{ "best_metric": null, "best_model_checkpoint": null, "epoch": 4.749841671944268, "global_step": 150000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.02, "learning_rate": 9.968334388853705e-05, "loss": 2.2174, "step": 500 }, { "epoch": 0.03, "learning_rate": 9.936668777707411e-05, "loss": 2.0257, "step": 1000 }, { "epoch": 0.05, "learning_rate": 9.905003166561115e-05, "loss": 1.9624, "step": 1500 }, { "epoch": 0.06, "learning_rate": 9.87333755541482e-05, "loss": 1.921, "step": 2000 }, { "epoch": 0.08, "learning_rate": 9.841671944268525e-05, "loss": 1.8888, "step": 2500 }, { "epoch": 0.09, "learning_rate": 9.81000633312223e-05, "loss": 1.8642, "step": 3000 }, { "epoch": 0.11, "learning_rate": 9.778340721975935e-05, "loss": 1.8481, "step": 3500 }, { "epoch": 0.13, "learning_rate": 9.74667511082964e-05, "loss": 1.8294, "step": 4000 }, { "epoch": 0.14, "learning_rate": 9.715009499683344e-05, "loss": 1.8145, "step": 4500 }, { "epoch": 0.16, "learning_rate": 9.68334388853705e-05, "loss": 1.8031, "step": 5000 }, { "epoch": 0.17, "learning_rate": 9.651678277390754e-05, "loss": 1.7909, "step": 5500 }, { "epoch": 0.19, "learning_rate": 9.620012666244459e-05, "loss": 1.7796, "step": 6000 }, { "epoch": 0.21, "learning_rate": 9.588347055098163e-05, "loss": 1.7726, "step": 6500 }, { "epoch": 0.22, "learning_rate": 9.556681443951869e-05, "loss": 1.7616, "step": 7000 }, { "epoch": 0.24, "learning_rate": 9.525015832805574e-05, "loss": 1.7524, "step": 7500 }, { "epoch": 0.25, "learning_rate": 9.493350221659278e-05, "loss": 1.7438, "step": 8000 }, { "epoch": 0.27, "learning_rate": 9.461684610512982e-05, "loss": 1.7382, "step": 8500 }, { "epoch": 0.28, "learning_rate": 9.430018999366689e-05, "loss": 1.7336, "step": 9000 }, { "epoch": 0.3, "learning_rate": 9.398353388220393e-05, "loss": 1.7286, "step": 9500 }, { "epoch": 0.32, "learning_rate": 9.366687777074097e-05, "loss": 1.7181, "step": 10000 }, { "epoch": 0.33, "learning_rate": 9.335022165927803e-05, "loss": 1.7131, "step": 10500 }, { "epoch": 0.35, "learning_rate": 9.303356554781508e-05, "loss": 1.7103, "step": 11000 }, { "epoch": 0.36, "learning_rate": 9.271690943635212e-05, "loss": 1.7029, "step": 11500 }, { "epoch": 0.38, "learning_rate": 9.240025332488918e-05, "loss": 1.6999, "step": 12000 }, { "epoch": 0.4, "learning_rate": 9.208359721342622e-05, "loss": 1.6927, "step": 12500 }, { "epoch": 0.41, "learning_rate": 9.176694110196327e-05, "loss": 1.6873, "step": 13000 }, { "epoch": 0.43, "learning_rate": 9.145028499050033e-05, "loss": 1.6841, "step": 13500 }, { "epoch": 0.44, "learning_rate": 9.113362887903737e-05, "loss": 1.6808, "step": 14000 }, { "epoch": 0.46, "learning_rate": 9.081697276757441e-05, "loss": 1.6772, "step": 14500 }, { "epoch": 0.47, "learning_rate": 9.050031665611148e-05, "loss": 1.6761, "step": 15000 }, { "epoch": 0.49, "learning_rate": 9.018366054464852e-05, "loss": 1.6673, "step": 15500 }, { "epoch": 0.51, "learning_rate": 8.986700443318556e-05, "loss": 1.6643, "step": 16000 }, { "epoch": 0.52, "learning_rate": 8.955034832172262e-05, "loss": 1.6645, "step": 16500 }, { "epoch": 0.54, "learning_rate": 8.923369221025967e-05, "loss": 1.6588, "step": 17000 }, { "epoch": 0.55, "learning_rate": 8.891703609879671e-05, "loss": 1.6529, "step": 17500 }, { "epoch": 0.57, "learning_rate": 8.860037998733377e-05, "loss": 1.6512, "step": 18000 }, { "epoch": 0.59, "learning_rate": 8.82837238758708e-05, "loss": 