{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "global_step": 323979, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.992283450470556e-05, "loss": 7.2271, "step": 500 }, { "epoch": 0.01, "learning_rate": 4.98456690094111e-05, "loss": 6.7719, "step": 1000 }, { "epoch": 0.01, "learning_rate": 4.976850351411666e-05, "loss": 6.6629, "step": 1500 }, { "epoch": 0.02, "learning_rate": 4.969133801882221e-05, "loss": 6.6057, "step": 2000 }, { "epoch": 0.02, "learning_rate": 4.961417252352776e-05, "loss": 6.5557, "step": 2500 }, { "epoch": 0.03, "learning_rate": 4.953700702823331e-05, "loss": 6.5339, "step": 3000 }, { "epoch": 0.03, "learning_rate": 4.945984153293887e-05, "loss": 6.4981, "step": 3500 }, { "epoch": 0.04, "learning_rate": 4.938267603764441e-05, "loss": 6.4864, "step": 4000 }, { "epoch": 0.04, "learning_rate": 4.930551054234997e-05, "loss": 6.4665, "step": 4500 }, { "epoch": 0.05, "learning_rate": 4.922834504705553e-05, "loss": 6.4375, "step": 5000 }, { "epoch": 0.05, "learning_rate": 4.915117955176107e-05, "loss": 6.4367, "step": 5500 }, { "epoch": 0.06, "learning_rate": 4.907401405646663e-05, "loss": 6.4276, "step": 6000 }, { "epoch": 0.06, "learning_rate": 4.899684856117218e-05, "loss": 6.4075, "step": 6500 }, { "epoch": 0.06, "learning_rate": 4.891968306587773e-05, "loss": 6.3899, "step": 7000 }, { "epoch": 0.07, "learning_rate": 4.884251757058328e-05, "loss": 6.3858, "step": 7500 }, { "epoch": 0.07, "learning_rate": 4.8765352075288836e-05, "loss": 6.3791, "step": 8000 }, { "epoch": 0.08, "learning_rate": 4.868818657999438e-05, "loss": 6.3676, "step": 8500 }, { "epoch": 0.08, "learning_rate": 4.861102108469994e-05, "loss": 6.3539, "step": 9000 }, { "epoch": 0.09, "learning_rate": 4.853385558940549e-05, "loss": 6.3511, "step": 9500 }, { "epoch": 0.09, "learning_rate": 4.845669009411104e-05, "loss": 6.3434, "step": 10000 }, { "epoch": 0.1, "learning_rate": 4.8379524598816595e-05, "loss": 6.3393, "step": 10500 }, { "epoch": 0.1, "learning_rate": 4.8302359103522146e-05, "loss": 6.3319, "step": 11000 }, { "epoch": 0.11, "learning_rate": 4.8225193608227696e-05, "loss": 6.3237, "step": 11500 }, { "epoch": 0.11, "learning_rate": 4.8148028112933246e-05, "loss": 6.3252, "step": 12000 }, { "epoch": 0.12, "learning_rate": 4.8070862617638804e-05, "loss": 6.31, "step": 12500 }, { "epoch": 0.12, "learning_rate": 4.799369712234435e-05, "loss": 6.3092, "step": 13000 }, { "epoch": 0.13, "learning_rate": 4.7916531627049905e-05, "loss": 6.3098, "step": 13500 }, { "epoch": 0.13, "learning_rate": 4.7839366131755455e-05, "loss": 6.3044, "step": 14000 }, { "epoch": 0.13, "learning_rate": 4.7762200636461005e-05, "loss": 6.297, "step": 14500 }, { "epoch": 0.14, "learning_rate": 4.768503514116656e-05, "loss": 6.2934, "step": 15000 }, { "epoch": 0.14, "learning_rate": 4.760786964587211e-05, "loss": 6.2945, "step": 15500 }, { "epoch": 0.15, "learning_rate": 4.7530704150577663e-05, "loss": 6.2808, "step": 16000 }, { "epoch": 0.15, "learning_rate": 4.7453538655283214e-05, "loss": 6.2806, "step": 16500 }, { "epoch": 0.16, "learning_rate": 4.737637315998877e-05, "loss": 6.2805, "step": 17000 }, { "epoch": 0.16, "learning_rate": 4.7299207664694315e-05, "loss": 6.2715, "step": 17500 }, { "epoch": 0.17, "learning_rate": 4.722204216939987e-05, "loss": 6.264, "step": 18000 }, { "epoch": 0.17, "learning_rate": 4.714487667410542e-05, "loss": 6.2593, "step": 18500 }, { "epoch": 0.18, "learning_rate": 4.706771117881097e-05, "loss": 6.2522, "step": 19000 }, { "epoch": 0.18, "learning_rate": 4.699054568351653e-05, "loss": 6.2625, "step": 19500 }, { "epoch": 0.19, "learning_rate": 4.691338018822208e-05, "loss": 6.2572, "step": 20000 }, { "epoch": 0.19, "learning_rate": 4.683621469292763e-05, "loss": 6.2562, "step": 20500 }, { "epoch": 0.19, "learning_rate": 4.675904919763318e-05, "loss": 6.2434, "step": 21000 }, { "epoch": 0.2, "learning_rate": 4.668188370233874e-05, "loss": 6.2465, "step": 21500 }, { "epoch": 0.2, "learning_rate": 4.660471820704428e-05, "loss": 6.2381, "step": 22000 }, { "epoch": 0.21, "learning_rate": 4.652755271174984e-05, "loss": 6.2361, "step": 22500 }, { "epoch": 0.21, "learning_rate": 4.645038721645539e-05, "loss": 6.2318, "step": 23000 }, { "epoch": 0.22, "learning_rate": 4.637322172116094e-05, "loss": 6.2407, "step": 23500 }, { "epoch": 0.22, "learning_rate": 4.629605622586649e-05, "loss": 6.2403, "step": 24000 }, { "epoch": 0.23, "learning_rate": 4.621889073057205e-05, "loss": 6.2229, "step": 24500 }, { "epoch": 0.23, "learning_rate": 4.61417252352776e-05, "loss": 6.231, "step": 25000 }, { "epoch": 0.24, "learning_rate": 4.606455973998315e-05, "loss": 6.2295, "step": 25500 }, { "epoch": 0.24, "learning_rate": 4.5987394244688706e-05, "loss": 6.2117, "step": 26000 }, { "epoch": 0.25, "learning_rate": 4.591022874939425e-05, "loss": 6.2233, "step": 26500 }, { "epoch": 0.25, "learning_rate": 4.583306325409981e-05, "loss": 6.2193, "step": 27000 }, { "epoch": 0.25, "learning_rate": 4.575589775880536e-05, "loss": 6.2166, "step": 27500 }, { "epoch": 0.26, "learning_rate": 4.567873226351091e-05, "loss": 6.2093, "step": 28000 }, { "epoch": 0.26, "learning_rate": 4.560156676821646e-05, "loss": 6.2025, "step": 28500 }, { "epoch": 0.27, "learning_rate": 4.5524401272922015e-05, "loss": 6.2056, "step": 29000 }, { "epoch": 0.27, "learning_rate": 4.5447235777627566e-05, "loss": 6.2018, "step": 29500 }, { "epoch": 0.28, "learning_rate": 4.5370070282333116e-05, "loss": 6.2111, "step": 30000 }, { "epoch": 0.28, "learning_rate": 4.529290478703867e-05, "loss": 6.2045, "step": 30500 }, { "epoch": 0.29, "learning_rate": 4.521573929174422e-05, "loss": 6.1887, "step": 31000 }, { "epoch": 0.29, "learning_rate": 4.5138573796449774e-05, "loss": 6.1956, "step": 31500 }, { "epoch": 0.3, "learning_rate": 4.5061408301155324e-05, "loss": 6.2002, "step": 32000 }, { "epoch": 0.3, "learning_rate": 4.4984242805860875e-05, "loss": 6.2027, "step": 32500 }, { "epoch": 0.31, "learning_rate": 4.4907077310566425e-05, "loss": 6.1877, "step": 33000 }, { "epoch": 0.31, "learning_rate": 4.482991181527198e-05, "loss": 6.1894, "step": 33500 }, { "epoch": 0.31, "learning_rate": 4.475274631997753e-05, "loss": 6.1861, "step": 34000 }, { "epoch": 0.32, "learning_rate": 4.467558082468308e-05, "loss": 6.1839, "step": 34500 }, { "epoch": 0.32, "learning_rate": 4.459841532938864e-05, "loss": 6.1783, "step": 35000 }, { "epoch": 0.33, "learning_rate": 4.4521249834094184e-05, "loss": 6.1894, "step": 35500 }, { "epoch": 0.33, "learning_rate": 4.444408433879974e-05, "loss": 6.1811, "step": 36000 }, { "epoch": 0.34, "learning_rate": 4.436691884350529e-05, "loss": 6.1822, "step": 36500 }, { "epoch": 0.34, "learning_rate": 4.428975334821084e-05, "loss": 6.1739, "step": 37000 }, { "epoch": 0.35, "learning_rate": 4.421258785291639e-05, "loss": 6.1715, "step": 37500 }, { "epoch": 0.35, "learning_rate": 4.413542235762195e-05, "loss": 6.1779, "step": 38000 }, { "epoch": 0.36, "learning_rate": 4.4058256862327493e-05, "loss": 6.1735, "step": 38500 }, { "epoch": 0.36, "learning_rate": 4.398109136703305e-05, "loss": 6.1723, "step": 39000 }, { "epoch": 0.37, "learning_rate": 4.390392587173861e-05, "loss": 6.1703, "step": 39500 }, { "epoch": 0.37, "learning_rate": 4.382676037644415e-05, "loss": 6.1738, "step": 40000 }, { "epoch": 0.38, "learning_rate": 4.374959488114971e-05, "loss": 6.1656, "step": 40500 }, { "epoch": 0.38, "learning_rate": 