{ "best_metric": null, "best_model_checkpoint": null, "epoch": 20.0, "eval_steps": 2.0, "global_step": 40920, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 1.9987781036168134e-05, "loss": 1.0996, "step": 25 }, { "epoch": 0.02, "learning_rate": 1.9975562072336266e-05, "loss": 1.0049, "step": 50 }, { "epoch": 0.04, "learning_rate": 1.99633431085044e-05, "loss": 0.8837, "step": 75 }, { "epoch": 0.05, "learning_rate": 1.9951124144672535e-05, "loss": 0.7945, "step": 100 }, { "epoch": 0.06, "learning_rate": 1.9938905180840667e-05, "loss": 0.7541, "step": 125 }, { "epoch": 0.07, "learning_rate": 1.99266862170088e-05, "loss": 0.694, "step": 150 }, { "epoch": 0.09, "learning_rate": 1.9914467253176932e-05, "loss": 0.6936, "step": 175 }, { "epoch": 0.1, "learning_rate": 1.9902248289345064e-05, "loss": 0.6625, "step": 200 }, { "epoch": 0.11, "learning_rate": 1.9890029325513197e-05, "loss": 0.661, "step": 225 }, { "epoch": 0.12, "learning_rate": 1.9877810361681332e-05, "loss": 0.6525, "step": 250 }, { "epoch": 0.13, "learning_rate": 1.9865591397849465e-05, "loss": 0.6298, "step": 275 }, { "epoch": 0.15, "learning_rate": 1.9853372434017597e-05, "loss": 0.6111, "step": 300 }, { "epoch": 0.16, "learning_rate": 1.984115347018573e-05, "loss": 0.62, "step": 325 }, { "epoch": 0.17, "learning_rate": 1.9828934506353862e-05, "loss": 0.611, "step": 350 }, { "epoch": 0.18, "learning_rate": 1.9816715542521994e-05, "loss": 0.5751, "step": 375 }, { "epoch": 0.2, "learning_rate": 1.980449657869013e-05, "loss": 0.6004, "step": 400 }, { "epoch": 0.21, "learning_rate": 1.9792277614858263e-05, "loss": 0.5816, "step": 425 }, { "epoch": 0.22, "learning_rate": 1.9780058651026395e-05, "loss": 0.5813, "step": 450 }, { "epoch": 0.23, "learning_rate": 1.9767839687194527e-05, "loss": 0.5784, "step": 475 }, { "epoch": 0.24, "learning_rate": 1.975562072336266e-05, "loss": 0.5869, "step": 500 }, { "epoch": 0.26, "learning_rate": 1.9743401759530792e-05, "loss": 0.5793, "step": 525 }, { "epoch": 0.27, "learning_rate": 1.9731182795698928e-05, "loss": 0.5782, "step": 550 }, { "epoch": 0.28, "learning_rate": 1.971896383186706e-05, "loss": 0.5638, "step": 575 }, { "epoch": 0.29, "learning_rate": 1.9706744868035193e-05, "loss": 0.5645, "step": 600 }, { "epoch": 0.31, "learning_rate": 1.9694525904203325e-05, "loss": 0.552, "step": 625 }, { "epoch": 0.32, "learning_rate": 1.9682306940371458e-05, "loss": 0.539, "step": 650 }, { "epoch": 0.33, "learning_rate": 1.967008797653959e-05, "loss": 0.5464, "step": 675 }, { "epoch": 0.34, "learning_rate": 1.9657869012707726e-05, "loss": 0.538, "step": 700 }, { "epoch": 0.35, "learning_rate": 1.9645650048875858e-05, "loss": 0.5394, "step": 725 }, { "epoch": 0.37, "learning_rate": 1.963343108504399e-05, "loss": 0.5586, "step": 750 }, { "epoch": 0.38, "learning_rate": 1.9621212121212123e-05, "loss": 0.5241, "step": 775 }, { "epoch": 0.39, "learning_rate": 1.9608993157380255e-05, "loss": 0.5304, "step": 800 }, { "epoch": 0.4, "learning_rate": 1.9596774193548388e-05, "loss": 0.5313, "step": 825 }, { "epoch": 0.42, "learning_rate": 1.9584555229716524e-05, "loss": 0.5423, "step": 850 }, { "epoch": 0.43, "learning_rate": 1.9572336265884656e-05, "loss": 0.5086, "step": 875 }, { "epoch": 0.44, "learning_rate": 1.956011730205279e-05, "loss": 0.5172, "step": 900 }, { "epoch": 0.45, "learning_rate": 1.954789833822092e-05, "loss": 0.5081, "step": 925 }, { "epoch": 0.46, "learning_rate": 1.9535679374389053e-05, "loss": 0.5244, "step": 950 }, { "epoch": 0.48, "learning_rate": 1.9523460410557186e-05, "loss": 0.5371, "step": 975 }, { "epoch": 0.49, "learning_rate": 1.951124144672532e-05, "loss": 0.5278, "step": 1000 }, { "epoch": 0.5, "learning_rate": 1.9499022482893454e-05, "loss": 0.5325, "step": 1025 }, { "epoch": 0.51, "learning_rate": 1.9486803519061586e-05, "loss": 0.5019, "step": 1050 }, { "epoch": 0.53, "learning_rate": 1.947458455522972e-05, "loss": 0.504, "step": 1075 }, { "epoch": 0.54, "learning_rate": 1.946236559139785e-05, "loss": 0.5225, "step": 1100 }, { "epoch": 0.55, "learning_rate": 1.9450146627565983e-05, "loss": 0.5232, "step": 1125 }, { "epoch": 0.56, "learning_rate": 1.9437927663734116e-05, "loss": 0.5282, "step": 1150 }, { "epoch": 0.57, "learning_rate": 1.942570869990225e-05, "loss": 0.514, "step": 1175 }, { "epoch": 0.59, "learning_rate": 1.9413489736070384e-05, "loss": 0.4916, "step": 1200 }, { "epoch": 0.6, "learning_rate": 1.9401270772238516e-05, "loss": 0.5079, "step": 1225 }, { "epoch": 0.61, "learning_rate": 1.938905180840665e-05, "loss": 0.5221, "step": 1250 }, { "epoch": 0.62, "learning_rate": 1.937683284457478e-05, "loss": 0.5097, "step": 1275 }, { "epoch": 0.64, "learning_rate": 1.9364613880742914e-05, "loss": 0.512, "step": 1300 }, { "epoch": 0.65, "learning_rate": 1.935239491691105e-05, "loss": 0.5001, "step": 1325 }, { "epoch": 0.66, "learning_rate": 1.9340175953079182e-05, "loss": 0.5021, "step": 1350 }, { "epoch": 0.67, "learning_rate": 1.9327956989247314e-05, "loss": 0.5044, "step": 1375 }, { "epoch": 0.68, "learning_rate": 1.9315738025415447e-05, "loss": 0.4943, "step": 1400 }, { "epoch": 0.7, "learning_rate": 1.930351906158358e-05, "loss": 0.4935, "step": 1425 }, { "epoch": 0.71, "learning_rate": 1.929130009775171e-05, "loss": 0.5144, "step": 1450 }, { "epoch": 0.72, "learning_rate": 1.9279081133919847e-05, "loss": 0.506, "step": 1475 }, { "epoch": 0.73, "learning_rate": 1.926686217008798e-05, "loss": 0.5079, "step": 1500 }, { "epoch": 0.75, "learning_rate": 1.9254643206256112e-05, "loss": 0.4939, "step": 1525 }, { "epoch": 0.76, "learning_rate": 1.9242424242424244e-05, "loss": 0.4874, "step": 1550 }, { "epoch": 0.77, "learning_rate": 1.9230205278592377e-05, "loss": 0.4994, "step": 1575 }, { "epoch": 0.78, "learning_rate": 1.921798631476051e-05, "loss": 0.4818, "step": 1600 }, { "epoch": 0.79, "learning_rate": 1.9205767350928645e-05, "loss": 0.5026, "step": 1625 }, { "epoch": 0.81, "learning_rate": 1.9193548387096777e-05, "loss": 0.4634, "step": 1650 }, { "epoch": 0.82, "learning_rate": 1.918132942326491e-05, "loss": 0.4824, "step": 1675 }, { "epoch": 0.83, "learning_rate": 1.9169110459433042e-05, "loss": 0.472, "step": 1700 }, { "epoch": 0.84, "learning_rate": 1.9156891495601175e-05, "loss": 0.4811, "step": 1725 }, { "epoch": 0.86, "learning_rate": 1.9144672531769307e-05, "loss": 0.4763, "step": 1750 }, { "epoch": 0.87, "learning_rate": 1.9132453567937443e-05, "loss": 0.4809, "step": 1775 }, { "epoch": 0.88, "learning_rate": 1.9120234604105575e-05, "loss": 0.5, "step": 1800 }, { "epoch": 0.89, "learning_rate": 1.9108015640273708e-05, "loss": 0.4846, "step": 1825 }, { "epoch": 0.9, "learning_rate": 1.909579667644184e-05, "loss": 0.4624, "step": 1850 }, { "epoch": 0.92, "learning_rate": 1.9083577712609972e-05, "loss": 0.4824, "step": 1875 }, { "epoch": 0.93, "learning_rate": 1.9071358748778105e-05, "loss": 0.4708, "step": 1900 }, { "epoch": 0.94, "learning_rate": 1.905913978494624e-05, "loss": 0.4884, "step": 1925 }, { "epoch": 0.95, "learning_rate": 1.9046920821114373e-05, "loss": 0.4768, "step": 1950 }, { "epoch": 0.97, "learning_rate": 1.9034701857282505e-05, "loss": 0.4583, "step": 1975 }, { "epoch": 0.98, "learning_rate": 1.9022482893450638e-05, "loss": 0.4634, "step": 2000 }, { "epoch": 0.99, "learning_rate": 1.901026392961877e-05, "loss": 0.4741, "step": 2025 }, { "epoch": 1.0, "learning_rate": 1.8998044965786903e-05, "loss": 0.458, "step": 2050 }, { "epoch": 1.01, "learning_rate": 1.8985826001955038e-05, "loss": 0.3961, "step": 2075 }, { "epoch": 1.03, "learning_rate": 1.897360703812317e-05, "loss": 0.3929, "step": 2100 }, { "epoch": 1.04, "learning_rate": 1.8961388074291303e-05, "loss": 0.4265, "step": 2125 }, { "epoch": 1.05, "learning_rate": 1.8949169110459435e-05, "loss": 0.4049, "step": 2150 }, { "epoch": 1.06, "learning_rate": 1.8936950146627568e-05, "loss": 0.4171, "step": 2175 }, { "epoch": 1.08, "learning_rate": 1.89247311827957e-05, "loss": 0.4024, "step": 2200 }, { "epoch": 1.09, "learning_rate": 1.8912512218963833e-05, "loss": 0.403, "step": 2225 }, { "epoch": 1.1, "learning_rate": 1.890029325513197e-05, "loss": 0.4105, "step": 2250 }, { "epoch": 1.11, "learning_rate": 1.88880742913001e-05, "loss": 0.3956, "step": 2275 }, { "epoch": 1.12, "learning_rate": 1.8875855327468233e-05, "loss": 0.397, "step": 2300 }, { "epoch": 1.14, "learning_rate": 1.8863636363636366e-05, "loss": 0.3883, "step": 2325 }, { "epoch": 1.15, "learning_rate": 1.8851417399804498e-05, "loss": 0.4056, "step": 2350 }, { "epoch": 1.16, "learning_rate": 1.883919843597263e-05, "loss": 0.3884, "step": 2375 }, { "epoch": 1.17, "learning_rate": 1.8826979472140766e-05, "loss": 0.4129, "step": 2400 }, { "epoch": 1.19, "learning_rate": 1.88147605083089e-05, "loss": 0.4097, "step": 2425 }, { "epoch": 1.2, "learning_rate": 1.880254154447703e-05, "loss": 0.4036, "step": 2450 }, { "epoch": 1.21, "learning_rate": 1.8790322580645163e-05, "loss": 0.3894, "step": 2475 }, { "epoch": 1.22, "learning_rate": 1.8778103616813296e-05, "loss": 0.4147, "step": 2500 }, { "epoch": 1.23, "learning_rate": 1.8765884652981428e-05, "loss": 0.409, "step": 2525 }, { "epoch": 1.25, "learning_rate": 1.8753665689149564e-05, "loss": 0.4106, "step": 2550 }, { "epoch": 1.26, "learning_rate": 1.8741446725317693e-05, "loss": 0.3971, "step": 2575 }, { "epoch": 1.27, "learning_rate": 1.8729227761485825e-05, "loss": 0.3995, "step": 2600 }, { "epoch": 1.28, "learning_rate": 1.8717008797653958e-05, "loss": 0.3947, "step": 2625 }, { "epoch": 1.3, "learning_rate": 1.870527859237537e-05, "loss": 0.4141, "step": 2650 }, { "epoch": 1.31, "learning_rate": 1.8693059628543502e-05, "loss": 0.4024, "step": 2675 }, { "epoch": 1.32, "learning_rate": 1.8680840664711635e-05, "loss": 0.4074, "step": 2700 }, { "epoch": 1.33, "learning_rate": 1.8668621700879767e-05, "loss": 0.3953, "step": 2725 }, { "epoch": 1.34, "learning_rate": 1.86564027370479e-05, "loss": 0.3861, "step": 2750 }, { "epoch": 1.36, "learning_rate": 1.8644183773216032e-05, "loss": 0.3939, "step": 2775 }, { "epoch": 1.37, "learning_rate": 1.8631964809384168e-05, "loss": 0.4088, "step": 2800 }, { "epoch": 1.38, "learning_rate": 1.86197458455523e-05, "loss": 0.4089, "step": 2825 }, { "epoch": 1.39, "learning_rate": 1.8607526881720433e-05, "loss": 0.3987, "step": 2850 }, { "epoch": 1.41, "learning_rate": 1.8595307917888565e-05, "loss": 0.3881, "step": 2875 }, { "epoch": 1.42, "learning_rate": 1.8583088954056697e-05, "loss": 0.3877, "step": 2900 }, { "epoch": 1.43, "learning_rate": 1.857086999022483e-05, "loss": 0.3996, "step": 2925 }, { "epoch": 1.44, "learning_rate": 1.8558651026392966e-05, "loss": 0.3979, "step": 2950 }, { "epoch": 1.45, "learning_rate": 1.8546432062561098e-05, "loss": 0.4009, "step": 2975 }, { "epoch": 1.47, "learning_rate": 1.853421309872923e-05, "loss": 0.4069, "step": 3000 }, { "epoch": 1.48, "learning_rate": 1.8521994134897363e-05, "loss": 0.3725, "step": 3025 }, { "epoch": 1.49, "learning_rate": 1.8509775171065495e-05, "loss": 0.3899, "step": 3050 }, { "epoch": 1.5, "learning_rate": 1.8497556207233628e-05, "loss": 0.3916, "step": 3075 }, { "epoch": 1.52, "learning_rate": 1.8485337243401763e-05, "loss": 0.4114, "step": 3100 }, { "epoch": 1.53, "learning_rate": 1.8473118279569896e-05, "loss": 0.3945, "step": 3125 }, { "epoch": 1.54, "learning_rate": 1.8460899315738028e-05, "loss": 0.3984, "step": 3150 }, { "epoch": 1.55, "learning_rate": 1.844868035190616e-05, "loss": 0.3871, "step": 3175 }, { "epoch": 1.56, "learning_rate": 1.8436461388074293e-05, "loss": 0.3859, "step": 3200 }, { "epoch": 1.58, "learning_rate": 1.8424242424242425e-05, "loss": 0.3926, "step": 3225 }, { "epoch": 1.59, "learning_rate": 1.841202346041056e-05, "loss": 0.3941, "step": 3250 }, { "epoch": 1.6, "learning_rate": 1.8399804496578694e-05, "loss": 0.3922, "step": 3275 }, { "epoch": 1.61, "learning_rate": 1.8387585532746826e-05, "loss": 0.3847, "step": 3300 }, { "epoch": 1.63, "learning_rate": 1.837536656891496e-05, "loss": 0.3709, "step": 3325 }, { "epoch": 1.64, "learning_rate": 1.836314760508309e-05, "loss": 0.3963, "step": 3350 }, { "epoch": 1.65, "learning_rate": 1.8350928641251223e-05, "loss": 0.3639, "step": 3375 }, { "epoch": 1.66, "learning_rate": 1.833870967741936e-05, "loss": 0.4072, "step": 3400 }, { "epoch": 1.67, "learning_rate": 1.832649071358749e-05, "loss": 0.3918, "step": 3425 }, { "epoch": 1.69, "learning_rate": 1.8314271749755624e-05, "loss": 0.3847, "step": 3450 }, { "epoch": 1.7, "learning_rate": 1.8302052785923756e-05, "loss": 0.393, "step": 3475 }, { "epoch": 1.71, "learning_rate": 1.828983382209189e-05, "loss": 0.4028, "step": 3500 }, { "epoch": 1.72, "learning_rate": 1.827761485826002e-05, "loss": 0.3967, "step": 3525 }, { "epoch": 1.74, "learning_rate": 1.8265395894428157e-05, "loss": 0.3993, "step": 3550 }, { "epoch": 1.75, "learning_rate": 1.825317693059629e-05, "loss": 0.3854, "step": 3575 }, { "epoch": 1.76, "learning_rate": 1.8240957966764418e-05, "loss": 0.3644, "step": 3600 }, { "epoch": 1.77, "learning_rate": 1.822873900293255e-05, "loss": 0.4081, "step": 3625 }, { "epoch": 1.78, "learning_rate": 1.8216520039100686e-05, "loss": 0.3836, "step": 3650 }, { "epoch": 1.8, "learning_rate": 1.820430107526882e-05, "loss": 0.3927, "step": 3675 }, { "epoch": 1.81, "learning_rate": 1.819208211143695e-05, "loss": 0.4068, "step": 3700 }, { "epoch": 1.82, "learning_rate": 1.8179863147605084e-05, "loss": 0.4027, "step": 3725 }, { "epoch": 1.83, "learning_rate": 1.8167644183773216e-05, "loss": 0.3736, "step": 3750 }, { "epoch": 1.85, "learning_rate": 1.815542521994135e-05, "loss": 0.3916, "step": 3775 }, { "epoch": 1.86, "learning_rate": 1.814320625610948e-05, "loss": 0.3872, "step": 3800 }, { "epoch": 1.87, "learning_rate": 1.8130987292277617e-05, "loss": 0.3867, "step": 3825 }, { "epoch": 1.88, "learning_rate": 1.811876832844575e-05, "loss": 0.3879, "step": 3850 }, { "epoch": 1.89, "learning_rate": 1.810654936461388e-05, "loss": 0.4013, "step": 3875 }, { "epoch": 1.91, "learning_rate": 1.8094330400782014e-05, "loss": 0.3964, "step": 3900 }, { "epoch": 1.92, "learning_rate": 1.8082111436950146e-05, "loss": 0.4086, "step": 3925 }, { "epoch": 1.93, "learning_rate": 1.806989247311828e-05, "loss": 0.3772, "step": 3950 }, { "epoch": 1.94, "learning_rate": 1.8057673509286414e-05, "loss": 0.3886, "step": 3975 }, { "epoch": 1.96, "learning_rate": 1.8045454545454547e-05, "loss": 0.3952, "step": 4000 }, { "epoch": 1.97, "learning_rate": 1.803323558162268e-05, "loss": 0.3962, "step": 4025 }, { "epoch": 1.98, "learning_rate": 1.802101661779081e-05, "loss": 0.3772, "step": 4050 }, { "epoch": 1.99, "learning_rate": 1.8008797653958944e-05, "loss": 0.3781, "step": 4075 }, { "epoch": 2.0, "learning_rate": 1.7996578690127076e-05, "loss": 0.3505, "step": 4100 }, { "epoch": 2.02, "learning_rate": 1.7984359726295212e-05, "loss": 0.3105, "step": 4125 }, { "epoch": 2.03, "learning_rate": 1.7972140762463344e-05, "loss": 0.3035, "step": 4150 }, { "epoch": 2.04, "learning_rate": 1.7959921798631477e-05, "loss": 0.3081, "step": 4175 }, { "epoch": 2.05, "learning_rate": 1.794770283479961e-05, "loss": 0.3007, "step": 4200 }, { "epoch": 2.07, "learning_rate": 1.7935483870967742e-05, "loss": 0.3166, "step": 4225 }, { "epoch": 2.08, "learning_rate": 1.7923264907135874e-05, "loss": 0.3119, "step": 4250 }, { "epoch": 2.09, "learning_rate": 1.791104594330401e-05, "loss": 0.2894, "step": 4275 }, { "epoch": 2.1, "learning_rate": 1.7898826979472142e-05, "loss": 0.3103, "step": 4300 }, { "epoch": 2.11, "learning_rate": 1.7886608015640275e-05, "loss": 0.3049, "step": 4325 }, { "epoch": 2.13, "learning_rate": 1.7874389051808407e-05, "loss": 0.2996, "step": 4350 }, { "epoch": 2.14, "learning_rate": 1.786217008797654e-05, "loss": 0.3012, "step": 4375 }, { "epoch": 2.15, "learning_rate": 1.7849951124144672e-05, "loss": 0.2915, "step": 4400 }, { "epoch": 2.16, "learning_rate": 1.7837732160312808e-05, "loss": 0.3193, "step": 4425 }, { "epoch": 2.17, "learning_rate": 1.782551319648094e-05, "loss": 0.3132, "step": 4450 }, { "epoch": 2.19, "learning_rate": 1.7813294232649072e-05, "loss": 0.3209, "step": 4475 }, { "epoch": 2.2, "learning_rate": 1.7801075268817205e-05, "loss": 0.2968, "step": 4500 }, { "epoch": 2.21, "learning_rate": 1.7788856304985337e-05, "loss": 0.3092, "step": 4525 }, { "epoch": 2.22, "learning_rate": 1.777663734115347e-05, "loss": 0.2926, "step": 4550 }, { "epoch": 2.24, "learning_rate": 1.7764418377321605e-05, "loss": 0.3175, "step": 4575 }, { "epoch": 2.25, "learning_rate": 1.7752199413489738e-05, "loss": 0.3062, "step": 4600 }, { "epoch": 2.26, "learning_rate": 1.773998044965787e-05, "loss": 0.3107, "step": 4625 }, { "epoch": 2.27, "learning_rate": 1.7727761485826003e-05, "loss": 0.3072, "step": 4650 }, { "epoch": 2.28, "learning_rate": 1.7715542521994135e-05, "loss": 0.3166, "step": 4675 }, { "epoch": 2.3, "learning_rate": 1.7703323558162267e-05, "loss": 0.3181, "step": 4700 }, { "epoch": 2.31, "learning_rate": 1.7691104594330403e-05, "loss": 0.2843, "step": 4725 }, { "epoch": 2.32, "learning_rate": 1.7678885630498536e-05, "loss": 0.3103, "step": 4750 }, { "epoch": 2.33, "learning_rate": 1.7666666666666668e-05, "loss": 0.3039, "step": 4775 }, { "epoch": 2.35, "learning_rate": 1.76544477028348e-05, "loss": 0.3099, "step": 4800 }, { "epoch": 2.36, "learning_rate": 1.7642228739002933e-05, "loss": 0.2989, "step": 4825 }, { "epoch": 2.37, "learning_rate": 1.7630009775171065e-05, "loss": 0.3132, "step": 4850 }, { "epoch": 2.38, "learning_rate": 1.7617790811339198e-05, "loss": 0.2937, "step": 4875 }, { "epoch": 2.39, "learning_rate": 1.7605571847507333e-05, "loss": 0.3067, "step": 4900 }, { "epoch": 2.41, "learning_rate": 1.7593352883675466e-05, "loss": 0.3163, "step": 4925 }, { "epoch": 2.42, "learning_rate": 1.7581133919843598e-05, "loss": 0.3078, "step": 4950 }, { "epoch": 2.43, "learning_rate": 1.756891495601173e-05, "loss": 0.3114, "step": 4975 }, { "epoch": 2.44, "learning_rate": 1.7556695992179863e-05, "loss": 0.3121, "step": 5000 }, { "epoch": 2.46, "learning_rate": 1.7544965786901272e-05, "loss": 0.3018, "step": 5025 }, { "epoch": 2.47, "learning_rate": 1.7532746823069404e-05, "loss": 0.3122, "step": 5050 }, { "epoch": 2.48, "learning_rate": 1.7520527859237537e-05, "loss": 0.3052, "step": 5075 }, { "epoch": 2.49, "learning_rate": 1.750830889540567e-05, "loss": 0.3206, "step": 5100 }, { "epoch": 2.5, "learning_rate": 1.7496089931573805e-05, "loss": 0.3136, "step": 5125 }, { "epoch": 2.52, "learning_rate": 1.7483870967741937e-05, "loss": 0.316, "step": 5150 }, { "epoch": 2.53, "learning_rate": 