1.6486, "step": 18500 }, { "epoch": 0.6, "learning_rate": 8.796706776440786e-05, "loss": 1.6475, "step": 19000 }, { "epoch": 0.62, "learning_rate": 8.76504116529449e-05, "loss": 1.6436, "step": 19500 }, { "epoch": 0.63, "learning_rate": 8.733375554148196e-05, "loss": 1.6433, "step": 20000 }, { "epoch": 0.65, "learning_rate": 8.7017099430019e-05, "loss": 1.6343, "step": 20500 }, { "epoch": 0.66, "learning_rate": 8.670044331855605e-05, "loss": 1.632, "step": 21000 }, { "epoch": 0.68, "learning_rate": 8.638378720709311e-05, "loss": 1.633, "step": 21500 }, { "epoch": 0.7, "learning_rate": 8.606713109563015e-05, "loss": 1.6275, "step": 22000 }, { "epoch": 0.71, "learning_rate": 8.575047498416719e-05, "loss": 1.6299, "step": 22500 }, { "epoch": 0.73, "learning_rate": 8.543381887270426e-05, "loss": 1.6206, "step": 23000 }, { "epoch": 0.74, "learning_rate": 8.51171627612413e-05, "loss": 1.624, "step": 23500 }, { "epoch": 0.76, "learning_rate": 8.480050664977834e-05, "loss": 1.6231, "step": 24000 }, { "epoch": 0.78, "learning_rate": 8.44838505383154e-05, "loss": 1.6209, "step": 24500 }, { "epoch": 0.79, "learning_rate": 8.416719442685245e-05, "loss": 1.616, "step": 25000 }, { "epoch": 0.81, "learning_rate": 8.385053831538949e-05, "loss": 1.6126, "step": 25500 }, { "epoch": 0.82, "learning_rate": 8.353388220392654e-05, "loss": 1.6108, "step": 26000 }, { "epoch": 0.84, "learning_rate": 8.321722609246358e-05, "loss": 1.608, "step": 26500 }, { "epoch": 0.85, "learning_rate": 8.290056998100064e-05, "loss": 1.6043, "step": 27000 }, { "epoch": 0.87, "learning_rate": 8.25839138695377e-05, "loss": 1.6054, "step": 27500 }, { "epoch": 0.89, "learning_rate": 8.226725775807473e-05, "loss": 1.5996, "step": 28000 }, { "epoch": 0.9, "learning_rate": 8.195060164661178e-05, "loss": 1.6003, "step": 28500 }, { "epoch": 0.92, "learning_rate": 8.163394553514883e-05, "loss": 1.5999, "step": 29000 }, { "epoch": 0.93, "learning_rate": 8.131728942368589e-05, "loss": 1.5989, "step": 29500 }, { "epoch": 0.95, "learning_rate": 8.100063331222293e-05, "loss": 1.5923, "step": 30000 }, { "epoch": 0.97, "learning_rate": 8.068397720075998e-05, "loss": 1.5925, "step": 30500 }, { "epoch": 0.98, "learning_rate": 8.036732108929704e-05, "loss": 1.5924, "step": 31000 }, { "epoch": 1.0, "learning_rate": 8.005066497783408e-05, "loss": 1.5856, "step": 31500 }, { "epoch": 1.01, "learning_rate": 7.973400886637112e-05, "loss": 1.586, "step": 32000 }, { "epoch": 1.03, "learning_rate": 7.941735275490817e-05, "loss": 1.5794, "step": 32500 }, { "epoch": 1.04, "learning_rate": 7.910069664344523e-05, "loss": 1.5744, "step": 33000 }, { "epoch": 1.06, "learning_rate": 7.878404053198227e-05, "loss": 1.5747, "step": 33500 }, { "epoch": 1.08, "learning_rate": 7.846738442051932e-05, "loss": 1.5787, "step": 34000 }, { "epoch": 1.09, "learning_rate": 7.815072830905636e-05, "loss": 1.5752, "step": 34500 }, { "epoch": 1.11, "learning_rate": 7.783407219759342e-05, "loss": 1.5735, "step": 35000 }, { "epoch": 1.12, "learning_rate": 7.751741608613047e-05, "loss": 1.5682, "step": 35500 }, { "epoch": 1.14, "learning_rate": 7.720075997466751e-05, "loss": 1.571, "step": 36000 }, { "epoch": 1.16, "learning_rate": 7.688410386320455e-05, "loss": 1.5647, "step": 36500 }, { "epoch": 1.17, "learning_rate": 7.656744775174162e-05, "loss": 1.5661, "step": 37000 }, { "epoch": 1.19, "learning_rate": 7.625079164027866e-05, "loss": 1.5652, "step": 