4.367242938585526e-05, "loss": 6.1722, "step": 41000 }, { "epoch": 0.38, "learning_rate": 4.359526389056081e-05, "loss": 6.1643, "step": 41500 }, { "epoch": 0.39, "learning_rate": 4.351809839526636e-05, "loss": 6.1708, "step": 42000 }, { "epoch": 0.39, "learning_rate": 4.344093289997192e-05, "loss": 6.1624, "step": 42500 }, { "epoch": 0.4, "learning_rate": 4.336376740467746e-05, "loss": 6.1644, "step": 43000 }, { "epoch": 0.4, "learning_rate": 4.328660190938302e-05, "loss": 6.1659, "step": 43500 }, { "epoch": 0.41, "learning_rate": 4.3209436414088575e-05, "loss": 6.1557, "step": 44000 }, { "epoch": 0.41, "learning_rate": 4.313227091879412e-05, "loss": 6.1448, "step": 44500 }, { "epoch": 0.42, "learning_rate": 4.3055105423499676e-05, "loss": 6.1532, "step": 45000 }, { "epoch": 0.42, "learning_rate": 4.2977939928205227e-05, "loss": 6.1445, "step": 45500 }, { "epoch": 0.43, "learning_rate": 4.290077443291078e-05, "loss": 6.1419, "step": 46000 }, { "epoch": 0.43, "learning_rate": 4.282360893761633e-05, "loss": 6.1518, "step": 46500 }, { "epoch": 0.44, "learning_rate": 4.2746443442321885e-05, "loss": 6.1501, "step": 47000 }, { "epoch": 0.44, "learning_rate": 4.266927794702743e-05, "loss": 6.142, "step": 47500 }, { "epoch": 0.44, "learning_rate": 4.2592112451732985e-05, "loss": 6.1523, "step": 48000 }, { "epoch": 0.45, "learning_rate": 4.2514946956438536e-05, "loss": 6.1383, "step": 48500 }, { "epoch": 0.45, "learning_rate": 4.2437781461144086e-05, "loss": 6.1358, "step": 49000 }, { "epoch": 0.46, "learning_rate": 4.2360615965849643e-05, "loss": 6.1378, "step": 49500 }, { "epoch": 0.46, "learning_rate": 4.2283450470555194e-05, "loss": 6.1415, "step": 50000 }, { "epoch": 0.47, "learning_rate": 4.2206284975260744e-05, "loss": 6.1407, "step": 50500 }, { "epoch": 0.47, "learning_rate": 4.2129119479966295e-05, "loss": 6.1405, "step": 51000 }, { "epoch": 0.48, "learning_rate": 4.205195398467185e-05, "loss": 6.1407, "step": 51500 }, { "epoch": 0.48, "learning_rate": 4.1974788489377396e-05, "loss": 6.1412, "step": 52000 }, { "epoch": 0.49, "learning_rate": 4.189762299408295e-05, "loss": 6.1413, "step": 52500 }, { "epoch": 0.49, "learning_rate": 4.18204574987885e-05, "loss": 6.1347, "step": 53000 }, { "epoch": 0.5, "learning_rate": 4.1743292003494054e-05, "loss": 6.1364, "step": 53500 }, { "epoch": 0.5, "learning_rate": 4.166612650819961e-05, "loss": 6.1401, "step": 54000 }, { "epoch": 0.5, "learning_rate": 4.158896101290516e-05, "loss": 6.1357, "step": 54500 }, { "epoch": 0.51, "learning_rate": 4.151179551761071e-05, "loss": 6.1317, "step": 55000 }, { "epoch": 0.51, "learning_rate": 4.143463002231626e-05, "loss": 6.1287, "step": 55500 }, { "epoch": 0.52, "learning_rate": 4.135746452702182e-05, "loss": 6.1378, "step": 56000 }, { "epoch": 0.52, "learning_rate": 4.128029903172736e-05, "loss": 6.1365, "step": 56500 }, { "epoch": 0.53, "learning_rate": 4.120313353643292e-05, "loss": 6.1393, "step": 57000 }, { "epoch": 0.53, "learning_rate": 4.112596804113847e-05, "loss": 6.1246, "step": 57500 }, { "epoch": 0.54, "learning_rate": 4.104880254584402e-05, "loss": 6.1184, "step": 58000 }, { "epoch": 0.54, "learning_rate": 4.097163705054958e-05, "loss": 6.1164, "step": 58500 }, { "epoch": 0.55, "learning_rate": 4.089447155525513e-05, "loss": 6.1308, "step": 59000 }, { "epoch": 0.55, "learning_rate": 4.081730605996068e-05, "loss": 6.1179, "step": 59500 }, { "epoch": 0.56, "learning_rate": 4.074014056466623e-05, "loss": 6.125, "step": 60000 }, { "epoch": 0.56, "learning_rate": 4.066297506937179e-05, "loss": 6.123, "step": 60500 }, { "epoch": 0.56, "learning_rate": 4.058580957407733e-05, "loss": 6.1191, "step": 61000 }, { "epoch": 0.57, "learning_rate": 4.050864407878289e-05, "loss": 6.1176, "step": 61500 }, { "epoch": 0.57, "learning_rate": 4.043147858348844e-05, "loss": 6.1186, "step": 62000 }, { "epoch": 0.58, "learning_rate": 4.035431308819399e-05, "loss": 6.1197, "step": 62500 }, { "epoch": 0.58, "learning_rate": 4.027714759289954e-05, "loss": 6.1212, "step": 63000 }, { "epoch": 0.59, "learning_rate": 4.0199982097605096e-05, "loss": 6.1219, "step": 63500 }, { "epoch": 0.59, "learning_rate": 4.0122816602310646e-05, "loss": 6.1169, "step": 64000 }, { "epoch": 0.6, "learning_rate": 4.00456511070162e-05, "loss": 6.1149, "step": 64500 }, { "epoch": 0.6, "learning_rate": 3.9968485611721754e-05, "loss": 6.118, "step": 65000 }, { "epoch": 0.61, "learning_rate": 3.98913201164273e-05, "loss": 6.1218, "step": 65500 }, { "epoch": 0.61, "learning_rate": 3.9814154621132855e-05, "loss": 6.11, "step": 66000 }, { "epoch": 0.62, "learning_rate": 3.9736989125838405e-05, "loss": 6.1164, "step": 66500 }, { "epoch": 0.62, "learning_rate": 3.9659823630543956e-05, "loss": 6.1142, "step": 67000 }, { "epoch": 0.63, "learning_rate": 3.9582658135249506e-05, "loss": 6.1067, "step": 67500 }, { "epoch": 0.63, "learning_rate": 3.9505492639955063e-05, "loss": 6.1179, "step": 68000 }, { "epoch": 0.63, "learning_rate": 3.9428327144660614e-05, "loss": 6.1031, "step": 68500 }, { "epoch": 0.64, "learning_rate": 3.9351161649366164e-05, "loss": 6.1052, "step": 69000 }, { "epoch": 0.64, "learning_rate": 3.927399615407172e-05, "loss": 6.1057, "step": 69500 }, { "epoch": 0.65, "learning_rate": 3.9196830658777265e-05, "loss": 6.1035, "step": 70000 }, { "epoch": 0.65, "learning_rate": 3.911966516348282e-05, "loss": 6.1028, "step": 70500 }, { "epoch": 0.66, "learning_rate": 3.904249966818837e-05, "loss": 6.1133, "step": 71000 }, { "epoch": 0.66, "learning_rate": 3.896533417289392e-05, "loss": 6.1124, "step": 71500 }, { "epoch": 0.67, "learning_rate": 3.8888168677599474e-05, "loss": 6.1048, "step": 72000 }, { "epoch": 0.67, "learning_rate": 3.881100318230503e-05, "loss": 6.0969, "step": 72500 }, { "epoch": 0.68, "learning_rate": 3.873383768701058e-05, "loss": 6.1106, "step": 73000 }, { "epoch": 0.68, "learning_rate": 3.865667219171613e-05, "loss": 6.095, "step": 73500 }, { "epoch": 0.69, "learning_rate": 3.857950669642169e-05, "loss": 6.1035, "step": 74000 }, { "epoch": 0.69, "learning_rate": 3.850234120112723e-05, "loss": 6.1016, "step": 74500 }, { "epoch": 0.69, "learning_rate": 3.842517570583279e-05, "loss": 6.1119, "step": 75000 }, { "epoch": 0.7, "learning_rate": 3.834801021053834e-05, "loss": 6.1011, "step": 75500 }, { "epoch": 0.7, "learning_rate": 3.827084471524389e-05, "loss": 6.097, "step": 76000 }, { "epoch": 0.71, "learning_rate": 3.819367921994944e-05, "loss": 6.1007, "step": 76500 }, { "epoch": 0.71, "learning_rate": 3.8116513724655e-05, "loss": 6.0952, "step": 77000 }, { "epoch": 0.72, "learning_rate": 3.803934822936054e-05, "loss": 6.0097, "step": 77500 }, { "epoch": 0.72, "learning_rate": 3.79621827340661e-05, "loss": 5.9232, "step": 78000 }, { "epoch": 0.73, "learning_rate": 3.7885017238771656e-05, "loss": 5.8762, "step": 78500 }, { "epoch": 0.73, "learning_rate": 3.78078517434772e-05, "loss": 5.8308, "step": 79000 }, { "epoch": 0.74, "learning_rate": 3.773068624818276e-05, "loss": 5.7992, "step": 79500 }, { "epoch": 0.74, "learning_rate": 3.765352075288831e-05, "loss": 5.761, "step": 80000 }, { "epoch": 0.75, "learning_rate": 3.757635525759386e-05, "loss": 5.7269, "step": 80500 }, { "epoch": 0.75, "learning_rate": 3.749918976229941e-05, "loss": 5.7042, "step": 81000 }, { "epoch": 0.75, "learning_rate": 3.7422024267004966e-05, "loss": 5.6826, "step": 81500 }, { "epoch": 0.76, "learning_rate": 