "epoch": 17.15, + "learning_rate": 7.126355073751555e-06, + "loss": 1.2849, + "step": 1254500 + }, + { + "epoch": 17.16, + "learning_rate": 7.1092671323700285e-06, + "loss": 1.2809, + "step": 1255000 + }, + { + "epoch": 17.16, + "learning_rate": 7.092179190988503e-06, + "loss": 1.2854, + "step": 1255500 + }, + { + "epoch": 17.17, + "learning_rate": 7.075091249606977e-06, + "loss": 1.2558, + "step": 1256000 + }, + { + "epoch": 17.18, + "learning_rate": 7.058003308225451e-06, + "loss": 1.2569, + "step": 1256500 + }, + { + "epoch": 17.18, + "learning_rate": 7.040915366843925e-06, + "loss": 1.2829, + "step": 1257000 + }, + { + "epoch": 17.19, + "learning_rate": 7.0238274254624e-06, + "loss": 1.2746, + "step": 1257500 + }, + { + "epoch": 17.2, + "learning_rate": 7.006739484080874e-06, + "loss": 1.2548, + "step": 1258000 + }, + { + "epoch": 17.2, + "learning_rate": 6.989651542699348e-06, + "loss": 1.253, + "step": 1258500 + }, + { + "epoch": 17.21, + "learning_rate": 6.9725636013178215e-06, + "loss": 1.2458, + "step": 1259000 + }, + { + "epoch": 17.22, + "learning_rate": 6.955475659936296e-06, + "loss": 1.2607, + "step": 1259500 + }, + { + "epoch": 17.22, + "learning_rate": 6.93838771855477e-06, + "loss": 1.2795, + "step": 1260000 + }, + { + "epoch": 17.23, + "learning_rate": 6.921299777173244e-06, + "loss": 1.2709, + "step": 1260500 + }, + { + "epoch": 17.24, + "learning_rate": 6.904211835791719e-06, + "loss": 1.2835, + "step": 1261000 + }, + { + "epoch": 17.25, + "learning_rate": 6.887123894410193e-06, + "loss": 1.2724, + "step": 1261500 + }, + { + "epoch": 17.25, + "learning_rate": 6.870035953028667e-06, + "loss": 1.2592, + "step": 1262000 + }, + { + "epoch": 17.26, + "learning_rate": 6.852948011647141e-06, + "loss": 1.2899, + "step": 1262500 + }, + { + "epoch": 17.27, + "learning_rate": 6.835860070265615e-06, + "loss": 1.2648, + "step": 1263000 + }, + { + "epoch": 17.27, + "learning_rate": 6.818772128884089e-06, + "loss": 1.2713, + "step": 1263500 + }, + { + "epoch": 17.28, + "learning_rate": 6.801684187502563e-06, + "loss": 1.2514, + "step": 1264000 + }, + { + "epoch": 17.29, + "learning_rate": 6.784596246121037e-06, + "loss": 1.2771, + "step": 1264500 + }, + { + "epoch": 17.29, + "learning_rate": 6.767508304739512e-06, + "loss": 1.2676, + "step": 1265000 + }, + { + "epoch": 17.3, + "learning_rate": 6.750420363357986e-06, + "loss": 1.2863, + "step": 1265500 + }, + { + "epoch": 17.31, + "learning_rate": 6.73333242197646e-06, + "loss": 1.2753, + "step": 1266000 + }, + { + "epoch": 17.31, + "learning_rate": 6.716244480594934e-06, + "loss": 1.2752, + "step": 1266500 + }, + { + "epoch": 17.32, + "learning_rate": 6.699156539213408e-06, + "loss": 1.2601, + "step": 1267000 + }, + { + "epoch": 17.33, + "learning_rate": 6.682068597831882e-06, + "loss": 1.2652, + "step": 1267500 + }, + { + "epoch": 17.33, + "learning_rate": 6.664980656450356e-06, + "loss": 1.2678, + "step": 1268000 + }, + { + "epoch": 17.34, + "learning_rate": 6.64789271506883e-06, + "loss": 1.2837, + "step": 1268500 + }, + { + "epoch": 17.35, + "learning_rate": 6.630804773687305e-06, + "loss": 1.2889, + "step": 1269000 + }, + { + "epoch": 17.35, + "learning_rate": 6.613716832305779e-06, + "loss": 1.2646, + "step": 1269500 + }, + { + "epoch": 17.36, + "learning_rate": 6.596628890924253e-06, + "loss": 1.2754, + "step": 1270000 + }, + { + "epoch": 17.37, + "learning_rate": 6.579540949542727e-06, + "loss": 1.2735, + "step": 1270500 + }, + { + "epoch": 17.38, + "learning_rate": 6.5624530081612014e-06, + "loss": 1.2886, + "step": 1271000 + }, + { + "epoch": 17.38, + "learning_rate": 6.545365066779675e-06, + "loss": 1.2797, + "step": 1271500 + }, + { + "epoch": 17.39, + "learning_rate": 6.528277125398149e-06, + "loss": 1.2788, + "step": 1272000 + }, + { + "epoch": 17.4, + "learning_rate": 6.511189184016624e-06, + "loss": 1.2842, + "step": 1272500 + }, + { + "epoch": 17.4, + "learning_rate": 6.494101242635098e-06, + "loss": 1.2614, + "step": 1273000 + }, + { + "epoch": 17.41, + "learning_rate": 6.477013301253572e-06, + "loss": 1.2719, + "step": 1273500 + }, + { + "epoch": 17.42, + "learning_rate": 6.459925359872046e-06, + "loss": 1.2772, + "step": 1274000 + }, + { + "epoch": 17.42, + "learning_rate": 6.4428374184905205e-06, + "loss": 1.2612, + "step": 1274500 + }, + { + "epoch": 17.43, + "learning_rate": 6.4257494771089945e-06, + "loss": 1.2702, + "step": 1275000 + }, + { + "epoch": 17.44, + "learning_rate": 6.408661535727468e-06, + "loss": 1.3107, + "step": 1275500 + }, + { + "epoch": 17.44, + "learning_rate": 6.391573594345942e-06, + "loss": 1.2714, + "step": 1276000 + }, + { + "epoch": 17.45, + "learning_rate": 6.374485652964417e-06, + "loss": 1.278, + "step": 1276500 + }, + { + "epoch": 17.46, + "learning_rate": 6.357397711582891e-06, + "loss": 1.27, + "step": 1277000 + }, + { + "epoch": 17.46, + "learning_rate": 6.340309770201365e-06, + "loss": 1.2773, + "step": 1277500 + }, + { + "epoch": 17.47, + "learning_rate": 6.323221828819839e-06, + "loss": 1.2719, + "step": 1278000 + }, + { + "epoch": 17.48, + "learning_rate": 6.3061338874383136e-06, + "loss": 1.27, + "step": 1278500 + }, + { + "epoch": 17.48, + "learning_rate": 6.2890459460567875e-06, + "loss": 1.2671, + "step": 1279000 + }, + { + "epoch": 17.49, + "learning_rate": 6.271958004675261e-06, + "loss": 1.2431, + "step": 1279500 + }, + { + "epoch": 17.5, + "learning_rate": 6.254870063293735e-06, + "loss": 1.273, + "step": 1280000 + }, + { + "epoch": 17.5, + "learning_rate": 6.237782121912209e-06, + "loss": 