1.747165200391007e-05, "loss": 0.3142, "step": 5175 }, { "epoch": 2.54, "learning_rate": 1.7459433040078202e-05, "loss": 0.3104, "step": 5200 }, { "epoch": 2.55, "learning_rate": 1.7447214076246334e-05, "loss": 0.3, "step": 5225 }, { "epoch": 2.57, "learning_rate": 1.7434995112414467e-05, "loss": 0.3027, "step": 5250 }, { "epoch": 2.58, "learning_rate": 1.74227761485826e-05, "loss": 0.3097, "step": 5275 }, { "epoch": 2.59, "learning_rate": 1.7410557184750735e-05, "loss": 0.314, "step": 5300 }, { "epoch": 2.6, "learning_rate": 1.7398338220918867e-05, "loss": 0.3252, "step": 5325 }, { "epoch": 2.61, "learning_rate": 1.7386119257087e-05, "loss": 0.3158, "step": 5350 }, { "epoch": 2.63, "learning_rate": 1.7373900293255132e-05, "loss": 0.3041, "step": 5375 }, { "epoch": 2.64, "learning_rate": 1.7361681329423265e-05, "loss": 0.3214, "step": 5400 }, { "epoch": 2.65, "learning_rate": 1.7349462365591397e-05, "loss": 0.3352, "step": 5425 }, { "epoch": 2.66, "learning_rate": 1.7337243401759533e-05, "loss": 0.3188, "step": 5450 }, { "epoch": 2.68, "learning_rate": 1.7325024437927665e-05, "loss": 0.3087, "step": 5475 }, { "epoch": 2.69, "learning_rate": 1.7312805474095798e-05, "loss": 0.3064, "step": 5500 }, { "epoch": 2.7, "learning_rate": 1.730058651026393e-05, "loss": 0.3153, "step": 5525 }, { "epoch": 2.71, "learning_rate": 1.7288367546432062e-05, "loss": 0.3084, "step": 5550 }, { "epoch": 2.72, "learning_rate": 1.7276148582600195e-05, "loss": 0.292, "step": 5575 }, { "epoch": 2.74, "learning_rate": 1.726392961876833e-05, "loss": 0.3167, "step": 5600 }, { "epoch": 2.75, "learning_rate": 1.7251710654936463e-05, "loss": 0.3141, "step": 5625 }, { "epoch": 2.76, "learning_rate": 1.7239491691104595e-05, "loss": 0.2913, "step": 5650 }, { "epoch": 2.77, "learning_rate": 1.7227272727272728e-05, "loss": 0.3279, "step": 5675 }, { "epoch": 2.79, "learning_rate": 1.721505376344086e-05, "loss": 0.3225, "step": 5700 }, { "epoch": 2.8, "learning_rate": 1.7202834799608993e-05, "loss": 0.306, "step": 5725 }, { "epoch": 2.81, "learning_rate": 1.719061583577713e-05, "loss": 0.3322, "step": 5750 }, { "epoch": 2.82, "learning_rate": 1.717839687194526e-05, "loss": 0.3102, "step": 5775 }, { "epoch": 2.83, "learning_rate": 1.7166177908113393e-05, "loss": 0.3123, "step": 5800 }, { "epoch": 2.85, "learning_rate": 1.7153958944281526e-05, "loss": 0.3077, "step": 5825 }, { "epoch": 2.86, "learning_rate": 1.7141739980449658e-05, "loss": 0.3223, "step": 5850 }, { "epoch": 2.87, "learning_rate": 1.712952101661779e-05, "loss": 0.3178, "step": 5875 }, { "epoch": 2.88, "learning_rate": 1.7117302052785926e-05, "loss": 0.3097, "step": 5900 }, { "epoch": 2.9, "learning_rate": 1.710508308895406e-05, "loss": 0.3047, "step": 5925 }, { "epoch": 2.91, "learning_rate": 1.709286412512219e-05, "loss": 0.3316, "step": 5950 }, { "epoch": 2.92, "learning_rate": 1.7080645161290323e-05, "loss": 0.3107, "step": 5975 }, { "epoch": 2.93, "learning_rate": 1.7068426197458456e-05, "loss": 0.3037, "step": 6000 }, { "epoch": 2.94, "learning_rate": 1.7056207233626588e-05, "loss": 0.3075, "step": 6025 }, { "epoch": 2.96, "learning_rate": 1.7043988269794724e-05, "loss": 0.3249, "step": 6050 }, { "epoch": 2.97, "learning_rate": 1.7031769305962856e-05, "loss": 0.3078, "step": 6075 }, { "epoch": 2.98, "learning_rate": 1.701955034213099e-05, "loss": 0.3015, "step": 6100 }, { "epoch": 2.99, "learning_rate": 1.700733137829912e-05, "loss": 0.3186, "step": 6125 }, { "epoch": 3.01, "learning_rate": 1.6995112414467254e-05, "loss": 0.2948, "step": 6150 }, { "epoch": 3.02, "learning_rate": 1.6982893450635386e-05, "loss": 0.2312, "step": 6175 }, { "epoch": 3.03, "learning_rate": 1.697067448680352e-05, "loss": 0.2379, "step": 6200 }, { "epoch": 3.04, "learning_rate": 1.6958455522971654e-05, "loss": 0.24, "step": 6225 }, { "epoch": 3.05, "learning_rate": 1.6946236559139786e-05, "loss": 0.2301, "step": 6250 }, { "epoch": 3.07, "learning_rate": 1.693401759530792e-05, "loss": 0.2257, "step": 6275 }, { "epoch": 3.08, "learning_rate": 1.692179863147605e-05, "loss": 0.2318, "step": 6300 }, { "epoch": 3.09, "learning_rate": 1.6909579667644184e-05, "loss": 0.2417, "step": 6325 }, { "epoch": 3.1, "learning_rate": 1.6897360703812316e-05, "loss": 0.2334, "step": 6350 }, { "epoch": 3.12, "learning_rate": 1.6885141739980452e-05, "loss": 0.2507, "step": 6375 }, { "epoch": 3.13, "learning_rate": 1.6872922776148584e-05, "loss": 0.2401, "step": 6400 }, { "epoch": 3.14, "learning_rate": 1.6860703812316717e-05, "loss": 0.2264, "step": 6425 }, { "epoch": 3.15, "learning_rate": 1.684848484848485e-05, "loss": 0.2395, "step": 6450 }, { "epoch": 3.16, "learning_rate": 1.683626588465298e-05, "loss": 0.2282, "step": 6475 }, { "epoch": 3.18, "learning_rate": 1.6824046920821114e-05, "loss": 0.237, "step": 6500 }, { "epoch": 3.19, "learning_rate": 1.681182795698925e-05, "loss": 0.2148, "step": 6525 }, { "epoch": 3.2, "learning_rate": 1.6799608993157382e-05, "loss": 0.2392, "step": 6550 }, { "epoch": 3.21, "learning_rate": 1.6787390029325514e-05, "loss": 0.2331, "step": 6575 }, { "epoch": 3.23, "learning_rate": 1.6775171065493647e-05, "loss": 0.2397, "step": 6600 }, { "epoch": 3.24, "learning_rate": 1.676295210166178e-05, "loss": 0.2398, "step": 6625 }, { "epoch": 3.25, "learning_rate": 1.675073313782991e-05, "loss": 0.2326, "step": 6650 }, { "epoch": 3.26, "learning_rate": 1.6738514173998047e-05, "loss": 0.2549, "step": 6675 }, { "epoch": 3.27, "learning_rate": 1.672629521016618e-05, "loss": 0.2261, "step": 6700 }, { "epoch": 3.29, "learning_rate": 1.6714076246334312e-05, "loss": 0.2366, "step": 6725 }, { "epoch": 3.3, "learning_rate": 1.6701857282502445e-05, "loss": 0.2504, "step": 6750 }, { "epoch": 3.31, "learning_rate": 1.6689638318670577e-05, "loss": 0.2381, "step": 6775 }, { "epoch": 3.32, "learning_rate": 1.667741935483871e-05, "loss": 0.2503, "step": 6800 }, { "epoch": 3.34, "learning_rate": 1.6665200391006845e-05, "loss": 0.2169, "step": 6825 }, { "epoch": 3.35, "learning_rate": 1.6652981427174978e-05, "loss": 0.2441, "step": 6850 }, { "epoch": 3.36, "learning_rate": 1.664076246334311e-05, "loss": 0.2348, "step": 6875 }, { "epoch": 3.37, "learning_rate": 1.6628543499511242e-05, "loss": 0.2479, "step": 6900 }, { "epoch": 3.38, "learning_rate": 1.6616324535679375e-05, "loss": 0.2479, "step": 6925 }, { "epoch": 3.4, "learning_rate": 1.6604105571847507e-05, "loss": 0.258, "step": 6950 }, { "epoch": 3.41, "learning_rate": 1.6591886608015643e-05, "loss": 0.2269, "step": 6975 }, { "epoch": 3.42, "learning_rate": 1.6579667644183775e-05, "loss": 0.2448, "step": 7000 }, { "epoch": 3.43, "learning_rate": 1.6567448680351908e-05, "loss": 0.2293, "step": 7025 }, { "epoch": 3.45, "learning_rate": 1.655522971652004e-05, "loss": 0.2407, "step": 7050 }, { "epoch": 3.46, "learning_rate": 1.6543010752688173e-05, "loss": 0.2318, "step": 7075 }, { "epoch": 3.47, "learning_rate": 1.6530791788856305e-05, "loss": 0.2363, "step": 7100 }, { "epoch": 3.48, "learning_rate": 1.651857282502444e-05, "loss": 0.2474, "step": 7125 }, { "epoch": 3.49, "learning_rate": 1.6506353861192573e-05, "loss": 0.2394, "step": 7150 }, { "epoch": 3.51, "learning_rate": 1.6494134897360706e-05, "loss": 0.2434, "step": 7175 }, { "epoch": 3.52, "learning_rate": 1.6481915933528838e-05, "loss": 0.2403, "step": 7200 }, { "epoch": 3.53, "learning_rate": 1.646969696969697e-05, "loss": 0.2456, "step": 7225 }, { "epoch": 3.54, "learning_rate": 1.6457478005865103e-05, "loss": 0.2441, "step": 7250 }, { "epoch": 3.56, "learning_rate": 1.644525904203324e-05, "loss": 0.2332, "step": 7275 }, { "epoch": 3.57, "learning_rate": 1.643304007820137e-05, "loss": 0.23, "step": 7300 }, { "epoch": 3.58, "learning_rate": 1.6420821114369503e-05, "loss": 0.2293, "step": 7325 }, { "epoch": 3.59, "learning_rate": 1.640909090909091e-05, "loss": 0.2224, "step": 7350 }, { "epoch": 3.6, "learning_rate": 1.6396871945259045e-05, "loss": 0.2324, "step": 7375 }, { "epoch": 3.62, "learning_rate": 1.6384652981427177e-05, "loss": 0.2409, "step": 7400 }, { "epoch": 3.63, "learning_rate": 1.637243401759531e-05, "loss": 0.2462, "step": 7425 }, { "epoch": 3.64, "learning_rate": 1.6360215053763442e-05, "loss": 0.2375, "step": 7450 }, { "epoch": 3.65, "learning_rate": 1.6347996089931574e-05, "loss": 0.2445, "step": 7475 }, { "epoch": 3.67, "learning_rate": 1.6335777126099707e-05, "loss": 0.2364, "step": 7500 }, { "epoch": 3.68, "learning_rate": 1.6323558162267842e-05, "loss": 0.2374, "step": 7525 }, { "epoch": 3.69, "learning_rate": 1.6311339198435975e-05, "loss": 0.2383, "step": 7550 }, { "epoch": 3.7, "learning_rate": 1.6299120234604107e-05, "loss": 0.2312, "step": 7575 }, { "epoch": 3.71, "learning_rate": 1.628690127077224e-05, "loss": 0.2475, "step": 7600 }, { "epoch": 3.73, "learning_rate": 1.6274682306940372e-05, "loss": 0.2552, "step": 7625 }, { "epoch": 3.74, "learning_rate": 1.6262463343108504e-05, "loss": 0.2507, "step": 7650 }, { "epoch": 3.75, "learning_rate": 1.625024437927664e-05, "loss": 0.248, "step": 7675 }, { "epoch": 3.76, "learning_rate": 1.6238025415444773e-05, "loss": 0.2333, "step": 7700 }, { "epoch": 3.78, "learning_rate": 1.6225806451612905e-05, "loss": 0.2424, "step": 7725 }, { "epoch": 3.79, "learning_rate": 1.6213587487781037e-05, "loss": 0.2565, "step": 7750 }, { "epoch": 3.8, "learning_rate": 1.620136852394917e-05, "loss": 0.2568, "step": 7775 }, { "epoch": 3.81, "learning_rate": 1.6189149560117302e-05, "loss": 0.2409, "step": 7800 }, { "epoch": 3.82, "learning_rate": 1.6176930596285435e-05, "loss": 0.2444, "step": 7825 }, { "epoch": 3.84, "learning_rate": 1.616471163245357e-05, "loss": 0.243, "step": 7850 }, { "epoch": 3.85, "learning_rate": 1.6152492668621703e-05, "loss": 0.2417, "step": 7875 }, { "epoch": 3.86, "learning_rate": 1.6140273704789835e-05, "loss": 0.2246, "step": 7900 }, { "epoch": 3.87, "learning_rate": 1.6128054740957968e-05, "loss": 0.2533, "step": 7925 }, { "epoch": 3.89, "learning_rate": 1.61158357771261e-05, "loss": 0.2508, "step": 7950 }, { "epoch": 3.9, "learning_rate": 1.6103616813294232e-05, "loss": 0.2384, "step": 7975 }, { "epoch": 3.91, "learning_rate": 1.6091397849462368e-05, "loss": 0.2531, "step": 8000 }, { "epoch": 3.92, "learning_rate": 1.60791788856305e-05, "loss": 0.2448, "step": 8025 }, { "epoch": 3.93, "learning_rate": 1.6066959921798633e-05, "loss": 0.241, "step": 8050 }, { "epoch": 3.95, "learning_rate": 1.6054740957966765e-05, "loss": 0.2566, "step": 8075 }, { "epoch": 3.96, "learning_rate": 1.6042521994134898e-05, "loss": 0.2297, "step": 8100 }, { "epoch": 3.97, "learning_rate": 1.603030303030303e-05, "loss": 0.2446, "step": 8125 }, { "epoch": 3.98, "learning_rate": 1.6018084066471166e-05, "loss": 0.2305, "step": 8150 }, { "epoch": 4.0, "learning_rate": 1.6005865102639298e-05, "loss": 0.2438, "step": 8175 }, { "epoch": 4.01, "learning_rate": 1.599364613880743e-05, "loss": 0.2059, "step": 8200 }, { "epoch": 4.02, "learning_rate": 1.5981427174975563e-05, "loss": 0.1706, "step": 8225 }, { "epoch": 4.03, "learning_rate": 1.5969208211143695e-05, "loss": 0.1822, "step": 8250 }, { "epoch": 4.04, "learning_rate": 1.5956989247311828e-05, "loss": 0.1947, "step": 8275 }, { "epoch": 4.06, "learning_rate": 1.5944770283479964e-05, "loss": 0.1714, "step": 8300 }, { "epoch": 4.07, "learning_rate": 1.5932551319648096e-05, "loss": 0.1776, "step": 8325 }, { "epoch": 4.08, "learning_rate": 1.592033235581623e-05, "loss": 0.1974, "step": 8350 }, { "epoch": 4.09, "learning_rate": 1.590811339198436e-05, "loss": 0.1898, "step": 8375 }, { "epoch": 4.11, "learning_rate": 1.5895894428152493e-05, "loss": 0.1747, "step": 8400 }, { "epoch": 4.12, "learning_rate": 1.5883675464320626e-05, "loss": 0.1692, "step": 8425 }, { "epoch": 4.13, "learning_rate": 1.587145650048876e-05, "loss": 0.1773, "step": 8450 }, { "epoch": 4.14, "learning_rate": 1.5859237536656894e-05, "loss": 0.1605, "step": 8475 }, { "epoch": 4.15, "learning_rate": 1.5847018572825026e-05, "loss": 0.1773, "step": 8500 }, { "epoch": 4.17, "learning_rate": 1.583479960899316e-05, "loss": 0.1778, "step": 8525 }, { "epoch": 4.18, "learning_rate": 1.582258064516129e-05, "loss": 0.1865, "step": 8550 }, { "epoch": 4.19, "learning_rate": 1.5810361681329423e-05, "loss": 0.1844, "step": 8575 }, { "epoch": 4.2, "learning_rate": 1.579814271749756e-05, "loss": 0.1694, "step": 8600 }, { "epoch": 4.22, "learning_rate": 1.578592375366569e-05, "loss": 0.1813, "step": 8625 }, { "epoch": 4.23, "learning_rate": 1.5773704789833824e-05, "loss": 0.1899, "step": 8650 }, { "epoch": 4.24, "learning_rate": 1.5761485826001956e-05, "loss": 0.1691, "step": 8675 }, { "epoch": 4.25, "learning_rate": 1.574926686217009e-05, "loss": 0.1777, "step": 8700 }, { "epoch": 4.26, "learning_rate": 1.573704789833822e-05, "loss": 0.1877, "step": 8725 }, { "epoch": 4.28, "learning_rate": 1.5724828934506357e-05, "loss": 0.1846, "step": 8750 }, { "epoch": 4.29, "learning_rate": 1.571260997067449e-05, "loss": 0.1963, "step": 8775 }, { "epoch": 4.3, "learning_rate": 1.5700391006842622e-05, "loss": 0.1771, "step": 8800 }, { "epoch": 4.31, "learning_rate": 1.5688172043010754e-05, "loss": 0.1865, "step": 8825 }, { "epoch": 4.33, "learning_rate": 1.5675953079178887e-05, "loss": 0.1763, "step": 8850 }, { "epoch": 4.34, "learning_rate": 1.566373411534702e-05, "loss": 0.1768, "step": 8875 }, { "epoch": 4.35, "learning_rate": 1.565151515151515e-05, "loss": 0.1645, "step": 8900 }, { "epoch": 4.36, "learning_rate": 1.5639296187683287e-05, "loss": 0.1816, "step": 8925 }, { "epoch": 4.37, "learning_rate": 1.562707722385142e-05, "loss": 0.1793, "step": 8950 }, { "epoch": 4.39, "learning_rate": 1.5614858260019552e-05, "loss": 0.1888, "step": 8975 }, { "epoch": 4.4, "learning_rate": 1.5602639296187684e-05, "loss": 0.1769, "step": 9000 }, { "epoch": 4.41, "learning_rate": 1.5590420332355817e-05, "loss": 0.1791, "step": 9025 }, { "epoch": 4.42, "learning_rate": 1.557820136852395e-05, "loss": 0.1823, "step": 9050 }, { "epoch": 4.44, "learning_rate": 1.5565982404692085e-05, "loss": 0.1835, "step": 9075 }, { "epoch": 4.45, "learning_rate": 1.5553763440860217e-05, "loss": 0.1875, "step": 9100 }, { "epoch": 4.46, "learning_rate": 1.554154447702835e-05, "loss": 0.1761, "step": 9125 }, { "epoch": 4.47, "learning_rate": 1.5529325513196482e-05, "loss": 0.1838, "step": 9150 }, { "epoch": 4.48, "learning_rate": 1.5517106549364615e-05, "loss": 0.1966, "step": 9175 }, { "epoch": 4.5, "learning_rate": 1.5504887585532747e-05, "loss": 0.1874, "step": 9200 }, { "epoch": 4.51, "learning_rate": 1.5492668621700883e-05, "loss": 0.1841, "step": 9225 }, { "epoch": 4.52, "learning_rate": 1.5480449657869015e-05, "loss": 0.2053, "step": 9250 }, { "epoch": 4.53, "learning_rate": 1.5468230694037148e-05, "loss": 0.1925, "step": 9275 }, { "epoch": 4.55, "learning_rate": 1.545601173020528e-05, "loss": 0.1784, "step": 9300 }, { "epoch": 4.56, "learning_rate": 1.5443792766373412e-05, "loss": 0.1914, "step": 9325 }, { "epoch": 4.57, "learning_rate": 1.5431573802541545e-05, "loss": 0.1875, "step": 9350 }, { "epoch": 4.58, "learning_rate": 1.541935483870968e-05, "loss": 0.183, "step": 9375 }, { "epoch": 4.59, "learning_rate": 1.5407135874877813e-05, "loss": 0.1964, "step": 9400 }, { "epoch": 4.61, "learning_rate": 1.539540566959922e-05, "loss": 0.1753, "step": 9425 }, { "epoch": 4.62, "learning_rate": 1.538318670576735e-05, "loss": 0.1911, "step": 9450 }, { "epoch": 4.63, "learning_rate": 1.5370967741935487e-05, "loss": 0.1957, "step": 9475 }, { "epoch": 4.64, "learning_rate": 1.535874877810362e-05, "loss": 0.1861, "step": 9500 }, { "epoch": 4.66, "learning_rate": 1.534652981427175e-05, "loss": 0.191, "step": 9525 }, { "epoch": 4.67, "learning_rate": 1.5334310850439884e-05, "loss": 0.1844, "step": 9550 }, { "epoch": 4.68, "learning_rate": 1.5322091886608016e-05, "loss": 0.2053, "step": 9575 }, { "epoch": 4.69, "learning_rate": 1.530987292277615e-05, "loss": 0.1857, "step": 9600 }, { "epoch": 4.7, "learning_rate": 1.5297653958944284e-05, "loss": 0.1783, "step": 9625 }, { "epoch": 4.72, "learning_rate": 1.5285434995112417e-05, "loss": 0.1917, "step": 9650 }, { "epoch": 4.73, "learning_rate": 1.527321603128055e-05, "loss": 0.1816, "step": 9675 }, { "epoch": 4.74, "learning_rate": 1.526099706744868e-05, "loss": 0.1925, "step": 9700 }, { "epoch": 4.75, "learning_rate": 1.5248778103616814e-05, "loss": 0.2035, "step": 9725 }, { "epoch": 4.77, "learning_rate": 1.5236559139784948e-05, "loss": 0.1839, "step": 9750 }, { "epoch": 4.78, "learning_rate": 1.522434017595308e-05, "loss": 0.1897, "step": 9775 }, { "epoch": 4.79, "learning_rate": 1.5212121212121213e-05, "loss": 0.188, "step": 9800 }, { "epoch": 4.8, "learning_rate": 1.5199902248289347e-05, "loss": 0.1791, "step": 9825 }, { "epoch": 4.81, "learning_rate": 1.518768328445748e-05, "loss": 0.1894, "step": 9850 }, { "epoch": 4.83, "learning_rate": 1.5175464320625612e-05, "loss": 0.1975, "step": 9875 }, { "epoch": 4.84, "learning_rate": 1.5163245356793746e-05, "loss": 0.181, "step": 9900 }, { "epoch": 4.85, "learning_rate": 1.5151026392961878e-05, "loss": 0.2079, "step": 9925 }, { "epoch": 4.86, "learning_rate": 1.513880742913001e-05, "loss": 0.1866, "step": 9950 }, { "epoch": 4.88, "learning_rate": 1.5126588465298145e-05, "loss": 0.1868, "step": 9975 }, { "epoch": 4.89, "learning_rate": 1.5114369501466277e-05, "loss": 0.1956, "step": 10000 }, { "epoch": 4.9, "learning_rate": 1.510215053763441e-05, "loss": 0.1871, "step": 10025 }, { "epoch": 4.91, "learning_rate": 1.5089931573802544e-05, "loss": 0.1954, "step": 10050 }, { "epoch": 4.92, "learning_rate": 1.5077712609970676e-05, "loss": 0.1872, "step": 10075 }, { "epoch": 4.94, "learning_rate": 1.5065493646138808e-05, "loss": 0.2023, "step": 10100 }, { "epoch": 4.95, "learning_rate": 1.5053274682306943e-05, "loss": 0.2043, "step": 10125 }, { "epoch": 4.96, "learning_rate": 1.5041055718475075e-05, "loss": 0.1863, "step": 10150 }, { "epoch": 4.97, "learning_rate": 1.5028836754643207e-05, "loss": 0.1987, "step": 10175 }, { "epoch": 4.99, "learning_rate": 1.5016617790811341e-05, "loss": 0.2, "step": 10200 }, { "epoch": 5.0, "learning_rate": 1.5004398826979474e-05, "loss": 0.1963, "step": 10225 }, { "epoch": 5.01, "learning_rate": 1.4992179863147606e-05, "loss": 0.1458, "step": 10250 }, { "epoch": 5.02, "learning_rate": 1.497996089931574e-05, "loss": 0.1321, "step": 10275 }, { "epoch": 5.03, "learning_rate": 1.4967741935483873e-05, "loss": 0.1246, "step": 10300 }, { "epoch": 5.05, "learning_rate": 1.4955522971652005e-05, "loss": 0.1298, "step": 10325 }, { "epoch": 5.06, "learning_rate": 1.494330400782014e-05, "loss": 0.1401, "step": 