3.734485877171051e-05, "loss": 5.6541, "step": 82000 }, { "epoch": 0.76, "learning_rate": 3.7267693276416066e-05, "loss": 5.6237, "step": 82500 }, { "epoch": 0.77, "learning_rate": 3.719052778112162e-05, "loss": 5.6091, "step": 83000 }, { "epoch": 0.77, "learning_rate": 3.711336228582717e-05, "loss": 5.5772, "step": 83500 }, { "epoch": 0.78, "learning_rate": 3.7036196790532724e-05, "loss": 5.5444, "step": 84000 }, { "epoch": 0.78, "learning_rate": 3.6959031295238275e-05, "loss": 5.5268, "step": 84500 }, { "epoch": 0.79, "learning_rate": 3.6881865799943825e-05, "loss": 5.4856, "step": 85000 }, { "epoch": 0.79, "learning_rate": 3.6804700304649376e-05, "loss": 5.4607, "step": 85500 }, { "epoch": 0.8, "learning_rate": 3.672753480935493e-05, "loss": 5.4276, "step": 86000 }, { "epoch": 0.8, "learning_rate": 3.6650369314060477e-05, "loss": 5.4144, "step": 86500 }, { "epoch": 0.81, "learning_rate": 3.6573203818766034e-05, "loss": 5.3782, "step": 87000 }, { "epoch": 0.81, "learning_rate": 3.6496038323471584e-05, "loss": 5.3337, "step": 87500 }, { "epoch": 0.81, "learning_rate": 3.6418872828177135e-05, "loss": 5.2709, "step": 88000 }, { "epoch": 0.82, "learning_rate": 3.634170733288269e-05, "loss": 5.2095, "step": 88500 }, { "epoch": 0.82, "learning_rate": 3.626454183758824e-05, "loss": 5.138, "step": 89000 }, { "epoch": 0.83, "learning_rate": 3.618737634229379e-05, "loss": 4.9959, "step": 89500 }, { "epoch": 0.83, "learning_rate": 3.611021084699934e-05, "loss": 4.8709, "step": 90000 }, { "epoch": 0.84, "learning_rate": 3.60330453517049e-05, "loss": 4.7619, "step": 90500 }, { "epoch": 0.84, "learning_rate": 3.5955879856410444e-05, "loss": 4.6625, "step": 91000 }, { "epoch": 0.85, "learning_rate": 3.5878714361116e-05, "loss": 4.5585, "step": 91500 }, { "epoch": 0.85, "learning_rate": 3.580154886582155e-05, "loss": 4.4651, "step": 92000 }, { "epoch": 0.86, "learning_rate": 3.57243833705271e-05, "loss": 4.3757, "step": 92500 }, { "epoch": 0.86, "learning_rate": 3.564721787523266e-05, "loss": 4.298, "step": 93000 }, { "epoch": 0.87, "learning_rate": 3.557005237993821e-05, "loss": 4.2163, "step": 93500 }, { "epoch": 0.87, "learning_rate": 3.549288688464376e-05, "loss": 4.1503, "step": 94000 }, { "epoch": 0.88, "learning_rate": 3.541572138934931e-05, "loss": 4.0701, "step": 94500 }, { "epoch": 0.88, "learning_rate": 3.533855589405487e-05, "loss": 4.0051, "step": 95000 }, { "epoch": 0.88, "learning_rate": 3.526139039876041e-05, "loss": 3.9148, "step": 95500 }, { "epoch": 0.89, "learning_rate": 3.518422490346597e-05, "loss": 3.7963, "step": 96000 }, { "epoch": 0.89, "learning_rate": 3.510705940817152e-05, "loss": 3.6717, "step": 96500 }, { "epoch": 0.9, "learning_rate": 3.502989391287707e-05, "loss": 3.5825, "step": 97000 }, { "epoch": 0.9, "learning_rate": 3.4952728417582627e-05, "loss": 3.5186, "step": 97500 }, { "epoch": 0.91, "learning_rate": 3.487556292228818e-05, "loss": 3.4746, "step": 98000 }, { "epoch": 0.91, "learning_rate": 3.479839742699373e-05, "loss": 3.3886, "step": 98500 }, { "epoch": 0.92, "learning_rate": 3.472123193169928e-05, "loss": 3.3322, "step": 99000 }, { "epoch": 0.92, "learning_rate": 3.4644066436404835e-05, "loss": 3.2667, "step": 99500 }, { "epoch": 0.93, "learning_rate": 3.456690094111038e-05, "loss": 3.2286, "step": 100000 }, { "epoch": 0.93, "learning_rate": 3.4489735445815936e-05, "loss": 3.1693, "step": 100500 }, { "epoch": 0.94, "learning_rate": 3.4412569950521486e-05, "loss": 3.1297, "step": 101000 }, { "epoch": 0.94, "learning_rate": 3.433540445522704e-05, "loss": 3.095, "step": 101500 }, { "epoch": 0.94, "learning_rate": 3.4258238959932594e-05, "loss": 3.0555, "step": 102000 }, { "epoch": 0.95, "learning_rate": 3.4181073464638144e-05, "loss": 3.0215, "step": 102500 }, { "epoch": 0.95, "learning_rate": 3.4103907969343695e-05, "loss": 2.9936, "step": 103000 }, { "epoch": 0.96, "learning_rate": 3.4026742474049245e-05, "loss": 2.9553, "step": 103500 }, { "epoch": 0.96, "learning_rate": 3.39495769787548e-05, "loss": 2.9263, "step": 104000 }, { "epoch": 0.97, "learning_rate": 3.3872411483460346e-05, "loss": 2.9042, "step": 104500 }, { "epoch": 0.97, "learning_rate": 3.37952459881659e-05, "loss": 2.8779, "step": 105000 }, { "epoch": 0.98, "learning_rate": 3.3718080492871454e-05, "loss": 2.8483, "step": 105500 }, { "epoch": 0.98, "learning_rate": 3.3640914997577004e-05, "loss": 2.8281, "step": 106000 }, { "epoch": 0.99, "learning_rate": 3.3563749502282555e-05, "loss": 2.8081, "step": 106500 }, { "epoch": 0.99, "learning_rate": 3.348658400698811e-05, "loss": 2.7802, "step": 107000 }, { "epoch": 1.0, "learning_rate": 3.340941851169366e-05, "loss": 2.7555, "step": 107500 }, { "epoch": 1.0, "learning_rate": 3.333225301639921e-05, "loss": 2.7429, "step": 108000 }, { "epoch": 1.0, "learning_rate": 3.325508752110476e-05, "loss": 2.723, "step": 108500 }, { "epoch": 1.01, "learning_rate": 3.3177922025810313e-05, "loss": 2.7106, "step": 109000 }, { "epoch": 1.01, "learning_rate": 3.310075653051587e-05, "loss": 2.6834, "step": 109500 }, { "epoch": 1.02, "learning_rate": 3.302359103522142e-05, "loss": 2.6625, "step": 110000 }, { "epoch": 1.02, "learning_rate": 3.294642553992697e-05, "loss": 2.6493, "step": 110500 }, { "epoch": 1.03, "learning_rate": 3.286926004463252e-05, "loss": 2.6323, "step": 111000 }, { "epoch": 1.03, "learning_rate": 3.279209454933808e-05, "loss": 2.6196, "step": 111500 }, { "epoch": 1.04, "learning_rate": 3.271492905404363e-05, "loss": 2.6093, "step": 112000 }, { "epoch": 1.04, "learning_rate": 3.263776355874918e-05, "loss": 2.5883, "step": 112500 }, { "epoch": 1.05, "learning_rate": 3.256059806345473e-05, "loss": 2.5722, "step": 113000 }, { "epoch": 1.05, "learning_rate": 3.248343256816028e-05, "loss": 2.5621, "step": 113500 }, { "epoch": 1.06, "learning_rate": 3.240626707286584e-05, "loss": 2.5431, "step": 114000 }, { "epoch": 1.06, "learning_rate": 3.232910157757139e-05, "loss": 2.5357, "step": 114500 }, { "epoch": 1.06, "learning_rate": 3.225193608227694e-05, "loss": 2.5253, "step": 115000 }, { "epoch": 1.07, "learning_rate": 3.217477058698249e-05, "loss": 2.5091, "step": 115500 }, { "epoch": 1.07, "learning_rate": 3.2097605091688046e-05, "loss": 2.5008, "step": 116000 }, { "epoch": 1.08, "learning_rate": 3.20204395963936e-05, "loss": 2.4897, "step": 116500 }, { "epoch": 1.08, "learning_rate": 3.194327410109915e-05, "loss": 2.4621, "step": 117000 }, { "epoch": 1.09, "learning_rate": 3.18661086058047e-05, "loss": 2.4584, "step": 117500 }, { "epoch": 1.09, "learning_rate": 3.178894311051025e-05, "loss": 2.4491, "step": 118000 }, { "epoch": 1.1, "learning_rate": 3.1711777615215805e-05, "loss": 2.4336, "step": 118500 }, { "epoch": 1.1, "learning_rate": 3.1634612119921356e-05, "loss": 2.4147, "step": 119000 }, { "epoch": 1.11, "learning_rate": 3.1557446624626906e-05, "loss": 2.4093, "step": 119500 }, { "epoch": 1.11, "learning_rate": 3.148028112933246e-05, "loss": 2.4001, "step": 120000 }, { "epoch": 1.12, "learning_rate": 3.1403115634038014e-05, "loss": 2.3902, "step": 120500 }, { "epoch": 1.12, "learning_rate": 3.132595013874356e-05, "loss": 2.3675, "step": 121000 }, { "epoch": 1.13, "learning_rate": 3.1248784643449115e-05, "loss": 2.3749, "step": 121500 }, { "epoch": 1.13, "learning_rate": 3.1171619148154665e-05, "loss": 2.3599, "step": 122000 }, { "epoch": 1.13, "learning_rate": 3.1094453652860216e-05, "loss": 2.3526, "step": 122500 }, { "epoch": 