1.302, + "step": 1280500 + }, + { + "epoch": 17.51, + "learning_rate": 6.220694180530683e-06, + "loss": 1.2908, + "step": 1281000 + }, + { + "epoch": 17.52, + "learning_rate": 6.203606239149157e-06, + "loss": 1.2839, + "step": 1281500 + }, + { + "epoch": 17.53, + "learning_rate": 6.186518297767632e-06, + "loss": 1.2634, + "step": 1282000 + }, + { + "epoch": 17.53, + "learning_rate": 6.169430356386106e-06, + "loss": 1.2995, + "step": 1282500 + }, + { + "epoch": 17.54, + "learning_rate": 6.15234241500458e-06, + "loss": 1.273, + "step": 1283000 + }, + { + "epoch": 17.55, + "learning_rate": 6.135254473623054e-06, + "loss": 1.2706, + "step": 1283500 + }, + { + "epoch": 17.55, + "learning_rate": 6.118166532241528e-06, + "loss": 1.2635, + "step": 1284000 + }, + { + "epoch": 17.56, + "learning_rate": 6.101078590860002e-06, + "loss": 1.2662, + "step": 1284500 + }, + { + "epoch": 17.57, + "learning_rate": 6.083990649478476e-06, + "loss": 1.2821, + "step": 1285000 + }, + { + "epoch": 17.57, + "learning_rate": 6.066902708096951e-06, + "loss": 1.2625, + "step": 1285500 + }, + { + "epoch": 17.58, + "learning_rate": 6.049814766715425e-06, + "loss": 1.2683, + "step": 1286000 + }, + { + "epoch": 17.59, + "learning_rate": 6.032726825333899e-06, + "loss": 1.2711, + "step": 1286500 + }, + { + "epoch": 17.59, + "learning_rate": 6.015638883952373e-06, + "loss": 1.2908, + "step": 1287000 + }, + { + "epoch": 17.6, + "learning_rate": 5.9985509425708474e-06, + "loss": 1.2803, + "step": 1287500 + }, + { + "epoch": 17.61, + "learning_rate": 5.981463001189321e-06, + "loss": 1.2816, + "step": 1288000 + }, + { + "epoch": 17.61, + "learning_rate": 5.964375059807795e-06, + "loss": 1.2878, + "step": 1288500 + }, + { + "epoch": 17.62, + "learning_rate": 5.947287118426269e-06, + "loss": 1.2811, + "step": 1289000 + }, + { + "epoch": 17.63, + "learning_rate": 5.930199177044744e-06, + "loss": 1.2778, + "step": 1289500 + }, + { + "epoch": 17.63, + "learning_rate": 5.913111235663218e-06, + "loss": 1.2982, + "step": 1290000 + }, + { + "epoch": 17.64, + "learning_rate": 5.896023294281692e-06, + "loss": 1.2937, + "step": 1290500 + }, + { + "epoch": 17.65, + "learning_rate": 5.878935352900166e-06, + "loss": 1.2616, + "step": 1291000 + }, + { + "epoch": 17.66, + "learning_rate": 5.86184741151864e-06, + "loss": 1.2378, + "step": 1291500 + }, + { + "epoch": 17.66, + "learning_rate": 5.8447594701371135e-06, + "loss": 1.2683, + "step": 1292000 + }, + { + "epoch": 17.67, + "learning_rate": 5.8276715287555875e-06, + "loss": 1.2333, + "step": 1292500 + }, + { + "epoch": 17.68, + "learning_rate": 5.810583587374062e-06, + "loss": 1.2728, + "step": 1293000 + }, + { + "epoch": 17.68, + "learning_rate": 5.793495645992536e-06, + "loss": 1.2752, + "step": 1293500 + }, + { + "epoch": 17.69, + "learning_rate": 5.77640770461101e-06, + "loss": 1.2793, + "step": 1294000 + }, + { + "epoch": 17.7, + "learning_rate": 5.759319763229484e-06, + "loss": 1.2491, + "step": 1294500 + }, + { + "epoch": 17.7, + "learning_rate": 5.742231821847959e-06, + "loss": 1.2741, + "step": 1295000 + }, + { + "epoch": 17.71, + "learning_rate": 5.725143880466433e-06, + "loss": 1.2592, + "step": 1295500 + }, + { + "epoch": 17.72, + "learning_rate": 5.7080559390849066e-06, + "loss": 1.291, + "step": 1296000 + }, + { + "epoch": 17.72, + "learning_rate": 5.6909679977033805e-06, + "loss": 1.2765, + "step": 1296500 + }, + { + "epoch": 17.73, + "learning_rate": 5.673880056321855e-06, + "loss": 1.2892, + "step": 1297000 + }, + { + "epoch": 17.74, + "learning_rate": 5.656792114940329e-06, + "loss": 1.2667, + "step": 1297500 + }, + { + "epoch": 17.74, + "learning_rate": 5.639704173558803e-06, + "loss": 1.2882, + "step": 1298000 + }, + { + "epoch": 17.75, + "learning_rate": 5.622616232177277e-06, + "loss": 1.2626, + "step": 1298500 + }, + { + "epoch": 17.76, + "learning_rate": 5.605528290795752e-06, + "loss": 1.2539, + "step": 1299000 + }, + { + "epoch": 17.76, + "learning_rate": 5.588440349414226e-06, + "loss": 1.2396, + "step": 1299500 + }, + { + "epoch": 17.77, + "learning_rate": 5.5713524080327e-06, + "loss": 1.2785, + "step": 1300000 + }, + { + "epoch": 17.78, + "learning_rate": 5.554264466651174e-06, + "loss": 1.2661, + "step": 1300500 + }, + { + "epoch": 17.79, + "learning_rate": 5.537176525269648e-06, + "loss": 1.2936, + "step": 1301000 + }, + { + "epoch": 17.79, + "learning_rate": 5.520088583888122e-06, + "loss": 1.289, + "step": 1301500 + }, + { + "epoch": 17.8, + "learning_rate": 5.503000642506596e-06, + "loss": 1.2579, + "step": 1302000 + }, + { + "epoch": 17.81, + "learning_rate": 5.485912701125071e-06, + "loss": 1.2468, + "step": 1302500 + }, + { + "epoch": 17.81, + "learning_rate": 5.468824759743545e-06, + "loss": 1.2733, + "step": 1303000 + }, + { + "epoch": 17.82, + "learning_rate": 5.451736818362019e-06, + "loss": 1.2459, + "step": 1303500 + }, + { + "epoch": 17.83, + "learning_rate": 5.434648876980493e-06, + "loss": 1.2607, + "step": 1304000 + }, + { + "epoch": 17.83, + "learning_rate": 5.417560935598967e-06, + "loss": 1.2641, + "step": 1304500 + }, + { + "epoch": 17.84, + "learning_rate": 5.400472994217441e-06, + "loss": 1.2564, + "step": 1305000 + }, + { + "epoch": 17.85, + "learning_rate": 5.383385052835915e-06, + "loss": 1.2816, + "step": 1305500 + }, + { + "epoch": 17.85, + "learning_rate": 5.366297111454389e-06, + "loss": 1.2818, + "step": 1306000 + }, + { + "epoch": 17.86, + "learning_rate": 5.349209170072863e-06, + "loss": 1.2594, + "step": 1306500 + }, + { + "epoch": 