10350 }, { "epoch": 5.07, "learning_rate": 1.4931085043988272e-05, "loss": 0.1299, "step": 10375 }, { "epoch": 5.08, "learning_rate": 1.4918866080156404e-05, "loss": 0.128, "step": 10400 }, { "epoch": 5.1, "learning_rate": 1.4906647116324538e-05, "loss": 0.134, "step": 10425 }, { "epoch": 5.11, "learning_rate": 1.489442815249267e-05, "loss": 0.137, "step": 10450 }, { "epoch": 5.12, "learning_rate": 1.4882209188660803e-05, "loss": 0.1393, "step": 10475 }, { "epoch": 5.13, "learning_rate": 1.4869990224828937e-05, "loss": 0.1303, "step": 10500 }, { "epoch": 5.14, "learning_rate": 1.485777126099707e-05, "loss": 0.1412, "step": 10525 }, { "epoch": 5.16, "learning_rate": 1.4845552297165202e-05, "loss": 0.1454, "step": 10550 }, { "epoch": 5.17, "learning_rate": 1.4833333333333336e-05, "loss": 0.1464, "step": 10575 }, { "epoch": 5.18, "learning_rate": 1.4821114369501468e-05, "loss": 0.138, "step": 10600 }, { "epoch": 5.19, "learning_rate": 1.48088954056696e-05, "loss": 0.1256, "step": 10625 }, { "epoch": 5.21, "learning_rate": 1.4796676441837735e-05, "loss": 0.1352, "step": 10650 }, { "epoch": 5.22, "learning_rate": 1.4784457478005867e-05, "loss": 0.138, "step": 10675 }, { "epoch": 5.23, "learning_rate": 1.4772238514174e-05, "loss": 0.1388, "step": 10700 }, { "epoch": 5.24, "learning_rate": 1.4760019550342134e-05, "loss": 0.1364, "step": 10725 }, { "epoch": 5.25, "learning_rate": 1.4747800586510266e-05, "loss": 0.1414, "step": 10750 }, { "epoch": 5.27, "learning_rate": 1.4735581622678398e-05, "loss": 0.1409, "step": 10775 }, { "epoch": 5.28, "learning_rate": 1.4723362658846531e-05, "loss": 0.1475, "step": 10800 }, { "epoch": 5.29, "learning_rate": 1.4711143695014665e-05, "loss": 0.1465, "step": 10825 }, { "epoch": 5.3, "learning_rate": 1.4698924731182797e-05, "loss": 0.1565, "step": 10850 }, { "epoch": 5.32, "learning_rate": 1.468670576735093e-05, "loss": 0.1534, "step": 10875 }, { "epoch": 5.33, "learning_rate": 1.4674486803519064e-05, "loss": 0.1549, "step": 10900 }, { "epoch": 5.34, "learning_rate": 1.4662267839687196e-05, "loss": 0.1404, "step": 10925 }, { "epoch": 5.35, "learning_rate": 1.4650048875855329e-05, "loss": 0.1352, "step": 10950 }, { "epoch": 5.36, "learning_rate": 1.4637829912023463e-05, "loss": 0.1455, "step": 10975 }, { "epoch": 5.38, "learning_rate": 1.4625610948191595e-05, "loss": 0.1359, "step": 11000 }, { "epoch": 5.39, "learning_rate": 1.4613391984359728e-05, "loss": 0.1395, "step": 11025 }, { "epoch": 5.4, "learning_rate": 1.4601173020527862e-05, "loss": 0.1544, "step": 11050 }, { "epoch": 5.41, "learning_rate": 1.4588954056695994e-05, "loss": 0.1603, "step": 11075 }, { "epoch": 5.43, "learning_rate": 1.4576735092864126e-05, "loss": 0.141, "step": 11100 }, { "epoch": 5.44, "learning_rate": 1.456451612903226e-05, "loss": 0.1459, "step": 11125 }, { "epoch": 5.45, "learning_rate": 1.4552297165200393e-05, "loss": 0.1327, "step": 11150 }, { "epoch": 5.46, "learning_rate": 1.4540078201368525e-05, "loss": 0.1393, "step": 11175 }, { "epoch": 5.47, "learning_rate": 1.452785923753666e-05, "loss": 0.1569, "step": 11200 }, { "epoch": 5.49, "learning_rate": 1.4515640273704792e-05, "loss": 0.1387, "step": 11225 }, { "epoch": 5.5, "learning_rate": 1.4503421309872924e-05, "loss": 0.1332, "step": 11250 }, { "epoch": 5.51, "learning_rate": 1.4491202346041058e-05, "loss": 0.1439, "step": 11275 }, { "epoch": 5.52, "learning_rate": 1.447898338220919e-05, "loss": 0.1252, "step": 11300 }, { "epoch": 5.54, "learning_rate": 1.4466764418377323e-05, "loss": 0.1516, "step": 11325 }, { "epoch": 5.55, "learning_rate": 1.4454545454545457e-05, "loss": 0.1378, "step": 11350 }, { "epoch": 5.56, "learning_rate": 1.4442815249266864e-05, "loss": 0.1571, "step": 11375 }, { "epoch": 5.57, "learning_rate": 1.4430596285434997e-05, "loss": 0.1519, "step": 11400 }, { "epoch": 5.58, "learning_rate": 1.4418377321603129e-05, "loss": 0.1498, "step": 11425 }, { "epoch": 5.6, "learning_rate": 1.4406158357771263e-05, "loss": 0.1408, "step": 11450 }, { "epoch": 5.61, "learning_rate": 1.4393939393939396e-05, "loss": 0.1336, "step": 11475 }, { "epoch": 5.62, "learning_rate": 1.4381720430107528e-05, "loss": 0.1423, "step": 11500 }, { "epoch": 5.63, "learning_rate": 1.4369501466275662e-05, "loss": 0.1498, "step": 11525 }, { "epoch": 5.65, "learning_rate": 1.4357282502443794e-05, "loss": 0.1476, "step": 11550 }, { "epoch": 5.66, "learning_rate": 1.4345063538611927e-05, "loss": 0.1363, "step": 11575 }, { "epoch": 5.67, "learning_rate": 1.4332844574780061e-05, "loss": 0.141, "step": 11600 }, { "epoch": 5.68, "learning_rate": 1.4320625610948193e-05, "loss": 0.1329, "step": 11625 }, { "epoch": 5.69, "learning_rate": 1.4308406647116326e-05, "loss": 0.153, "step": 11650 }, { "epoch": 5.71, "learning_rate": 1.429618768328446e-05, "loss": 0.1587, "step": 11675 }, { "epoch": 5.72, "learning_rate": 1.4283968719452592e-05, "loss": 0.1518, "step": 11700 }, { "epoch": 5.73, "learning_rate": 1.4271749755620725e-05, "loss": 0.1449, "step": 11725 }, { "epoch": 5.74, "learning_rate": 1.4259530791788859e-05, "loss": 0.154, "step": 11750 }, { "epoch": 5.76, "learning_rate": 1.4247311827956991e-05, "loss": 0.1519, "step": 11775 }, { "epoch": 5.77, "learning_rate": 1.4235092864125124e-05, "loss": 0.1417, "step": 11800 }, { "epoch": 5.78, "learning_rate": 1.4222873900293258e-05, "loss": 0.1479, "step": 11825 }, { "epoch": 5.79, "learning_rate": 1.421065493646139e-05, "loss": 0.1499, "step": 11850 }, { "epoch": 5.8, "learning_rate": 1.4198435972629522e-05, "loss": 0.1362, "step": 11875 }, { "epoch": 5.82, "learning_rate": 1.4186217008797657e-05, "loss": 0.1522, "step": 11900 }, { "epoch": 5.83, "learning_rate": 1.4173998044965789e-05, "loss": 0.149, "step": 11925 }, { "epoch": 5.84, "learning_rate": 1.4161779081133921e-05, "loss": 0.1448, "step": 11950 }, { "epoch": 5.85, "learning_rate": 1.4149560117302055e-05, "loss": 0.1602, "step": 11975 }, { "epoch": 5.87, "learning_rate": 1.4137341153470188e-05, "loss": 0.1481, "step": 12000 }, { "epoch": 5.88, "learning_rate": 1.412512218963832e-05, "loss": 0.1631, "step": 12025 }, { "epoch": 5.89, "learning_rate": 1.4112903225806454e-05, "loss": 0.1516, "step": 12050 }, { "epoch": 5.9, "learning_rate": 1.4100684261974587e-05, "loss": 0.1449, "step": 12075 }, { "epoch": 5.91, "learning_rate": 1.4088465298142719e-05, "loss": 0.1549, "step": 12100 }, { "epoch": 5.93, "learning_rate": 1.4076246334310853e-05, "loss": 0.1456, "step": 12125 }, { "epoch": 5.94, "learning_rate": 1.4064027370478986e-05, "loss": 0.1443, "step": 12150 }, { "epoch": 5.95, "learning_rate": 1.4051808406647118e-05, "loss": 0.1445, "step": 12175 }, { "epoch": 5.96, "learning_rate": 1.4039589442815252e-05, "loss": 0.15, "step": 12200 }, { "epoch": 5.98, "learning_rate": 1.4027370478983385e-05, "loss": 0.1459, "step": 12225 }, { "epoch": 5.99, "learning_rate": 1.4015151515151517e-05, "loss": 0.1594, "step": 12250 }, { "epoch": 6.0, "learning_rate": 1.400293255131965e-05, "loss": 0.1511, "step": 12275 }, { "epoch": 6.01, "learning_rate": 1.3990713587487783e-05, "loss": 0.1114, "step": 12300 }, { "epoch": 6.02, "learning_rate": 1.3978494623655916e-05, "loss": 0.1002, "step": 12325 }, { "epoch": 6.04, "learning_rate": 1.3966275659824048e-05, "loss": 0.117, "step": 12350 }, { "epoch": 6.05, "learning_rate": 1.3954056695992182e-05, "loss": 0.1082, "step": 12375 }, { "epoch": 6.06, "learning_rate": 1.3941837732160315e-05, "loss": 0.1215, "step": 12400 }, { "epoch": 6.07, "learning_rate": 1.3929618768328447e-05, "loss": 0.1046, "step": 12425 }, { "epoch": 6.09, "learning_rate": 1.3917399804496581e-05, "loss": 0.1063, "step": 12450 }, { "epoch": 6.1, "learning_rate": 1.3905180840664714e-05, "loss": 0.1178, "step": 12475 }, { "epoch": 6.11, "learning_rate": 1.3892961876832846e-05, "loss": 0.1093, "step": 12500 }, { "epoch": 6.12, "learning_rate": 1.388074291300098e-05, "loss": 0.1085, "step": 12525 }, { "epoch": 6.13, "learning_rate": 1.3868523949169112e-05, "loss": 0.1171, "step": 12550 }, { "epoch": 6.15, "learning_rate": 1.3856304985337245e-05, "loss": 0.1145, "step": 12575 }, { "epoch": 6.16, "learning_rate": 1.3844086021505379e-05, "loss": 0.1042, "step": 12600 }, { "epoch": 6.17, "learning_rate": 1.3831867057673511e-05, "loss": 0.1059, "step": 12625 }, { "epoch": 6.18, "learning_rate": 1.3819648093841644e-05, "loss": 0.1144, "step": 12650 }, { "epoch": 6.2, "learning_rate": 1.3807429130009778e-05, "loss": 0.1023, "step": 12675 }, { "epoch": 6.21, "learning_rate": 1.379521016617791e-05, "loss": 0.1188, "step": 12700 }, { "epoch": 6.22, "learning_rate": 1.3782991202346043e-05, "loss": 0.1028, "step": 12725 }, { "epoch": 6.23, "learning_rate": 1.3770772238514177e-05, "loss": 0.1174, "step": 12750 }, { "epoch": 6.24, "learning_rate": 1.375855327468231e-05, "loss": 0.1113, "step": 12775 }, { "epoch": 6.26, "learning_rate": 1.374633431085044e-05, "loss": 0.1183, "step": 12800 }, { "epoch": 6.27, "learning_rate": 1.3734115347018572e-05, "loss": 0.1145, "step": 12825 }, { "epoch": 6.28, "learning_rate": 1.3721896383186706e-05, "loss": 0.1075, "step": 12850 }, { "epoch": 6.29, "learning_rate": 1.3709677419354839e-05, "loss": 0.1083, "step": 12875 }, { "epoch": 6.3, "learning_rate": 1.3697458455522971e-05, "loss": 0.1123, "step": 12900 }, { "epoch": 6.32, "learning_rate": 1.3685239491691105e-05, "loss": 0.1097, "step": 12925 }, { "epoch": 6.33, "learning_rate": 1.3673020527859238e-05, "loss": 0.1262, "step": 12950 }, { "epoch": 6.34, "learning_rate": 1.366080156402737e-05, "loss": 0.1156, "step": 12975 }, { "epoch": 6.35, "learning_rate": 1.3648582600195504e-05, "loss": 0.1229, "step": 13000 }, { "epoch": 6.37, "learning_rate": 1.3636363636363637e-05, "loss": 0.1173, "step": 13025 }, { "epoch": 6.38, "learning_rate": 1.3624144672531769e-05, "loss": 0.113, "step": 13050 }, { "epoch": 6.39, "learning_rate": 1.3611925708699903e-05, "loss": 0.1119, "step": 13075 }, { "epoch": 6.4, "learning_rate": 1.3599706744868035e-05, "loss": 0.1205, "step": 13100 }, { "epoch": 6.41, "learning_rate": 1.3587487781036168e-05, "loss": 0.1074, "step": 13125 }, { "epoch": 6.43, "learning_rate": 1.3575268817204302e-05, "loss": 0.1134, "step": 13150 }, { "epoch": 6.44, "learning_rate": 1.3563049853372434e-05, "loss": 0.1146, "step": 13175 }, { "epoch": 6.45, "learning_rate": 1.3550830889540567e-05, "loss": 0.1055, "step": 13200 }, { "epoch": 6.46, "learning_rate": 1.35386119257087e-05, "loss": 0.1067, "step": 13225 }, { "epoch": 6.48, "learning_rate": 1.3526392961876833e-05, "loss": 0.1141, "step": 13250 }, { "epoch": 6.49, "learning_rate": 1.3514173998044966e-05, "loss": 0.1143, "step": 13275 }, { "epoch": 6.5, "learning_rate": 1.35019550342131e-05, "loss": 0.115, "step": 13300 }, { "epoch": 6.51, "learning_rate": 1.3489736070381232e-05, "loss": 0.1158, "step": 13325 }, { "epoch": 6.52, "learning_rate": 1.3477517106549365e-05, "loss": 0.1119, "step": 13350 }, { "epoch": 6.54, "learning_rate": 1.3465298142717499e-05, "loss": 0.1225, "step": 13375 }, { "epoch": 6.55, "learning_rate": 1.3453079178885631e-05, "loss": 0.1171, "step": 13400 }, { "epoch": 6.56, "learning_rate": 1.3440860215053763e-05, "loss": 0.1235, "step": 13425 }, { "epoch": 6.57, "learning_rate": 1.3428641251221896e-05, "loss": 0.1156, "step": 13450 }, { "epoch": 6.59, "learning_rate": 1.341642228739003e-05, "loss": 0.1129, "step": 13475 }, { "epoch": 6.6, "learning_rate": 1.3404203323558162e-05, "loss": 0.1219, "step": 13500 }, { "epoch": 6.61, "learning_rate": 1.3391984359726295e-05, "loss": 0.1131, "step": 13525 }, { "epoch": 6.62, "learning_rate": 1.3379765395894429e-05, "loss": 0.1163, "step": 13550 }, { "epoch": 6.63, "learning_rate": 1.3367546432062561e-05, "loss": 0.1073, "step": 13575 }, { "epoch": 6.65, "learning_rate": 1.3355327468230694e-05, "loss": 0.1082, "step": 13600 }, { "epoch": 6.66, "learning_rate": 1.3343108504398828e-05, "loss": 0.1161, "step": 13625 }, { "epoch": 6.67, "learning_rate": 1.333088954056696e-05, "loss": 0.1021, "step": 13650 }, { "epoch": 6.68, "learning_rate": 1.3318670576735093e-05, "loss": 0.117, "step": 13675 }, { "epoch": 6.7, "learning_rate": 1.3306451612903227e-05, "loss": 0.1294, "step": 13700 }, { "epoch": 6.71, "learning_rate": 1.3294232649071359e-05, "loss": 0.1148, "step": 13725 }, { "epoch": 6.72, "learning_rate": 1.3282013685239491e-05, "loss": 0.1037, "step": 13750 }, { "epoch": 6.73, "learning_rate": 1.3269794721407626e-05, "loss": 0.1125, "step": 13775 }, { "epoch": 6.74, "learning_rate": 1.3257575757575758e-05, "loss": 0.119, "step": 13800 }, { "epoch": 6.76, "learning_rate": 1.324535679374389e-05, "loss": 0.1161, "step": 13825 }, { "epoch": 6.77, "learning_rate": 1.3233137829912024e-05, "loss": 0.112, "step": 13850 }, { "epoch": 6.78, "learning_rate": 1.3220918866080157e-05, "loss": 0.1124, "step": 13875 }, { "epoch": 6.79, "learning_rate": 1.320869990224829e-05, "loss": 0.1352, "step": 13900 }, { "epoch": 6.81, "learning_rate": 1.3196480938416423e-05, "loss": 0.1252, "step": 13925 }, { "epoch": 6.82, "learning_rate": 1.3184261974584556e-05, "loss": 0.1054, "step": 13950 }, { "epoch": 6.83, "learning_rate": 1.3172043010752688e-05, "loss": 0.1153, "step": 13975 }, { "epoch": 6.84, "learning_rate": 1.3159824046920822e-05, "loss": 0.1323, "step": 14000 }, { "epoch": 6.85, "learning_rate": 1.3147605083088955e-05, "loss": 0.1227, "step": 14025 }, { "epoch": 6.87, "learning_rate": 1.3135386119257087e-05, "loss": 0.1197, "step": 14050 }, { "epoch": 6.88, "learning_rate": 1.3123167155425221e-05, "loss": 0.1202, "step": 14075 }, { "epoch": 6.89, "learning_rate": 1.3110948191593353e-05, "loss": 0.1119, "step": 14100 }, { "epoch": 6.9, "learning_rate": 1.3098729227761486e-05, "loss": 0.115, "step": 14125 }, { "epoch": 6.92, "learning_rate": 1.308651026392962e-05, "loss": 0.1209, "step": 14150 }, { "epoch": 6.93, "learning_rate": 1.3074291300097752e-05, "loss": 0.1201, "step": 14175 }, { "epoch": 6.94, "learning_rate": 1.3062072336265885e-05, "loss": 0.1227, "step": 14200 }, { "epoch": 6.95, "learning_rate": 1.3049853372434019e-05, "loss": 0.1132, "step": 14225 }, { "epoch": 6.96, "learning_rate": 1.3037634408602151e-05, "loss": 0.1326, "step": 14250 }, { "epoch": 6.98, "learning_rate": 1.3025415444770284e-05, "loss": 0.1098, "step": 14275 }, { "epoch": 6.99, "learning_rate": 1.3013196480938418e-05, "loss": 0.1168, "step": 14300 }, { "epoch": 7.0, "learning_rate": 1.300097751710655e-05, "loss": 0.132, "step": 14325 }, { "epoch": 7.01, "learning_rate": 1.2988758553274683e-05, "loss": 0.0912, "step": 14350 }, { "epoch": 7.03, "learning_rate": 1.2976539589442817e-05, "loss": 0.091, "step": 14375 }, { "epoch": 7.04, "learning_rate": 1.2964320625610949e-05, "loss": 0.0789, "step": 14400 }, { "epoch": 7.05, "learning_rate": 1.2952101661779081e-05, "loss": 0.0815, "step": 14425 }, { "epoch": 7.06, "learning_rate": 1.2939882697947216e-05, "loss": 0.0935, "step": 14450 }, { "epoch": 7.07, "learning_rate": 1.2927663734115348e-05, "loss": 0.0863, "step": 14475 }, { "epoch": 7.09, "learning_rate": 1.291544477028348e-05, "loss": 0.0877, "step": 14500 }, { "epoch": 7.1, "learning_rate": 1.2903225806451613e-05, "loss": 0.0832, "step": 14525 }, { "epoch": 7.11, "learning_rate": 1.2891006842619747e-05, "loss": 0.0888, "step": 14550 }, { "epoch": 7.12, "learning_rate": 1.287878787878788e-05, "loss": 0.0901, "step": 14575 }, { "epoch": 7.14, "learning_rate": 1.2866568914956012e-05, "loss": 0.0891, "step": 14600 }, { "epoch": 7.15, "learning_rate": 1.2854349951124146e-05, "loss": 0.0887, "step": 14625 }, { "epoch": 7.16, "learning_rate": 1.2842130987292278e-05, "loss": 0.0916, "step": 14650 }, { "epoch": 7.17, "learning_rate": 1.282991202346041e-05, "loss": 0.0853, "step": 14675 }, { "epoch": 7.18, "learning_rate": 1.2817693059628545e-05, "loss": 0.0858, "step": 14700 }, { "epoch": 7.2, "learning_rate": 1.2805474095796677e-05, "loss": 0.0876, "step": 14725 }, { "epoch": 7.21, "learning_rate": 1.279325513196481e-05, "loss": 0.1036, "step": 14750 }, { "epoch": 7.22, "learning_rate": 1.2781036168132944e-05, "loss": 0.0959, "step": 14775 }, { "epoch": 7.23, "learning_rate": 1.2768817204301076e-05, "loss": 0.0789, "step": 14800 }, { "epoch": 7.25, "learning_rate": 1.2756598240469208e-05, "loss": 0.0843, "step": 14825 }, { "epoch": 7.26, "learning_rate": 1.2744379276637342e-05, "loss": 0.0895, "step": 14850 }, { "epoch": 7.27, "learning_rate": 1.2732160312805475e-05, "loss": 0.0985, "step": 14875 }, { "epoch": 7.28, "learning_rate": 1.2719941348973607e-05, "loss": 0.0793, "step": 14900 }, { "epoch": 7.29, "learning_rate": 1.2707722385141741e-05, "loss": 0.0808, "step": 14925 }, { "epoch": 7.31, "learning_rate": 1.2695503421309874e-05, "loss": 0.0843, "step": 14950 }, { "epoch": 7.32, "learning_rate": 1.2683284457478006e-05, "loss": 0.0979, "step": 14975 }, { "epoch": 7.33, "learning_rate": 1.267106549364614e-05, "loss": 0.1021, "step": 15000 }, { "epoch": 7.34, "learning_rate": 1.2658846529814273e-05, "loss": 0.0892, "step": 15025 }, { "epoch": 7.36, "learning_rate": 1.2646627565982405e-05, "loss": 0.0842, "step": 15050 }, { "epoch": 7.37, "learning_rate": 1.2634408602150539e-05, "loss": 0.0836, "step": 15075 }, { "epoch": 7.38, "learning_rate": 1.2622189638318672e-05, "loss": 0.0844, "step": 15100 }, { "epoch": 7.39, "learning_rate": 1.2609970674486804e-05, "loss": 0.1078, "step": 15125 }, { "epoch": 7.4, "learning_rate": 1.2597751710654938e-05, "loss": 0.0929, "step": 15150 }, { "epoch": 7.42, "learning_rate": 1.258553274682307e-05, "loss": 0.089, "step": 15175 }, { "epoch": 7.43, "learning_rate": 1.2573313782991203e-05, "loss": 0.1009, "step": 15200 }, { "epoch": 7.44, "learning_rate": 1.2561094819159337e-05, "loss": 0.0925, "step": 15225 }, { "epoch": 7.45, "learning_rate": 1.254887585532747e-05, "loss": 0.0892, "step": 15250 }, { "epoch": 7.47, "learning_rate": 1.2536656891495602e-05, "loss": 0.0887, "step": 15275 }, { "epoch": 7.48, "learning_rate": 1.2524437927663736e-05, "loss": 0.0964, "step": 15300 }, { "epoch": 7.49, "learning_rate": 1.2512218963831868e-05, "loss": 0.0894, "step": 15325 }, { "epoch": 7.5, "learning_rate": 1.25e-05, "loss": 0.0946, "step": 15350 }, { "epoch": 7.51, "learning_rate": 1.2487781036168135e-05, "loss": 0.0921, "step": 15375 }, { "epoch": 7.53, "learning_rate": 1.2475562072336267e-05, "loss": 0.0926, "step": 15400 }, { "epoch": 7.54, "learning_rate": 1.24633431085044e-05, "loss": 0.0959, "step": 15425 }, { "epoch": 7.55, "learning_rate": 1.2451124144672534e-05, "loss": 0.0936, "step": 15450 }, { "epoch": 7.56, "learning_rate": 1.243939393939394e-05, "loss": 0.0971, "step": 