37500 }, { "epoch": 1.2, "learning_rate": 7.59341355288157e-05, "loss": 1.5601, "step": 38000 }, { "epoch": 1.22, "learning_rate": 7.561747941735276e-05, "loss": 1.56, "step": 38500 }, { "epoch": 1.23, "learning_rate": 7.530082330588981e-05, "loss": 1.5614, "step": 39000 }, { "epoch": 1.25, "learning_rate": 7.498416719442685e-05, "loss": 1.5619, "step": 39500 }, { "epoch": 1.27, "learning_rate": 7.466751108296391e-05, "loss": 1.5608, "step": 40000 }, { "epoch": 1.28, "learning_rate": 7.435085497150095e-05, "loss": 1.5553, "step": 40500 }, { "epoch": 1.3, "learning_rate": 7.4034198860038e-05, "loss": 1.5573, "step": 41000 }, { "epoch": 1.31, "learning_rate": 7.371754274857506e-05, "loss": 1.5524, "step": 41500 }, { "epoch": 1.33, "learning_rate": 7.34008866371121e-05, "loss": 1.555, "step": 42000 }, { "epoch": 1.35, "learning_rate": 7.308423052564914e-05, "loss": 1.5511, "step": 42500 }, { "epoch": 1.36, "learning_rate": 7.27675744141862e-05, "loss": 1.5479, "step": 43000 }, { "epoch": 1.38, "learning_rate": 7.245091830272325e-05, "loss": 1.5468, "step": 43500 }, { "epoch": 1.39, "learning_rate": 7.213426219126029e-05, "loss": 1.5462, "step": 44000 }, { "epoch": 1.41, "learning_rate": 7.181760607979735e-05, "loss": 1.544, "step": 44500 }, { "epoch": 1.42, "learning_rate": 7.15009499683344e-05, "loss": 1.5454, "step": 45000 }, { "epoch": 1.44, "learning_rate": 7.118429385687144e-05, "loss": 1.541, "step": 45500 }, { "epoch": 1.46, "learning_rate": 7.086763774540848e-05, "loss": 1.5408, "step": 46000 }, { "epoch": 1.47, "learning_rate": 7.055098163394554e-05, "loss": 1.5363, "step": 46500 }, { "epoch": 1.49, "learning_rate": 7.023432552248259e-05, "loss": 1.5372, "step": 47000 }, { "epoch": 1.5, "learning_rate": 6.991766941101963e-05, "loss": 1.5402, "step": 47500 }, { "epoch": 1.52, "learning_rate": 6.960101329955669e-05, "loss": 1.5334, "step": 48000 }, { "epoch": 1.54, "learning_rate": 6.928435718809373e-05, "loss": 1.538, "step": 48500 }, { "epoch": 1.55, "learning_rate": 6.896770107663078e-05, "loss": 1.5398, "step": 49000 }, { "epoch": 1.57, "learning_rate": 6.865104496516784e-05, "loss": 1.5366, "step": 49500 }, { "epoch": 1.58, "learning_rate": 6.833438885370488e-05, "loss": 1.5316, "step": 50000 }, { "epoch": 1.6, "learning_rate": 6.801773274224192e-05, "loss": 1.5312, "step": 50500 }, { "epoch": 1.61, "learning_rate": 6.770107663077899e-05, "loss": 1.5289, "step": 51000 }, { "epoch": 1.63, "learning_rate": 6.738442051931603e-05, "loss": 1.5259, "step": 51500 }, { "epoch": 1.65, "learning_rate": 6.706776440785307e-05, "loss": 1.5277, "step": 52000 }, { "epoch": 1.66, "learning_rate": 6.675110829639012e-05, "loss": 1.5232, "step": 52500 }, { "epoch": 1.68, "learning_rate": 6.643445218492718e-05, "loss": 1.5246, "step": 53000 }, { "epoch": 1.69, "learning_rate": 6.611779607346422e-05, "loss": 1.5192, "step": 53500 }, { "epoch": 1.71, "learning_rate": 6.580113996200127e-05, "loss": 1.5224, "step": 54000 }, { "epoch": 1.73, "learning_rate": 6.548448385053831e-05, "loss": 1.5194, "step": 54500 }, { "epoch": 1.74, "learning_rate": 6.516782773907537e-05, "loss": 1.5183, "step": 55000 }, { "epoch": 1.76, "learning_rate": 6.485117162761241e-05, "loss": 1.5183, "step": 55500 }, { "epoch": 1.77, "learning_rate": 6.453451551614947e-05, "loss": 1.5164, "step": 56000 }, { "epoch": 1.79, "learning_rate": 6.42178594046865e-05, "loss": 1.5173, "step": 56500 }, { "epoch": 1.8, "learning_rate": 6.390120329322356e-05, "loss": 1.5144, "step": 57000 }, { "epoch": 1.82, "learning_rate": 6.358454718176062e-05, "loss": 1.5138, "step": 57500 }, { "epoch": 1.84, "learning_rate": 6.326789107029766e-05, "loss": 1.5117, "step": 58000 }, { "epoch": 1.85, "learning_rate": 6.29512349588347e-05, "loss": 1.5129, "step": 58500 }, { "epoch": 1.87, "learning_rate": 6.263457884737177e-05, "loss": 1.5102, "step": 59000 }, { "epoch": 1.88, "learning_rate": 6.23179227359088e-05, "loss": 1.5083, "step": 59500 }, { "epoch": 1.9, "learning_rate": 6.200126662444585e-05, "loss": 1.5079, "step": 60000 }, { "epoch": 1.92, "learning_rate": 6.16846105129829e-05, "loss": 1.5065, "step": 60500 }, { "epoch": 1.93, "learning_rate": 6.136795440151996e-05, "loss": 1.507, "step": 61000 }, { "epoch": 1.95, "learning_rate": 6.1051298290057e-05, "loss": 1.5056, "step": 61500 }, { "epoch": 1.96, "learning_rate": 6.0734642178594045e-05, "loss": 1.5039, "step": 62000 }, { "epoch": 1.98, "learning_rate": 6.041798606713109e-05, "loss": 1.5013, "step": 62500 }, { "epoch": 1.99, "learning_rate": 6.010132995566815e-05, "loss": 1.5026, "step": 63000 }, { "epoch": 2.01, "learning_rate": 5.9784673844205195e-05, "loss": 1.5002, "step": 63500 }, { "epoch": 2.03, "learning_rate": 5.946801773274224e-05, "loss": 1.4947, "step": 64000 }, { "epoch": 2.04, "learning_rate": 5.915136162127929e-05, "loss": 1.4905, "step": 64500 }, { "epoch": 2.06, "learning_rate": 5.8834705509816345e-05, "loss": 1.4922, "step": 65000 }, { "epoch": 2.07, "learning_rate": 5.851804939835339e-05, "loss": 1.4918, "step": 65500 }, { "epoch": 2.09, "learning_rate": 5.820139328689044e-05, "loss": 1.4919, "step": 66000 }, { "epoch": 2.11, "learning_rate": 5.788473717542748e-05, "loss": 1.4885, "step": 66500 }, { "epoch": 2.12, "learning_rate": 5.7568081063964543e-05, "loss": 1.4895, "step": 67000 }, { "epoch": 2.14, "learning_rate": 5.7251424952501584e-05, "loss": 1.4913, "step": 67500 }, { "epoch": 2.15, "learning_rate": 5.693476884103863e-05, "loss": 1.4878, "step": 68000 }, { "epoch": 2.17, "learning_rate": 5.661811272957568e-05, "loss": 1.4846, "step": 68500 }, { "epoch": 2.18, "learning_rate": 5.6301456618112735e-05, "loss": 1.4845, "step": 69000 }, { "epoch": 2.2, "learning_rate": 5.598480050664978e-05, "loss": 1.4861, "step": 69500 }, { "epoch": 2.22, "learning_rate": 5.566814439518683e-05, "loss": 1.4872, "step": 70000 }, { "epoch": 2.23, "learning_rate": 5.535148828372387e-05, "loss": 1.4821, "step": 70500 }, { "epoch": 2.25, "learning_rate": 5.503483217226093e-05, "loss": 1.484, "step": 71000 }, { "epoch": 2.26, "learning_rate": 5.4718176060797973e-05, "loss": 1.483, "step": 71500 }, { "epoch": 2.28, "learning_rate": 5.440151994933502e-05, "loss": 1.4782, "step": 72000 }, { "epoch": 2.3, "learning_rate": 5.408486383787207e-05, "loss": 1.486, "step": 72500 }, { "epoch": 2.31, "learning_rate": 5.3768207726409124e-05, "loss": 1.4793, "step": 73000 }, { "epoch": 2.33, "learning_rate": 5.345155161494617e-05, "loss": 1.4749, "step": 73500 }, { "epoch": 2.34, "learning_rate": 5.313489550348322e-05, "loss": 1.4787, "step": 74000 }, { "epoch": 2.36, "learning_rate": 5.281823939202026e-05, "loss": 1.4756, "step": 74500 }, { "epoch": 2.37, "learning_rate": 5.250158328055732e-05, "loss": 1.4739, "step": 75000 }, { "epoch": 2.39, "learning_rate": 5.218492716909437e-05, "loss": 1.4759, "step": 75500 }, { "epoch": 