17.87, + "learning_rate": 5.332121228691337e-06, + "loss": 1.2834, + "step": 1307000 + }, + { + "epoch": 17.87, + "learning_rate": 5.315033287309811e-06, + "loss": 1.2597, + "step": 1307500 + }, + { + "epoch": 17.88, + "learning_rate": 5.297945345928286e-06, + "loss": 1.269, + "step": 1308000 + }, + { + "epoch": 17.89, + "learning_rate": 5.2808574045467595e-06, + "loss": 1.2616, + "step": 1308500 + }, + { + "epoch": 17.89, + "learning_rate": 5.2637694631652335e-06, + "loss": 1.2625, + "step": 1309000 + }, + { + "epoch": 17.9, + "learning_rate": 5.246681521783707e-06, + "loss": 1.2569, + "step": 1309500 + }, + { + "epoch": 17.91, + "learning_rate": 5.229593580402182e-06, + "loss": 1.2458, + "step": 1310000 + }, + { + "epoch": 17.91, + "learning_rate": 5.212505639020656e-06, + "loss": 1.2663, + "step": 1310500 + }, + { + "epoch": 17.92, + "learning_rate": 5.19541769763913e-06, + "loss": 1.2603, + "step": 1311000 + }, + { + "epoch": 17.93, + "learning_rate": 5.178329756257604e-06, + "loss": 1.244, + "step": 1311500 + }, + { + "epoch": 17.94, + "learning_rate": 5.161241814876079e-06, + "loss": 1.253, + "step": 1312000 + }, + { + "epoch": 17.94, + "learning_rate": 5.1441538734945526e-06, + "loss": 1.2727, + "step": 1312500 + }, + { + "epoch": 17.95, + "learning_rate": 5.1270659321130265e-06, + "loss": 1.2569, + "step": 1313000 + }, + { + "epoch": 17.96, + "learning_rate": 5.109977990731501e-06, + "loss": 1.271, + "step": 1313500 + }, + { + "epoch": 17.96, + "learning_rate": 5.092890049349975e-06, + "loss": 1.2637, + "step": 1314000 + }, + { + "epoch": 17.97, + "learning_rate": 5.075802107968449e-06, + "loss": 1.2829, + "step": 1314500 + }, + { + "epoch": 17.98, + "learning_rate": 5.058714166586923e-06, + "loss": 1.2677, + "step": 1315000 + }, + { + "epoch": 17.98, + "learning_rate": 5.041626225205398e-06, + "loss": 1.2554, + "step": 1315500 + }, + { + "epoch": 17.99, + "learning_rate": 5.024538283823872e-06, + "loss": 1.2851, + "step": 1316000 + }, + { + "epoch": 18.0, + "learning_rate": 5.007450342442346e-06, + "loss": 1.2571, + "step": 1316500 + }, + { + "epoch": 18.0, + "learning_rate": 4.9903624010608195e-06, + "loss": 1.2701, + "step": 1317000 + }, + { + "epoch": 18.01, + "learning_rate": 4.973274459679294e-06, + "loss": 1.2223, + "step": 1317500 + }, + { + "epoch": 18.02, + "learning_rate": 4.956186518297768e-06, + "loss": 1.2657, + "step": 1318000 + }, + { + "epoch": 18.02, + "learning_rate": 4.939098576916242e-06, + "loss": 1.2459, + "step": 1318500 + }, + { + "epoch": 18.03, + "learning_rate": 4.922010635534716e-06, + "loss": 1.2478, + "step": 1319000 + }, + { + "epoch": 18.04, + "learning_rate": 4.904922694153191e-06, + "loss": 1.2301, + "step": 1319500 + }, + { + "epoch": 18.04, + "learning_rate": 4.887834752771665e-06, + "loss": 1.2567, + "step": 1320000 + }, + { + "epoch": 18.05, + "learning_rate": 4.870746811390139e-06, + "loss": 1.2359, + "step": 1320500 + }, + { + "epoch": 18.06, + "learning_rate": 4.8536588700086125e-06, + "loss": 1.2517, + "step": 1321000 + }, + { + "epoch": 18.07, + "learning_rate": 4.8365709286270864e-06, + "loss": 1.2588, + "step": 1321500 + }, + { + "epoch": 18.07, + "learning_rate": 4.81948298724556e-06, + "loss": 1.2756, + "step": 1322000 + }, + { + "epoch": 18.08, + "learning_rate": 4.802395045864034e-06, + "loss": 1.2639, + "step": 1322500 + }, + { + "epoch": 18.09, + "learning_rate": 4.785307104482509e-06, + "loss": 1.2727, + "step": 1323000 + }, + { + "epoch": 18.09, + "learning_rate": 4.768219163100983e-06, + "loss": 1.2454, + "step": 1323500 + }, + { + "epoch": 18.1, + "learning_rate": 4.751131221719457e-06, + "loss": 1.2658, + "step": 1324000 + }, + { + "epoch": 18.11, + "learning_rate": 4.734043280337931e-06, + "loss": 1.2411, + "step": 1324500 + }, + { + "epoch": 18.11, + "learning_rate": 4.7169553389564055e-06, + "loss": 1.2471, + "step": 1325000 + }, + { + "epoch": 18.12, + "learning_rate": 4.6998673975748795e-06, + "loss": 1.2687, + "step": 1325500 + }, + { + "epoch": 18.13, + "learning_rate": 4.682779456193353e-06, + "loss": 1.2472, + "step": 1326000 + }, + { + "epoch": 18.13, + "learning_rate": 4.665691514811827e-06, + "loss": 1.2759, + "step": 1326500 + }, + { + "epoch": 18.14, + "learning_rate": 4.648603573430302e-06, + "loss": 1.2665, + "step": 1327000 + }, + { + "epoch": 18.15, + "learning_rate": 4.631515632048776e-06, + "loss": 1.2923, + "step": 1327500 + }, + { + "epoch": 18.15, + "learning_rate": 4.61442769066725e-06, + "loss": 1.2595, + "step": 1328000 + }, + { + "epoch": 18.16, + "learning_rate": 4.597339749285725e-06, + "loss": 1.2556, + "step": 1328500 + }, + { + "epoch": 18.17, + "learning_rate": 4.5802518079041986e-06, + "loss": 1.2446, + "step": 1329000 + }, + { + "epoch": 18.17, + "learning_rate": 4.5631638665226725e-06, + "loss": 1.2288, + "step": 1329500 + }, + { + "epoch": 18.18, + "learning_rate": 4.546075925141146e-06, + "loss": 1.2569, + "step": 1330000 + }, + { + "epoch": 18.19, + "learning_rate": 4.528987983759621e-06, + "loss": 1.2304, + "step": 1330500 + }, + { + "epoch": 18.2, + "learning_rate": 4.511900042378095e-06, + "loss": 1.2598, + "step": 1331000 + }, + { + "epoch": 18.2, + "learning_rate": 4.494812100996569e-06, + "loss": 1.2752, + "step": 1331500 + }, + { + "epoch": 18.21, + "learning_rate": 4.477724159615043e-06, + "loss": 1.2706, + "step": 1332000 + }, + { + "epoch": 18.22, + "learning_rate": 4.460636218233518e-06, + "loss": 1.2532, + "step": 1332500 + }, + { + "epoch": 18.22, + "learning_rate": 4.443548276851992e-06, + "loss": 