15475 }, { "epoch": 7.58, "learning_rate": 1.2427174975562073e-05, "loss": 0.0852, "step": 15500 }, { "epoch": 7.59, "learning_rate": 1.2414956011730205e-05, "loss": 0.0913, "step": 15525 }, { "epoch": 7.6, "learning_rate": 1.240273704789834e-05, "loss": 0.0922, "step": 15550 }, { "epoch": 7.61, "learning_rate": 1.2390518084066472e-05, "loss": 0.0918, "step": 15575 }, { "epoch": 7.62, "learning_rate": 1.2378299120234604e-05, "loss": 0.1045, "step": 15600 }, { "epoch": 7.64, "learning_rate": 1.2366080156402738e-05, "loss": 0.1003, "step": 15625 }, { "epoch": 7.65, "learning_rate": 1.235386119257087e-05, "loss": 0.087, "step": 15650 }, { "epoch": 7.66, "learning_rate": 1.2341642228739003e-05, "loss": 0.0965, "step": 15675 }, { "epoch": 7.67, "learning_rate": 1.232991202346041e-05, "loss": 0.0913, "step": 15700 }, { "epoch": 7.69, "learning_rate": 1.2317693059628544e-05, "loss": 0.0866, "step": 15725 }, { "epoch": 7.7, "learning_rate": 1.2305474095796677e-05, "loss": 0.091, "step": 15750 }, { "epoch": 7.71, "learning_rate": 1.2293255131964809e-05, "loss": 0.101, "step": 15775 }, { "epoch": 7.72, "learning_rate": 1.2281036168132943e-05, "loss": 0.1011, "step": 15800 }, { "epoch": 7.73, "learning_rate": 1.2268817204301076e-05, "loss": 0.0854, "step": 15825 }, { "epoch": 7.75, "learning_rate": 1.2256598240469208e-05, "loss": 0.0872, "step": 15850 }, { "epoch": 7.76, "learning_rate": 1.2244379276637342e-05, "loss": 0.0904, "step": 15875 }, { "epoch": 7.77, "learning_rate": 1.2232160312805475e-05, "loss": 0.1043, "step": 15900 }, { "epoch": 7.78, "learning_rate": 1.2219941348973607e-05, "loss": 0.0815, "step": 15925 }, { "epoch": 7.8, "learning_rate": 1.2207722385141741e-05, "loss": 0.0909, "step": 15950 }, { "epoch": 7.81, "learning_rate": 1.2195503421309873e-05, "loss": 0.0998, "step": 15975 }, { "epoch": 7.82, "learning_rate": 1.2183284457478006e-05, "loss": 0.0947, "step": 16000 }, { "epoch": 7.83, "learning_rate": 1.217106549364614e-05, "loss": 0.1099, "step": 16025 }, { "epoch": 7.84, "learning_rate": 1.2158846529814272e-05, "loss": 0.0971, "step": 16050 }, { "epoch": 7.86, "learning_rate": 1.2146627565982405e-05, "loss": 0.0944, "step": 16075 }, { "epoch": 7.87, "learning_rate": 1.2134408602150539e-05, "loss": 0.1006, "step": 16100 }, { "epoch": 7.88, "learning_rate": 1.2122189638318671e-05, "loss": 0.1007, "step": 16125 }, { "epoch": 7.89, "learning_rate": 1.2109970674486804e-05, "loss": 0.1004, "step": 16150 }, { "epoch": 7.91, "learning_rate": 1.2097751710654938e-05, "loss": 0.0946, "step": 16175 }, { "epoch": 7.92, "learning_rate": 1.208553274682307e-05, "loss": 0.0988, "step": 16200 }, { "epoch": 7.93, "learning_rate": 1.2073313782991203e-05, "loss": 0.0994, "step": 16225 }, { "epoch": 7.94, "learning_rate": 1.2061094819159337e-05, "loss": 0.0912, "step": 16250 }, { "epoch": 7.95, "learning_rate": 1.2048875855327469e-05, "loss": 0.094, "step": 16275 }, { "epoch": 7.97, "learning_rate": 1.2036656891495601e-05, "loss": 0.1037, "step": 16300 }, { "epoch": 7.98, "learning_rate": 1.2024437927663734e-05, "loss": 0.1082, "step": 16325 }, { "epoch": 7.99, "learning_rate": 1.2012218963831868e-05, "loss": 0.0959, "step": 16350 }, { "epoch": 8.0, "learning_rate": 1.2e-05, "loss": 0.0848, "step": 16375 }, { "epoch": 8.02, "learning_rate": 1.1987781036168133e-05, "loss": 0.07, "step": 16400 }, { "epoch": 8.03, "learning_rate": 1.1975562072336267e-05, "loss": 0.0713, "step": 16425 }, { "epoch": 8.04, "learning_rate": 1.19633431085044e-05, "loss": 0.0816, "step": 16450 }, { "epoch": 8.05, "learning_rate": 1.1951124144672532e-05, "loss": 0.0823, "step": 16475 }, { "epoch": 8.06, "learning_rate": 1.1938905180840666e-05, "loss": 0.0624, "step": 16500 }, { "epoch": 8.08, "learning_rate": 1.1926686217008798e-05, "loss": 0.0606, "step": 16525 }, { "epoch": 8.09, "learning_rate": 1.191446725317693e-05, "loss": 0.063, "step": 16550 }, { "epoch": 8.1, "learning_rate": 1.1902248289345065e-05, "loss": 0.0679, "step": 16575 }, { "epoch": 8.11, "learning_rate": 1.1890029325513197e-05, "loss": 0.0748, "step": 16600 }, { "epoch": 8.13, "learning_rate": 1.187781036168133e-05, "loss": 0.0666, "step": 16625 }, { "epoch": 8.14, "learning_rate": 1.1865591397849463e-05, "loss": 0.0703, "step": 16650 }, { "epoch": 8.15, "learning_rate": 1.1853372434017596e-05, "loss": 0.0756, "step": 16675 }, { "epoch": 8.16, "learning_rate": 1.1841153470185728e-05, "loss": 0.0611, "step": 16700 }, { "epoch": 8.17, "learning_rate": 1.1828934506353862e-05, "loss": 0.0766, "step": 16725 }, { "epoch": 8.19, "learning_rate": 1.1816715542521995e-05, "loss": 0.067, "step": 16750 }, { "epoch": 8.2, "learning_rate": 1.1804496578690127e-05, "loss": 0.075, "step": 16775 }, { "epoch": 8.21, "learning_rate": 1.1792277614858261e-05, "loss": 0.0887, "step": 16800 }, { "epoch": 8.22, "learning_rate": 1.1780058651026394e-05, "loss": 0.0652, "step": 16825 }, { "epoch": 8.24, "learning_rate": 1.1767839687194526e-05, "loss": 0.0787, "step": 16850 }, { "epoch": 8.25, "learning_rate": 1.175562072336266e-05, "loss": 0.0753, "step": 16875 }, { "epoch": 8.26, "learning_rate": 1.1743401759530793e-05, "loss": 0.0747, "step": 16900 }, { "epoch": 8.27, "learning_rate": 1.1731182795698925e-05, "loss": 0.0681, "step": 16925 }, { "epoch": 8.28, "learning_rate": 1.1718963831867059e-05, "loss": 0.0716, "step": 16950 }, { "epoch": 8.3, "learning_rate": 1.1706744868035191e-05, "loss": 0.0696, "step": 16975 }, { "epoch": 8.31, "learning_rate": 1.1694525904203324e-05, "loss": 0.0787, "step": 17000 }, { "epoch": 8.32, "learning_rate": 1.1682306940371458e-05, "loss": 0.0835, "step": 17025 }, { "epoch": 8.33, "learning_rate": 1.167008797653959e-05, "loss": 0.0712, "step": 17050 }, { "epoch": 8.35, "learning_rate": 1.1657869012707723e-05, "loss": 0.0728, "step": 17075 }, { "epoch": 8.36, "learning_rate": 1.1645650048875857e-05, "loss": 0.0765, "step": 17100 }, { "epoch": 8.37, "learning_rate": 1.163343108504399e-05, "loss": 0.0742, "step": 17125 }, { "epoch": 8.38, "learning_rate": 1.1621212121212122e-05, "loss": 0.0949, "step": 17150 }, { "epoch": 8.39, "learning_rate": 1.1608993157380256e-05, "loss": 0.0709, "step": 17175 }, { "epoch": 8.41, "learning_rate": 1.1596774193548388e-05, "loss": 0.0712, "step": 17200 }, { "epoch": 8.42, "learning_rate": 1.158455522971652e-05, "loss": 0.082, "step": 17225 }, { "epoch": 8.43, "learning_rate": 1.1572336265884655e-05, "loss": 0.0783, "step": 17250 }, { "epoch": 8.44, "learning_rate": 1.1560117302052787e-05, "loss": 0.068, "step": 17275 }, { "epoch": 8.46, "learning_rate": 1.154789833822092e-05, "loss": 0.0788, "step": 17300 }, { "epoch": 8.47, "learning_rate": 1.1535679374389054e-05, "loss": 0.0783, "step": 17325 }, { "epoch": 8.48, "learning_rate": 1.1523460410557186e-05, "loss": 0.0796, "step": 17350 }, { "epoch": 8.49, "learning_rate": 1.1511241446725318e-05, "loss": 0.0751, "step": 17375 }, { "epoch": 8.5, "learning_rate": 1.149902248289345e-05, "loss": 0.0754, "step": 17400 }, { "epoch": 8.52, "learning_rate": 1.1486803519061585e-05, "loss": 0.0855, "step": 17425 }, { "epoch": 8.53, "learning_rate": 1.1474584555229717e-05, "loss": 0.0773, "step": 17450 }, { "epoch": 8.54, "learning_rate": 1.146236559139785e-05, "loss": 0.0844, "step": 17475 }, { "epoch": 8.55, "learning_rate": 1.1450146627565984e-05, "loss": 0.0797, "step": 17500 }, { "epoch": 8.57, "learning_rate": 1.1437927663734116e-05, "loss": 0.0675, "step": 17525 }, { "epoch": 8.58, "learning_rate": 1.1425708699902249e-05, "loss": 0.0742, "step": 17550 }, { "epoch": 8.59, "learning_rate": 1.1413489736070383e-05, "loss": 0.0761, "step": 17575 }, { "epoch": 8.6, "learning_rate": 1.1401270772238515e-05, "loss": 0.0806, "step": 17600 }, { "epoch": 8.61, "learning_rate": 1.1389051808406647e-05, "loss": 0.071, "step": 17625 }, { "epoch": 8.63, "learning_rate": 1.1376832844574782e-05, "loss": 0.0782, "step": 17650 }, { "epoch": 8.64, "learning_rate": 1.1364613880742914e-05, "loss": 0.067, "step": 17675 }, { "epoch": 8.65, "learning_rate": 1.1352394916911046e-05, "loss": 0.0778, "step": 17700 }, { "epoch": 8.66, "learning_rate": 1.134017595307918e-05, "loss": 0.0729, "step": 17725 }, { "epoch": 8.68, "learning_rate": 1.1327956989247313e-05, "loss": 0.0775, "step": 17750 }, { "epoch": 8.69, "learning_rate": 1.1315738025415445e-05, "loss": 0.0726, "step": 17775 }, { "epoch": 8.7, "learning_rate": 1.130351906158358e-05, "loss": 0.0761, "step": 17800 }, { "epoch": 8.71, "learning_rate": 1.1291300097751712e-05, "loss": 0.0824, "step": 17825 }, { "epoch": 8.72, "learning_rate": 1.1279081133919844e-05, "loss": 0.0753, "step": 17850 }, { "epoch": 8.74, "learning_rate": 1.1266862170087978e-05, "loss": 0.069, "step": 17875 }, { "epoch": 8.75, "learning_rate": 1.125464320625611e-05, "loss": 0.0848, "step": 17900 }, { "epoch": 8.76, "learning_rate": 1.1242424242424243e-05, "loss": 0.0845, "step": 17925 }, { "epoch": 8.77, "learning_rate": 1.1230205278592377e-05, "loss": 0.0823, "step": 17950 }, { "epoch": 8.79, "learning_rate": 1.121798631476051e-05, "loss": 0.0764, "step": 17975 }, { "epoch": 8.8, "learning_rate": 1.1205767350928642e-05, "loss": 0.0899, "step": 18000 }, { "epoch": 8.81, "learning_rate": 1.1193548387096776e-05, "loss": 0.0819, "step": 18025 }, { "epoch": 8.82, "learning_rate": 1.1181329423264908e-05, "loss": 0.0823, "step": 18050 }, { "epoch": 8.83, "learning_rate": 1.1169599217986315e-05, "loss": 0.0717, "step": 18075 }, { "epoch": 8.85, "learning_rate": 1.1157380254154448e-05, "loss": 0.0776, "step": 18100 }, { "epoch": 8.86, "learning_rate": 1.1145161290322582e-05, "loss": 0.0786, "step": 18125 }, { "epoch": 8.87, "learning_rate": 1.1132942326490714e-05, "loss": 0.0759, "step": 18150 }, { "epoch": 8.88, "learning_rate": 1.1120723362658847e-05, "loss": 0.0759, "step": 18175 }, { "epoch": 8.9, "learning_rate": 1.110850439882698e-05, "loss": 0.0659, "step": 18200 }, { "epoch": 8.91, "learning_rate": 1.1096285434995113e-05, "loss": 0.0849, "step": 18225 }, { "epoch": 8.92, "learning_rate": 1.1084066471163246e-05, "loss": 0.0795, "step": 18250 }, { "epoch": 8.93, "learning_rate": 1.107184750733138e-05, "loss": 0.0871, "step": 18275 }, { "epoch": 8.94, "learning_rate": 1.1059628543499512e-05, "loss": 0.0786, "step": 18300 }, { "epoch": 8.96, "learning_rate": 1.1047409579667645e-05, "loss": 0.0857, "step": 18325 }, { "epoch": 8.97, "learning_rate": 1.1035190615835779e-05, "loss": 0.0847, "step": 18350 }, { "epoch": 8.98, "learning_rate": 1.1022971652003911e-05, "loss": 0.0869, "step": 18375 }, { "epoch": 8.99, "learning_rate": 1.1010752688172043e-05, "loss": 0.0711, "step": 18400 }, { "epoch": 9.01, "learning_rate": 1.0998533724340177e-05, "loss": 0.0767, "step": 18425 }, { "epoch": 9.02, "learning_rate": 1.098631476050831e-05, "loss": 0.063, "step": 18450 }, { "epoch": 9.03, "learning_rate": 1.0974095796676442e-05, "loss": 0.0544, "step": 18475 }, { "epoch": 9.04, "learning_rate": 1.0961876832844576e-05, "loss": 0.0664, "step": 18500 }, { "epoch": 9.05, "learning_rate": 1.0949657869012709e-05, "loss": 0.0577, "step": 18525 }, { "epoch": 9.07, "learning_rate": 1.0937438905180841e-05, "loss": 0.0518, "step": 18550 }, { "epoch": 9.08, "learning_rate": 1.0925219941348975e-05, "loss": 0.0689, "step": 18575 }, { "epoch": 9.09, "learning_rate": 1.0913000977517108e-05, "loss": 0.0557, "step": 18600 }, { "epoch": 9.1, "learning_rate": 1.090078201368524e-05, "loss": 0.0646, "step": 18625 }, { "epoch": 9.12, "learning_rate": 1.0888563049853374e-05, "loss": 0.0705, "step": 18650 }, { "epoch": 9.13, "learning_rate": 1.0876344086021507e-05, "loss": 0.049, "step": 18675 }, { "epoch": 9.14, "learning_rate": 1.0864125122189639e-05, "loss": 0.0568, "step": 18700 }, { "epoch": 9.15, "learning_rate": 1.0851906158357773e-05, "loss": 0.0547, "step": 18725 }, { "epoch": 9.16, "learning_rate": 1.0839687194525905e-05, "loss": 0.0629, "step": 18750 }, { "epoch": 9.18, "learning_rate": 1.0827468230694038e-05, "loss": 0.0647, "step": 18775 }, { "epoch": 9.19, "learning_rate": 1.0815249266862172e-05, "loss": 0.0692, "step": 18800 }, { "epoch": 9.2, "learning_rate": 1.0803030303030304e-05, "loss": 0.0495, "step": 18825 }, { "epoch": 9.21, "learning_rate": 1.0790811339198437e-05, "loss": 0.0628, "step": 18850 }, { "epoch": 9.23, "learning_rate": 1.077859237536657e-05, "loss": 0.0563, "step": 18875 }, { "epoch": 9.24, "learning_rate": 1.0766373411534703e-05, "loss": 0.0657, "step": 18900 }, { "epoch": 9.25, "learning_rate": 1.0754154447702836e-05, "loss": 0.0535, "step": 18925 }, { "epoch": 9.26, "learning_rate": 1.0741935483870968e-05, "loss": 0.0667, "step": 18950 }, { "epoch": 9.27, "learning_rate": 1.0729716520039102e-05, "loss": 0.0594, "step": 18975 }, { "epoch": 9.29, "learning_rate": 1.0717497556207235e-05, "loss": 0.0628, "step": 19000 }, { "epoch": 9.3, "learning_rate": 1.0705278592375367e-05, "loss": 0.058, "step": 19025 }, { "epoch": 9.31, "learning_rate": 1.0693059628543501e-05, "loss": 0.0677, "step": 19050 }, { "epoch": 9.32, "learning_rate": 1.0680840664711633e-05, "loss": 0.0623, "step": 19075 }, { "epoch": 9.34, "learning_rate": 1.0668621700879766e-05, "loss": 0.0631, "step": 19100 }, { "epoch": 9.35, "learning_rate": 1.06564027370479e-05, "loss": 0.0616, "step": 19125 }, { "epoch": 9.36, "learning_rate": 1.0644183773216032e-05, "loss": 0.0613, "step": 19150 }, { "epoch": 9.37, "learning_rate": 1.0631964809384165e-05, "loss": 0.066, "step": 19175 }, { "epoch": 9.38, "learning_rate": 1.0619745845552299e-05, "loss": 0.0653, "step": 19200 }, { "epoch": 9.4, "learning_rate": 1.0607526881720431e-05, "loss": 0.0701, "step": 19225 }, { "epoch": 9.41, "learning_rate": 1.0595307917888564e-05, "loss": 0.0612, "step": 19250 }, { "epoch": 9.42, "learning_rate": 1.0583088954056698e-05, "loss": 0.055, "step": 19275 }, { "epoch": 9.43, "learning_rate": 1.057086999022483e-05, "loss": 0.0603, "step": 19300 }, { "epoch": 9.45, "learning_rate": 1.0558651026392963e-05, "loss": 0.0648, "step": 19325 }, { "epoch": 9.46, "learning_rate": 1.0546432062561097e-05, "loss": 0.0659, "step": 19350 }, { "epoch": 9.47, "learning_rate": 1.0534213098729229e-05, "loss": 0.0548, "step": 19375 }, { "epoch": 9.48, "learning_rate": 1.0521994134897361e-05, "loss": 0.0605, "step": 19400 }, { "epoch": 9.49, "learning_rate": 1.0509775171065496e-05, "loss": 0.0542, "step": 19425 }, { "epoch": 9.51, "learning_rate": 1.0497556207233628e-05, "loss": 0.0647, "step": 19450 }, { "epoch": 9.52, "learning_rate": 1.048533724340176e-05, "loss": 0.0677, "step": 19475 }, { "epoch": 9.53, "learning_rate": 1.0473118279569894e-05, "loss": 0.06, "step": 19500 }, { "epoch": 9.54, "learning_rate": 1.0460899315738027e-05, "loss": 0.0664, "step": 19525 }, { "epoch": 9.56, "learning_rate": 1.044868035190616e-05, "loss": 0.0666, "step": 19550 }, { "epoch": 9.57, "learning_rate": 1.0436461388074293e-05, "loss": 0.0576, "step": 19575 }, { "epoch": 9.58, "learning_rate": 1.0424242424242426e-05, "loss": 0.0586, "step": 19600 }, { "epoch": 9.59, "learning_rate": 1.0412023460410558e-05, "loss": 0.0631, "step": 19625 }, { "epoch": 9.6, "learning_rate": 1.0399804496578692e-05, "loss": 0.0688, "step": 19650 }, { "epoch": 9.62, "learning_rate": 1.0387585532746825e-05, "loss": 0.0619, "step": 19675 }, { "epoch": 9.63, "learning_rate": 1.0375366568914957e-05, "loss": 0.0695, "step": 19700 }, { "epoch": 9.64, "learning_rate": 1.0363147605083091e-05, "loss": 0.0694, "step": 19725 }, { "epoch": 9.65, "learning_rate": 1.0350928641251224e-05, "loss": 0.0654, "step": 19750 }, { "epoch": 9.67, "learning_rate": 1.0338709677419356e-05, "loss": 0.0688, "step": 19775 }, { "epoch": 9.68, "learning_rate": 1.032649071358749e-05, "loss": 0.0657, "step": 19800 }, { "epoch": 9.69, "learning_rate": 1.0314271749755622e-05, "loss": 0.0575, "step": 19825 }, { "epoch": 9.7, "learning_rate": 1.0302052785923755e-05, "loss": 0.0545, "step": 19850 }, { "epoch": 9.71, "learning_rate": 1.0289833822091889e-05, "loss": 0.0694, "step": 19875 }, { "epoch": 9.73, "learning_rate": 1.0277614858260021e-05, "loss": 0.0706, "step": 19900 }, { "epoch": 9.74, "learning_rate": 1.0265395894428154e-05, "loss": 0.061, "step": 19925 }, { "epoch": 9.75, "learning_rate": 1.0253176930596286e-05, "loss": 0.064, "step": 19950 }, { "epoch": 9.76, "learning_rate": 1.024095796676442e-05, "loss": 0.0629, "step": 19975 }, { "epoch": 9.78, "learning_rate": 1.0228739002932553e-05, "loss": 0.0654, "step": 20000 }, { "epoch": 9.79, "learning_rate": 1.0216520039100685e-05, "loss": 0.0594, "step": 20025 }, { "epoch": 9.8, "learning_rate": 1.0204301075268819e-05, "loss": 0.0644, "step": 20050 }, { "epoch": 9.81, "learning_rate": 1.0192082111436951e-05, "loss": 0.0781, "step": 20075 }, { "epoch": 9.82, "learning_rate": 1.0179863147605084e-05, "loss": 0.0659, "step": 20100 }, { "epoch": 9.84, "learning_rate": 1.0167644183773218e-05, "loss": 0.072, "step": 20125 }, { "epoch": 9.85, "learning_rate": 1.015542521994135e-05, "loss": 0.0661, "step": 20150 }, { "epoch": 9.86, "learning_rate": 1.0143206256109483e-05, "loss": 0.0615, "step": 20175 }, { "epoch": 9.87, "learning_rate": 1.0130987292277617e-05, "loss": 0.0625, "step": 20200 }, { "epoch": 9.89, "learning_rate": 1.011876832844575e-05, "loss": 0.0634, "step": 20225 }, { "epoch": 9.9, "learning_rate": 1.0106549364613882e-05, "loss": 0.0616, "step": 20250 }, { "epoch": 9.91, "learning_rate": 1.0094330400782016e-05, "loss": 0.0602, "step": 20275 }, { "epoch": 9.92, "learning_rate": 1.0082111436950148e-05, "loss": 0.0739, "step": 20300 }, { "epoch": 9.93, "learning_rate": 1.006989247311828e-05, "loss": 0.0654, "step": 20325 }, { "epoch": 9.95, "learning_rate": 1.0057673509286415e-05, "loss": 0.0724, "step": 20350 }, { "epoch": 9.96, "learning_rate": 1.0045454545454547e-05, "loss": 0.0735, "step": 20375 }, { "epoch": 9.97, "learning_rate": 1.003323558162268e-05, "loss": 0.0652, "step": 20400 }, { "epoch": 9.98, "learning_rate": 1.0021016617790814e-05, "loss": 0.0694, "step": 20425 }, { "epoch": 10.0, "learning_rate": 1.0008797653958946e-05, "loss": 0.071, "step": 20450 }, { "epoch": 10.01, "learning_rate": 9.996578690127078e-06, "loss": 0.0609, "step": 20475 }, { "epoch": 10.02, "learning_rate": 9.98435972629521e-06, "loss": 0.0383, "step": 20500 }, { "epoch": 10.03, "learning_rate": 9.972140762463343e-06, "loss": 0.0541, "step": 20525 }, { "epoch": 10.04, "learning_rate": 9.959921798631477e-06, "loss": 0.0552, "step": 20550 }, { "epoch": 10.06, "learning_rate": 9.94770283479961e-06, "loss": 0.052, "step": 20575 }, { "epoch": 10.07, "learning_rate": 9.935483870967742e-06, "loss": 0.0512, "step": 