1.14, "learning_rate": 3.101728815756577e-05, "loss": 2.3341, "step": 123000 }, { "epoch": 1.14, "learning_rate": 3.094012266227132e-05, "loss": 2.3356, "step": 123500 }, { "epoch": 1.15, "learning_rate": 3.0862957166976874e-05, "loss": 2.3146, "step": 124000 }, { "epoch": 1.15, "learning_rate": 3.0785791671682424e-05, "loss": 2.2939, "step": 124500 }, { "epoch": 1.16, "learning_rate": 3.070862617638798e-05, "loss": 2.2914, "step": 125000 }, { "epoch": 1.16, "learning_rate": 3.0631460681093525e-05, "loss": 2.2835, "step": 125500 }, { "epoch": 1.17, "learning_rate": 3.055429518579908e-05, "loss": 2.2782, "step": 126000 }, { "epoch": 1.17, "learning_rate": 3.0477129690504636e-05, "loss": 2.272, "step": 126500 }, { "epoch": 1.18, "learning_rate": 3.0399964195210183e-05, "loss": 2.2664, "step": 127000 }, { "epoch": 1.18, "learning_rate": 3.032279869991574e-05, "loss": 2.2529, "step": 127500 }, { "epoch": 1.19, "learning_rate": 3.0245633204621287e-05, "loss": 2.2433, "step": 128000 }, { "epoch": 1.19, "learning_rate": 3.016846770932684e-05, "loss": 2.2377, "step": 128500 }, { "epoch": 1.19, "learning_rate": 3.009130221403239e-05, "loss": 2.226, "step": 129000 }, { "epoch": 1.2, "learning_rate": 3.0014136718737945e-05, "loss": 2.2204, "step": 129500 }, { "epoch": 1.2, "learning_rate": 2.9936971223443496e-05, "loss": 2.2053, "step": 130000 }, { "epoch": 1.21, "learning_rate": 2.985980572814905e-05, "loss": 2.2028, "step": 130500 }, { "epoch": 1.21, "learning_rate": 2.9782640232854596e-05, "loss": 2.1984, "step": 131000 }, { "epoch": 1.22, "learning_rate": 2.970547473756015e-05, "loss": 2.1928, "step": 131500 }, { "epoch": 1.22, "learning_rate": 2.9628309242265707e-05, "loss": 2.1778, "step": 132000 }, { "epoch": 1.23, "learning_rate": 2.9551143746971255e-05, "loss": 2.1823, "step": 132500 }, { "epoch": 1.23, "learning_rate": 2.947397825167681e-05, "loss": 2.1643, "step": 133000 }, { "epoch": 1.24, "learning_rate": 2.939681275638236e-05, "loss": 2.1642, "step": 133500 }, { "epoch": 1.24, "learning_rate": 2.9319647261087913e-05, "loss": 2.1537, "step": 134000 }, { "epoch": 1.25, "learning_rate": 2.9242481765793463e-05, "loss": 2.1488, "step": 134500 }, { "epoch": 1.25, "learning_rate": 2.9165316270499017e-05, "loss": 2.1471, "step": 135000 }, { "epoch": 1.25, "learning_rate": 2.9088150775204564e-05, "loss": 2.1342, "step": 135500 }, { "epoch": 1.26, "learning_rate": 2.9010985279910118e-05, "loss": 2.1308, "step": 136000 }, { "epoch": 1.26, "learning_rate": 2.8933819784615675e-05, "loss": 2.126, "step": 136500 }, { "epoch": 1.27, "learning_rate": 2.8856654289321222e-05, "loss": 2.1168, "step": 137000 }, { "epoch": 1.27, "learning_rate": 2.8779488794026776e-05, "loss": 2.1099, "step": 137500 }, { "epoch": 1.28, "learning_rate": 2.8702323298732326e-05, "loss": 2.1055, "step": 138000 }, { "epoch": 1.28, "learning_rate": 2.862515780343788e-05, "loss": 2.1085, "step": 138500 }, { "epoch": 1.29, "learning_rate": 2.8547992308143427e-05, "loss": 2.0995, "step": 139000 }, { "epoch": 1.29, "learning_rate": 2.8470826812848984e-05, "loss": 2.0969, "step": 139500 }, { "epoch": 1.3, "learning_rate": 2.839366131755453e-05, "loss": 2.0776, "step": 140000 }, { "epoch": 1.3, "learning_rate": 2.8316495822260085e-05, "loss": 2.0799, "step": 140500 }, { "epoch": 1.31, "learning_rate": 2.8239330326965642e-05, "loss": 2.0776, "step": 141000 }, { "epoch": 1.31, "learning_rate": 2.816216483167119e-05, "loss": 2.0776, "step": 141500 }, { "epoch": 1.31, "learning_rate": 2.8084999336376743e-05, "loss": 2.0655, "step": 142000 }, { "epoch": 1.32, "learning_rate": 2.8007833841082293e-05, "loss": 2.0537, "step": 142500 }, { "epoch": 1.32, "learning_rate": 2.7930668345787847e-05, "loss": 2.0508, "step": 143000 }, { "epoch": 1.33, "learning_rate": 2.7853502850493394e-05, "loss": 2.0511, "step": 143500 }, { "epoch": 1.33, "learning_rate": 2.777633735519895e-05, "loss": 2.04, "step": 144000 }, { "epoch": 1.34, "learning_rate": 2.76991718599045e-05, "loss": 2.0409, "step": 144500 }, { "epoch": 1.34, "learning_rate": 2.7622006364610052e-05, "loss": 2.037, "step": 145000 }, { "epoch": 1.35, "learning_rate": 2.7544840869315603e-05, "loss": 2.0314, "step": 145500 }, { "epoch": 1.35, "learning_rate": 2.7467675374021157e-05, "loss": 2.0304, "step": 146000 }, { "epoch": 1.36, "learning_rate": 2.739050987872671e-05, "loss": 2.0209, "step": 146500 }, { "epoch": 1.36, "learning_rate": 2.731334438343226e-05, "loss": 2.0227, "step": 147000 }, { "epoch": 1.37, "learning_rate": 2.7236178888137815e-05, "loss": 2.0162, "step": 147500 }, { "epoch": 1.37, "learning_rate": 2.7159013392843362e-05, "loss": 2.013, "step": 148000 }, { "epoch": 1.38, "learning_rate": 2.708184789754892e-05, "loss": 2.0051, "step": 148500 }, { "epoch": 1.38, "learning_rate": 2.7004682402254466e-05, "loss": 1.9949, "step": 149000 }, { "epoch": 1.38, "learning_rate": 2.692751690696002e-05, "loss": 2.0048, "step": 149500 }, { "epoch": 1.39, "learning_rate": 2.685035141166557e-05, "loss": 1.9992, "step": 150000 }, { "epoch": 1.39, "learning_rate": 2.6773185916371124e-05, "loss": 1.9944, "step": 150500 }, { "epoch": 1.4, "learning_rate": 2.6696020421076678e-05, "loss": 1.9888, "step": 151000 }, { "epoch": 1.4, "learning_rate": 2.6618854925782228e-05, "loss": 1.9888, "step": 151500 }, { "epoch": 1.41, "learning_rate": 2.6541689430487782e-05, "loss": 1.9869, "step": 152000 }, { "epoch": 1.41, "learning_rate": 2.646452393519333e-05, "loss": 1.988, "step": 152500 }, { "epoch": 1.42, "learning_rate": 2.6387358439898886e-05, "loss": 1.9722, "step": 153000 }, { "epoch": 1.42, "learning_rate": 2.6310192944604433e-05, "loss": 1.9684, "step": 153500 }, { "epoch": 1.43, "learning_rate": 2.6233027449309987e-05, "loss": 1.9693, "step": 154000 }, { "epoch": 1.43, "learning_rate": 2.6155861954015538e-05, "loss": 1.9706, "step": 154500 }, { "epoch": 1.44, "learning_rate": 2.607869645872109e-05, "loss": 1.9631, "step": 155000 }, { "epoch": 1.44, "learning_rate": 2.6001530963426645e-05, "loss": 1.9571, "step": 155500 }, { "epoch": 1.44, "learning_rate": 2.5924365468132196e-05, "loss": 1.9525, "step": 156000 }, { "epoch": 1.45, "learning_rate": 2.584719997283775e-05, "loss": 2.0859, "step": 156500 }, { "epoch": 1.45, "learning_rate": 2.5770034477543296e-05, "loss": 1.986, "step": 157000 }, { "epoch": 1.46, "learning_rate": 2.5692868982248854e-05, "loss": 1.9676, "step": 157500 }, { "epoch": 1.46, "learning_rate": 2.56157034869544e-05, "loss": 1.9541, "step": 158000 }, { "epoch": 1.47, "learning_rate": 2.5538537991659954e-05, "loss": 1.9392, "step": 158500 }, { "epoch": 1.47, "learning_rate": 2.5461372496365505e-05, "loss": 1.9335, "step": 159000 }, { "epoch": 1.48, "learning_rate": 2.538420700107106e-05, "loss": 1.9308, "step": 159500 }, { "epoch": 1.48, "learning_rate": 2.530704150577661e-05, "loss": 1.9314, "step": 160000 }, { "epoch": 1.49, "learning_rate": 2.5229876010482163e-05, "loss": 1.9259, "step": 160500 }, { "epoch": 1.49, "learning_rate": 2.5152710515187717e-05, "loss": 1.9202, "step": 161000 }, { "epoch": 1.5, "learning_rate": 2.5075545019893264e-05, "loss": 1.9131, "step": 161500 }, { "epoch": 1.5, "learning_rate": 2.4998379524598818e-05, "loss": 1.9097, "step": 162000 }, { "epoch": 1.5, "learning_rate": 2.4921214029304368e-05, "loss": 1.9126, "step": 162500 }, { "epoch": 1.51, "learning_rate": 2.4844048534009922e-05, "loss": 1.9031, "step": 163000 }, { "epoch": 