1.2692, + "step": 1333000 + }, + { + "epoch": 18.23, + "learning_rate": 4.4264603354704655e-06, + "loss": 1.254, + "step": 1333500 + }, + { + "epoch": 18.24, + "learning_rate": 4.409372394088939e-06, + "loss": 1.2525, + "step": 1334000 + }, + { + "epoch": 18.24, + "learning_rate": 4.392284452707414e-06, + "loss": 1.267, + "step": 1334500 + }, + { + "epoch": 18.25, + "learning_rate": 4.375196511325888e-06, + "loss": 1.2849, + "step": 1335000 + }, + { + "epoch": 18.26, + "learning_rate": 4.358108569944362e-06, + "loss": 1.2371, + "step": 1335500 + }, + { + "epoch": 18.26, + "learning_rate": 4.341020628562836e-06, + "loss": 1.2632, + "step": 1336000 + }, + { + "epoch": 18.27, + "learning_rate": 4.32393268718131e-06, + "loss": 1.2666, + "step": 1336500 + }, + { + "epoch": 18.28, + "learning_rate": 4.306844745799784e-06, + "loss": 1.2421, + "step": 1337000 + }, + { + "epoch": 18.28, + "learning_rate": 4.289756804418258e-06, + "loss": 1.2549, + "step": 1337500 + }, + { + "epoch": 18.29, + "learning_rate": 4.2726688630367324e-06, + "loss": 1.2562, + "step": 1338000 + }, + { + "epoch": 18.3, + "learning_rate": 4.255580921655206e-06, + "loss": 1.2713, + "step": 1338500 + }, + { + "epoch": 18.3, + "learning_rate": 4.23849298027368e-06, + "loss": 1.2417, + "step": 1339000 + }, + { + "epoch": 18.31, + "learning_rate": 4.221405038892154e-06, + "loss": 1.2444, + "step": 1339500 + }, + { + "epoch": 18.32, + "learning_rate": 4.204317097510629e-06, + "loss": 1.2648, + "step": 1340000 + }, + { + "epoch": 18.33, + "learning_rate": 4.187229156129103e-06, + "loss": 1.2708, + "step": 1340500 + }, + { + "epoch": 18.33, + "learning_rate": 4.170141214747577e-06, + "loss": 1.2545, + "step": 1341000 + }, + { + "epoch": 18.34, + "learning_rate": 4.1530532733660515e-06, + "loss": 1.2176, + "step": 1341500 + }, + { + "epoch": 18.35, + "learning_rate": 4.1359653319845255e-06, + "loss": 1.243, + "step": 1342000 + }, + { + "epoch": 18.35, + "learning_rate": 4.118877390602999e-06, + "loss": 1.2454, + "step": 1342500 + }, + { + "epoch": 18.36, + "learning_rate": 4.101789449221473e-06, + "loss": 1.2452, + "step": 1343000 + }, + { + "epoch": 18.37, + "learning_rate": 4.084701507839948e-06, + "loss": 1.2675, + "step": 1343500 + }, + { + "epoch": 18.37, + "learning_rate": 4.067613566458422e-06, + "loss": 1.232, + "step": 1344000 + }, + { + "epoch": 18.38, + "learning_rate": 4.050525625076896e-06, + "loss": 1.2174, + "step": 1344500 + }, + { + "epoch": 18.39, + "learning_rate": 4.03343768369537e-06, + "loss": 1.2573, + "step": 1345000 + }, + { + "epoch": 18.39, + "learning_rate": 4.0163497423138446e-06, + "loss": 1.2683, + "step": 1345500 + }, + { + "epoch": 18.4, + "learning_rate": 3.9992618009323185e-06, + "loss": 1.2483, + "step": 1346000 + }, + { + "epoch": 18.41, + "learning_rate": 3.982173859550792e-06, + "loss": 1.2692, + "step": 1346500 + }, + { + "epoch": 18.41, + "learning_rate": 3.965085918169266e-06, + "loss": 1.257, + "step": 1347000 + }, + { + "epoch": 18.42, + "learning_rate": 3.947997976787741e-06, + "loss": 1.2828, + "step": 1347500 + }, + { + "epoch": 18.43, + "learning_rate": 3.930910035406215e-06, + "loss": 1.2508, + "step": 1348000 + }, + { + "epoch": 18.43, + "learning_rate": 3.913822094024689e-06, + "loss": 1.2549, + "step": 1348500 + }, + { + "epoch": 18.44, + "learning_rate": 3.896734152643163e-06, + "loss": 1.2588, + "step": 1349000 + }, + { + "epoch": 18.45, + "learning_rate": 3.879646211261638e-06, + "loss": 1.2442, + "step": 1349500 + }, + { + "epoch": 18.45, + "learning_rate": 3.8625582698801115e-06, + "loss": 1.2581, + "step": 1350000 + }, + { + "epoch": 18.46, + "learning_rate": 3.845470328498585e-06, + "loss": 1.2612, + "step": 1350500 + }, + { + "epoch": 18.47, + "learning_rate": 3.828382387117059e-06, + "loss": 1.29, + "step": 1351000 + }, + { + "epoch": 18.48, + "learning_rate": 3.8112944457355337e-06, + "loss": 1.2701, + "step": 1351500 + }, + { + "epoch": 18.48, + "learning_rate": 3.794206504354007e-06, + "loss": 1.259, + "step": 1352000 + }, + { + "epoch": 18.49, + "learning_rate": 3.7771185629724815e-06, + "loss": 1.2587, + "step": 1352500 + }, + { + "epoch": 18.5, + "learning_rate": 3.7600306215909554e-06, + "loss": 1.2236, + "step": 1353000 + }, + { + "epoch": 18.5, + "learning_rate": 3.7429426802094298e-06, + "loss": 1.273, + "step": 1353500 + }, + { + "epoch": 18.51, + "learning_rate": 3.7258547388279037e-06, + "loss": 1.2286, + "step": 1354000 + }, + { + "epoch": 18.52, + "learning_rate": 3.708766797446378e-06, + "loss": 1.2695, + "step": 1354500 + }, + { + "epoch": 18.52, + "learning_rate": 3.6916788560648523e-06, + "loss": 1.2466, + "step": 1355000 + }, + { + "epoch": 18.53, + "learning_rate": 3.6745909146833263e-06, + "loss": 1.2588, + "step": 1355500 + }, + { + "epoch": 18.54, + "learning_rate": 3.6575029733018006e-06, + "loss": 1.225, + "step": 1356000 + }, + { + "epoch": 18.54, + "learning_rate": 3.6404150319202745e-06, + "loss": 1.2432, + "step": 1356500 + }, + { + "epoch": 18.55, + "learning_rate": 3.623327090538749e-06, + "loss": 1.2672, + "step": 1357000 + }, + { + "epoch": 18.56, + "learning_rate": 3.6062391491572228e-06, + "loss": 1.2876, + "step": 1357500 + }, + { + "epoch": 18.56, + "learning_rate": 3.589151207775697e-06, + "loss": 1.2336, + "step": 1358000 + }, + { + "epoch": 18.57, + "learning_rate": 3.572063266394171e-06, + "loss": 1.2462, + "step": 1358500 + }, + { + "epoch": 18.58, + "learning_rate": 3.5549753250126454e-06, + "loss": 1.217, + "step": 1359000 + }, + { + "epoch": 18.58, + "learning_rate": 3.5378873836311193e-06, + "loss": 1.252, + "step": 1359500 + }, + { + "epoch": 18.59, + "learning_rate": 3.5207994422495936e-06, + "loss": 1.271, + "step": 1360000 + }, + { + "epoch": 18.6, + "learning_rate": 3.5037115008680675e-06, + "loss": 1.2529, + "step": 1360500 + }, + { + "epoch": 18.61, + "learning_rate": 3.486623559486542e-06, + "loss": 1.2531, + "step": 1361000 + }, + { + "epoch": 18.61, + "learning_rate": 3.469535618105016e-06, + "loss": 1.2569, + "step": 1361500 + }, + { + "epoch": 18.62, + "learning_rate": 3.45244767672349e-06, + "loss": 1.2408, + "step": 1362000 + }, + { + "epoch": 18.63, + "learning_rate": 3.435359735341964e-06, + "loss": 1.2626, + "step": 1362500 + }, + { + "epoch": 18.63, + "learning_rate": 3.4182717939604384e-06, + "loss": 1.2358, + "step": 1363000 + }, + { + "epoch": 18.64, + "learning_rate": 3.4011838525789123e-06, + "loss": 1.2461, + "step": 1363500 + }, + { + "epoch": 18.65, + "learning_rate": 3.3840959111973866e-06, + "loss": 1.2681, + "step": 1364000 + }, + { + "epoch": 18.65, + "learning_rate": 3.3670079698158606e-06, + "loss": 1.2701, + "step": 1364500 + }, + { + "epoch": 18.66, + "learning_rate": 3.349920028434335e-06, + "loss": 1.2509, + "step": 1365000 + }, + { + "epoch": 18.67, + "learning_rate": 3.332832087052809e-06, + "loss": 1.2404, + "step": 1365500 + }, + { + "epoch": 18.67, + "learning_rate": 3.315744145671283e-06, + "loss": 1.2353, + "step": 1366000 + }, + { + "epoch": 18.68, + "learning_rate": 3.2986562042897575e-06, + "loss": 1.244, + "step": 1366500 + }, + { + "epoch": 18.69, + "learning_rate": 3.2815682629082306e-06, + "loss": 1.2517, + "step": 1367000 + }, + { + "epoch": 18.69, + "learning_rate": 3.264480321526705e-06, + "loss": 1.2549, + "step": 1367500 + }, + { + "epoch": 18.7, + "learning_rate": 3.247392380145179e-06, + "loss": 1.2261, + "step": 1368000 + }, + { + "epoch": 18.71, + "learning_rate": 3.230304438763653e-06, + "loss": 1.2665, + "step": 1368500 + }, + { + "epoch": 18.71, + "learning_rate": 3.2132164973821275e-06, + "loss": 1.2491, + "step": 1369000 + }, + { + "epoch": 18.72, + "learning_rate": 3.1961285560006014e-06, + "loss": 1.2358, + "step": 1369500 + }, + { + "epoch": 18.73, + "learning_rate": 3.1790406146190758e-06, + "loss": 1.2416, + "step": 1370000 + }, + { + "epoch": 18.74, + "learning_rate": 3.1619526732375497e-06, + "loss": 1.2317, + "step": 1370500 + }, + { + "epoch": 18.74, + "learning_rate": 3.144864731856024e-06, + "loss": 1.257, + "step": 1371000 + }, + { + "epoch": 18.75, + "learning_rate": 3.127776790474498e-06, + "loss": 1.2447, + "step": 1371500 + }, + { + "epoch": 18.76, + "learning_rate": 3.1106888490929723e-06, + "loss": 1.2693, + "step": 1372000 + }, + { + "epoch": 18.76, + "learning_rate": 3.093600907711446e-06, + "loss": 1.2531, + "step": 1372500 + }, + { + "epoch": 18.77, + "learning_rate": 3.0765129663299205e-06, + "loss": 1.2717, + "step": 1373000 + }, + { + "epoch": 18.78, + "learning_rate": 3.0594250249483944e-06, + "loss": 1.2516, + "step": 1373500 + }, + { + "epoch": 18.78, + "learning_rate": 3.0423370835668688e-06, + "loss": 1.2224, + "step": 1374000 + }, + { + "epoch": 18.79, + "learning_rate": 3.0252491421853427e-06, + "loss": 1.2453, + "step": 1374500 + }, + { + "epoch": 18.8, + "learning_rate": 3.008161200803817e-06, + "loss": 1.2483, + "step": 1375000 + }, + { + "epoch": 18.8, + "learning_rate": 2.991073259422291e-06, + "loss": 1.2632, + "step": 1375500 + }, + { + "epoch": 18.81, + "learning_rate": 2.9739853180407653e-06, + "loss": 1.2789, + "step": 1376000 + }, + { + "epoch": 18.82, + "learning_rate": 2.956897376659239e-06, + "loss": 1.245, + "step": 1376500 + }, + { + "epoch": 18.82, + "learning_rate": 2.9398094352777135e-06, + "loss": 1.2535, + "step": 1377000 + }, + { + "epoch": 18.83, + "learning_rate": 2.9227214938961875e-06, + "loss": 1.2657, + "step": 1377500 + }, + { + "epoch": 18.84, + "learning_rate": 2.9056335525146614e-06, + "loss": 1.2606, + "step": 1378000 + }, + { + "epoch": 18.84, + "learning_rate": 2.8885456111331357e-06, + "loss": 1.2608, + "step": 1378500 + }, + { + "epoch": 18.85, + "learning_rate": 2.8714576697516096e-06, + "loss": 1.2476, + "step": 1379000 + }, + { + "epoch": 18.86, + "learning_rate": 2.854369728370084e-06, + "loss": 1.2612, + "step": 1379500 + }, + { + "epoch": 18.87, + "learning_rate": 2.837281786988558e-06, + "loss": 1.23, + "step": 1380000 + }, + { + "epoch": 18.87, + "learning_rate": 2.8201938456070322e-06, + "loss": 1.2513, + "step": 1380500 + }, + { + "epoch": 18.88, + "learning_rate": 2.803105904225506e-06, + "loss": 1.2685, + "step": 1381000 + }, + { + "epoch": 18.89, + "learning_rate": 2.7860179628439805e-06, + "loss": 1.2384, + "step": 1381500 + }, + { + "epoch": 18.89, + "learning_rate": 2.7689300214624544e-06, + "loss": 1.2447, + "step": 1382000 + }, + { + "epoch": 18.9, + "learning_rate": 2.7518420800809287e-06, + "loss": 1.2487, + "step": 1382500 + }, + { + "epoch": 18.91, + "learning_rate": 2.7347541386994027e-06, + "loss": 1.2596, + "step": 1383000 + }, + { + "epoch": 18.91, + "learning_rate": 2.717666197317877e-06, + "loss": 1.2505, + "step": 1383500 + }, + { + "epoch": 18.92, + "learning_rate": 2.700578255936351e-06, + "loss": 1.2289, + "step": 1384000 + }, + { + "epoch": 18.93, + "learning_rate": 2.6834903145548252e-06, + "loss": 1.2644, + "step": 1384500 + }, + { + "epoch": 18.93, + "learning_rate": 2.666402373173299e-06, + "loss": 1.2667, + "step": 1385000 + }, + { + "epoch": 18.94, + "learning_rate": 