20600 }, { "epoch": 10.08, "learning_rate": 9.923264907135876e-06, "loss": 0.0453, "step": 20625 }, { "epoch": 10.09, "learning_rate": 9.911045943304009e-06, "loss": 0.0465, "step": 20650 }, { "epoch": 10.11, "learning_rate": 9.898826979472141e-06, "loss": 0.0522, "step": 20675 }, { "epoch": 10.12, "learning_rate": 9.886608015640275e-06, "loss": 0.054, "step": 20700 }, { "epoch": 10.13, "learning_rate": 9.874389051808407e-06, "loss": 0.0575, "step": 20725 }, { "epoch": 10.14, "learning_rate": 9.86217008797654e-06, "loss": 0.0526, "step": 20750 }, { "epoch": 10.15, "learning_rate": 9.849951124144674e-06, "loss": 0.0572, "step": 20775 }, { "epoch": 10.17, "learning_rate": 9.837732160312806e-06, "loss": 0.0539, "step": 20800 }, { "epoch": 10.18, "learning_rate": 9.825513196480939e-06, "loss": 0.0505, "step": 20825 }, { "epoch": 10.19, "learning_rate": 9.813294232649073e-06, "loss": 0.0607, "step": 20850 }, { "epoch": 10.2, "learning_rate": 9.801075268817205e-06, "loss": 0.0487, "step": 20875 }, { "epoch": 10.22, "learning_rate": 9.788856304985338e-06, "loss": 0.0438, "step": 20900 }, { "epoch": 10.23, "learning_rate": 9.776637341153472e-06, "loss": 0.0585, "step": 20925 }, { "epoch": 10.24, "learning_rate": 9.764418377321604e-06, "loss": 0.0608, "step": 20950 }, { "epoch": 10.25, "learning_rate": 9.752199413489737e-06, "loss": 0.0421, "step": 20975 }, { "epoch": 10.26, "learning_rate": 9.73998044965787e-06, "loss": 0.0511, "step": 21000 }, { "epoch": 10.28, "learning_rate": 9.727761485826003e-06, "loss": 0.0517, "step": 21025 }, { "epoch": 10.29, "learning_rate": 9.715542521994135e-06, "loss": 0.0639, "step": 21050 }, { "epoch": 10.3, "learning_rate": 9.703323558162268e-06, "loss": 0.0544, "step": 21075 }, { "epoch": 10.31, "learning_rate": 9.691104594330402e-06, "loss": 0.049, "step": 21100 }, { "epoch": 10.33, "learning_rate": 9.678885630498534e-06, "loss": 0.0522, "step": 21125 }, { "epoch": 10.34, "learning_rate": 9.666666666666667e-06, "loss": 0.0596, "step": 21150 }, { "epoch": 10.35, "learning_rate": 9.6544477028348e-06, "loss": 0.0452, "step": 21175 }, { "epoch": 10.36, "learning_rate": 9.642228739002933e-06, "loss": 0.0555, "step": 21200 }, { "epoch": 10.37, "learning_rate": 9.630009775171066e-06, "loss": 0.0507, "step": 21225 }, { "epoch": 10.39, "learning_rate": 9.6177908113392e-06, "loss": 0.0371, "step": 21250 }, { "epoch": 10.4, "learning_rate": 9.605571847507332e-06, "loss": 0.0534, "step": 21275 }, { "epoch": 10.41, "learning_rate": 9.593352883675465e-06, "loss": 0.0482, "step": 21300 }, { "epoch": 10.42, "learning_rate": 9.581133919843599e-06, "loss": 0.0473, "step": 21325 }, { "epoch": 10.43, "learning_rate": 9.568914956011731e-06, "loss": 0.0538, "step": 21350 }, { "epoch": 10.45, "learning_rate": 9.556695992179863e-06, "loss": 0.0545, "step": 21375 }, { "epoch": 10.46, "learning_rate": 9.544477028347998e-06, "loss": 0.0528, "step": 21400 }, { "epoch": 10.47, "learning_rate": 9.53225806451613e-06, "loss": 0.0476, "step": 21425 }, { "epoch": 10.48, "learning_rate": 9.520039100684262e-06, "loss": 0.0582, "step": 21450 }, { "epoch": 10.5, "learning_rate": 9.507820136852396e-06, "loss": 0.05, "step": 21475 }, { "epoch": 10.51, "learning_rate": 9.495601173020529e-06, "loss": 0.0498, "step": 21500 }, { "epoch": 10.52, "learning_rate": 9.483382209188661e-06, "loss": 0.0537, "step": 21525 }, { "epoch": 10.53, "learning_rate": 9.471163245356795e-06, "loss": 0.0596, "step": 21550 }, { "epoch": 10.54, "learning_rate": 9.458944281524928e-06, "loss": 0.0495, "step": 21575 }, { "epoch": 10.56, "learning_rate": 9.44672531769306e-06, "loss": 0.056, "step": 21600 }, { "epoch": 10.57, "learning_rate": 9.434506353861194e-06, "loss": 0.0493, "step": 21625 }, { "epoch": 10.58, "learning_rate": 9.422776148582601e-06, "loss": 0.0561, "step": 21650 }, { "epoch": 10.59, "learning_rate": 9.410557184750734e-06, "loss": 0.0483, "step": 21675 }, { "epoch": 10.61, "learning_rate": 9.398338220918866e-06, "loss": 0.0545, "step": 21700 }, { "epoch": 10.62, "learning_rate": 9.386119257087e-06, "loss": 0.0519, "step": 21725 }, { "epoch": 10.63, "learning_rate": 9.373900293255133e-06, "loss": 0.051, "step": 21750 }, { "epoch": 10.64, "learning_rate": 9.361681329423265e-06, "loss": 0.054, "step": 21775 }, { "epoch": 10.65, "learning_rate": 9.349462365591399e-06, "loss": 0.0594, "step": 21800 }, { "epoch": 10.67, "learning_rate": 9.337243401759531e-06, "loss": 0.0561, "step": 21825 }, { "epoch": 10.68, "learning_rate": 9.325024437927664e-06, "loss": 0.0607, "step": 21850 }, { "epoch": 10.69, "learning_rate": 9.312805474095798e-06, "loss": 0.0416, "step": 21875 }, { "epoch": 10.7, "learning_rate": 9.30058651026393e-06, "loss": 0.053, "step": 21900 }, { "epoch": 10.72, "learning_rate": 9.288367546432063e-06, "loss": 0.0555, "step": 21925 }, { "epoch": 10.73, "learning_rate": 9.276148582600197e-06, "loss": 0.0633, "step": 21950 }, { "epoch": 10.74, "learning_rate": 9.26392961876833e-06, "loss": 0.0621, "step": 21975 }, { "epoch": 10.75, "learning_rate": 9.251710654936462e-06, "loss": 0.052, "step": 22000 }, { "epoch": 10.76, "learning_rate": 9.239491691104596e-06, "loss": 0.0548, "step": 22025 }, { "epoch": 10.78, "learning_rate": 9.227272727272728e-06, "loss": 0.06, "step": 22050 }, { "epoch": 10.79, "learning_rate": 9.21505376344086e-06, "loss": 0.0528, "step": 22075 }, { "epoch": 10.8, "learning_rate": 9.202834799608995e-06, "loss": 0.06, "step": 22100 }, { "epoch": 10.81, "learning_rate": 9.190615835777127e-06, "loss": 0.0561, "step": 22125 }, { "epoch": 10.83, "learning_rate": 9.17839687194526e-06, "loss": 0.057, "step": 22150 }, { "epoch": 10.84, "learning_rate": 9.166177908113393e-06, "loss": 0.0546, "step": 22175 }, { "epoch": 10.85, "learning_rate": 9.153958944281526e-06, "loss": 0.0549, "step": 22200 }, { "epoch": 10.86, "learning_rate": 9.141739980449658e-06, "loss": 0.0615, "step": 22225 }, { "epoch": 10.87, "learning_rate": 9.129521016617792e-06, "loss": 0.0557, "step": 22250 }, { "epoch": 10.89, "learning_rate": 9.117302052785925e-06, "loss": 0.0605, "step": 22275 }, { "epoch": 10.9, "learning_rate": 9.105083088954057e-06, "loss": 0.0611, "step": 22300 }, { "epoch": 10.91, "learning_rate": 9.092864125122191e-06, "loss": 0.0596, "step": 22325 }, { "epoch": 10.92, "learning_rate": 9.080645161290324e-06, "loss": 0.057, "step": 22350 }, { "epoch": 10.94, "learning_rate": 9.068426197458456e-06, "loss": 0.0543, "step": 22375 }, { "epoch": 10.95, "learning_rate": 9.05620723362659e-06, "loss": 0.0604, "step": 22400 }, { "epoch": 10.96, "learning_rate": 9.043988269794723e-06, "loss": 0.0546, "step": 22425 }, { "epoch": 10.97, "learning_rate": 9.031769305962855e-06, "loss": 0.0552, "step": 22450 }, { "epoch": 10.98, "learning_rate": 9.019550342130989e-06, "loss": 0.065, "step": 22475 }, { "epoch": 11.0, "learning_rate": 9.007331378299121e-06, "loss": 0.0511, "step": 22500 }, { "epoch": 11.01, "learning_rate": 8.995112414467254e-06, "loss": 0.0367, "step": 22525 }, { "epoch": 11.02, "learning_rate": 8.982893450635386e-06, "loss": 0.0531, "step": 22550 }, { "epoch": 11.03, "learning_rate": 8.97067448680352e-06, "loss": 0.0524, "step": 22575 }, { "epoch": 11.05, "learning_rate": 8.958455522971653e-06, "loss": 0.0442, "step": 22600 }, { "epoch": 11.06, "learning_rate": 8.946236559139785e-06, "loss": 0.0443, "step": 22625 }, { "epoch": 11.07, "learning_rate": 8.93401759530792e-06, "loss": 0.0395, "step": 22650 }, { "epoch": 11.08, "learning_rate": 8.921798631476052e-06, "loss": 0.0429, "step": 22675 }, { "epoch": 11.09, "learning_rate": 8.909579667644184e-06, "loss": 0.0411, "step": 22700 }, { "epoch": 11.11, "learning_rate": 8.897360703812318e-06, "loss": 0.0516, "step": 22725 }, { "epoch": 11.12, "learning_rate": 8.88514173998045e-06, "loss": 0.0428, "step": 22750 }, { "epoch": 11.13, "learning_rate": 8.872922776148583e-06, "loss": 0.0367, "step": 22775 }, { "epoch": 11.14, "learning_rate": 8.860703812316717e-06, "loss": 0.045, "step": 22800 }, { "epoch": 11.16, "learning_rate": 8.84848484848485e-06, "loss": 0.0447, "step": 22825 }, { "epoch": 11.17, "learning_rate": 8.836265884652982e-06, "loss": 0.0475, "step": 22850 }, { "epoch": 11.18, "learning_rate": 8.824046920821116e-06, "loss": 0.0441, "step": 22875 }, { "epoch": 11.19, "learning_rate": 8.811827956989248e-06, "loss": 0.0416, "step": 22900 }, { "epoch": 11.2, "learning_rate": 8.79960899315738e-06, "loss": 0.0555, "step": 22925 }, { "epoch": 11.22, "learning_rate": 8.787390029325515e-06, "loss": 0.035, "step": 22950 }, { "epoch": 11.23, "learning_rate": 8.775171065493647e-06, "loss": 0.0486, "step": 22975 }, { "epoch": 11.24, "learning_rate": 8.76295210166178e-06, "loss": 0.0478, "step": 23000 }, { "epoch": 11.25, "learning_rate": 8.750733137829914e-06, "loss": 0.0494, "step": 23025 }, { "epoch": 11.27, "learning_rate": 8.738514173998046e-06, "loss": 0.0452, "step": 23050 }, { "epoch": 11.28, "learning_rate": 8.726295210166179e-06, "loss": 0.0398, "step": 23075 }, { "epoch": 11.29, "learning_rate": 8.714076246334313e-06, "loss": 0.0438, "step": 23100 }, { "epoch": 11.3, "learning_rate": 8.701857282502445e-06, "loss": 0.0479, "step": 23125 }, { "epoch": 11.31, "learning_rate": 8.689638318670577e-06, "loss": 0.0484, "step": 23150 }, { "epoch": 11.33, "learning_rate": 8.677419354838712e-06, "loss": 0.04, "step": 23175 }, { "epoch": 11.34, "learning_rate": 8.665200391006844e-06, "loss": 0.043, "step": 23200 }, { "epoch": 11.35, "learning_rate": 8.652981427174976e-06, "loss": 0.0434, "step": 23225 }, { "epoch": 11.36, "learning_rate": 8.64076246334311e-06, "loss": 0.0462, "step": 23250 }, { "epoch": 11.38, "learning_rate": 8.628543499511243e-06, "loss": 0.0479, "step": 23275 }, { "epoch": 11.39, "learning_rate": 8.616324535679375e-06, "loss": 0.0419, "step": 23300 }, { "epoch": 11.4, "learning_rate": 8.604105571847508e-06, "loss": 0.0459, "step": 23325 }, { "epoch": 11.41, "learning_rate": 8.59188660801564e-06, "loss": 0.0449, "step": 23350 }, { "epoch": 11.42, "learning_rate": 8.579667644183774e-06, "loss": 0.0398, "step": 23375 }, { "epoch": 11.44, "learning_rate": 8.567448680351907e-06, "loss": 0.0443, "step": 23400 }, { "epoch": 11.45, "learning_rate": 8.555229716520039e-06, "loss": 0.0518, "step": 23425 }, { "epoch": 11.46, "learning_rate": 8.543010752688173e-06, "loss": 0.0377, "step": 23450 }, { "epoch": 11.47, "learning_rate": 8.530791788856305e-06, "loss": 0.0547, "step": 23475 }, { "epoch": 11.49, "learning_rate": 8.518572825024438e-06, "loss": 0.045, "step": 23500 }, { "epoch": 11.5, "learning_rate": 8.506353861192572e-06, "loss": 0.0444, "step": 23525 }, { "epoch": 11.51, "learning_rate": 8.494134897360704e-06, "loss": 0.0483, "step": 23550 }, { "epoch": 11.52, "learning_rate": 8.481915933528837e-06, "loss": 0.0436, "step": 23575 }, { "epoch": 11.53, "learning_rate": 8.46969696969697e-06, "loss": 0.0505, "step": 23600 }, { "epoch": 11.55, "learning_rate": 8.457478005865103e-06, "loss": 0.0395, "step": 23625 }, { "epoch": 11.56, "learning_rate": 8.445259042033236e-06, "loss": 0.0417, "step": 23650 }, { "epoch": 11.57, "learning_rate": 8.433040078201368e-06, "loss": 0.0577, "step": 23675 }, { "epoch": 11.58, "learning_rate": 8.420821114369502e-06, "loss": 0.0476, "step": 23700 }, { "epoch": 11.6, "learning_rate": 8.408602150537634e-06, "loss": 0.0415, "step": 23725 }, { "epoch": 11.61, "learning_rate": 8.396383186705767e-06, "loss": 0.0455, "step": 23750 }, { "epoch": 11.62, "learning_rate": 8.384164222873901e-06, "loss": 0.0572, "step": 23775 }, { "epoch": 11.63, "learning_rate": 8.371945259042033e-06, "loss": 0.0442, "step": 23800 }, { "epoch": 11.64, "learning_rate": 8.359726295210166e-06, "loss": 0.0376, "step": 23825 }, { "epoch": 11.66, "learning_rate": 8.3475073313783e-06, "loss": 0.0455, "step": 23850 }, { "epoch": 11.67, "learning_rate": 8.335288367546432e-06, "loss": 0.0514, "step": 23875 }, { "epoch": 11.68, "learning_rate": 8.323069403714565e-06, "loss": 0.0496, "step": 23900 }, { "epoch": 11.69, "learning_rate": 8.310850439882699e-06, "loss": 0.0498, "step": 23925 }, { "epoch": 11.71, "learning_rate": 8.298631476050831e-06, "loss": 0.0479, "step": 23950 }, { "epoch": 11.72, "learning_rate": 8.286412512218964e-06, "loss": 0.0423, "step": 23975 }, { "epoch": 11.73, "learning_rate": 8.274193548387098e-06, "loss": 0.0444, "step": 24000 }, { "epoch": 11.74, "learning_rate": 8.26197458455523e-06, "loss": 0.0482, "step": 24025 }, { "epoch": 11.75, "learning_rate": 8.249755620723362e-06, "loss": 0.0502, "step": 24050 }, { "epoch": 11.77, "learning_rate": 8.237536656891497e-06, "loss": 0.0503, "step": 24075 }, { "epoch": 11.78, "learning_rate": 8.225317693059629e-06, "loss": 0.0503, "step": 24100 }, { "epoch": 11.79, "learning_rate": 8.213098729227761e-06, "loss": 0.0424, "step": 24125 }, { "epoch": 11.8, "learning_rate": 8.200879765395895e-06, "loss": 0.0543, "step": 24150 }, { "epoch": 11.82, "learning_rate": 8.188660801564028e-06, "loss": 0.0505, "step": 24175 }, { "epoch": 11.83, "learning_rate": 8.17644183773216e-06, "loss": 0.0513, "step": 24200 }, { "epoch": 11.84, "learning_rate": 8.164222873900294e-06, "loss": 0.0537, "step": 24225 }, { "epoch": 11.85, "learning_rate": 8.152003910068427e-06, "loss": 0.0449, "step": 24250 }, { "epoch": 11.86, "learning_rate": 8.139784946236559e-06, "loss": 0.0469, "step": 24275 }, { "epoch": 11.88, "learning_rate": 8.127565982404693e-06, "loss": 0.055, "step": 24300 }, { "epoch": 11.89, "learning_rate": 8.115347018572826e-06, "loss": 0.0455, "step": 24325 }, { "epoch": 11.9, "learning_rate": 8.103128054740958e-06, "loss": 0.049, "step": 24350 }, { "epoch": 11.91, "learning_rate": 8.090909090909092e-06, "loss": 0.0457, "step": 24375 }, { "epoch": 11.93, "learning_rate": 8.078690127077225e-06, "loss": 0.0512, "step": 24400 }, { "epoch": 11.94, "learning_rate": 8.066471163245357e-06, "loss": 0.052, "step": 24425 }, { "epoch": 11.95, "learning_rate": 8.054252199413491e-06, "loss": 0.0468, "step": 24450 }, { "epoch": 11.96, "learning_rate": 8.042033235581623e-06, "loss": 0.0451, "step": 24475 }, { "epoch": 11.97, "learning_rate": 8.029814271749756e-06, "loss": 0.0443, "step": 24500 }, { "epoch": 11.99, "learning_rate": 8.01759530791789e-06, "loss": 0.0454, "step": 24525 }, { "epoch": 12.0, "learning_rate": 8.005376344086022e-06, "loss": 0.0415, "step": 24550 }, { "epoch": 12.01, "learning_rate": 7.993157380254155e-06, "loss": 0.0313, "step": 24575 }, { "epoch": 12.02, "learning_rate": 7.980938416422289e-06, "loss": 0.0428, "step": 24600 }, { "epoch": 12.04, "learning_rate": 7.968719452590421e-06, "loss": 0.0386, "step": 24625 }, { "epoch": 12.05, "learning_rate": 7.956500488758554e-06, "loss": 0.0357, "step": 24650 }, { "epoch": 12.06, "learning_rate": 7.944281524926688e-06, "loss": 0.0376, "step": 24675 }, { "epoch": 12.07, "learning_rate": 7.93206256109482e-06, "loss": 0.0398, "step": 24700 }, { "epoch": 12.08, "learning_rate": 7.919843597262953e-06, "loss": 0.0413, "step": 24725 }, { "epoch": 12.1, "learning_rate": 7.907624633431087e-06, "loss": 0.0261, "step": 24750 }, { "epoch": 12.11, "learning_rate": 7.895405669599219e-06, "loss": 0.0375, "step": 24775 }, { "epoch": 12.12, "learning_rate": 7.883186705767351e-06, "loss": 0.0396, "step": 24800 }, { "epoch": 12.13, "learning_rate": 7.870967741935484e-06, "loss": 0.0356, "step": 24825 }, { "epoch": 12.15, "learning_rate": 7.858748778103618e-06, "loss": 0.0284, "step": 24850 }, { "epoch": 12.16, "learning_rate": 7.84652981427175e-06, "loss": 0.0369, "step": 24875 }, { "epoch": 12.17, "learning_rate": 7.834310850439883e-06, "loss": 0.033, "step": 24900 }, { "epoch": 12.18, "learning_rate": 7.822091886608017e-06, "loss": 0.0361, "step": 24925 }, { "epoch": 12.19, "learning_rate": 7.80987292277615e-06, "loss": 0.0384, "step": 24950 }, { "epoch": 12.21, "learning_rate": 7.797653958944282e-06, "loss": 0.0418, "step": 24975 }, { "epoch": 12.22, "learning_rate": 7.785434995112416e-06, "loss": 0.0395, "step": 25000 }, { "epoch": 12.23, "learning_rate": 7.773216031280548e-06, "loss": 0.0356, "step": 25025 }, { "epoch": 12.24, "learning_rate": 7.76099706744868e-06, "loss": 0.0326, "step": 25050 }, { "epoch": 12.26, "learning_rate": 7.748778103616815e-06, "loss": 0.0402, "step": 25075 }, { "epoch": 12.27, "learning_rate": 7.736559139784947e-06, "loss": 0.0372, "step": 25100 }, { "epoch": 12.28, "learning_rate": 7.72434017595308e-06, "loss": 0.0486, "step": 25125 }, { "epoch": 12.29, "learning_rate": 7.712121212121213e-06, "loss": 0.0351, "step": 25150 }, { "epoch": 12.3, "learning_rate": 7.699902248289346e-06, "loss": 0.0388, "step": 25175 }, { "epoch": 12.32, "learning_rate": 7.687683284457478e-06, "loss": 0.05, "step": 25200 }, { "epoch": 12.33, "learning_rate": 7.675464320625612e-06, "loss": 0.0448, "step": 25225 }, { "epoch": 12.34, "learning_rate": 7.663245356793745e-06, "loss": 0.0338, "step": 25250 }, { "epoch": 12.35, "learning_rate": 7.651026392961877e-06, "loss": 0.0393, "step": 25275 }, { "epoch": 12.37, "learning_rate": 7.638807429130011e-06, "loss": 0.0363, "step": 25300 }, { "epoch": 12.38, "learning_rate": 7.626588465298144e-06, "loss": 0.039, "step": 25325 }, { "epoch": 12.39, "learning_rate": 7.614369501466277e-06, "loss": 0.0343, "step": 25350 }, { "epoch": 12.4, "learning_rate": 7.602150537634409e-06, "loss": 0.0416, "step": 25375 }, { "epoch": 12.41, "learning_rate": 7.5899315738025426e-06, "loss": 0.0472, "step": 25400 }, { "epoch": 12.43, "learning_rate": 7.577712609970676e-06, "loss": 0.0384, "step": 25425 }, { "epoch": 12.44, "learning_rate": 7.565493646138808e-06, "loss": 0.0376, "step": 25450 }, { "epoch": 12.45, "learning_rate": 7.5532746823069415e-06, "loss": 0.0386, "step": 25475 }, { "epoch": 12.46, "learning_rate": 7.541055718475075e-06, "loss": 0.0413, "step": 25500 }, { "epoch": 12.48, "learning_rate": 7.528836754643207e-06, "loss": 0.0372, "step": 25525 }, { "epoch": 12.49, "learning_rate": 7.51661779081134e-06, "loss": 0.04, "step": 25550 }, { "epoch": 12.5, "learning_rate": 7.504398826979474e-06, "loss": 0.0416, "step": 25575 }, { "epoch": 12.51, "learning_rate": 7.492179863147606e-06, "loss": 0.0443, "step": 25600 }, { "epoch": 12.52, "learning_rate": 7.479960899315739e-06, "loss": 0.038, "step": 25625 }, { "epoch": 12.54, "learning_rate": 7.467741935483872e-06, "loss": 0.0452, "step": 25650 }, { "epoch": 12.55, "learning_rate": 7.455522971652005e-06, "loss": 0.0366, "step": 25675 }, { "epoch": 12.56, "learning_rate": 7.443304007820138e-06, "loss": 0.0463, "step": 25700 }, { "epoch": 12.57, "learning_rate": 7.4310850439882706e-06, "loss": 