2.41, "learning_rate": 5.186827105763141e-05, "loss": 1.4725, "step": 76000 }, { "epoch": 2.42, "learning_rate": 5.155161494616846e-05, "loss": 1.4745, "step": 76500 }, { "epoch": 2.44, "learning_rate": 5.123495883470551e-05, "loss": 1.4759, "step": 77000 }, { "epoch": 2.45, "learning_rate": 5.091830272324256e-05, "loss": 1.4744, "step": 77500 }, { "epoch": 2.47, "learning_rate": 5.060164661177961e-05, "loss": 1.4726, "step": 78000 }, { "epoch": 2.49, "learning_rate": 5.0284990500316656e-05, "loss": 1.4721, "step": 78500 }, { "epoch": 2.5, "learning_rate": 4.9968334388853704e-05, "loss": 1.4704, "step": 79000 }, { "epoch": 2.52, "learning_rate": 4.965167827739076e-05, "loss": 1.4687, "step": 79500 }, { "epoch": 2.53, "learning_rate": 4.93350221659278e-05, "loss": 1.4663, "step": 80000 }, { "epoch": 2.55, "learning_rate": 4.9018366054464854e-05, "loss": 1.4684, "step": 80500 }, { "epoch": 2.56, "learning_rate": 4.87017099430019e-05, "loss": 1.4685, "step": 81000 }, { "epoch": 2.58, "learning_rate": 4.838505383153895e-05, "loss": 1.4665, "step": 81500 }, { "epoch": 2.6, "learning_rate": 4.8068397720076e-05, "loss": 1.4646, "step": 82000 }, { "epoch": 2.61, "learning_rate": 4.775174160861305e-05, "loss": 1.4639, "step": 82500 }, { "epoch": 2.63, "learning_rate": 4.743508549715009e-05, "loss": 1.4638, "step": 83000 }, { "epoch": 2.64, "learning_rate": 4.711842938568715e-05, "loss": 1.4595, "step": 83500 }, { "epoch": 2.66, "learning_rate": 4.6801773274224195e-05, "loss": 1.4613, "step": 84000 }, { "epoch": 2.68, "learning_rate": 4.648511716276124e-05, "loss": 1.4589, "step": 84500 }, { "epoch": 2.69, "learning_rate": 4.616846105129829e-05, "loss": 1.4571, "step": 85000 }, { "epoch": 2.71, "learning_rate": 4.585180493983534e-05, "loss": 1.4619, "step": 85500 }, { "epoch": 2.72, "learning_rate": 4.5535148828372386e-05, "loss": 1.4614, "step": 86000 }, { "epoch": 2.74, "learning_rate": 4.521849271690944e-05, "loss": 1.4542, "step": 86500 }, { "epoch": 2.75, "learning_rate": 4.490183660544648e-05, "loss": 1.4571, "step": 87000 }, { "epoch": 2.77, "learning_rate": 4.4585180493983536e-05, "loss": 1.4544, "step": 87500 }, { "epoch": 2.79, "learning_rate": 4.4268524382520584e-05, "loss": 1.4499, "step": 88000 }, { "epoch": 2.8, "learning_rate": 4.395186827105763e-05, "loss": 1.4543, "step": 88500 }, { "epoch": 2.82, "learning_rate": 4.3635212159594687e-05, "loss": 1.4536, "step": 89000 }, { "epoch": 2.83, "learning_rate": 4.331855604813173e-05, "loss": 1.4525, "step": 89500 }, { "epoch": 2.85, "learning_rate": 4.300189993666878e-05, "loss": 1.4521, "step": 90000 }, { "epoch": 2.87, "learning_rate": 4.268524382520583e-05, "loss": 1.4548, "step": 90500 }, { "epoch": 2.88, "learning_rate": 4.236858771374288e-05, "loss": 1.4519, "step": 91000 }, { "epoch": 2.9, "learning_rate": 4.2051931602279925e-05, "loss": 1.4509, "step": 91500 }, { "epoch": 2.91, "learning_rate": 4.173527549081698e-05, "loss": 1.4498, "step": 92000 }, { "epoch": 2.93, "learning_rate": 4.141861937935402e-05, "loss": 1.4517, "step": 92500 }, { "epoch": 2.94, "learning_rate": 4.1101963267891076e-05, "loss": 1.4504, "step": 93000 }, { "epoch": 2.96, "learning_rate": 4.0785307156428123e-05, "loss": 1.4453, "step": 93500 }, { "epoch": 2.98, "learning_rate": 4.046865104496517e-05, "loss": 1.4467, "step": 94000 }, { "epoch": 2.99, "learning_rate": 4.015199493350222e-05, "loss": 1.4442, "step": 94500 }, { "epoch": 3.01, "learning_rate": 3.983533882203927e-05, "loss": 1.4394, "step": 95000 }, { "epoch": 3.02, "learning_rate": 3.9518682710576315e-05, "loss": 1.4389, "step": 95500 }, { "epoch": 3.04, "learning_rate": 3.920202659911337e-05, "loss": 1.4357, "step": 96000 }, { "epoch": 3.06, "learning_rate": 3.888537048765041e-05, "loss": 1.4373, "step": 96500 }, { "epoch": 3.07, "learning_rate": 3.8568714376187465e-05, "loss": 1.4373, "step": 97000 }, { "epoch": 3.09, "learning_rate": 3.825205826472451e-05, "loss": 1.4371, "step": 97500 }, { "epoch": 3.1, "learning_rate": 3.793540215326156e-05, "loss": 1.4355, "step": 98000 }, { "epoch": 3.12, "learning_rate": 3.761874604179861e-05, "loss": 1.4365, "step": 98500 }, { "epoch": 3.13, "learning_rate": 3.730208993033566e-05, "loss": 1.4345, "step": 99000 }, { "epoch": 3.15, "learning_rate": 3.6985433818872704e-05, "loss": 1.4364, "step": 99500 }, { "epoch": 3.17, "learning_rate": 3.666877770740976e-05, "loss": 1.434, "step": 100000 }, { "epoch": 3.18, "learning_rate": 3.6352121595946806e-05, "loss": 1.4322, "step": 100500 }, { "epoch": 3.2, "learning_rate": 3.6035465484483854e-05, "loss": 1.4323, "step": 101000 }, { "epoch": 3.21, "learning_rate": 3.57188093730209e-05, "loss": 1.4275, "step": 101500 }, { "epoch": 3.23, "learning_rate": 3.540215326155795e-05, "loss": 1.4307, "step": 102000 }, { "epoch": 3.25, "learning_rate": 3.5085497150095e-05, "loss": 1.4314, "step": 102500 }, { "epoch": 3.26, "learning_rate": 3.476884103863205e-05, "loss": 1.4298, "step": 103000 }, { "epoch": 3.28, "learning_rate": 3.445218492716909e-05, "loss": 1.4331, "step": 103500 }, { "epoch": 3.29, "learning_rate": 3.413552881570615e-05, "loss": 1.4295, "step": 104000 }, { "epoch": 3.31, "learning_rate": 3.3818872704243195e-05, "loss": 1.4279, "step": 104500 }, { "epoch": 3.32, "learning_rate": 3.350221659278024e-05, "loss": 1.4282, "step": 105000 }, { "epoch": 3.34, "learning_rate": 3.318556048131729e-05, "loss": 1.4276, "step": 105500 }, { "epoch": 3.36, "learning_rate": 3.2868904369854345e-05, "loss": 1.4255, "step": 106000 }, { "epoch": 3.37, "learning_rate": 3.2552248258391386e-05, "loss": 1.425, "step": 106500 }, { "epoch": 3.39, "learning_rate": 3.223559214692844e-05, "loss": 1.426, "step": 107000 }, { "epoch": 3.4, "learning_rate": 3.191893603546549e-05, "loss": 1.4277, "step": 107500 }, { "epoch": 3.42, "learning_rate": 3.1602279924002536e-05, "loss": 1.4247, "step": 108000 }, { "epoch": 3.44, "learning_rate": 3.1285623812539584e-05, "loss": 1.4271, "step": 108500 }, { "epoch": 3.45, "learning_rate": 3.096896770107663e-05, "loss": 1.421, "step": 109000 }, { "epoch": 3.47, "learning_rate": 3.065231158961368e-05, "loss": 1.4226, "step": 109500 }, { "epoch": 3.48, "learning_rate": 3.033565547815073e-05, "loss": 1.4163, "step": 110000 }, { "epoch": 3.5, "learning_rate": 3.001899936668778e-05, "loss": 1.4214, "step": 110500 }, { "epoch": 3.51, "learning_rate": 2.970234325522483e-05, "loss": 1.4168, "step": 111000 }, { "epoch": 3.53, "learning_rate": 2.9385687143761874e-05, "loss": 1.4177, "step": 111500 }, { "epoch": 3.55, "learning_rate": 2.9069031032298925e-05, "loss": 1.4157, "step": 112000 }, { "epoch": 3.56, "learning_rate": 2.8752374920835973e-05, "loss": 1.4187, "step": 112500 }, { "epoch": 3.58, "learning_rate": 2.8435718809373024e-05, "loss": 1.4151, "step": 113000 }, { "epoch": 3.59, "learning_rate": 2.811906269791007e-05, "loss": 