2.6493144317917735e-06, + "loss": 1.2276, + "step": 1385500 + }, + { + "epoch": 18.95, + "learning_rate": 2.6322264904102474e-06, + "loss": 1.2522, + "step": 1386000 + }, + { + "epoch": 18.95, + "learning_rate": 2.6151385490287213e-06, + "loss": 1.2284, + "step": 1386500 + }, + { + "epoch": 18.96, + "learning_rate": 2.5980506076471957e-06, + "loss": 1.2192, + "step": 1387000 + }, + { + "epoch": 18.97, + "learning_rate": 2.5809626662656696e-06, + "loss": 1.2397, + "step": 1387500 + }, + { + "epoch": 18.97, + "learning_rate": 2.563874724884144e-06, + "loss": 1.2243, + "step": 1388000 + }, + { + "epoch": 18.98, + "learning_rate": 2.546786783502618e-06, + "loss": 1.2395, + "step": 1388500 + }, + { + "epoch": 18.99, + "learning_rate": 2.529698842121092e-06, + "loss": 1.2523, + "step": 1389000 + }, + { + "epoch": 18.99, + "learning_rate": 2.512610900739566e-06, + "loss": 1.249, + "step": 1389500 + }, + { + "epoch": 19.0, + "learning_rate": 2.4955229593580404e-06, + "loss": 1.2685, + "step": 1390000 + }, + { + "epoch": 19.01, + "learning_rate": 2.4784350179765144e-06, + "loss": 1.2218, + "step": 1390500 + }, + { + "epoch": 19.02, + "learning_rate": 2.4613470765949887e-06, + "loss": 1.2551, + "step": 1391000 + }, + { + "epoch": 19.02, + "learning_rate": 2.4442591352134626e-06, + "loss": 1.2672, + "step": 1391500 + }, + { + "epoch": 19.03, + "learning_rate": 2.427171193831937e-06, + "loss": 1.2742, + "step": 1392000 + }, + { + "epoch": 19.04, + "learning_rate": 2.410083252450411e-06, + "loss": 1.2048, + "step": 1392500 + }, + { + "epoch": 19.04, + "learning_rate": 2.392995311068885e-06, + "loss": 1.2382, + "step": 1393000 + }, + { + "epoch": 19.05, + "learning_rate": 2.375907369687359e-06, + "loss": 1.251, + "step": 1393500 + }, + { + "epoch": 19.06, + "learning_rate": 2.358819428305833e-06, + "loss": 1.2738, + "step": 1394000 + }, + { + "epoch": 19.06, + "learning_rate": 2.3417314869243074e-06, + "loss": 1.2506, + "step": 1394500 + }, + { + "epoch": 19.07, + "learning_rate": 2.3246435455427813e-06, + "loss": 1.2586, + "step": 1395000 + }, + { + "epoch": 19.08, + "learning_rate": 2.3075556041612556e-06, + "loss": 1.2722, + "step": 1395500 + }, + { + "epoch": 19.08, + "learning_rate": 2.2904676627797295e-06, + "loss": 1.2491, + "step": 1396000 + }, + { + "epoch": 19.09, + "learning_rate": 2.273379721398204e-06, + "loss": 1.2476, + "step": 1396500 + }, + { + "epoch": 19.1, + "learning_rate": 2.256291780016678e-06, + "loss": 1.2599, + "step": 1397000 + }, + { + "epoch": 19.1, + "learning_rate": 2.239203838635152e-06, + "loss": 1.2427, + "step": 1397500 + }, + { + "epoch": 19.11, + "learning_rate": 2.222115897253626e-06, + "loss": 1.2273, + "step": 1398000 + }, + { + "epoch": 19.12, + "learning_rate": 2.2050279558721004e-06, + "loss": 1.2378, + "step": 1398500 + }, + { + "epoch": 19.12, + "learning_rate": 2.1879400144905743e-06, + "loss": 1.2387, + "step": 1399000 + }, + { + "epoch": 19.13, + "learning_rate": 2.1708520731090487e-06, + "loss": 1.2307, + "step": 1399500 + }, + { + "epoch": 19.14, + "learning_rate": 2.1537641317275226e-06, + "loss": 1.2335, + "step": 1400000 + }, + { + "epoch": 19.15, + "learning_rate": 2.136676190345997e-06, + "loss": 1.2362, + "step": 1400500 + }, + { + "epoch": 19.15, + "learning_rate": 2.119588248964471e-06, + "loss": 1.2231, + "step": 1401000 + }, + { + "epoch": 19.16, + "learning_rate": 2.1025003075829447e-06, + "loss": 1.2188, + "step": 1401500 + }, + { + "epoch": 19.17, + "learning_rate": 2.085412366201419e-06, + "loss": 1.2285, + "step": 1402000 + }, + { + "epoch": 19.17, + "learning_rate": 2.068324424819893e-06, + "loss": 1.2387, + "step": 1402500 + }, + { + "epoch": 19.18, + "learning_rate": 2.0512364834383673e-06, + "loss": 1.2124, + "step": 1403000 + }, + { + "epoch": 19.19, + "learning_rate": 2.0341485420568412e-06, + "loss": 1.271, + "step": 1403500 + }, + { + "epoch": 19.19, + "learning_rate": 2.0170606006753156e-06, + "loss": 1.2487, + "step": 1404000 + }, + { + "epoch": 19.2, + "learning_rate": 1.9999726592937895e-06, + "loss": 1.2682, + "step": 1404500 + }, + { + "epoch": 19.21, + "learning_rate": 1.982884717912264e-06, + "loss": 1.2798, + "step": 1405000 + }, + { + "epoch": 19.21, + "learning_rate": 1.9657967765307378e-06, + "loss": 1.2199, + "step": 1405500 + }, + { + "epoch": 19.22, + "learning_rate": 1.948708835149212e-06, + "loss": 1.2505, + "step": 1406000 + }, + { + "epoch": 19.23, + "learning_rate": 1.931620893767686e-06, + "loss": 1.2326, + "step": 1406500 + }, + { + "epoch": 19.23, + "learning_rate": 1.9145329523861604e-06, + "loss": 1.2441, + "step": 1407000 + }, + { + "epoch": 19.24, + "learning_rate": 1.8974450110046345e-06, + "loss": 1.2316, + "step": 1407500 + }, + { + "epoch": 19.25, + "learning_rate": 1.8803570696231086e-06, + "loss": 1.2388, + "step": 1408000 + }, + { + "epoch": 19.25, + "learning_rate": 1.8632691282415823e-06, + "loss": 1.2681, + "step": 1408500 + }, + { + "epoch": 19.26, + "learning_rate": 1.8461811868600564e-06, + "loss": 1.2362, + "step": 1409000 + }, + { + "epoch": 19.27, + "learning_rate": 1.8290932454785308e-06, + "loss": 1.228, + "step": 1409500 + }, + { + "epoch": 19.28, + "learning_rate": 1.812005304097005e-06, + "loss": 1.257, + "step": 1410000 + }, + { + "epoch": 19.28, + "learning_rate": 1.794917362715479e-06, + "loss": 1.2239, + "step": 1410500 + }, + { + "epoch": 19.29, + "learning_rate": 1.7778294213339532e-06, + "loss": 1.221, + "step": 1411000 + }, + { + "epoch": 19.3, + "learning_rate": 1.7607414799524273e-06, + "loss": 