1.51, "learning_rate": 2.4766883038715476e-05, "loss": 1.9042, "step": 163500 }, { "epoch": 1.52, "learning_rate": 2.4689717543421026e-05, "loss": 1.899, "step": 164000 }, { "epoch": 1.52, "learning_rate": 2.4612552048126577e-05, "loss": 1.8939, "step": 164500 }, { "epoch": 1.53, "learning_rate": 2.453538655283213e-05, "loss": 1.8871, "step": 165000 }, { "epoch": 1.53, "learning_rate": 2.445822105753768e-05, "loss": 1.8973, "step": 165500 }, { "epoch": 1.54, "learning_rate": 2.438105556224323e-05, "loss": 1.8879, "step": 166000 }, { "epoch": 1.54, "learning_rate": 2.4303890066948785e-05, "loss": 1.8853, "step": 166500 }, { "epoch": 1.55, "learning_rate": 2.4226724571654335e-05, "loss": 1.8851, "step": 167000 }, { "epoch": 1.55, "learning_rate": 2.414955907635989e-05, "loss": 1.8871, "step": 167500 }, { "epoch": 1.56, "learning_rate": 2.4072393581065443e-05, "loss": 1.8749, "step": 168000 }, { "epoch": 1.56, "learning_rate": 2.3995228085770993e-05, "loss": 1.8756, "step": 168500 }, { "epoch": 1.56, "learning_rate": 2.3918062590476544e-05, "loss": 1.8784, "step": 169000 }, { "epoch": 1.57, "learning_rate": 2.3840897095182098e-05, "loss": 1.8693, "step": 169500 }, { "epoch": 1.57, "learning_rate": 2.3763731599887648e-05, "loss": 1.8639, "step": 170000 }, { "epoch": 1.58, "learning_rate": 2.36865661045932e-05, "loss": 1.8625, "step": 170500 }, { "epoch": 1.58, "learning_rate": 2.3609400609298752e-05, "loss": 1.8659, "step": 171000 }, { "epoch": 1.59, "learning_rate": 2.3532235114004303e-05, "loss": 1.8591, "step": 171500 }, { "epoch": 1.59, "learning_rate": 2.3455069618709853e-05, "loss": 1.8524, "step": 172000 }, { "epoch": 1.6, "learning_rate": 2.337790412341541e-05, "loss": 1.8551, "step": 172500 }, { "epoch": 1.6, "learning_rate": 2.330073862812096e-05, "loss": 1.8547, "step": 173000 }, { "epoch": 1.61, "learning_rate": 2.322357313282651e-05, "loss": 1.8481, "step": 173500 }, { "epoch": 1.61, "learning_rate": 2.3146407637532065e-05, "loss": 1.8484, "step": 174000 }, { "epoch": 1.62, "learning_rate": 2.3069242142237616e-05, "loss": 1.8412, "step": 174500 }, { "epoch": 1.62, "learning_rate": 2.2992076646943166e-05, "loss": 1.8404, "step": 175000 }, { "epoch": 1.63, "learning_rate": 2.291491115164872e-05, "loss": 1.8344, "step": 175500 }, { "epoch": 1.63, "learning_rate": 2.283774565635427e-05, "loss": 1.8417, "step": 176000 }, { "epoch": 1.63, "learning_rate": 2.276058016105982e-05, "loss": 1.837, "step": 176500 }, { "epoch": 1.64, "learning_rate": 2.2683414665765374e-05, "loss": 1.8251, "step": 177000 }, { "epoch": 1.64, "learning_rate": 2.2606249170470928e-05, "loss": 1.825, "step": 177500 }, { "epoch": 1.65, "learning_rate": 2.252908367517648e-05, "loss": 1.8313, "step": 178000 }, { "epoch": 1.65, "learning_rate": 2.2451918179882032e-05, "loss": 1.8324, "step": 178500 }, { "epoch": 1.66, "learning_rate": 2.2374752684587583e-05, "loss": 1.8244, "step": 179000 }, { "epoch": 1.66, "learning_rate": 2.2297587189293133e-05, "loss": 1.8218, "step": 179500 }, { "epoch": 1.67, "learning_rate": 2.2220421693998687e-05, "loss": 1.825, "step": 180000 }, { "epoch": 1.67, "learning_rate": 2.2143256198704238e-05, "loss": 1.8204, "step": 180500 }, { "epoch": 1.68, "learning_rate": 2.2066090703409788e-05, "loss": 1.8138, "step": 181000 }, { "epoch": 1.68, "learning_rate": 2.1988925208115342e-05, "loss": 1.8059, "step": 181500 }, { "epoch": 1.69, "learning_rate": 2.1911759712820896e-05, "loss": 1.8126, "step": 182000 }, { "epoch": 1.69, "learning_rate": 2.1834594217526446e-05, "loss": 1.8102, "step": 182500 }, { "epoch": 1.69, "learning_rate": 2.1757428722232e-05, "loss": 1.8102, "step": 183000 }, { "epoch": 1.7, "learning_rate": 2.168026322693755e-05, "loss": 1.8105, "step": 183500 }, { "epoch": 1.7, "learning_rate": 2.16030977316431e-05, "loss": 1.8027, "step": 184000 }, { "epoch": 1.71, "learning_rate": 2.1525932236348654e-05, "loss": 1.8073, "step": 184500 }, { "epoch": 1.71, "learning_rate": 2.1448766741054205e-05, "loss": 1.8022, "step": 185000 }, { "epoch": 1.72, "learning_rate": 2.1371601245759755e-05, "loss": 1.7926, "step": 185500 }, { "epoch": 1.72, "learning_rate": 2.129443575046531e-05, "loss": 1.7932, "step": 186000 }, { "epoch": 1.73, "learning_rate": 2.121727025517086e-05, "loss": 1.7964, "step": 186500 }, { "epoch": 1.73, "learning_rate": 2.1140104759876413e-05, "loss": 1.7932, "step": 187000 }, { "epoch": 1.74, "learning_rate": 2.1062939264581967e-05, "loss": 1.7926, "step": 187500 }, { "epoch": 1.74, "learning_rate": 2.0985773769287518e-05, "loss": 1.7899, "step": 188000 }, { "epoch": 1.75, "learning_rate": 2.0908608273993068e-05, "loss": 1.7837, "step": 188500 }, { "epoch": 1.75, "learning_rate": 2.0831442778698622e-05, "loss": 1.7836, "step": 189000 }, { "epoch": 1.75, "learning_rate": 2.0754277283404172e-05, "loss": 1.7863, "step": 189500 }, { "epoch": 1.76, "learning_rate": 2.0677111788109723e-05, "loss": 1.7759, "step": 190000 }, { "epoch": 1.76, "learning_rate": 2.0599946292815277e-05, "loss": 1.7749, "step": 190500 }, { "epoch": 1.77, "learning_rate": 2.0522780797520827e-05, "loss": 1.7745, "step": 191000 }, { "epoch": 1.77, "learning_rate": 2.0445615302226377e-05, "loss": 1.7678, "step": 191500 }, { "epoch": 1.78, "learning_rate": 2.0368449806931935e-05, "loss": 1.778, "step": 192000 }, { "epoch": 1.78, "learning_rate": 2.0291284311637485e-05, "loss": 1.7771, "step": 192500 }, { "epoch": 1.79, "learning_rate": 2.0214118816343035e-05, "loss": 1.7682, "step": 193000 }, { "epoch": 1.79, "learning_rate": 2.013695332104859e-05, "loss": 1.7679, "step": 193500 }, { "epoch": 1.8, "learning_rate": 2.005978782575414e-05, "loss": 1.7712, "step": 194000 }, { "epoch": 1.8, "learning_rate": 1.998262233045969e-05, "loss": 1.7615, "step": 194500 }, { "epoch": 1.81, "learning_rate": 1.9905456835165244e-05, "loss": 1.7664, "step": 195000 }, { "epoch": 1.81, "learning_rate": 1.9828291339870794e-05, "loss": 1.7618, "step": 195500 }, { "epoch": 1.81, "learning_rate": 1.9751125844576345e-05, "loss": 1.7604, "step": 196000 }, { "epoch": 1.82, "learning_rate": 1.96739603492819e-05, "loss": 1.754, "step": 196500 }, { "epoch": 1.82, "learning_rate": 1.9596794853987452e-05, "loss": 1.7563, "step": 197000 }, { "epoch": 1.83, "learning_rate": 1.9519629358693003e-05, "loss": 1.7573, "step": 197500 }, { "epoch": 1.83, "learning_rate": 1.9442463863398557e-05, "loss": 1.7473, "step": 198000 }, { "epoch": 1.84, "learning_rate": 1.9365298368104107e-05, "loss": 1.743, "step": 198500 }, { "epoch": 1.84, "learning_rate": 1.9288132872809657e-05, "loss": 1.7482, "step": 199000 }, { "epoch": 1.85, "learning_rate": 1.921096737751521e-05, "loss": 1.7395, "step": 199500 }, { "epoch": 1.85, "learning_rate": 1.9133801882220762e-05, "loss": 1.7428, "step": 200000 }, { "epoch": 1.86, "learning_rate": 1.9056636386926312e-05, "loss": 1.7448, "step": 200500 }, { "epoch": 1.86, "learning_rate": 1.8979470891631866e-05, "loss": 1.7461, "step": 201000 }, { "epoch": 1.87, "learning_rate": 1.890230539633742e-05, "loss": 1.7332, "step": 201500 }, { "epoch": 1.87, "learning_rate": 1.882513990104297e-05, "loss": 1.7409, "step": 202000 }, { "epoch": 1.88, "learning_rate": 1.8747974405748524e-05, "loss": 1.7361, "step": 202500 }, { "epoch": 1.88, "learning_rate": 1.8670808910454074e-05, "loss": 1.7425, "step": 203000 }, { "epoch": 1.88, "learning_rate": 1.8593643415159625e-05, "loss": 1.7305, "step": 203500 }, { "epoch": 