0.0447, "step": 25725 }, { "epoch": 12.59, "learning_rate": 7.418866080156404e-06, "loss": 0.0411, "step": 25750 }, { "epoch": 12.6, "learning_rate": 7.406647116324537e-06, "loss": 0.0413, "step": 25775 }, { "epoch": 12.61, "learning_rate": 7.3944281524926694e-06, "loss": 0.0322, "step": 25800 }, { "epoch": 12.62, "learning_rate": 7.382697947214077e-06, "loss": 0.0464, "step": 25825 }, { "epoch": 12.63, "learning_rate": 7.37047898338221e-06, "loss": 0.0337, "step": 25850 }, { "epoch": 12.65, "learning_rate": 7.358260019550343e-06, "loss": 0.0365, "step": 25875 }, { "epoch": 12.66, "learning_rate": 7.346041055718476e-06, "loss": 0.0432, "step": 25900 }, { "epoch": 12.67, "learning_rate": 7.333822091886609e-06, "loss": 0.0457, "step": 25925 }, { "epoch": 12.68, "learning_rate": 7.321603128054742e-06, "loss": 0.0368, "step": 25950 }, { "epoch": 12.7, "learning_rate": 7.309384164222874e-06, "loss": 0.0385, "step": 25975 }, { "epoch": 12.71, "learning_rate": 7.2971652003910075e-06, "loss": 0.044, "step": 26000 }, { "epoch": 12.72, "learning_rate": 7.284946236559141e-06, "loss": 0.0417, "step": 26025 }, { "epoch": 12.73, "learning_rate": 7.272727272727273e-06, "loss": 0.0418, "step": 26050 }, { "epoch": 12.74, "learning_rate": 7.260508308895406e-06, "loss": 0.0328, "step": 26075 }, { "epoch": 12.76, "learning_rate": 7.24828934506354e-06, "loss": 0.0343, "step": 26100 }, { "epoch": 12.77, "learning_rate": 7.236070381231672e-06, "loss": 0.0349, "step": 26125 }, { "epoch": 12.78, "learning_rate": 7.223851417399805e-06, "loss": 0.0364, "step": 26150 }, { "epoch": 12.79, "learning_rate": 7.2116324535679386e-06, "loss": 0.0358, "step": 26175 }, { "epoch": 12.81, "learning_rate": 7.199413489736071e-06, "loss": 0.0399, "step": 26200 }, { "epoch": 12.82, "learning_rate": 7.187194525904204e-06, "loss": 0.0376, "step": 26225 }, { "epoch": 12.83, "learning_rate": 7.1749755620723375e-06, "loss": 0.0396, "step": 26250 }, { "epoch": 12.84, "learning_rate": 7.16275659824047e-06, "loss": 0.0377, "step": 26275 }, { "epoch": 12.85, "learning_rate": 7.150537634408603e-06, "loss": 0.0274, "step": 26300 }, { "epoch": 12.87, "learning_rate": 7.138318670576736e-06, "loss": 0.0409, "step": 26325 }, { "epoch": 12.88, "learning_rate": 7.126099706744869e-06, "loss": 0.0409, "step": 26350 }, { "epoch": 12.89, "learning_rate": 7.113880742913002e-06, "loss": 0.0351, "step": 26375 }, { "epoch": 12.9, "learning_rate": 7.101661779081135e-06, "loss": 0.0372, "step": 26400 }, { "epoch": 12.92, "learning_rate": 7.089442815249268e-06, "loss": 0.0394, "step": 26425 }, { "epoch": 12.93, "learning_rate": 7.077223851417401e-06, "loss": 0.0409, "step": 26450 }, { "epoch": 12.94, "learning_rate": 7.065004887585533e-06, "loss": 0.0328, "step": 26475 }, { "epoch": 12.95, "learning_rate": 7.0527859237536665e-06, "loss": 0.0348, "step": 26500 }, { "epoch": 12.96, "learning_rate": 7.0405669599218e-06, "loss": 0.0355, "step": 26525 }, { "epoch": 12.98, "learning_rate": 7.028347996089932e-06, "loss": 0.043, "step": 26550 }, { "epoch": 12.99, "learning_rate": 7.0161290322580654e-06, "loss": 0.0468, "step": 26575 }, { "epoch": 13.0, "learning_rate": 7.003910068426199e-06, "loss": 0.0427, "step": 26600 }, { "epoch": 13.01, "learning_rate": 6.991691104594331e-06, "loss": 0.0343, "step": 26625 }, { "epoch": 13.03, "learning_rate": 6.979472140762464e-06, "loss": 0.0318, "step": 26650 }, { "epoch": 13.04, "learning_rate": 6.9672531769305976e-06, "loss": 0.0302, "step": 26675 }, { "epoch": 13.05, "learning_rate": 6.95503421309873e-06, "loss": 0.032, "step": 26700 }, { "epoch": 13.06, "learning_rate": 6.942815249266863e-06, "loss": 0.0327, "step": 26725 }, { "epoch": 13.07, "learning_rate": 6.9305962854349965e-06, "loss": 0.0335, "step": 26750 }, { "epoch": 13.09, "learning_rate": 6.918377321603129e-06, "loss": 0.0319, "step": 26775 }, { "epoch": 13.1, "learning_rate": 6.906158357771262e-06, "loss": 0.036, "step": 26800 }, { "epoch": 13.11, "learning_rate": 6.893939393939395e-06, "loss": 0.0346, "step": 26825 }, { "epoch": 13.12, "learning_rate": 6.881720430107528e-06, "loss": 0.0309, "step": 26850 }, { "epoch": 13.14, "learning_rate": 6.86950146627566e-06, "loss": 0.0306, "step": 26875 }, { "epoch": 13.15, "learning_rate": 6.8572825024437926e-06, "loss": 0.0311, "step": 26900 }, { "epoch": 13.16, "learning_rate": 6.845063538611926e-06, "loss": 0.0334, "step": 26925 }, { "epoch": 13.17, "learning_rate": 6.832844574780059e-06, "loss": 0.0331, "step": 26950 }, { "epoch": 13.18, "learning_rate": 6.8206256109481915e-06, "loss": 0.0408, "step": 26975 }, { "epoch": 13.2, "learning_rate": 6.808406647116325e-06, "loss": 0.0338, "step": 27000 }, { "epoch": 13.21, "learning_rate": 6.796187683284458e-06, "loss": 0.0267, "step": 27025 }, { "epoch": 13.22, "learning_rate": 6.78396871945259e-06, "loss": 0.0303, "step": 27050 }, { "epoch": 13.23, "learning_rate": 6.771749755620724e-06, "loss": 0.0434, "step": 27075 }, { "epoch": 13.25, "learning_rate": 6.759530791788856e-06, "loss": 0.029, "step": 27100 }, { "epoch": 13.26, "learning_rate": 6.747311827956989e-06, "loss": 0.032, "step": 27125 }, { "epoch": 13.27, "learning_rate": 6.7350928641251225e-06, "loss": 0.0319, "step": 27150 }, { "epoch": 13.28, "learning_rate": 6.722873900293255e-06, "loss": 0.0304, "step": 27175 }, { "epoch": 13.29, "learning_rate": 6.710654936461388e-06, "loss": 0.0315, "step": 27200 }, { "epoch": 13.31, "learning_rate": 6.698435972629521e-06, "loss": 0.0414, "step": 27225 }, { "epoch": 13.32, "learning_rate": 6.686217008797654e-06, "loss": 0.0367, "step": 27250 }, { "epoch": 13.33, "learning_rate": 6.673998044965787e-06, "loss": 0.0297, "step": 27275 }, { "epoch": 13.34, "learning_rate": 6.66177908113392e-06, "loss": 0.0301, "step": 27300 }, { "epoch": 13.36, "learning_rate": 6.649560117302053e-06, "loss": 0.033, "step": 27325 }, { "epoch": 13.37, "learning_rate": 6.637341153470186e-06, "loss": 0.0324, "step": 27350 }, { "epoch": 13.38, "learning_rate": 6.625122189638319e-06, "loss": 0.0251, "step": 27375 }, { "epoch": 13.39, "learning_rate": 6.612903225806452e-06, "loss": 0.035, "step": 27400 }, { "epoch": 13.4, "learning_rate": 6.600684261974585e-06, "loss": 0.0311, "step": 27425 }, { "epoch": 13.42, "learning_rate": 6.588465298142718e-06, "loss": 0.0347, "step": 27450 }, { "epoch": 13.43, "learning_rate": 6.5762463343108505e-06, "loss": 0.0305, "step": 27475 }, { "epoch": 13.44, "learning_rate": 6.564027370478984e-06, "loss": 0.0298, "step": 27500 }, { "epoch": 13.45, "learning_rate": 6.551808406647117e-06, "loss": 0.0313, "step": 27525 }, { "epoch": 13.47, "learning_rate": 6.539589442815249e-06, "loss": 0.0454, "step": 27550 }, { "epoch": 13.48, "learning_rate": 6.527370478983383e-06, "loss": 0.029, "step": 27575 }, { "epoch": 13.49, "learning_rate": 6.515151515151516e-06, "loss": 0.0318, "step": 27600 }, { "epoch": 13.5, "learning_rate": 6.502932551319648e-06, "loss": 0.0381, "step": 27625 }, { "epoch": 13.51, "learning_rate": 6.4907135874877815e-06, "loss": 0.03, "step": 27650 }, { "epoch": 13.53, "learning_rate": 6.478494623655914e-06, "loss": 0.0414, "step": 27675 }, { "epoch": 13.54, "learning_rate": 6.466275659824047e-06, "loss": 0.035, "step": 27700 }, { "epoch": 13.55, "learning_rate": 6.45405669599218e-06, "loss": 0.0411, "step": 27725 }, { "epoch": 13.56, "learning_rate": 6.441837732160313e-06, "loss": 0.0394, "step": 27750 }, { "epoch": 13.58, "learning_rate": 6.429618768328446e-06, "loss": 0.0287, "step": 27775 }, { "epoch": 13.59, "learning_rate": 6.417399804496579e-06, "loss": 0.029, "step": 27800 }, { "epoch": 13.6, "learning_rate": 6.405180840664712e-06, "loss": 0.0396, "step": 27825 }, { "epoch": 13.61, "learning_rate": 6.392961876832845e-06, "loss": 0.0378, "step": 27850 }, { "epoch": 13.62, "learning_rate": 6.380742913000978e-06, "loss": 0.0365, "step": 27875 }, { "epoch": 13.64, "learning_rate": 6.368523949169111e-06, "loss": 0.0385, "step": 27900 }, { "epoch": 13.65, "learning_rate": 6.356304985337244e-06, "loss": 0.04, "step": 27925 }, { "epoch": 13.66, "learning_rate": 6.344086021505377e-06, "loss": 0.0302, "step": 27950 }, { "epoch": 13.67, "learning_rate": 6.332355816226784e-06, "loss": 0.0293, "step": 27975 }, { "epoch": 13.69, "learning_rate": 6.3201368523949165e-06, "loss": 0.0422, "step": 28000 }, { "epoch": 13.7, "learning_rate": 6.30791788856305e-06, "loss": 0.035, "step": 28025 }, { "epoch": 13.71, "learning_rate": 6.295698924731183e-06, "loss": 0.0323, "step": 28050 }, { "epoch": 13.72, "learning_rate": 6.283479960899315e-06, "loss": 0.0275, "step": 28075 }, { "epoch": 13.73, "learning_rate": 6.271260997067449e-06, "loss": 0.0374, "step": 28100 }, { "epoch": 13.75, "learning_rate": 6.259042033235582e-06, "loss": 0.034, "step": 28125 }, { "epoch": 13.76, "learning_rate": 6.246823069403714e-06, "loss": 0.0326, "step": 28150 }, { "epoch": 13.77, "learning_rate": 6.2346041055718476e-06, "loss": 0.0338, "step": 28175 }, { "epoch": 13.78, "learning_rate": 6.222385141739981e-06, "loss": 0.031, "step": 28200 }, { "epoch": 13.8, "learning_rate": 6.210166177908113e-06, "loss": 0.0345, "step": 28225 }, { "epoch": 13.81, "learning_rate": 6.1979472140762465e-06, "loss": 0.0375, "step": 28250 }, { "epoch": 13.82, "learning_rate": 6.18572825024438e-06, "loss": 0.0296, "step": 28275 }, { "epoch": 13.83, "learning_rate": 6.173509286412512e-06, "loss": 0.0329, "step": 28300 }, { "epoch": 13.84, "learning_rate": 6.161290322580645e-06, "loss": 0.0301, "step": 28325 }, { "epoch": 13.86, "learning_rate": 6.149071358748779e-06, "loss": 0.034, "step": 28350 }, { "epoch": 13.87, "learning_rate": 6.136852394916911e-06, "loss": 0.034, "step": 28375 }, { "epoch": 13.88, "learning_rate": 6.124633431085044e-06, "loss": 0.0303, "step": 28400 }, { "epoch": 13.89, "learning_rate": 6.1124144672531775e-06, "loss": 0.0304, "step": 28425 }, { "epoch": 13.91, "learning_rate": 6.10019550342131e-06, "loss": 0.0281, "step": 28450 }, { "epoch": 13.92, "learning_rate": 6.087976539589443e-06, "loss": 0.0342, "step": 28475 }, { "epoch": 13.93, "learning_rate": 6.0757575757575755e-06, "loss": 0.0438, "step": 28500 }, { "epoch": 13.94, "learning_rate": 6.063538611925709e-06, "loss": 0.0241, "step": 28525 }, { "epoch": 13.95, "learning_rate": 6.051319648093842e-06, "loss": 0.0268, "step": 28550 }, { "epoch": 13.97, "learning_rate": 6.0391006842619744e-06, "loss": 0.0462, "step": 28575 }, { "epoch": 13.98, "learning_rate": 6.026881720430108e-06, "loss": 0.0292, "step": 28600 }, { "epoch": 13.99, "learning_rate": 6.014662756598241e-06, "loss": 0.0365, "step": 28625 }, { "epoch": 14.0, "learning_rate": 6.002443792766373e-06, "loss": 0.0397, "step": 28650 }, { "epoch": 14.02, "learning_rate": 5.990224828934507e-06, "loss": 0.0296, "step": 28675 }, { "epoch": 14.03, "learning_rate": 5.97800586510264e-06, "loss": 0.0288, "step": 28700 }, { "epoch": 14.04, "learning_rate": 5.965786901270772e-06, "loss": 0.0268, "step": 28725 }, { "epoch": 14.05, "learning_rate": 5.9535679374389055e-06, "loss": 0.0244, "step": 28750 }, { "epoch": 14.06, "learning_rate": 5.941348973607039e-06, "loss": 0.0288, "step": 28775 }, { "epoch": 14.08, "learning_rate": 5.929130009775171e-06, "loss": 0.0301, "step": 28800 }, { "epoch": 14.09, "learning_rate": 5.916911045943304e-06, "loss": 0.0244, "step": 28825 }, { "epoch": 14.1, "learning_rate": 5.904692082111438e-06, "loss": 0.0289, "step": 28850 }, { "epoch": 14.11, "learning_rate": 5.89247311827957e-06, "loss": 0.0195, "step": 28875 }, { "epoch": 14.13, "learning_rate": 5.880254154447703e-06, "loss": 0.0257, "step": 28900 }, { "epoch": 14.14, "learning_rate": 5.8680351906158365e-06, "loss": 0.0256, "step": 28925 }, { "epoch": 14.15, "learning_rate": 5.855816226783969e-06, "loss": 0.0317, "step": 28950 }, { "epoch": 14.16, "learning_rate": 5.843597262952102e-06, "loss": 0.0346, "step": 28975 }, { "epoch": 14.17, "learning_rate": 5.831378299120235e-06, "loss": 0.0306, "step": 29000 }, { "epoch": 14.19, "learning_rate": 5.819159335288368e-06, "loss": 0.0262, "step": 29025 }, { "epoch": 14.2, "learning_rate": 5.806940371456501e-06, "loss": 0.0255, "step": 29050 }, { "epoch": 14.21, "learning_rate": 5.7947214076246335e-06, "loss": 0.027, "step": 29075 }, { "epoch": 14.22, "learning_rate": 5.782502443792767e-06, "loss": 0.0419, "step": 29100 }, { "epoch": 14.24, "learning_rate": 5.7702834799609e-06, "loss": 0.0211, "step": 29125 }, { "epoch": 14.25, "learning_rate": 5.758064516129032e-06, "loss": 0.0332, "step": 29150 }, { "epoch": 14.26, "learning_rate": 5.745845552297166e-06, "loss": 0.0282, "step": 29175 }, { "epoch": 14.27, "learning_rate": 5.733626588465299e-06, "loss": 0.0274, "step": 29200 }, { "epoch": 14.28, "learning_rate": 5.721407624633431e-06, "loss": 0.0271, "step": 29225 }, { "epoch": 14.3, "learning_rate": 5.7091886608015645e-06, "loss": 0.0276, "step": 29250 }, { "epoch": 14.31, "learning_rate": 5.696969696969698e-06, "loss": 0.0331, "step": 29275 }, { "epoch": 14.32, "learning_rate": 5.68475073313783e-06, "loss": 0.0321, "step": 29300 }, { "epoch": 14.33, "learning_rate": 5.672531769305963e-06, "loss": 0.0253, "step": 29325 }, { "epoch": 14.35, "learning_rate": 5.660312805474097e-06, "loss": 0.0265, "step": 29350 }, { "epoch": 14.36, "learning_rate": 5.648093841642229e-06, "loss": 0.0305, "step": 29375 }, { "epoch": 14.37, "learning_rate": 5.635874877810362e-06, "loss": 0.0266, "step": 29400 }, { "epoch": 14.38, "learning_rate": 5.6236559139784955e-06, "loss": 0.0336, "step": 29425 }, { "epoch": 14.39, "learning_rate": 5.611436950146628e-06, "loss": 0.0267, "step": 29450 }, { "epoch": 14.41, "learning_rate": 5.599217986314761e-06, "loss": 0.0252, "step": 29475 }, { "epoch": 14.42, "learning_rate": 5.586999022482894e-06, "loss": 0.0246, "step": 29500 }, { "epoch": 14.43, "learning_rate": 5.574780058651027e-06, "loss": 0.0302, "step": 29525 }, { "epoch": 14.44, "learning_rate": 5.56256109481916e-06, "loss": 0.0255, "step": 29550 }, { "epoch": 14.46, "learning_rate": 5.550342130987293e-06, "loss": 0.0313, "step": 29575 }, { "epoch": 14.47, "learning_rate": 5.538123167155426e-06, "loss": 0.0289, "step": 29600 }, { "epoch": 14.48, "learning_rate": 5.525904203323559e-06, "loss": 0.0266, "step": 29625 }, { "epoch": 14.49, "learning_rate": 5.513685239491691e-06, "loss": 0.0326, "step": 29650 }, { "epoch": 14.5, "learning_rate": 5.501466275659825e-06, "loss": 0.0298, "step": 29675 }, { "epoch": 14.52, "learning_rate": 5.489247311827958e-06, "loss": 0.0218, "step": 29700 }, { "epoch": 14.53, "learning_rate": 5.47702834799609e-06, "loss": 0.0281, "step": 29725 }, { "epoch": 14.54, "learning_rate": 5.4648093841642235e-06, "loss": 0.0258, "step": 29750 }, { "epoch": 14.55, "learning_rate": 5.452590420332357e-06, "loss": 0.0259, "step": 29775 }, { "epoch": 14.57, "learning_rate": 5.440371456500489e-06, "loss": 0.0255, "step": 29800 }, { "epoch": 14.58, "learning_rate": 5.428152492668622e-06, "loss": 0.0321, "step": 29825 }, { "epoch": 14.59, "learning_rate": 5.415933528836756e-06, "loss": 0.0337, "step": 29850 }, { "epoch": 14.6, "learning_rate": 5.403714565004888e-06, "loss": 0.032, "step": 29875 }, { "epoch": 14.61, "learning_rate": 5.391495601173021e-06, "loss": 0.0374, "step": 29900 }, { "epoch": 14.63, "learning_rate": 5.3792766373411545e-06, "loss": 0.0285, "step": 29925 }, { "epoch": 14.64, "learning_rate": 5.367057673509287e-06, "loss": 0.0283, "step": 29950 }, { "epoch": 14.65, "learning_rate": 5.35483870967742e-06, "loss": 0.0249, "step": 29975 }, { "epoch": 14.66, "learning_rate": 5.3426197458455534e-06, "loss": 0.0305, "step": 30000 }, { "epoch": 14.67, "learning_rate": 5.330400782013686e-06, "loss": 0.035, "step": 30025 }, { "epoch": 14.69, "learning_rate": 5.318181818181819e-06, "loss": 0.0392, "step": 30050 }, { "epoch": 14.7, "learning_rate": 5.305962854349952e-06, "loss": 0.0269, "step": 30075 }, { "epoch": 14.71, "learning_rate": 5.294232649071359e-06, "loss": 0.0258, "step": 30100 }, { "epoch": 14.72, "learning_rate": 5.282013685239492e-06, "loss": 0.0358, "step": 30125 }, { "epoch": 14.74, "learning_rate": 5.269794721407625e-06, "loss": 0.0332, "step": 30150 }, { "epoch": 14.75, "learning_rate": 5.257575757575758e-06, "loss": 0.0273, "step": 30175 }, { "epoch": 14.76, "learning_rate": 5.245356793743891e-06, "loss": 0.0275, "step": 30200 }, { "epoch": 14.77, "learning_rate": 5.233137829912024e-06, "loss": 0.0282, "step": 30225 }, { "epoch": 14.78, "learning_rate": 5.220918866080157e-06, "loss": 0.0232, "step": 30250 }, { "epoch": 14.8, "learning_rate": 5.2086999022482896e-06, "loss": 0.0323, "step": 30275 }, { "epoch": 14.81, "learning_rate": 5.196480938416423e-06, "loss": 0.0354, "step": 30300 }, { "epoch": 14.82, "learning_rate": 5.184261974584556e-06, "loss": 0.0288, "step": 30325 }, { "epoch": 14.83, "learning_rate": 5.1720430107526885e-06, "loss": 0.032, "step": 30350 }, { "epoch": 14.85, "learning_rate": 5.159824046920822e-06, "loss": 0.0333, "step": 30375 }, { "epoch": 14.86, "learning_rate": 5.147605083088955e-06, "loss": 0.0299, "step": 30400 }, { "epoch": 14.87, "learning_rate": 5.135386119257087e-06, "loss": 0.0395, "step": 30425 }, { "epoch": 14.88, "learning_rate": 5.123167155425221e-06, "loss": 0.0292, "step": 30450 }, { "epoch": 14.89, "learning_rate": 5.110948191593354e-06, "loss": 0.0283, "step": 30475 }, { "epoch": 14.91, "learning_rate": 5.098729227761486e-06, "loss": 0.0315, "step": 30500 }, { "epoch": 14.92, "learning_rate": 5.0865102639296195e-06, "loss": 0.028, "step": 30525 }, { "epoch": 14.93, "learning_rate": 5.074291300097752e-06, "loss": 0.0343, "step": 30550 }, { "epoch": 14.94, "learning_rate": 5.062072336265885e-06, "loss": 0.0243, "step": 30575 }, { "epoch": 14.96, "learning_rate": 5.049853372434018e-06, "loss": 0.0299, "step": 30600 }, { "epoch": 14.97, "learning_rate": 5.037634408602151e-06, "loss": 0.0295, "step": 30625 }, { "epoch": 14.98, "learning_rate": 5.025415444770284e-06, "loss": 0.0278, "step": 30650 }, { "epoch": 14.99, "learning_rate": 5.013196480938417e-06, "loss": 0.038, "step": 30675 }, { "epoch": 15.0, "learning_rate": 5.00097751710655e-06, "loss": 0.038, "step": 30700 }, { "epoch": 15.02, "learning_rate": 4.988758553274683e-06, "loss": 0.0213, "step": 30725 }, { "epoch": 15.03, "learning_rate": 4.976539589442816e-06, "loss": 0.0222, "step": 30750 }, { "epoch": 15.04, "learning_rate": 4.9643206256109486e-06, "loss": 0.0252, "step": 30775 }, { "epoch": 15.05, "learning_rate": 4.952101661779082e-06, "loss": 0.026, "step": 30800 }, { "epoch": 15.07, "learning_rate": 4.939882697947215e-06, "loss": 0.0261, "step": 30825 }, { "epoch": 