1.2161, + "step": 1411500 + }, + { + "epoch": 19.3, + "learning_rate": 1.7436535385709014e-06, + "loss": 1.2366, + "step": 1412000 + }, + { + "epoch": 19.31, + "learning_rate": 1.7265655971893755e-06, + "loss": 1.22, + "step": 1412500 + }, + { + "epoch": 19.32, + "learning_rate": 1.7094776558078497e-06, + "loss": 1.249, + "step": 1413000 + }, + { + "epoch": 19.32, + "learning_rate": 1.6923897144263238e-06, + "loss": 1.2145, + "step": 1413500 + }, + { + "epoch": 19.33, + "learning_rate": 1.675301773044798e-06, + "loss": 1.2361, + "step": 1414000 + }, + { + "epoch": 19.34, + "learning_rate": 1.658213831663272e-06, + "loss": 1.2139, + "step": 1414500 + }, + { + "epoch": 19.34, + "learning_rate": 1.6411258902817462e-06, + "loss": 1.2454, + "step": 1415000 + }, + { + "epoch": 19.35, + "learning_rate": 1.6240379489002203e-06, + "loss": 1.2303, + "step": 1415500 + }, + { + "epoch": 19.36, + "learning_rate": 1.606950007518694e-06, + "loss": 1.2386, + "step": 1416000 + }, + { + "epoch": 19.36, + "learning_rate": 1.5898620661371684e-06, + "loss": 1.2146, + "step": 1416500 + }, + { + "epoch": 19.37, + "learning_rate": 1.5727741247556425e-06, + "loss": 1.225, + "step": 1417000 + }, + { + "epoch": 19.38, + "learning_rate": 1.5556861833741166e-06, + "loss": 1.2363, + "step": 1417500 + }, + { + "epoch": 19.38, + "learning_rate": 1.5385982419925907e-06, + "loss": 1.2719, + "step": 1418000 + }, + { + "epoch": 19.39, + "learning_rate": 1.5215103006110649e-06, + "loss": 1.2023, + "step": 1418500 + }, + { + "epoch": 19.4, + "learning_rate": 1.504422359229539e-06, + "loss": 1.2272, + "step": 1419000 + }, + { + "epoch": 19.41, + "learning_rate": 1.4873344178480131e-06, + "loss": 1.2491, + "step": 1419500 + }, + { + "epoch": 19.41, + "learning_rate": 1.4702464764664873e-06, + "loss": 1.2184, + "step": 1420000 + }, + { + "epoch": 19.42, + "learning_rate": 1.4531585350849614e-06, + "loss": 1.2451, + "step": 1420500 + }, + { + "epoch": 19.43, + "learning_rate": 1.4360705937034355e-06, + "loss": 1.2349, + "step": 1421000 + }, + { + "epoch": 19.43, + "learning_rate": 1.4189826523219094e-06, + "loss": 1.2205, + "step": 1421500 + }, + { + "epoch": 19.44, + "learning_rate": 1.4018947109403835e-06, + "loss": 1.2365, + "step": 1422000 + }, + { + "epoch": 19.45, + "learning_rate": 1.3848067695588577e-06, + "loss": 1.2228, + "step": 1422500 + }, + { + "epoch": 19.45, + "learning_rate": 1.3677188281773318e-06, + "loss": 1.2473, + "step": 1423000 + }, + { + "epoch": 19.46, + "learning_rate": 1.350630886795806e-06, + "loss": 1.238, + "step": 1423500 + }, + { + "epoch": 19.47, + "learning_rate": 1.33354294541428e-06, + "loss": 1.2325, + "step": 1424000 + }, + { + "epoch": 19.47, + "learning_rate": 1.3164550040327542e-06, + "loss": 1.218, + "step": 1424500 + }, + { + "epoch": 19.48, + "learning_rate": 1.2993670626512285e-06, + "loss": 1.2097, + "step": 1425000 + }, + { + "epoch": 19.49, + "learning_rate": 1.2822791212697024e-06, + "loss": 1.2531, + "step": 1425500 + }, + { + "epoch": 19.49, + "learning_rate": 1.2651911798881766e-06, + "loss": 1.2261, + "step": 1426000 + }, + { + "epoch": 19.5, + "learning_rate": 1.2481032385066507e-06, + "loss": 1.2563, + "step": 1426500 + }, + { + "epoch": 19.51, + "learning_rate": 1.2310152971251248e-06, + "loss": 1.2304, + "step": 1427000 + }, + { + "epoch": 19.51, + "learning_rate": 1.213927355743599e-06, + "loss": 1.232, + "step": 1427500 + }, + { + "epoch": 19.52, + "learning_rate": 1.196839414362073e-06, + "loss": 1.2426, + "step": 1428000 + }, + { + "epoch": 19.53, + "learning_rate": 1.1797514729805472e-06, + "loss": 1.2331, + "step": 1428500 + }, + { + "epoch": 19.53, + "learning_rate": 1.1626635315990211e-06, + "loss": 1.2272, + "step": 1429000 + }, + { + "epoch": 19.54, + "learning_rate": 1.1455755902174953e-06, + "loss": 1.2167, + "step": 1429500 + }, + { + "epoch": 19.55, + "learning_rate": 1.1284876488359694e-06, + "loss": 1.2371, + "step": 1430000 + }, + { + "epoch": 19.56, + "learning_rate": 1.1113997074544435e-06, + "loss": 1.2449, + "step": 1430500 + }, + { + "epoch": 19.56, + "learning_rate": 1.0943117660729176e-06, + "loss": 1.2303, + "step": 1431000 + }, + { + "epoch": 19.57, + "learning_rate": 1.0772238246913918e-06, + "loss": 1.199, + "step": 1431500 + }, + { + "epoch": 19.58, + "learning_rate": 1.060135883309866e-06, + "loss": 1.2628, + "step": 1432000 + }, + { + "epoch": 19.58, + "learning_rate": 1.0430479419283402e-06, + "loss": 1.218, + "step": 1432500 + }, + { + "epoch": 19.59, + "learning_rate": 1.0259600005468141e-06, + "loss": 1.2361, + "step": 1433000 + }, + { + "epoch": 19.6, + "learning_rate": 1.0088720591652883e-06, + "loss": 1.207, + "step": 1433500 + }, + { + "epoch": 19.6, + "learning_rate": 9.917841177837624e-07, + "loss": 1.23, + "step": 1434000 + }, + { + "epoch": 19.61, + "learning_rate": 9.746961764022365e-07, + "loss": 1.2326, + "step": 1434500 + }, + { + "epoch": 19.62, + "learning_rate": 9.576082350207107e-07, + "loss": 1.2445, + "step": 1435000 + }, + { + "epoch": 19.62, + "learning_rate": 9.405202936391848e-07, + "loss": 1.2255, + "step": 1435500 + }, + { + "epoch": 19.63, + "learning_rate": 9.234323522576589e-07, + "loss": 1.235, + "step": 1436000 + }, + { + "epoch": 19.64, + "learning_rate": 9.06344410876133e-07, + "loss": 1.22, + "step": 1436500 + }, + { + "epoch": 19.64, + "learning_rate": 8.892564694946071e-07, + "loss": 1.254, + "step": 1437000 + }, + { + "epoch": 19.65, + "learning_rate": 8.721685281130812e-07, + "loss": 1.2544, + "step": 