1.4233, "step": 113500 }, { "epoch": 3.61, "learning_rate": 2.7802406586447123e-05, "loss": 1.4135, "step": 114000 }, { "epoch": 3.63, "learning_rate": 2.7485750474984168e-05, "loss": 1.4184, "step": 114500 }, { "epoch": 3.64, "learning_rate": 2.716909436352122e-05, "loss": 1.4151, "step": 115000 }, { "epoch": 3.66, "learning_rate": 2.6852438252058267e-05, "loss": 1.4122, "step": 115500 }, { "epoch": 3.67, "learning_rate": 2.6535782140595318e-05, "loss": 1.4132, "step": 116000 }, { "epoch": 3.69, "learning_rate": 2.6219126029132362e-05, "loss": 1.4149, "step": 116500 }, { "epoch": 3.7, "learning_rate": 2.5902469917669413e-05, "loss": 1.4156, "step": 117000 }, { "epoch": 3.72, "learning_rate": 2.558581380620646e-05, "loss": 1.4125, "step": 117500 }, { "epoch": 3.74, "learning_rate": 2.5269157694743512e-05, "loss": 1.4098, "step": 118000 }, { "epoch": 3.75, "learning_rate": 2.4952501583280557e-05, "loss": 1.4104, "step": 118500 }, { "epoch": 3.77, "learning_rate": 2.4635845471817608e-05, "loss": 1.4128, "step": 119000 }, { "epoch": 3.78, "learning_rate": 2.4319189360354656e-05, "loss": 1.4085, "step": 119500 }, { "epoch": 3.8, "learning_rate": 2.4002533248891703e-05, "loss": 1.4081, "step": 120000 }, { "epoch": 3.82, "learning_rate": 2.368587713742875e-05, "loss": 1.4113, "step": 120500 }, { "epoch": 3.83, "learning_rate": 2.3369221025965802e-05, "loss": 1.4078, "step": 121000 }, { "epoch": 3.85, "learning_rate": 2.305256491450285e-05, "loss": 1.406, "step": 121500 }, { "epoch": 3.86, "learning_rate": 2.2735908803039898e-05, "loss": 1.4044, "step": 122000 }, { "epoch": 3.88, "learning_rate": 2.2419252691576946e-05, "loss": 1.4063, "step": 122500 }, { "epoch": 3.89, "learning_rate": 2.2102596580113997e-05, "loss": 1.4076, "step": 123000 }, { "epoch": 3.91, "learning_rate": 2.1785940468651045e-05, "loss": 1.4041, "step": 123500 }, { "epoch": 3.93, "learning_rate": 2.1469284357188092e-05, "loss": 1.4071, "step": 124000 }, { "epoch": 3.94, "learning_rate": 2.1152628245725144e-05, "loss": 1.4048, "step": 124500 }, { "epoch": 3.96, "learning_rate": 2.083597213426219e-05, "loss": 1.4033, "step": 125000 }, { "epoch": 3.97, "learning_rate": 2.051931602279924e-05, "loss": 1.4038, "step": 125500 }, { "epoch": 3.99, "learning_rate": 2.0202659911336287e-05, "loss": 1.4059, "step": 126000 }, { "epoch": 4.01, "learning_rate": 1.9886003799873338e-05, "loss": 1.4011, "step": 126500 }, { "epoch": 4.02, "learning_rate": 1.9569347688410386e-05, "loss": 1.3957, "step": 127000 }, { "epoch": 4.04, "learning_rate": 1.9252691576947434e-05, "loss": 1.3968, "step": 127500 }, { "epoch": 4.05, "learning_rate": 1.8936035465484485e-05, "loss": 1.3945, "step": 128000 }, { "epoch": 4.07, "learning_rate": 1.8619379354021533e-05, "loss": 1.3927, "step": 128500 }, { "epoch": 4.08, "learning_rate": 1.830272324255858e-05, "loss": 1.3953, "step": 129000 }, { "epoch": 4.1, "learning_rate": 1.7986067131095628e-05, "loss": 1.3943, "step": 129500 }, { "epoch": 4.12, "learning_rate": 1.766941101963268e-05, "loss": 1.3951, "step": 130000 }, { "epoch": 4.13, "learning_rate": 1.7352754908169727e-05, "loss": 1.3936, "step": 130500 }, { "epoch": 4.15, "learning_rate": 1.7036098796706775e-05, "loss": 1.3976, "step": 131000 }, { "epoch": 4.16, "learning_rate": 1.6719442685243826e-05, "loss": 1.3921, "step": 131500 }, { "epoch": 4.18, "learning_rate": 1.6402786573780877e-05, "loss": 1.3929, "step": 