1.89, "learning_rate": 1.851647791986518e-05, "loss": 1.7352, "step": 204000 }, { "epoch": 1.89, "learning_rate": 1.843931242457073e-05, "loss": 1.7351, "step": 204500 }, { "epoch": 1.9, "learning_rate": 1.836214692927628e-05, "loss": 1.7237, "step": 205000 }, { "epoch": 1.9, "learning_rate": 1.8284981433981833e-05, "loss": 1.729, "step": 205500 }, { "epoch": 1.91, "learning_rate": 1.8207815938687384e-05, "loss": 1.7275, "step": 206000 }, { "epoch": 1.91, "learning_rate": 1.8130650443392938e-05, "loss": 1.725, "step": 206500 }, { "epoch": 1.92, "learning_rate": 1.805348494809849e-05, "loss": 1.7216, "step": 207000 }, { "epoch": 1.92, "learning_rate": 1.7976319452804042e-05, "loss": 1.7218, "step": 207500 }, { "epoch": 1.93, "learning_rate": 1.7899153957509592e-05, "loss": 1.7187, "step": 208000 }, { "epoch": 1.93, "learning_rate": 1.7821988462215146e-05, "loss": 1.7173, "step": 208500 }, { "epoch": 1.94, "learning_rate": 1.7744822966920696e-05, "loss": 1.7228, "step": 209000 }, { "epoch": 1.94, "learning_rate": 1.7667657471626247e-05, "loss": 1.7113, "step": 209500 }, { "epoch": 1.94, "learning_rate": 1.75904919763318e-05, "loss": 1.7072, "step": 210000 }, { "epoch": 1.95, "learning_rate": 1.751332648103735e-05, "loss": 1.7174, "step": 210500 }, { "epoch": 1.95, "learning_rate": 1.74361609857429e-05, "loss": 1.707, "step": 211000 }, { "epoch": 1.96, "learning_rate": 1.7358995490448455e-05, "loss": 1.7147, "step": 211500 }, { "epoch": 1.96, "learning_rate": 1.728182999515401e-05, "loss": 1.7156, "step": 212000 }, { "epoch": 1.97, "learning_rate": 1.720466449985956e-05, "loss": 1.7024, "step": 212500 }, { "epoch": 1.97, "learning_rate": 1.7127499004565113e-05, "loss": 1.7038, "step": 213000 }, { "epoch": 1.98, "learning_rate": 1.7050333509270664e-05, "loss": 1.7006, "step": 213500 }, { "epoch": 1.98, "learning_rate": 1.6973168013976214e-05, "loss": 1.6952, "step": 214000 }, { "epoch": 1.99, "learning_rate": 1.6896002518681768e-05, "loss": 1.7124, "step": 214500 }, { "epoch": 1.99, "learning_rate": 1.681883702338732e-05, "loss": 1.7015, "step": 215000 }, { "epoch": 2.0, "learning_rate": 1.674167152809287e-05, "loss": 1.6964, "step": 215500 }, { "epoch": 2.0, "learning_rate": 1.6664506032798423e-05, "loss": 1.6972, "step": 216000 }, { "epoch": 2.0, "learning_rate": 1.6587340537503977e-05, "loss": 1.6966, "step": 216500 }, { "epoch": 2.01, "learning_rate": 1.6510175042209527e-05, "loss": 1.6976, "step": 217000 }, { "epoch": 2.01, "learning_rate": 1.643300954691508e-05, "loss": 1.6848, "step": 217500 }, { "epoch": 2.02, "learning_rate": 1.635584405162063e-05, "loss": 1.6875, "step": 218000 }, { "epoch": 2.02, "learning_rate": 1.627867855632618e-05, "loss": 1.6929, "step": 218500 }, { "epoch": 2.03, "learning_rate": 1.6201513061031735e-05, "loss": 1.6858, "step": 219000 }, { "epoch": 2.03, "learning_rate": 1.6124347565737286e-05, "loss": 1.6863, "step": 219500 }, { "epoch": 2.04, "learning_rate": 1.6047182070442836e-05, "loss": 1.6858, "step": 220000 }, { "epoch": 2.04, "learning_rate": 1.597001657514839e-05, "loss": 1.6841, "step": 220500 }, { "epoch": 2.05, "learning_rate": 1.5892851079853944e-05, "loss": 1.6815, "step": 221000 }, { "epoch": 2.05, "learning_rate": 1.5815685584559494e-05, "loss": 1.6829, "step": 221500 }, { "epoch": 2.06, "learning_rate": 1.5738520089265045e-05, "loss": 1.6897, "step": 222000 }, { "epoch": 2.06, "learning_rate": 1.56613545939706e-05, "loss": 1.6817, "step": 222500 }, { "epoch": 2.06, "learning_rate": 1.558418909867615e-05, "loss": 1.6822, "step": 223000 }, { "epoch": 2.07, "learning_rate": 1.5507023603381703e-05, "loss": 1.6784, "step": 223500 }, { "epoch": 2.07, "learning_rate": 1.5429858108087253e-05, "loss": 1.6743, "step": 224000 }, { "epoch": 2.08, "learning_rate": 1.5352692612792804e-05, "loss": 1.6865, "step": 224500 }, { "epoch": 2.08, "learning_rate": 1.5275527117498357e-05, "loss": 1.671, "step": 225000 }, { "epoch": 2.09, "learning_rate": 1.5198361622203908e-05, "loss": 1.671, "step": 225500 }, { "epoch": 2.09, "learning_rate": 1.5121196126909462e-05, "loss": 1.6757, "step": 226000 }, { "epoch": 2.1, "learning_rate": 1.5044030631615014e-05, "loss": 1.6741, "step": 226500 }, { "epoch": 2.1, "learning_rate": 1.4966865136320566e-05, "loss": 1.6719, "step": 227000 }, { "epoch": 2.11, "learning_rate": 1.4889699641026116e-05, "loss": 1.671, "step": 227500 }, { "epoch": 2.11, "learning_rate": 1.4812534145731668e-05, "loss": 1.6658, "step": 228000 }, { "epoch": 2.12, "learning_rate": 1.473536865043722e-05, "loss": 1.6649, "step": 228500 }, { "epoch": 2.12, "learning_rate": 1.4658203155142771e-05, "loss": 1.6693, "step": 229000 }, { "epoch": 2.13, "learning_rate": 1.4581037659848323e-05, "loss": 1.6604, "step": 229500 }, { "epoch": 2.13, "learning_rate": 1.4503872164553875e-05, "loss": 1.6655, "step": 230000 }, { "epoch": 2.13, "learning_rate": 1.4426706669259429e-05, "loss": 1.6564, "step": 230500 }, { "epoch": 2.14, "learning_rate": 1.4349541173964981e-05, "loss": 1.6664, "step": 231000 }, { "epoch": 2.14, "learning_rate": 1.4272375678670533e-05, "loss": 1.6537, "step": 231500 }, { "epoch": 2.15, "learning_rate": 1.4195210183376084e-05, "loss": 1.6616, "step": 232000 }, { "epoch": 2.15, "learning_rate": 1.4118044688081636e-05, "loss": 1.6579, "step": 232500 }, { "epoch": 2.16, "learning_rate": 1.4040879192787188e-05, "loss": 1.6617, "step": 233000 }, { "epoch": 2.16, "learning_rate": 1.3963713697492738e-05, "loss": 1.652, "step": 233500 }, { "epoch": 2.17, "learning_rate": 1.388654820219829e-05, "loss": 1.6544, "step": 234000 }, { "epoch": 2.17, "learning_rate": 1.3809382706903843e-05, "loss": 1.6504, "step": 234500 }, { "epoch": 2.18, "learning_rate": 1.3732217211609393e-05, "loss": 1.6574, "step": 235000 }, { "epoch": 2.18, "learning_rate": 1.3655051716314949e-05, "loss": 1.6495, "step": 235500 }, { "epoch": 2.19, "learning_rate": 1.35778862210205e-05, "loss": 1.6519, "step": 236000 }, { "epoch": 2.19, "learning_rate": 1.3500720725726051e-05, "loss": 1.6521, "step": 236500 }, { "epoch": 2.19, "learning_rate": 1.3423555230431603e-05, "loss": 1.6535, "step": 237000 }, { "epoch": 2.2, "learning_rate": 1.3346389735137155e-05, "loss": 1.6527, "step": 237500 }, { "epoch": 2.2, "learning_rate": 1.3269224239842706e-05, "loss": 1.6459, "step": 238000 }, { "epoch": 2.21, "learning_rate": 1.3192058744548258e-05, "loss": 1.6444, "step": 238500 }, { "epoch": 2.21, "learning_rate": 1.311489324925381e-05, "loss": 1.6443, "step": 239000 }, { "epoch": 2.22, "learning_rate": 1.303772775395936e-05, "loss": 1.6448, "step": 239500 }, { "epoch": 2.22, "learning_rate": 1.2960562258664913e-05, "loss": 1.6469, "step": 240000 }, { "epoch": 2.23, "learning_rate": 1.2883396763370466e-05, "loss": 1.6368, "step": 240500 }, { "epoch": 2.23, "learning_rate": 1.2806231268076018e-05, "loss": 1.6462, "step": 241000 }, { "epoch": 2.24, "learning_rate": 1.272906577278157e-05, "loss": 1.6395, "step": 241500 }, { "epoch": 2.24, "learning_rate": 1.2651900277487123e-05, "loss": 1.6396, "step": 242000 }, { "epoch": 2.25, "learning_rate": 1.2574734782192673e-05, "loss": 1.6442, "step": 242500 }, { "epoch": 2.25, "learning_rate": 1.2497569286898225e-05, "loss": 1.6417, "step": 243000 }, { "epoch": 2.25, "learning_rate": 1.2420403791603777e-05, "loss": 1.6377, "step": 243500 }, { "epoch": 2.26, "learning_rate": 1.234323829630933e-05, "loss": 1.6415, "step": 244000 }, { "epoch": 2.26, "learning_rate": 1.2266072801014882e-05, "loss": 1.6313, "step": 244500 }, { "epoch": 2.27, "learning_rate": 1.2188907305720434e-05, "loss": 1.6325, "step": 245000 }, { "epoch": 2.27, "learning_rate": 1.2111741810425984e-05, "loss": 1.6376, "step": 245500 }, { "epoch": 2.28, "learning_rate": 1.2034576315131536e-05, "loss": 1.6412, "step": 246000 }, { "epoch": 2.28, "learning_rate": 1.195741081983709e-05, "loss": 1.6296, "step": 246500 }, { "epoch": 2.29, "learning_rate": 1.188024532454264e-05, "loss": 1.6288, "step": 247000 }, { "epoch": 2.29, "learning_rate": 1.1803079829248193e-05, "loss": 1.6349, "step": 247500 }, { "epoch": 2.3, "learning_rate": 1.1725914333953745e-05, "loss": 1.6317, "step": 248000 }, { "epoch": 2.3, "learning_rate": 1.1648748838659295e-05, "loss": 1.6306, "step": 248500 }, { "epoch": 2.31, "learning_rate": 1.1571583343364849e-05, "loss": 1.6322, "step": 249000 }, { "epoch": 2.31, "learning_rate": 1.1494417848070401e-05, "loss": 1.6334, "step": 249500 }, { "epoch": 2.31, "learning_rate": 1.1417252352775952e-05, "loss": 1.6261, "step": 250000 }, { "epoch": 2.32, "learning_rate": 1.1340086857481504e-05, "loss": 1.6245, "step": 250500 }, { "epoch": 2.32, "learning_rate": 1.1262921362187056e-05, "loss": 1.6254, "step": 251000 }, { "epoch": 2.33, "learning_rate": 1.1185755866892608e-05, "loss": 1.6244, "step": 251500 }, { "epoch": 2.33, "learning_rate": 1.110859037159816e-05, "loss": 1.6203, "step": 252000 }, { "epoch": 2.34, "learning_rate": 1.1031424876303712e-05, "loss": 1.6255, "step": 252500 }, { "epoch": 2.34, "learning_rate": 1.0954259381009263e-05, "loss": 1.6187, "step": 253000 }, { "epoch": 2.35, "learning_rate": 1.0877093885714815e-05, "loss": 1.6242, "step": 253500 }, { "epoch": 2.35, "learning_rate": 1.0799928390420368e-05, "loss": 1.6218, "step": 254000 }, { "epoch": 2.36, "learning_rate": 1.0722762895125919e-05, "loss": 1.6202, "step": 254500 }, { "epoch": 2.36, "learning_rate": 1.0645597399831471e-05, "loss": 1.6138, "step": 255000 }, { "epoch": 2.37, "learning_rate": 1.0568431904537023e-05, "loss": 1.6146, "step": 255500 }, { "epoch": 2.37, "learning_rate": 1.0491266409242575e-05, "loss": 1.608, "step": 256000 }, { "epoch": 2.38, "learning_rate": 1.0414100913948127e-05, "loss": 1.62, "step": 256500 }, { "epoch": 2.38, "learning_rate": 1.033693541865368e-05, "loss": 1.6238, "step": 257000 }, { "epoch": 2.38, "learning_rate": 1.025976992335923e-05, "loss": 1.6198, "step": 257500 }, { "epoch": 2.39, "learning_rate": 1.0182604428064782e-05, "loss": 1.6165, "step": 258000 }, { "epoch": 2.39, "learning_rate": 1.0105438932770334e-05, "loss": 1.615, "step": 258500 }, { "epoch": 2.4, "learning_rate": 1.0028273437475886e-05, "loss": 1.6119, "step": 259000 }, { "epoch": 2.4, "learning_rate": 9.951107942181438e-06, "loss": 1.6029, "step": 259500 }, { "epoch": 2.41, "learning_rate": 9.87394244688699e-06, "loss": 1.6108, "step": 260000 }, { "epoch": 2.41, "learning_rate": 9.796776951592541e-06, "loss": 1.6152, "step": 260500 }, { "epoch": 2.42, "learning_rate": 9.719611456298095e-06, "loss": 1.6118, "step": 261000 }, { "epoch": 2.42, "learning_rate": 9.642445961003647e-06, "loss": 1.6092, "step": 261500 }, { "epoch": 2.43, "learning_rate": 9.565280465709197e-06, "loss": 1.6072, "step": 262000 }, { "epoch": 2.43, "learning_rate": 9.48811497041475e-06, "loss": 1.6117, "step": 262500 }, { "epoch": 2.44, "learning_rate": 9.410949475120302e-06, "loss": 1.6077, "step": 263000 }, { "epoch": 2.44, "learning_rate": 9.333783979825854e-06, "loss": 1.6082, "step": 263500 }, { "epoch": 2.44, "learning_rate": 9.256618484531406e-06, "loss": 1.6054, "step": 264000 }, { "epoch": 2.45, "learning_rate": 9.179452989236958e-06, "loss": 1.6012, "step": 264500 }, { "epoch": 2.45, "learning_rate": 9.102287493942508e-06, "loss": 1.6091, "step": 265000 }, { "epoch": 2.46, "learning_rate": 9.02512199864806e-06, "loss": 1.6026, "step": 265500 }, { "epoch": 2.46, "learning_rate": 8.947956503353613e-06, "loss": 1.6038, "step": 266000 }, { "epoch": 2.47, "learning_rate": 8.870791008059165e-06, "loss": 1.5986, "step": 266500 }, { "epoch": 2.47, "learning_rate": 8.793625512764717e-06, "loss": 1.5989, "step": 267000 }, { "epoch": 2.48, "learning_rate": 8.716460017470269e-06, "loss": 1.6037, "step": 267500 }, { "epoch": 2.48, "learning_rate": 8.63929452217582e-06, "loss": 1.5999, "step": 268000 }, { "epoch": 2.49, "learning_rate": 8.562129026881373e-06, "loss": 1.5942, "step": 268500 }, { "epoch": 2.49, "learning_rate": 8.484963531586925e-06, "loss": 1.6006, "step": 269000 }, { "epoch": 2.5, "learning_rate": 8.407798036292476e-06, "loss": 1.5911, "step": 269500 }, { "epoch": 2.5, "learning_rate": 8.330632540998028e-06, "loss": 1.5997, "step": 270000 }, { "epoch": 2.5, "learning_rate": 8.25346704570358e-06, "loss": 1.5983, "step": 270500 }, { "epoch": 2.51, "learning_rate": 8.176301550409132e-06, "loss": 1.5966, "step": 271000 }, { "epoch": 2.51, "learning_rate": 8.099136055114684e-06, "loss": 1.5899, "step": 271500 }, { "epoch": 2.52, "learning_rate": 8.021970559820236e-06, "loss": 1.599, "step": 272000 }, { "epoch": 2.52, "learning_rate": 7.944805064525787e-06, "loss": 1.5894, "step": 272500 }, { "epoch": 2.53, "learning_rate": 7.86763956923134e-06, "loss": 1.5969, "step": 273000 }, { "epoch": 2.53, "learning_rate": 7.790474073936891e-06, "loss": 1.5976, "step": 273500 }, { "epoch": 2.54, "learning_rate": 7.713308578642443e-06, "loss": 1.5907, "step": 274000 }, { "epoch": 2.54, "learning_rate": 7.636143083347995e-06, "loss": 1.5812, "step": 274500 }, { "epoch": 2.55, "learning_rate": 7.5589775880535464e-06, "loss": 1.5889, "step": 275000 }, { "epoch": 2.55, "learning_rate": 7.481812092759099e-06, "loss": 1.5857, "step": 275500 }, { "epoch": 2.56, "learning_rate": 7.4046465974646515e-06, "loss": 1.5849, "step": 276000 }, { "epoch": 2.56, "learning_rate": 7.327481102170203e-06, "loss": 1.5893, "step": 276500 }, { "epoch": 2.56, "learning_rate": 7.250315606875754e-06, "loss": 1.5901, "step": 277000 }, { "epoch": 2.57, "learning_rate": 7.173150111581306e-06, "loss": 1.5812, "step": 277500 }, { "epoch": 2.57, "learning_rate": 7.095984616286859e-06, "loss": 1.5969, "step": 278000 }, { "epoch": 2.58, "learning_rate": 7.01881912099241e-06, "loss": 1.5894, "step": 278500 }, { "epoch": 2.58, "learning_rate": 6.9416536256979625e-06, "loss": 1.5812, "step": 279000 }, { "epoch": 2.59, "learning_rate": 6.864488130403514e-06, "loss": 1.5908, "step": 279500 }, { "epoch": 2.59, "learning_rate": 6.787322635109065e-06, "loss": 1.5869, "step": 280000 }, { "epoch": 2.6, "learning_rate": 6.710157139814618e-06, "loss": 1.5863, "step": 280500 }, { "epoch": 2.6, "learning_rate": 6.63299164452017e-06, "loss": 1.5842, "step": 281000 }, { "epoch": 2.61, "learning_rate": 6.5558261492257214e-06, "loss": 1.5794, "step": 281500 }, { "epoch": 2.61, "learning_rate": 6.4786606539312736e-06, "loss": 1.5896, "step": 282000 }, { "epoch": 2.62, "learning_rate": 6.401495158636825e-06, "loss": 1.5837, "step": 282500 }, { "epoch": 2.62, "learning_rate": 6.324329663342378e-06, "loss": 1.579, "step": 283000 }, { "epoch": 2.63, "learning_rate": 6.24716416804793e-06, "loss": 1.5811, "step": 283500 }, { "epoch": 2.63, "learning_rate": 6.169998672753481e-06, "loss": 1.5833, "step": 284000 }, { "epoch": 2.63, "learning_rate": 6.092833177459033e-06, "loss": 1.5814, "step": 284500 }, { "epoch": 2.64, "learning_rate": 