15.08, "learning_rate": 4.9276637341153475e-06, "loss": 0.0296, "step": 30850 }, { "epoch": 15.09, "learning_rate": 4.91544477028348e-06, "loss": 0.0262, "step": 30875 }, { "epoch": 15.1, "learning_rate": 4.903225806451613e-06, "loss": 0.0173, "step": 30900 }, { "epoch": 15.11, "learning_rate": 4.891006842619746e-06, "loss": 0.0303, "step": 30925 }, { "epoch": 15.13, "learning_rate": 4.878787878787879e-06, "loss": 0.0275, "step": 30950 }, { "epoch": 15.14, "learning_rate": 4.866568914956012e-06, "loss": 0.0257, "step": 30975 }, { "epoch": 15.15, "learning_rate": 4.854349951124145e-06, "loss": 0.0221, "step": 31000 }, { "epoch": 15.16, "learning_rate": 4.842130987292278e-06, "loss": 0.0224, "step": 31025 }, { "epoch": 15.18, "learning_rate": 4.829912023460411e-06, "loss": 0.0215, "step": 31050 }, { "epoch": 15.19, "learning_rate": 4.817693059628543e-06, "loss": 0.0303, "step": 31075 }, { "epoch": 15.2, "learning_rate": 4.8054740957966766e-06, "loss": 0.0222, "step": 31100 }, { "epoch": 15.21, "learning_rate": 4.79325513196481e-06, "loss": 0.0268, "step": 31125 }, { "epoch": 15.22, "learning_rate": 4.781036168132942e-06, "loss": 0.0276, "step": 31150 }, { "epoch": 15.24, "learning_rate": 4.7688172043010755e-06, "loss": 0.0245, "step": 31175 }, { "epoch": 15.25, "learning_rate": 4.757086999022483e-06, "loss": 0.0185, "step": 31200 }, { "epoch": 15.26, "learning_rate": 4.744868035190617e-06, "loss": 0.0214, "step": 31225 }, { "epoch": 15.27, "learning_rate": 4.732649071358749e-06, "loss": 0.0236, "step": 31250 }, { "epoch": 15.29, "learning_rate": 4.720430107526882e-06, "loss": 0.0289, "step": 31275 }, { "epoch": 15.3, "learning_rate": 4.7082111436950155e-06, "loss": 0.0335, "step": 31300 }, { "epoch": 15.31, "learning_rate": 4.695992179863148e-06, "loss": 0.0229, "step": 31325 }, { "epoch": 15.32, "learning_rate": 4.683773216031281e-06, "loss": 0.0237, "step": 31350 }, { "epoch": 15.33, "learning_rate": 4.6715542521994135e-06, "loss": 0.0186, "step": 31375 }, { "epoch": 15.35, "learning_rate": 4.659335288367547e-06, "loss": 0.0219, "step": 31400 }, { "epoch": 15.36, "learning_rate": 4.64711632453568e-06, "loss": 0.0292, "step": 31425 }, { "epoch": 15.37, "learning_rate": 4.634897360703812e-06, "loss": 0.0261, "step": 31450 }, { "epoch": 15.38, "learning_rate": 4.622678396871946e-06, "loss": 0.0301, "step": 31475 }, { "epoch": 15.4, "learning_rate": 4.610459433040079e-06, "loss": 0.0243, "step": 31500 }, { "epoch": 15.41, "learning_rate": 4.598240469208211e-06, "loss": 0.034, "step": 31525 }, { "epoch": 15.42, "learning_rate": 4.5860215053763446e-06, "loss": 0.0253, "step": 31550 }, { "epoch": 15.43, "learning_rate": 4.573802541544478e-06, "loss": 0.0265, "step": 31575 }, { "epoch": 15.44, "learning_rate": 4.56158357771261e-06, "loss": 0.0178, "step": 31600 }, { "epoch": 15.46, "learning_rate": 4.5493646138807435e-06, "loss": 0.0208, "step": 31625 }, { "epoch": 15.47, "learning_rate": 4.537145650048877e-06, "loss": 0.0264, "step": 31650 }, { "epoch": 15.48, "learning_rate": 4.524926686217009e-06, "loss": 0.0266, "step": 31675 }, { "epoch": 15.49, "learning_rate": 4.512707722385142e-06, "loss": 0.0267, "step": 31700 }, { "epoch": 15.51, "learning_rate": 4.500488758553276e-06, "loss": 0.0218, "step": 31725 }, { "epoch": 15.52, "learning_rate": 4.488269794721408e-06, "loss": 0.0252, "step": 31750 }, { "epoch": 15.53, "learning_rate": 4.476050830889541e-06, "loss": 0.023, "step": 31775 }, { "epoch": 15.54, "learning_rate": 4.4638318670576745e-06, "loss": 0.0216, "step": 31800 }, { "epoch": 15.55, "learning_rate": 4.451612903225807e-06, "loss": 0.0278, "step": 31825 }, { "epoch": 15.57, "learning_rate": 4.43939393939394e-06, "loss": 0.0224, "step": 31850 }, { "epoch": 15.58, "learning_rate": 4.4271749755620725e-06, "loss": 0.0302, "step": 31875 }, { "epoch": 15.59, "learning_rate": 4.414956011730206e-06, "loss": 0.0213, "step": 31900 }, { "epoch": 15.6, "learning_rate": 4.402737047898338e-06, "loss": 0.0243, "step": 31925 }, { "epoch": 15.62, "learning_rate": 4.3905180840664714e-06, "loss": 0.0263, "step": 31950 }, { "epoch": 15.63, "learning_rate": 4.378299120234604e-06, "loss": 0.018, "step": 31975 }, { "epoch": 15.64, "learning_rate": 4.366080156402737e-06, "loss": 0.0155, "step": 32000 }, { "epoch": 15.65, "learning_rate": 4.35386119257087e-06, "loss": 0.0368, "step": 32025 }, { "epoch": 15.66, "learning_rate": 4.341642228739003e-06, "loss": 0.0271, "step": 32050 }, { "epoch": 15.68, "learning_rate": 4.329423264907136e-06, "loss": 0.0309, "step": 32075 }, { "epoch": 15.69, "learning_rate": 4.317204301075269e-06, "loss": 0.0252, "step": 32100 }, { "epoch": 15.7, "learning_rate": 4.304985337243402e-06, "loss": 0.0353, "step": 32125 }, { "epoch": 15.71, "learning_rate": 4.292766373411535e-06, "loss": 0.0246, "step": 32150 }, { "epoch": 15.73, "learning_rate": 4.280547409579668e-06, "loss": 0.0315, "step": 32175 }, { "epoch": 15.74, "learning_rate": 4.2683284457478005e-06, "loss": 0.0209, "step": 32200 }, { "epoch": 15.75, "learning_rate": 4.256109481915934e-06, "loss": 0.0209, "step": 32225 }, { "epoch": 15.76, "learning_rate": 4.243890518084067e-06, "loss": 0.0264, "step": 32250 }, { "epoch": 15.77, "learning_rate": 4.231671554252199e-06, "loss": 0.0232, "step": 32275 }, { "epoch": 15.79, "learning_rate": 4.219452590420333e-06, "loss": 0.0259, "step": 32300 }, { "epoch": 15.8, "learning_rate": 4.207233626588466e-06, "loss": 0.0284, "step": 32325 }, { "epoch": 15.81, "learning_rate": 4.195014662756598e-06, "loss": 0.0319, "step": 32350 }, { "epoch": 15.82, "learning_rate": 4.1827956989247316e-06, "loss": 0.0208, "step": 32375 }, { "epoch": 15.84, "learning_rate": 4.170576735092865e-06, "loss": 0.0265, "step": 32400 }, { "epoch": 15.85, "learning_rate": 4.158357771260997e-06, "loss": 0.0241, "step": 32425 }, { "epoch": 15.86, "learning_rate": 4.1461388074291305e-06, "loss": 0.023, "step": 32450 }, { "epoch": 15.87, "learning_rate": 4.133919843597264e-06, "loss": 0.0263, "step": 32475 }, { "epoch": 15.88, "learning_rate": 4.121700879765396e-06, "loss": 0.0181, "step": 32500 }, { "epoch": 15.9, "learning_rate": 4.109481915933529e-06, "loss": 0.0224, "step": 32525 }, { "epoch": 15.91, "learning_rate": 4.097262952101662e-06, "loss": 0.0259, "step": 32550 }, { "epoch": 15.92, "learning_rate": 4.085043988269795e-06, "loss": 0.0242, "step": 32575 }, { "epoch": 15.93, "learning_rate": 4.072825024437928e-06, "loss": 0.0293, "step": 32600 }, { "epoch": 15.95, "learning_rate": 4.060606060606061e-06, "loss": 0.0252, "step": 32625 }, { "epoch": 15.96, "learning_rate": 4.048387096774194e-06, "loss": 0.0234, "step": 32650 }, { "epoch": 15.97, "learning_rate": 4.036168132942327e-06, "loss": 0.0224, "step": 32675 }, { "epoch": 15.98, "learning_rate": 4.0239491691104595e-06, "loss": 0.025, "step": 32700 }, { "epoch": 15.99, "learning_rate": 4.011730205278593e-06, "loss": 0.0226, "step": 32725 }, { "epoch": 16.01, "learning_rate": 3.999511241446726e-06, "loss": 0.0239, "step": 32750 }, { "epoch": 16.02, "learning_rate": 3.9872922776148584e-06, "loss": 0.0246, "step": 32775 }, { "epoch": 16.03, "learning_rate": 3.975073313782992e-06, "loss": 0.0157, "step": 32800 }, { "epoch": 16.04, "learning_rate": 3.962854349951125e-06, "loss": 0.022, "step": 32825 }, { "epoch": 16.06, "learning_rate": 3.950635386119257e-06, "loss": 0.0236, "step": 32850 }, { "epoch": 16.07, "learning_rate": 3.9384164222873906e-06, "loss": 0.023, "step": 32875 }, { "epoch": 16.08, "learning_rate": 3.926197458455524e-06, "loss": 0.034, "step": 32900 }, { "epoch": 16.09, "learning_rate": 3.913978494623656e-06, "loss": 0.0249, "step": 32925 }, { "epoch": 16.1, "learning_rate": 3.9017595307917895e-06, "loss": 0.0224, "step": 32950 }, { "epoch": 16.12, "learning_rate": 3.889540566959923e-06, "loss": 0.0261, "step": 32975 }, { "epoch": 16.13, "learning_rate": 3.877321603128055e-06, "loss": 0.0153, "step": 33000 }, { "epoch": 16.14, "learning_rate": 3.865102639296188e-06, "loss": 0.0267, "step": 33025 }, { "epoch": 16.15, "learning_rate": 3.852883675464321e-06, "loss": 0.0158, "step": 33050 }, { "epoch": 16.17, "learning_rate": 3.840664711632454e-06, "loss": 0.02, "step": 33075 }, { "epoch": 16.18, "learning_rate": 3.828445747800587e-06, "loss": 0.0227, "step": 33100 }, { "epoch": 16.19, "learning_rate": 3.81622678396872e-06, "loss": 0.0202, "step": 33125 }, { "epoch": 16.2, "learning_rate": 3.8040078201368525e-06, "loss": 0.024, "step": 33150 }, { "epoch": 16.21, "learning_rate": 3.7917888563049853e-06, "loss": 0.0255, "step": 33175 }, { "epoch": 16.23, "learning_rate": 3.7795698924731186e-06, "loss": 0.0202, "step": 33200 }, { "epoch": 16.24, "learning_rate": 3.7673509286412514e-06, "loss": 0.0299, "step": 33225 }, { "epoch": 16.25, "learning_rate": 3.755131964809384e-06, "loss": 0.0266, "step": 33250 }, { "epoch": 16.26, "learning_rate": 3.742913000977517e-06, "loss": 0.0232, "step": 33275 }, { "epoch": 16.28, "learning_rate": 3.7306940371456503e-06, "loss": 0.0193, "step": 33300 }, { "epoch": 16.29, "learning_rate": 3.718475073313783e-06, "loss": 0.0192, "step": 33325 }, { "epoch": 16.3, "learning_rate": 3.706256109481916e-06, "loss": 0.0195, "step": 33350 }, { "epoch": 16.31, "learning_rate": 3.694037145650049e-06, "loss": 0.0182, "step": 33375 }, { "epoch": 16.32, "learning_rate": 3.6823069403714566e-06, "loss": 0.0261, "step": 33400 }, { "epoch": 16.34, "learning_rate": 3.67008797653959e-06, "loss": 0.0219, "step": 33425 }, { "epoch": 16.35, "learning_rate": 3.6578690127077227e-06, "loss": 0.0293, "step": 33450 }, { "epoch": 16.36, "learning_rate": 3.6456500488758555e-06, "loss": 0.0318, "step": 33475 }, { "epoch": 16.37, "learning_rate": 3.6334310850439888e-06, "loss": 0.0198, "step": 33500 }, { "epoch": 16.39, "learning_rate": 3.6212121212121216e-06, "loss": 0.0257, "step": 33525 }, { "epoch": 16.4, "learning_rate": 3.6089931573802544e-06, "loss": 0.0192, "step": 33550 }, { "epoch": 16.41, "learning_rate": 3.5967741935483872e-06, "loss": 0.0204, "step": 33575 }, { "epoch": 16.42, "learning_rate": 3.5845552297165205e-06, "loss": 0.0221, "step": 33600 }, { "epoch": 16.43, "learning_rate": 3.5723362658846533e-06, "loss": 0.0229, "step": 33625 }, { "epoch": 16.45, "learning_rate": 3.560117302052786e-06, "loss": 0.0234, "step": 33650 }, { "epoch": 16.46, "learning_rate": 3.5478983382209194e-06, "loss": 0.0233, "step": 33675 }, { "epoch": 16.47, "learning_rate": 3.535679374389052e-06, "loss": 0.0248, "step": 33700 }, { "epoch": 16.48, "learning_rate": 3.523460410557185e-06, "loss": 0.0251, "step": 33725 }, { "epoch": 16.5, "learning_rate": 3.5112414467253183e-06, "loss": 0.0297, "step": 33750 }, { "epoch": 16.51, "learning_rate": 3.499022482893451e-06, "loss": 0.0256, "step": 33775 }, { "epoch": 16.52, "learning_rate": 3.486803519061584e-06, "loss": 0.0194, "step": 33800 }, { "epoch": 16.53, "learning_rate": 3.4745845552297167e-06, "loss": 0.026, "step": 33825 }, { "epoch": 16.54, "learning_rate": 3.46236559139785e-06, "loss": 0.0165, "step": 33850 }, { "epoch": 16.56, "learning_rate": 3.450146627565983e-06, "loss": 0.021, "step": 33875 }, { "epoch": 16.57, "learning_rate": 3.4379276637341156e-06, "loss": 0.0226, "step": 33900 }, { "epoch": 16.58, "learning_rate": 3.425708699902249e-06, "loss": 0.0273, "step": 33925 }, { "epoch": 16.59, "learning_rate": 3.4134897360703817e-06, "loss": 0.0209, "step": 33950 }, { "epoch": 16.61, "learning_rate": 3.4012707722385145e-06, "loss": 0.0228, "step": 33975 }, { "epoch": 16.62, "learning_rate": 3.3890518084066478e-06, "loss": 0.026, "step": 34000 }, { "epoch": 16.63, "learning_rate": 3.3768328445747806e-06, "loss": 0.0192, "step": 34025 }, { "epoch": 16.64, "learning_rate": 3.3646138807429134e-06, "loss": 0.0226, "step": 34050 }, { "epoch": 16.65, "learning_rate": 3.3523949169110463e-06, "loss": 0.021, "step": 34075 }, { "epoch": 16.67, "learning_rate": 3.3401759530791795e-06, "loss": 0.0185, "step": 34100 }, { "epoch": 16.68, "learning_rate": 3.3279569892473123e-06, "loss": 0.0279, "step": 34125 }, { "epoch": 16.69, "learning_rate": 3.315738025415445e-06, "loss": 0.0306, "step": 34150 }, { "epoch": 16.7, "learning_rate": 3.3035190615835775e-06, "loss": 0.033, "step": 34175 }, { "epoch": 16.72, "learning_rate": 3.291300097751711e-06, "loss": 0.0196, "step": 34200 }, { "epoch": 16.73, "learning_rate": 3.2790811339198436e-06, "loss": 0.0293, "step": 34225 }, { "epoch": 16.74, "learning_rate": 3.2668621700879764e-06, "loss": 0.0224, "step": 34250 }, { "epoch": 16.75, "learning_rate": 3.2546432062561097e-06, "loss": 0.0271, "step": 34275 }, { "epoch": 16.76, "learning_rate": 3.2424242424242425e-06, "loss": 0.0323, "step": 34300 }, { "epoch": 16.78, "learning_rate": 3.2302052785923753e-06, "loss": 0.0242, "step": 34325 }, { "epoch": 16.79, "learning_rate": 3.2179863147605086e-06, "loss": 0.0176, "step": 34350 }, { "epoch": 16.8, "learning_rate": 3.2057673509286414e-06, "loss": 0.0229, "step": 34375 }, { "epoch": 16.81, "learning_rate": 3.1935483870967742e-06, "loss": 0.022, "step": 34400 }, { "epoch": 16.83, "learning_rate": 3.181329423264907e-06, "loss": 0.0181, "step": 34425 }, { "epoch": 16.84, "learning_rate": 3.1691104594330403e-06, "loss": 0.0219, "step": 34450 }, { "epoch": 16.85, "learning_rate": 3.156891495601173e-06, "loss": 0.0192, "step": 34475 }, { "epoch": 16.86, "learning_rate": 3.144672531769306e-06, "loss": 0.0228, "step": 34500 }, { "epoch": 16.87, "learning_rate": 3.132453567937439e-06, "loss": 0.026, "step": 34525 }, { "epoch": 16.89, "learning_rate": 3.120234604105572e-06, "loss": 0.0213, "step": 34550 }, { "epoch": 16.9, "learning_rate": 3.108015640273705e-06, "loss": 0.0219, "step": 34575 }, { "epoch": 16.91, "learning_rate": 3.095796676441838e-06, "loss": 0.0241, "step": 34600 }, { "epoch": 16.92, "learning_rate": 3.083577712609971e-06, "loss": 0.0218, "step": 34625 }, { "epoch": 16.94, "learning_rate": 3.0713587487781037e-06, "loss": 0.0175, "step": 34650 }, { "epoch": 16.95, "learning_rate": 3.0591397849462366e-06, "loss": 0.0182, "step": 34675 }, { "epoch": 16.96, "learning_rate": 3.04692082111437e-06, "loss": 0.0235, "step": 34700 }, { "epoch": 16.97, "learning_rate": 3.0347018572825026e-06, "loss": 0.0176, "step": 34725 }, { "epoch": 16.98, "learning_rate": 3.0224828934506355e-06, "loss": 0.0187, "step": 34750 }, { "epoch": 17.0, "learning_rate": 3.0102639296187687e-06, "loss": 0.0206, "step": 34775 }, { "epoch": 17.01, "learning_rate": 2.9980449657869015e-06, "loss": 0.0201, "step": 34800 }, { "epoch": 17.02, "learning_rate": 2.9858260019550344e-06, "loss": 0.0172, "step": 34825 }, { "epoch": 17.03, "learning_rate": 2.9736070381231676e-06, "loss": 0.0219, "step": 34850 }, { "epoch": 17.05, "learning_rate": 2.9613880742913004e-06, "loss": 0.0164, "step": 34875 }, { "epoch": 17.06, "learning_rate": 2.9491691104594332e-06, "loss": 0.0212, "step": 34900 }, { "epoch": 17.07, "learning_rate": 2.936950146627566e-06, "loss": 0.0187, "step": 34925 }, { "epoch": 17.08, "learning_rate": 2.9247311827956993e-06, "loss": 0.0248, "step": 34950 }, { "epoch": 17.09, "learning_rate": 2.912512218963832e-06, "loss": 0.0203, "step": 34975 }, { "epoch": 17.11, "learning_rate": 2.900293255131965e-06, "loss": 0.0255, "step": 35000 }, { "epoch": 17.12, "learning_rate": 2.8880742913000982e-06, "loss": 0.0222, "step": 35025 }, { "epoch": 17.13, "learning_rate": 2.875855327468231e-06, "loss": 0.0234, "step": 35050 }, { "epoch": 17.14, "learning_rate": 2.863636363636364e-06, "loss": 0.0234, "step": 35075 }, { "epoch": 17.16, "learning_rate": 2.851417399804497e-06, "loss": 0.0174, "step": 35100 }, { "epoch": 17.17, "learning_rate": 2.83919843597263e-06, "loss": 0.0167, "step": 35125 }, { "epoch": 17.18, "learning_rate": 2.8269794721407628e-06, "loss": 0.0182, "step": 35150 }, { "epoch": 17.19, "learning_rate": 2.814760508308896e-06, "loss": 0.0146, "step": 35175 }, { "epoch": 17.2, "learning_rate": 2.802541544477029e-06, "loss": 0.0199, "step": 35200 }, { "epoch": 17.22, "learning_rate": 2.7903225806451617e-06, "loss": 0.021, "step": 35225 }, { "epoch": 17.23, "learning_rate": 2.7781036168132945e-06, "loss": 0.0173, "step": 35250 }, { "epoch": 17.24, "learning_rate": 2.7658846529814277e-06, "loss": 0.0164, "step": 35275 }, { "epoch": 17.25, "learning_rate": 2.7536656891495605e-06, "loss": 0.0172, "step": 35300 }, { "epoch": 17.27, "learning_rate": 2.7414467253176934e-06, "loss": 0.0202, "step": 35325 }, { "epoch": 17.28, "learning_rate": 2.7292277614858266e-06, "loss": 0.019, "step": 35350 }, { "epoch": 17.29, "learning_rate": 2.7170087976539594e-06, "loss": 0.0211, "step": 35375 }, { "epoch": 17.3, "learning_rate": 2.7047898338220923e-06, "loss": 0.0238, "step": 35400 }, { "epoch": 17.31, "learning_rate": 2.6925708699902255e-06, "loss": 0.0187, "step": 35425 }, { "epoch": 17.33, "learning_rate": 2.680351906158358e-06, "loss": 0.0218, "step": 35450 }, { "epoch": 17.34, "learning_rate": 2.6681329423264907e-06, "loss": 0.0138, "step": 35475 }, { "epoch": 17.35, "learning_rate": 2.6559139784946236e-06, "loss": 0.0146, "step": 35500 }, { "epoch": 17.36, "learning_rate": 2.6436950146627564e-06, "loss": 0.0183, "step": 35525 }, { "epoch": 17.38, "learning_rate": 2.6314760508308896e-06, "loss": 0.0126, "step": 35550 }, { "epoch": 17.39, "learning_rate": 2.6192570869990225e-06, "loss": 0.0127, "step": 35575 }, { "epoch": 17.4, "learning_rate": 2.6070381231671553e-06, "loss": 0.0194, "step": 35600 }, { "epoch": 17.41, "learning_rate": 2.5948191593352885e-06, "loss": 0.0206, "step": 35625 }, { "epoch": 17.42, "learning_rate": 2.5826001955034214e-06, "loss": 0.0194, "step": 35650 }, { "epoch": 17.44, "learning_rate": 2.570381231671554e-06, "loss": 0.0186, "step": 35675 }, { "epoch": 17.45, "learning_rate": 2.5581622678396874e-06, "loss": 0.031, "step": 35700 }, { "epoch": 17.46, "learning_rate": 2.5459433040078202e-06, "loss": 0.0235, "step": 35725 }, { "epoch": 17.47, "learning_rate": 2.533724340175953e-06, "loss": 0.02, "step": 35750 }, { "epoch": 17.49, "learning_rate": 2.521505376344086e-06, "loss": 0.0157, "step": 35775 }, { "epoch": 17.5, "learning_rate": 2.509286412512219e-06, "loss": 0.018, "step": 35800 }, { "epoch": 17.51, "learning_rate": 2.497067448680352e-06, "loss": 0.0209, "step": 35825 }, { "epoch": 17.52, "learning_rate": 2.4848484848484848e-06, "loss": 0.0201, "step": 35850 }, { "epoch": 17.53, "learning_rate": 2.472629521016618e-06, "loss": 0.0174, "step": 35875 }, { "epoch": 17.55, "learning_rate": 2.460410557184751e-06, "loss": 0.0223, "step": 35900 }, { "epoch": 17.56, "learning_rate": 2.4481915933528837e-06, "loss": 