132000 }, { "epoch": 4.2, "learning_rate": 1.6086130462317925e-05, "loss": 1.391, "step": 132500 }, { "epoch": 4.21, "learning_rate": 1.5769474350854973e-05, "loss": 1.3901, "step": 133000 }, { "epoch": 4.23, "learning_rate": 1.545281823939202e-05, "loss": 1.3903, "step": 133500 }, { "epoch": 4.24, "learning_rate": 1.5136162127929072e-05, "loss": 1.3864, "step": 134000 }, { "epoch": 4.26, "learning_rate": 1.481950601646612e-05, "loss": 1.3891, "step": 134500 }, { "epoch": 4.27, "learning_rate": 1.4502849905003169e-05, "loss": 1.3897, "step": 135000 }, { "epoch": 4.29, "learning_rate": 1.4186193793540217e-05, "loss": 1.3908, "step": 135500 }, { "epoch": 4.31, "learning_rate": 1.3869537682077266e-05, "loss": 1.3893, "step": 136000 }, { "epoch": 4.32, "learning_rate": 1.3552881570614314e-05, "loss": 1.3876, "step": 136500 }, { "epoch": 4.34, "learning_rate": 1.3236225459151364e-05, "loss": 1.3921, "step": 137000 }, { "epoch": 4.35, "learning_rate": 1.2919569347688411e-05, "loss": 1.3869, "step": 137500 }, { "epoch": 4.37, "learning_rate": 1.2602913236225461e-05, "loss": 1.3846, "step": 138000 }, { "epoch": 4.39, "learning_rate": 1.2286257124762509e-05, "loss": 1.3891, "step": 138500 }, { "epoch": 4.4, "learning_rate": 1.1969601013299556e-05, "loss": 1.3834, "step": 139000 }, { "epoch": 4.42, "learning_rate": 1.1652944901836606e-05, "loss": 1.3868, "step": 139500 }, { "epoch": 4.43, "learning_rate": 1.1336288790373654e-05, "loss": 1.3866, "step": 140000 }, { "epoch": 4.45, "learning_rate": 1.1019632678910703e-05, "loss": 1.385, "step": 140500 }, { "epoch": 4.46, "learning_rate": 1.0702976567447753e-05, "loss": 1.3841, "step": 141000 }, { "epoch": 4.48, "learning_rate": 1.03863204559848e-05, "loss": 1.3827, "step": 141500 }, { "epoch": 4.5, "learning_rate": 1.006966434452185e-05, "loss": 1.3826, "step": 142000 }, { "epoch": 4.51, "learning_rate": 9.753008233058898e-06, "loss": 1.3833, "step": 142500 }, { "epoch": 4.53, "learning_rate": 9.436352121595947e-06, "loss": 1.3801, "step": 143000 }, { "epoch": 4.54, "learning_rate": 9.119696010132995e-06, "loss": 1.3809, "step": 143500 }, { "epoch": 4.56, "learning_rate": 8.803039898670044e-06, "loss": 1.3808, "step": 144000 }, { "epoch": 4.58, "learning_rate": 8.486383787207092e-06, "loss": 1.3823, "step": 144500 }, { "epoch": 4.59, "learning_rate": 8.169727675744142e-06, "loss": 1.3835, "step": 145000 }, { "epoch": 4.61, "learning_rate": 7.853071564281191e-06, "loss": 1.3794, "step": 145500 }, { "epoch": 4.62, "learning_rate": 7.536415452818239e-06, "loss": 1.3812, "step": 146000 }, { "epoch": 4.64, "learning_rate": 7.219759341355288e-06, "loss": 1.3848, "step": 146500 }, { "epoch": 4.65, "learning_rate": 6.903103229892337e-06, "loss": 1.3784, "step": 147000 }, { "epoch": 4.67, "learning_rate": 6.586447118429386e-06, "loss": 1.3777, "step": 147500 }, { "epoch": 4.69, "learning_rate": 6.269791006966434e-06, "loss": 1.3841, "step": 148000 }, { "epoch": 4.7, "learning_rate": 5.953134895503484e-06, "loss": 1.3802, "step": 148500 }, { "epoch": 4.72, "learning_rate": 5.6364787840405325e-06, "loss": 1.3795, "step": 149000 }, { "epoch": 4.73, "learning_rate": 5.319822672577581e-06, "loss": 1.3818, "step": 149500 }, { "epoch": 4.75, "learning_rate": 5.00316656111463e-06, "loss": 1.3756, "step": 150000 } ], "max_steps": 157900, "num_train_epochs": 5, "total_flos": 2.2108746298825728e+18, "trial_name": null, "trial_params": null }