1437500 + }, + { + "epoch": 19.66, + "learning_rate": 8.550805867315553e-07, + "loss": 1.2482, + "step": 1438000 + }, + { + "epoch": 19.66, + "learning_rate": 8.379926453500294e-07, + "loss": 1.2497, + "step": 1438500 + }, + { + "epoch": 19.67, + "learning_rate": 8.209047039685036e-07, + "loss": 1.2635, + "step": 1439000 + }, + { + "epoch": 19.68, + "learning_rate": 8.038167625869777e-07, + "loss": 1.2418, + "step": 1439500 + }, + { + "epoch": 19.69, + "learning_rate": 7.867288212054518e-07, + "loss": 1.2417, + "step": 1440000 + }, + { + "epoch": 19.69, + "learning_rate": 7.696408798239258e-07, + "loss": 1.2575, + "step": 1440500 + }, + { + "epoch": 19.7, + "learning_rate": 7.525529384424001e-07, + "loss": 1.2035, + "step": 1441000 + }, + { + "epoch": 19.71, + "learning_rate": 7.354649970608741e-07, + "loss": 1.2341, + "step": 1441500 + }, + { + "epoch": 19.71, + "learning_rate": 7.183770556793482e-07, + "loss": 1.2516, + "step": 1442000 + }, + { + "epoch": 19.72, + "learning_rate": 7.012891142978224e-07, + "loss": 1.2352, + "step": 1442500 + }, + { + "epoch": 19.73, + "learning_rate": 6.842011729162965e-07, + "loss": 1.2333, + "step": 1443000 + }, + { + "epoch": 19.73, + "learning_rate": 6.671132315347705e-07, + "loss": 1.238, + "step": 1443500 + }, + { + "epoch": 19.74, + "learning_rate": 6.500252901532446e-07, + "loss": 1.2243, + "step": 1444000 + }, + { + "epoch": 19.75, + "learning_rate": 6.329373487717189e-07, + "loss": 1.2337, + "step": 1444500 + }, + { + "epoch": 19.75, + "learning_rate": 6.15849407390193e-07, + "loss": 1.2405, + "step": 1445000 + }, + { + "epoch": 19.76, + "learning_rate": 5.98761466008667e-07, + "loss": 1.2355, + "step": 1445500 + }, + { + "epoch": 19.77, + "learning_rate": 5.816735246271411e-07, + "loss": 1.2146, + "step": 1446000 + }, + { + "epoch": 19.77, + "learning_rate": 5.645855832456153e-07, + "loss": 1.2083, + "step": 1446500 + }, + { + "epoch": 19.78, + "learning_rate": 5.474976418640893e-07, + "loss": 1.1832, + "step": 1447000 + }, + { + "epoch": 19.79, + "learning_rate": 5.304097004825634e-07, + "loss": 1.2482, + "step": 1447500 + }, + { + "epoch": 19.79, + "learning_rate": 5.133217591010377e-07, + "loss": 1.2286, + "step": 1448000 + }, + { + "epoch": 19.8, + "learning_rate": 4.962338177195118e-07, + "loss": 1.2254, + "step": 1448500 + }, + { + "epoch": 19.81, + "learning_rate": 4.791458763379858e-07, + "loss": 1.2187, + "step": 1449000 + }, + { + "epoch": 19.82, + "learning_rate": 4.6205793495645994e-07, + "loss": 1.2395, + "step": 1449500 + }, + { + "epoch": 19.82, + "learning_rate": 4.4496999357493406e-07, + "loss": 1.2756, + "step": 1450000 + }, + { + "epoch": 19.83, + "learning_rate": 4.278820521934082e-07, + "loss": 1.2393, + "step": 1450500 + }, + { + "epoch": 19.84, + "learning_rate": 4.1079411081188227e-07, + "loss": 1.2123, + "step": 1451000 + }, + { + "epoch": 19.84, + "learning_rate": 3.937061694303564e-07, + "loss": 1.2536, + "step": 1451500 + }, + { + "epoch": 19.85, + "learning_rate": 3.766182280488305e-07, + "loss": 1.2367, + "step": 1452000 + }, + { + "epoch": 19.86, + "learning_rate": 3.5953028666730465e-07, + "loss": 1.2187, + "step": 1452500 + }, + { + "epoch": 19.86, + "learning_rate": 3.424423452857788e-07, + "loss": 1.2271, + "step": 1453000 + }, + { + "epoch": 19.87, + "learning_rate": 3.2535440390425285e-07, + "loss": 1.2424, + "step": 1453500 + }, + { + "epoch": 19.88, + "learning_rate": 3.08266462522727e-07, + "loss": 1.2133, + "step": 1454000 + }, + { + "epoch": 19.88, + "learning_rate": 2.911785211412011e-07, + "loss": 1.2285, + "step": 1454500 + }, + { + "epoch": 19.89, + "learning_rate": 2.740905797596752e-07, + "loss": 1.2163, + "step": 1455000 + }, + { + "epoch": 19.9, + "learning_rate": 2.570026383781493e-07, + "loss": 1.2487, + "step": 1455500 + }, + { + "epoch": 19.9, + "learning_rate": 2.3991469699662344e-07, + "loss": 1.2114, + "step": 1456000 + }, + { + "epoch": 19.91, + "learning_rate": 2.2282675561509756e-07, + "loss": 1.238, + "step": 1456500 + }, + { + "epoch": 19.92, + "learning_rate": 2.0573881423357164e-07, + "loss": 1.2134, + "step": 1457000 + }, + { + "epoch": 19.92, + "learning_rate": 1.886508728520458e-07, + "loss": 1.2535, + "step": 1457500 + }, + { + "epoch": 19.93, + "learning_rate": 1.715629314705199e-07, + "loss": 1.206, + "step": 1458000 + }, + { + "epoch": 19.94, + "learning_rate": 1.54474990088994e-07, + "loss": 1.2526, + "step": 1458500 + }, + { + "epoch": 19.95, + "learning_rate": 1.3738704870746812e-07, + "loss": 1.2433, + "step": 1459000 + }, + { + "epoch": 19.95, + "learning_rate": 1.2029910732594222e-07, + "loss": 1.2504, + "step": 1459500 + }, + { + "epoch": 19.96, + "learning_rate": 1.0321116594441634e-07, + "loss": 1.2399, + "step": 1460000 + }, + { + "epoch": 19.97, + "learning_rate": 8.612322456289047e-08, + "loss": 1.2388, + "step": 1460500 + }, + { + "epoch": 19.97, + "learning_rate": 6.903528318136458e-08, + "loss": 1.2225, + "step": 1461000 + }, + { + "epoch": 19.98, + "learning_rate": 5.194734179983869e-08, + "loss": 1.2189, + "step": 1461500 + }, + { + "epoch": 19.99, + "learning_rate": 3.485940041831281e-08, + "loss": 1.2224, + "step": 1462000 + }, + { + "epoch": 19.99, + "learning_rate": 1.7771459036786924e-08, + "loss": 1.223, + "step": 1462500 + }, + { + "epoch": 20.0, + "learning_rate": 6.835176552610355e-10, + "loss": 1.2238, + "step": 1463000 + } + ], + "max_steps": 1463020, + "num_train_epochs": 20, + "total_flos": 8.246728127625382e+17, + "trial_name": null, + "trial_params": null +}