0.0249, "step": 35925 }, { "epoch": 17.57, "learning_rate": 2.435972629521017e-06, "loss": 0.0181, "step": 35950 }, { "epoch": 17.58, "learning_rate": 2.4237536656891498e-06, "loss": 0.0239, "step": 35975 }, { "epoch": 17.6, "learning_rate": 2.4115347018572826e-06, "loss": 0.0204, "step": 36000 }, { "epoch": 17.61, "learning_rate": 2.399315738025416e-06, "loss": 0.028, "step": 36025 }, { "epoch": 17.62, "learning_rate": 2.3870967741935486e-06, "loss": 0.0179, "step": 36050 }, { "epoch": 17.63, "learning_rate": 2.3748778103616815e-06, "loss": 0.0211, "step": 36075 }, { "epoch": 17.64, "learning_rate": 2.3626588465298143e-06, "loss": 0.016, "step": 36100 }, { "epoch": 17.66, "learning_rate": 2.3504398826979475e-06, "loss": 0.0202, "step": 36125 }, { "epoch": 17.67, "learning_rate": 2.3382209188660804e-06, "loss": 0.0211, "step": 36150 }, { "epoch": 17.68, "learning_rate": 2.326001955034213e-06, "loss": 0.0198, "step": 36175 }, { "epoch": 17.69, "learning_rate": 2.3137829912023464e-06, "loss": 0.0241, "step": 36200 }, { "epoch": 17.71, "learning_rate": 2.3015640273704793e-06, "loss": 0.0189, "step": 36225 }, { "epoch": 17.72, "learning_rate": 2.289345063538612e-06, "loss": 0.0179, "step": 36250 }, { "epoch": 17.73, "learning_rate": 2.2771260997067453e-06, "loss": 0.0252, "step": 36275 }, { "epoch": 17.74, "learning_rate": 2.264907135874878e-06, "loss": 0.019, "step": 36300 }, { "epoch": 17.75, "learning_rate": 2.252688172043011e-06, "loss": 0.019, "step": 36325 }, { "epoch": 17.77, "learning_rate": 2.240469208211144e-06, "loss": 0.0178, "step": 36350 }, { "epoch": 17.78, "learning_rate": 2.228250244379277e-06, "loss": 0.0174, "step": 36375 }, { "epoch": 17.79, "learning_rate": 2.2160312805474095e-06, "loss": 0.0172, "step": 36400 }, { "epoch": 17.8, "learning_rate": 2.2038123167155427e-06, "loss": 0.0157, "step": 36425 }, { "epoch": 17.82, "learning_rate": 2.1915933528836755e-06, "loss": 0.0212, "step": 36450 }, { "epoch": 17.83, "learning_rate": 2.1793743890518083e-06, "loss": 0.0204, "step": 36475 }, { "epoch": 17.84, "learning_rate": 2.1671554252199416e-06, "loss": 0.0279, "step": 36500 }, { "epoch": 17.85, "learning_rate": 2.1549364613880744e-06, "loss": 0.0197, "step": 36525 }, { "epoch": 17.86, "learning_rate": 2.1427174975562072e-06, "loss": 0.0188, "step": 36550 }, { "epoch": 17.88, "learning_rate": 2.1304985337243405e-06, "loss": 0.0241, "step": 36575 }, { "epoch": 17.89, "learning_rate": 2.1182795698924733e-06, "loss": 0.0207, "step": 36600 }, { "epoch": 17.9, "learning_rate": 2.106060606060606e-06, "loss": 0.0158, "step": 36625 }, { "epoch": 17.91, "learning_rate": 2.093841642228739e-06, "loss": 0.0214, "step": 36650 }, { "epoch": 17.93, "learning_rate": 2.081622678396872e-06, "loss": 0.0178, "step": 36675 }, { "epoch": 17.94, "learning_rate": 2.069403714565005e-06, "loss": 0.0242, "step": 36700 }, { "epoch": 17.95, "learning_rate": 2.057184750733138e-06, "loss": 0.0229, "step": 36725 }, { "epoch": 17.96, "learning_rate": 2.044965786901271e-06, "loss": 0.0168, "step": 36750 }, { "epoch": 17.97, "learning_rate": 2.032746823069404e-06, "loss": 0.0209, "step": 36775 }, { "epoch": 17.99, "learning_rate": 2.0205278592375367e-06, "loss": 0.021, "step": 36800 }, { "epoch": 18.0, "learning_rate": 2.00830889540567e-06, "loss": 0.0219, "step": 36825 }, { "epoch": 18.01, "learning_rate": 1.996089931573803e-06, "loss": 0.0174, "step": 36850 }, { "epoch": 18.02, "learning_rate": 1.9838709677419356e-06, "loss": 0.0169, "step": 36875 }, { "epoch": 18.04, "learning_rate": 1.9716520039100685e-06, "loss": 0.0171, "step": 36900 }, { "epoch": 18.05, "learning_rate": 1.9594330400782017e-06, "loss": 0.0203, "step": 36925 }, { "epoch": 18.06, "learning_rate": 1.9472140762463345e-06, "loss": 0.0165, "step": 36950 }, { "epoch": 18.07, "learning_rate": 1.9349951124144674e-06, "loss": 0.012, "step": 36975 }, { "epoch": 18.08, "learning_rate": 1.9227761485826006e-06, "loss": 0.0178, "step": 37000 }, { "epoch": 18.1, "learning_rate": 1.9105571847507334e-06, "loss": 0.0165, "step": 37025 }, { "epoch": 18.11, "learning_rate": 1.898338220918866e-06, "loss": 0.0157, "step": 37050 }, { "epoch": 18.12, "learning_rate": 1.886119257086999e-06, "loss": 0.0202, "step": 37075 }, { "epoch": 18.13, "learning_rate": 1.873900293255132e-06, "loss": 0.0149, "step": 37100 }, { "epoch": 18.15, "learning_rate": 1.861681329423265e-06, "loss": 0.0185, "step": 37125 }, { "epoch": 18.16, "learning_rate": 1.849462365591398e-06, "loss": 0.0226, "step": 37150 }, { "epoch": 18.17, "learning_rate": 1.8372434017595308e-06, "loss": 0.0152, "step": 37175 }, { "epoch": 18.18, "learning_rate": 1.8250244379276638e-06, "loss": 0.0126, "step": 37200 }, { "epoch": 18.19, "learning_rate": 1.8128054740957969e-06, "loss": 0.0211, "step": 37225 }, { "epoch": 18.21, "learning_rate": 1.8005865102639297e-06, "loss": 0.017, "step": 37250 }, { "epoch": 18.22, "learning_rate": 1.7883675464320627e-06, "loss": 0.0217, "step": 37275 }, { "epoch": 18.23, "learning_rate": 1.7761485826001956e-06, "loss": 0.0195, "step": 37300 }, { "epoch": 18.24, "learning_rate": 1.7639296187683286e-06, "loss": 0.015, "step": 37325 }, { "epoch": 18.26, "learning_rate": 1.7517106549364616e-06, "loss": 0.016, "step": 37350 }, { "epoch": 18.27, "learning_rate": 1.7394916911045944e-06, "loss": 0.019, "step": 37375 }, { "epoch": 18.28, "learning_rate": 1.7272727272727275e-06, "loss": 0.0206, "step": 37400 }, { "epoch": 18.29, "learning_rate": 1.715542521994135e-06, "loss": 0.0207, "step": 37425 }, { "epoch": 18.3, "learning_rate": 1.7033235581622678e-06, "loss": 0.0232, "step": 37450 }, { "epoch": 18.32, "learning_rate": 1.6911045943304008e-06, "loss": 0.0165, "step": 37475 }, { "epoch": 18.33, "learning_rate": 1.6788856304985338e-06, "loss": 0.0152, "step": 37500 }, { "epoch": 18.34, "learning_rate": 1.6666666666666667e-06, "loss": 0.0144, "step": 37525 }, { "epoch": 18.35, "learning_rate": 1.6544477028347997e-06, "loss": 0.0149, "step": 37550 }, { "epoch": 18.37, "learning_rate": 1.6422287390029325e-06, "loss": 0.0158, "step": 37575 }, { "epoch": 18.38, "learning_rate": 1.6300097751710656e-06, "loss": 0.0144, "step": 37600 }, { "epoch": 18.39, "learning_rate": 1.6177908113391986e-06, "loss": 0.0224, "step": 37625 }, { "epoch": 18.4, "learning_rate": 1.6055718475073314e-06, "loss": 0.0198, "step": 37650 }, { "epoch": 18.41, "learning_rate": 1.5933528836754645e-06, "loss": 0.0116, "step": 37675 }, { "epoch": 18.43, "learning_rate": 1.5811339198435973e-06, "loss": 0.0169, "step": 37700 }, { "epoch": 18.44, "learning_rate": 1.5689149560117303e-06, "loss": 0.0183, "step": 37725 }, { "epoch": 18.45, "learning_rate": 1.5566959921798633e-06, "loss": 0.0183, "step": 37750 }, { "epoch": 18.46, "learning_rate": 1.5444770283479962e-06, "loss": 0.0264, "step": 37775 }, { "epoch": 18.48, "learning_rate": 1.5322580645161292e-06, "loss": 0.0174, "step": 37800 }, { "epoch": 18.49, "learning_rate": 1.5200391006842622e-06, "loss": 0.0193, "step": 37825 }, { "epoch": 18.5, "learning_rate": 1.507820136852395e-06, "loss": 0.0169, "step": 37850 }, { "epoch": 18.51, "learning_rate": 1.495601173020528e-06, "loss": 0.0188, "step": 37875 }, { "epoch": 18.52, "learning_rate": 1.483382209188661e-06, "loss": 0.0185, "step": 37900 }, { "epoch": 18.54, "learning_rate": 1.471163245356794e-06, "loss": 0.0167, "step": 37925 }, { "epoch": 18.55, "learning_rate": 1.458944281524927e-06, "loss": 0.0199, "step": 37950 }, { "epoch": 18.56, "learning_rate": 1.4467253176930598e-06, "loss": 0.02, "step": 37975 }, { "epoch": 18.57, "learning_rate": 1.4345063538611929e-06, "loss": 0.0184, "step": 38000 }, { "epoch": 18.59, "learning_rate": 1.4222873900293257e-06, "loss": 0.0182, "step": 38025 }, { "epoch": 18.6, "learning_rate": 1.4100684261974587e-06, "loss": 0.0142, "step": 38050 }, { "epoch": 18.61, "learning_rate": 1.3978494623655913e-06, "loss": 0.0195, "step": 38075 }, { "epoch": 18.62, "learning_rate": 1.3856304985337244e-06, "loss": 0.0112, "step": 38100 }, { "epoch": 18.63, "learning_rate": 1.3734115347018572e-06, "loss": 0.0183, "step": 38125 }, { "epoch": 18.65, "learning_rate": 1.3611925708699902e-06, "loss": 0.0187, "step": 38150 }, { "epoch": 18.66, "learning_rate": 1.3489736070381233e-06, "loss": 0.0149, "step": 38175 }, { "epoch": 18.67, "learning_rate": 1.336754643206256e-06, "loss": 0.0212, "step": 38200 }, { "epoch": 18.68, "learning_rate": 1.3245356793743891e-06, "loss": 0.0166, "step": 38225 }, { "epoch": 18.7, "learning_rate": 1.3123167155425222e-06, "loss": 0.0141, "step": 38250 }, { "epoch": 18.71, "learning_rate": 1.300097751710655e-06, "loss": 0.0147, "step": 38275 }, { "epoch": 18.72, "learning_rate": 1.287878787878788e-06, "loss": 0.0214, "step": 38300 }, { "epoch": 18.73, "learning_rate": 1.2756598240469208e-06, "loss": 0.0193, "step": 38325 }, { "epoch": 18.74, "learning_rate": 1.2634408602150539e-06, "loss": 0.0189, "step": 38350 }, { "epoch": 18.76, "learning_rate": 1.251221896383187e-06, "loss": 0.0193, "step": 38375 }, { "epoch": 18.77, "learning_rate": 1.2390029325513197e-06, "loss": 0.0208, "step": 38400 }, { "epoch": 18.78, "learning_rate": 1.2267839687194528e-06, "loss": 0.0179, "step": 38425 }, { "epoch": 18.79, "learning_rate": 1.2145650048875856e-06, "loss": 0.0149, "step": 38450 }, { "epoch": 18.8, "learning_rate": 1.2023460410557186e-06, "loss": 0.0181, "step": 38475 }, { "epoch": 18.82, "learning_rate": 1.1901270772238517e-06, "loss": 0.0233, "step": 38500 }, { "epoch": 18.83, "learning_rate": 1.1779081133919845e-06, "loss": 0.0138, "step": 38525 }, { "epoch": 18.84, "learning_rate": 1.1656891495601173e-06, "loss": 0.0209, "step": 38550 }, { "epoch": 18.85, "learning_rate": 1.1534701857282503e-06, "loss": 0.0222, "step": 38575 }, { "epoch": 18.87, "learning_rate": 1.1412512218963832e-06, "loss": 0.0217, "step": 38600 }, { "epoch": 18.88, "learning_rate": 1.1290322580645162e-06, "loss": 0.0217, "step": 38625 }, { "epoch": 18.89, "learning_rate": 1.1168132942326492e-06, "loss": 0.0175, "step": 38650 }, { "epoch": 18.9, "learning_rate": 1.104594330400782e-06, "loss": 0.0215, "step": 38675 }, { "epoch": 18.91, "learning_rate": 1.092375366568915e-06, "loss": 0.0189, "step": 38700 }, { "epoch": 18.93, "learning_rate": 1.080156402737048e-06, "loss": 0.0179, "step": 38725 }, { "epoch": 18.94, "learning_rate": 1.067937438905181e-06, "loss": 0.0128, "step": 38750 }, { "epoch": 18.95, "learning_rate": 1.055718475073314e-06, "loss": 0.0214, "step": 38775 }, { "epoch": 18.96, "learning_rate": 1.0434995112414468e-06, "loss": 0.0192, "step": 38800 }, { "epoch": 18.98, "learning_rate": 1.0312805474095798e-06, "loss": 0.016, "step": 38825 }, { "epoch": 18.99, "learning_rate": 1.0190615835777127e-06, "loss": 0.0184, "step": 38850 }, { "epoch": 19.0, "learning_rate": 1.0068426197458455e-06, "loss": 0.0146, "step": 38875 }, { "epoch": 19.01, "learning_rate": 9.946236559139785e-07, "loss": 0.0122, "step": 38900 }, { "epoch": 19.02, "learning_rate": 9.824046920821116e-07, "loss": 0.0165, "step": 38925 }, { "epoch": 19.04, "learning_rate": 9.701857282502444e-07, "loss": 0.0165, "step": 38950 }, { "epoch": 19.05, "learning_rate": 9.579667644183774e-07, "loss": 0.0188, "step": 38975 }, { "epoch": 19.06, "learning_rate": 9.457478005865104e-07, "loss": 0.0208, "step": 39000 }, { "epoch": 19.07, "learning_rate": 9.335288367546433e-07, "loss": 0.0188, "step": 39025 }, { "epoch": 19.09, "learning_rate": 9.213098729227762e-07, "loss": 0.0133, "step": 39050 }, { "epoch": 19.1, "learning_rate": 9.090909090909091e-07, "loss": 0.0088, "step": 39075 }, { "epoch": 19.11, "learning_rate": 8.968719452590422e-07, "loss": 0.0206, "step": 39100 }, { "epoch": 19.12, "learning_rate": 8.846529814271751e-07, "loss": 0.0149, "step": 39125 }, { "epoch": 19.13, "learning_rate": 8.72434017595308e-07, "loss": 0.0158, "step": 39150 }, { "epoch": 19.15, "learning_rate": 8.60215053763441e-07, "loss": 0.0162, "step": 39175 }, { "epoch": 19.16, "learning_rate": 8.479960899315738e-07, "loss": 0.0144, "step": 39200 }, { "epoch": 19.17, "learning_rate": 8.357771260997067e-07, "loss": 0.0171, "step": 39225 }, { "epoch": 19.18, "learning_rate": 8.235581622678398e-07, "loss": 0.0192, "step": 39250 }, { "epoch": 19.2, "learning_rate": 8.113391984359727e-07, "loss": 0.0156, "step": 39275 }, { "epoch": 19.21, "learning_rate": 7.991202346041056e-07, "loss": 0.017, "step": 39300 }, { "epoch": 19.22, "learning_rate": 7.869012707722385e-07, "loss": 0.0192, "step": 39325 }, { "epoch": 19.23, "learning_rate": 7.746823069403715e-07, "loss": 0.021, "step": 39350 }, { "epoch": 19.24, "learning_rate": 7.624633431085045e-07, "loss": 0.0211, "step": 39375 }, { "epoch": 19.26, "learning_rate": 7.502443792766374e-07, "loss": 0.0158, "step": 39400 }, { "epoch": 19.27, "learning_rate": 7.38514173998045e-07, "loss": 0.0166, "step": 39425 }, { "epoch": 19.28, "learning_rate": 7.262952101661779e-07, "loss": 0.0116, "step": 39450 }, { "epoch": 19.29, "learning_rate": 7.14076246334311e-07, "loss": 0.0179, "step": 39475 }, { "epoch": 19.31, "learning_rate": 7.018572825024439e-07, "loss": 0.0195, "step": 39500 }, { "epoch": 19.32, "learning_rate": 6.896383186705768e-07, "loss": 0.0198, "step": 39525 }, { "epoch": 19.33, "learning_rate": 6.774193548387098e-07, "loss": 0.0191, "step": 39550 }, { "epoch": 19.34, "learning_rate": 6.652003910068426e-07, "loss": 0.0181, "step": 39575 }, { "epoch": 19.35, "learning_rate": 6.529814271749755e-07, "loss": 0.0167, "step": 39600 }, { "epoch": 19.37, "learning_rate": 6.407624633431086e-07, "loss": 0.0167, "step": 39625 }, { "epoch": 19.38, "learning_rate": 6.285434995112415e-07, "loss": 0.0111, "step": 39650 }, { "epoch": 19.39, "learning_rate": 6.163245356793744e-07, "loss": 0.0152, "step": 39675 }, { "epoch": 19.4, "learning_rate": 6.041055718475073e-07, "loss": 0.0189, "step": 39700 }, { "epoch": 19.42, "learning_rate": 5.92375366568915e-07, "loss": 0.0158, "step": 39725 }, { "epoch": 19.43, "learning_rate": 5.801564027370479e-07, "loss": 0.0207, "step": 39750 }, { "epoch": 19.44, "learning_rate": 5.679374389051809e-07, "loss": 0.0189, "step": 39775 }, { "epoch": 19.45, "learning_rate": 5.557184750733138e-07, "loss": 0.0172, "step": 39800 }, { "epoch": 19.46, "learning_rate": 5.434995112414468e-07, "loss": 0.0173, "step": 39825 }, { "epoch": 19.48, "learning_rate": 5.312805474095798e-07, "loss": 0.0135, "step": 39850 }, { "epoch": 19.49, "learning_rate": 5.190615835777126e-07, "loss": 0.0126, "step": 39875 }, { "epoch": 19.5, "learning_rate": 5.068426197458456e-07, "loss": 0.0183, "step": 39900 }, { "epoch": 19.51, "learning_rate": 4.946236559139786e-07, "loss": 0.0121, "step": 39925 }, { "epoch": 19.53, "learning_rate": 4.824046920821115e-07, "loss": 0.0164, "step": 39950 }, { "epoch": 19.54, "learning_rate": 4.701857282502444e-07, "loss": 0.0227, "step": 39975 }, { "epoch": 19.55, "learning_rate": 4.579667644183774e-07, "loss": 0.0172, "step": 40000 }, { "epoch": 19.56, "learning_rate": 4.4574780058651033e-07, "loss": 0.0156, "step": 40025 }, { "epoch": 19.57, "learning_rate": 4.335288367546432e-07, "loss": 0.0212, "step": 40050 }, { "epoch": 19.59, "learning_rate": 4.213098729227762e-07, "loss": 0.0138, "step": 40075 }, { "epoch": 19.6, "learning_rate": 4.090909090909091e-07, "loss": 0.0179, "step": 40100 }, { "epoch": 19.61, "learning_rate": 3.968719452590421e-07, "loss": 0.0121, "step": 40125 }, { "epoch": 19.62, "learning_rate": 3.8465298142717503e-07, "loss": 0.0175, "step": 40150 }, { "epoch": 19.64, "learning_rate": 3.7243401759530796e-07, "loss": 0.0194, "step": 40175 }, { "epoch": 19.65, "learning_rate": 3.602150537634409e-07, "loss": 0.0119, "step": 40200 }, { "epoch": 19.66, "learning_rate": 3.479960899315738e-07, "loss": 0.0121, "step": 40225 }, { "epoch": 19.67, "learning_rate": 3.3577712609970675e-07, "loss": 0.0207, "step": 40250 }, { "epoch": 19.68, "learning_rate": 3.2355816226783973e-07, "loss": 0.0118, "step": 40275 }, { "epoch": 19.7, "learning_rate": 3.1133919843597266e-07, "loss": 0.0142, "step": 40300 }, { "epoch": 19.71, "learning_rate": 2.991202346041056e-07, "loss": 0.0228, "step": 40325 }, { "epoch": 19.72, "learning_rate": 2.869012707722385e-07, "loss": 0.0185, "step": 40350 }, { "epoch": 19.73, "learning_rate": 2.746823069403715e-07, "loss": 0.017, "step": 40375 }, { "epoch": 19.75, "learning_rate": 2.6246334310850443e-07, "loss": 0.0229, "step": 40400 }, { "epoch": 19.76, "learning_rate": 2.5024437927663736e-07, "loss": 0.0141, "step": 40425 }, { "epoch": 19.77, "learning_rate": 2.3802541544477032e-07, "loss": 0.0139, "step": 40450 }, { "epoch": 19.78, "learning_rate": 2.2580645161290322e-07, "loss": 0.0188, "step": 40475 }, { "epoch": 19.79, "learning_rate": 2.1358748778103618e-07, "loss": 0.0168, "step": 40500 }, { "epoch": 19.81, "learning_rate": 2.0136852394916913e-07, "loss": 0.0158, "step": 40525 }, { "epoch": 19.82, "learning_rate": 1.8914956011730206e-07, "loss": 0.0168, "step": 40550 }, { "epoch": 19.83, "learning_rate": 1.7693059628543502e-07, "loss": 0.0183, "step": 40575 }, { "epoch": 19.84, "learning_rate": 1.6471163245356795e-07, "loss": 0.019, "step": 40600 }, { "epoch": 19.86, "learning_rate": 1.524926686217009e-07, "loss": 0.0162, "step": 40625 }, { "epoch": 19.87, "learning_rate": 1.4027370478983383e-07, "loss": 0.0227, "step": 40650 }, { "epoch": 19.88, "learning_rate": 1.280547409579668e-07, "loss": 0.0169, "step": 40675 }, { "epoch": 19.89, "learning_rate": 1.1583577712609972e-07, "loss": 0.0106, "step": 40700 }, { "epoch": 19.9, "learning_rate": 1.0361681329423266e-07, "loss": 0.0175, "step": 40725 }, { "epoch": 19.92, "learning_rate": 9.13978494623656e-08, "loss": 0.0197, "step": 40750 }, { "epoch": 19.93, "learning_rate": 7.917888563049853e-08, "loss": 0.0176, "step": 40775 }, { "epoch": 19.94, "learning_rate": 6.695992179863148e-08, "loss": 0.0144, "step": 40800 }, { "epoch": 19.95, "learning_rate": 5.4740957966764426e-08, "loss": 0.0128, "step": 40825 }, { "epoch": 19.97, "learning_rate": 4.252199413489737e-08, "loss": 0.0136, "step": 40850 }, { "epoch": 19.98, "learning_rate": 3.0303030303030305e-08, "loss": 0.011, "step": 40875 }, { "epoch": 19.99, "learning_rate": 1.8084066471163248e-08, "loss": 0.0185, "step": 40900 }, { "epoch": 20.0, "step": 40920, "total_flos": 5.166258268431053e+17, "train_loss": 0.12400934287872191, "train_runtime": 16044.2941, "train_samples_per_second": 489.522, "train_steps_per_second": 2.55 } ], "logging_steps": 25, "max_steps": 40920, "num_train_epochs": 20, "save_steps": 500, "total_flos": 5.166258268431053e+17, "trial_name": null, "trial_params": null }