6.015667682164585e-06, "loss": 1.5826, "step": 285000 }, { "epoch": 2.64, "learning_rate": 5.938502186870137e-06, "loss": 1.5786, "step": 285500 }, { "epoch": 2.65, "learning_rate": 5.861336691575689e-06, "loss": 1.575, "step": 286000 }, { "epoch": 2.65, "learning_rate": 5.784171196281241e-06, "loss": 1.5753, "step": 286500 }, { "epoch": 2.66, "learning_rate": 5.707005700986793e-06, "loss": 1.5788, "step": 287000 }, { "epoch": 2.66, "learning_rate": 5.629840205692344e-06, "loss": 1.5777, "step": 287500 }, { "epoch": 2.67, "learning_rate": 5.5526747103978964e-06, "loss": 1.5719, "step": 288000 }, { "epoch": 2.67, "learning_rate": 5.4755092151034486e-06, "loss": 1.5772, "step": 288500 }, { "epoch": 2.68, "learning_rate": 5.398343719809e-06, "loss": 1.5733, "step": 289000 }, { "epoch": 2.68, "learning_rate": 5.321178224514552e-06, "loss": 1.5791, "step": 289500 }, { "epoch": 2.69, "learning_rate": 5.244012729220104e-06, "loss": 1.5711, "step": 290000 }, { "epoch": 2.69, "learning_rate": 5.166847233925655e-06, "loss": 1.5727, "step": 290500 }, { "epoch": 2.69, "learning_rate": 5.089681738631208e-06, "loss": 1.57, "step": 291000 }, { "epoch": 2.7, "learning_rate": 5.0125162433367596e-06, "loss": 1.5743, "step": 291500 }, { "epoch": 2.7, "learning_rate": 4.935350748042312e-06, "loss": 1.5749, "step": 292000 }, { "epoch": 2.71, "learning_rate": 4.858185252747864e-06, "loss": 1.5697, "step": 292500 }, { "epoch": 2.71, "learning_rate": 4.781019757453416e-06, "loss": 1.573, "step": 293000 }, { "epoch": 2.72, "learning_rate": 4.703854262158967e-06, "loss": 1.5754, "step": 293500 }, { "epoch": 2.72, "learning_rate": 4.626688766864519e-06, "loss": 1.5677, "step": 294000 }, { "epoch": 2.73, "learning_rate": 4.5495232715700714e-06, "loss": 1.5689, "step": 294500 }, { "epoch": 2.73, "learning_rate": 4.472357776275623e-06, "loss": 1.5688, "step": 295000 }, { "epoch": 2.74, "learning_rate": 4.395192280981175e-06, "loss": 1.5759, "step": 295500 }, { "epoch": 2.74, "learning_rate": 4.318026785686727e-06, "loss": 1.5662, "step": 296000 }, { "epoch": 2.75, "learning_rate": 4.240861290392278e-06, "loss": 1.5725, "step": 296500 }, { "epoch": 2.75, "learning_rate": 4.16369579509783e-06, "loss": 1.5712, "step": 297000 }, { "epoch": 2.75, "learning_rate": 4.0865302998033825e-06, "loss": 1.5632, "step": 297500 }, { "epoch": 2.76, "learning_rate": 4.0093648045089346e-06, "loss": 1.5631, "step": 298000 }, { "epoch": 2.76, "learning_rate": 3.932199309214486e-06, "loss": 1.5639, "step": 298500 }, { "epoch": 2.77, "learning_rate": 3.855033813920038e-06, "loss": 1.5672, "step": 299000 }, { "epoch": 2.77, "learning_rate": 3.77786831862559e-06, "loss": 1.5691, "step": 299500 }, { "epoch": 2.78, "learning_rate": 3.7007028233311418e-06, "loss": 1.5654, "step": 300000 }, { "epoch": 2.78, "learning_rate": 3.623537328036694e-06, "loss": 1.559, "step": 300500 }, { "epoch": 2.79, "learning_rate": 3.5463718327422456e-06, "loss": 1.5709, "step": 301000 }, { "epoch": 2.79, "learning_rate": 3.469206337447798e-06, "loss": 1.5669, "step": 301500 }, { "epoch": 2.8, "learning_rate": 3.39204084215335e-06, "loss": 1.5635, "step": 302000 }, { "epoch": 2.8, "learning_rate": 3.314875346858901e-06, "loss": 1.5659, "step": 302500 }, { "epoch": 2.81, "learning_rate": 3.2377098515644536e-06, "loss": 1.5623, "step": 303000 }, { "epoch": 2.81, "learning_rate": 3.1605443562700053e-06, "loss": 1.5638, "step": 303500 }, { "epoch": 2.81, "learning_rate": 3.083378860975557e-06, "loss": 1.5644, "step": 304000 }, { "epoch": 2.82, "learning_rate": 3.006213365681109e-06, "loss": 1.5593, "step": 304500 }, { "epoch": 2.82, "learning_rate": 2.929047870386661e-06, "loss": 1.5592, "step": 305000 }, { "epoch": 2.83, "learning_rate": 2.851882375092213e-06, "loss": 1.5571, "step": 305500 }, { "epoch": 2.83, "learning_rate": 2.774716879797765e-06, "loss": 1.5574, "step": 306000 }, { "epoch": 2.84, "learning_rate": 2.6975513845033168e-06, "loss": 1.557, "step": 306500 }, { "epoch": 2.84, "learning_rate": 2.6203858892088685e-06, "loss": 1.5555, "step": 307000 }, { "epoch": 2.85, "learning_rate": 2.5432203939144206e-06, "loss": 1.5559, "step": 307500 }, { "epoch": 2.85, "learning_rate": 2.4660548986199723e-06, "loss": 1.562, "step": 308000 }, { "epoch": 2.86, "learning_rate": 2.3888894033255244e-06, "loss": 1.5609, "step": 308500 }, { "epoch": 2.86, "learning_rate": 2.311723908031076e-06, "loss": 1.5584, "step": 309000 }, { "epoch": 2.87, "learning_rate": 2.234558412736628e-06, "loss": 1.5618, "step": 309500 }, { "epoch": 2.87, "learning_rate": 2.15739291744218e-06, "loss": 1.5539, "step": 310000 }, { "epoch": 2.88, "learning_rate": 2.080227422147732e-06, "loss": 1.5567, "step": 310500 }, { "epoch": 2.88, "learning_rate": 2.0030619268532837e-06, "loss": 1.5585, "step": 311000 }, { "epoch": 2.88, "learning_rate": 1.925896431558836e-06, "loss": 1.5562, "step": 311500 }, { "epoch": 2.89, "learning_rate": 1.8487309362643875e-06, "loss": 1.5575, "step": 312000 }, { "epoch": 2.89, "learning_rate": 1.7715654409699394e-06, "loss": 1.5587, "step": 312500 }, { "epoch": 2.9, "learning_rate": 1.6943999456754913e-06, "loss": 1.5602, "step": 313000 }, { "epoch": 2.9, "learning_rate": 1.6172344503810433e-06, "loss": 1.562, "step": 313500 }, { "epoch": 2.91, "learning_rate": 1.5400689550865952e-06, "loss": 1.5581, "step": 314000 }, { "epoch": 2.91, "learning_rate": 1.462903459792147e-06, "loss": 1.5536, "step": 314500 }, { "epoch": 2.92, "learning_rate": 1.385737964497699e-06, "loss": 1.5588, "step": 315000 }, { "epoch": 2.92, "learning_rate": 1.3085724692032509e-06, "loss": 1.5569, "step": 315500 }, { "epoch": 2.93, "learning_rate": 1.2314069739088028e-06, "loss": 1.5536, "step": 316000 }, { "epoch": 2.93, "learning_rate": 1.1542414786143547e-06, "loss": 1.5558, "step": 316500 }, { "epoch": 2.94, "learning_rate": 1.0770759833199066e-06, "loss": 1.5585, "step": 317000 }, { "epoch": 2.94, "learning_rate": 9.999104880254585e-07, "loss": 1.5542, "step": 317500 }, { "epoch": 2.94, "learning_rate": 9.227449927310104e-07, "loss": 1.5529, "step": 318000 }, { "epoch": 2.95, "learning_rate": 8.455794974365622e-07, "loss": 1.5605, "step": 318500 }, { "epoch": 2.95, "learning_rate": 7.684140021421142e-07, "loss": 1.5496, "step": 319000 }, { "epoch": 2.96, "learning_rate": 6.91248506847666e-07, "loss": 1.5489, "step": 319500 }, { "epoch": 2.96, "learning_rate": 6.140830115532179e-07, "loss": 1.5593, "step": 320000 }, { "epoch": 2.97, "learning_rate": 5.3691751625877e-07, "loss": 1.5583, "step": 320500 }, { "epoch": 2.97, "learning_rate": 4.597520209643218e-07, "loss": 1.5566, "step": 321000 }, { "epoch": 2.98, "learning_rate": 3.825865256698737e-07, "loss": 1.557, "step": 321500 }, { "epoch": 2.98, "learning_rate": 3.0542103037542556e-07, "loss": 1.5617, "step": 322000 }, { "epoch": 2.99, "learning_rate": 2.282555350809775e-07, "loss": 1.5574, "step": 322500 }, { "epoch": 2.99, "learning_rate": 1.5109003978652938e-07, "loss": 1.5566, "step": 323000 }, { "epoch": 3.0, "learning_rate": 7.392454449208128e-08, "loss": 1.551, "step": 323500 }, { "epoch": 3.0, "step": 323979, "total_flos": 2.728710874004779e+18, "train_loss": 3.0939811468297327, "train_runtime": 276726.3152, "train_samples_per_second": 37.464, "train_steps_per_second": 1.171 } ], "max_steps": 323979, "num_train_epochs": 3, "total_flos": 2.728710874004779e+18, "trial_name": null, "trial_params": null }