diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10816 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.033884630899556,
+  "global_step": 90000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998599345614274e-05,
+      "loss": 6.6897,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999719869122854e-05,
+      "loss": 5.258,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999579803684282e-05,
+      "loss": 4.5571,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999439738245709e-05,
+      "loss": 4.277,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992996728071354e-05,
+      "loss": 4.0698,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991596073685625e-05,
+      "loss": 3.8909,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9990195419299904e-05,
+      "loss": 3.7887,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998879476491417e-05,
+      "loss": 3.7145,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998739411052844e-05,
+      "loss": 3.6474,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998599345614271e-05,
+      "loss": 3.5479,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9984592801756984e-05,
+      "loss": 3.4806,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9983192147371255e-05,
+      "loss": 3.4368,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998179149298553e-05,
+      "loss": 3.4167,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998039083859979e-05,
+      "loss": 3.3564,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997899018421407e-05,
+      "loss": 3.3145,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997758952982834e-05,
+      "loss": 3.2817,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997618887544261e-05,
+      "loss": 3.2437,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997478822105688e-05,
+      "loss": 3.1763,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997338756667116e-05,
+      "loss": 3.1199,
+      "step": 950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997198691228542e-05,
+      "loss": 3.0686,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.997058625789969e-05,
+      "loss": 3.0862,
+      "step": 1050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9969185603513965e-05,
+      "loss": 3.0317,
+      "step": 1100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996778494912823e-05,
+      "loss": 2.9852,
+      "step": 1150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996638429474251e-05,
+      "loss": 2.9593,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.996498364035678e-05,
+      "loss": 2.9514,
+      "step": 1250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9963582985971045e-05,
+      "loss": 2.9288,
+      "step": 1300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9962182331585316e-05,
+      "loss": 2.8808,
+      "step": 1350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9960781677199595e-05,
+      "loss": 2.8553,
+      "step": 1400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995938102281386e-05,
+      "loss": 2.849,
+      "step": 1450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.995798036842813e-05,
+      "loss": 2.8272,
+      "step": 1500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.99565797140424e-05,
+      "loss": 2.8049,
+      "step": 1550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9955179059656674e-05,
+      "loss": 2.7629,
+      "step": 1600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9953778405270946e-05,
+      "loss": 2.7371,
+      "step": 1650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.995237775088522e-05,
+      "loss": 2.728,
+      "step": 1700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.995097709649948e-05,
+      "loss": 2.6995,
+      "step": 1750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.994957644211376e-05,
+      "loss": 2.6784,
+      "step": 1800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.994817578772803e-05,
+      "loss": 2.6507,
+      "step": 1850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.99467751333423e-05,
+      "loss": 2.6582,
+      "step": 1900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.994537447895657e-05,
+      "loss": 2.629,
+      "step": 1950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.994397382457085e-05,
+      "loss": 2.5755,
+      "step": 2000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.994257317018511e-05,
+      "loss": 2.6007,
+      "step": 2050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9941172515799384e-05,
+      "loss": 2.5898,
+      "step": 2100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9939771861413656e-05,
+      "loss": 2.5711,
+      "step": 2150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.993837120702793e-05,
+      "loss": 2.5769,
+      "step": 2200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.99369705526422e-05,
+      "loss": 2.5123,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.993556989825647e-05,
+      "loss": 2.471,
+      "step": 2300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9934169243870735e-05,
+      "loss": 2.4919,
+      "step": 2350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.993276858948501e-05,
+      "loss": 2.4754,
+      "step": 2400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9931367935099285e-05,
+      "loss": 2.4559,
+      "step": 2450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.992996728071355e-05,
+      "loss": 2.4689,
+      "step": 2500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.992856662632782e-05,
+      "loss": 2.3926,
+      "step": 2550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9927165971942094e-05,
+      "loss": 2.4048,
+      "step": 2600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9925765317556365e-05,
+      "loss": 2.3582,
+      "step": 2650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992436466317064e-05,
+      "loss": 2.3653,
+      "step": 2700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.992296400878491e-05,
+      "loss": 2.3437,
+      "step": 2750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.992156335439917e-05,
+      "loss": 2.3232,
+      "step": 2800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.992016270001345e-05,
+      "loss": 2.3,
+      "step": 2850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.991876204562772e-05,
+      "loss": 2.3017,
+      "step": 2900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.991736139124199e-05,
+      "loss": 2.2912,
+      "step": 2950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.991596073685626e-05,
+      "loss": 2.2834,
+      "step": 3000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.991456008247054e-05,
+      "loss": 2.2803,
+      "step": 3050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.99131594280848e-05,
+      "loss": 2.2308,
+      "step": 3100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9911758773699075e-05,
+      "loss": 2.2106,
+      "step": 3150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9910358119313346e-05,
+      "loss": 2.2169,
+      "step": 3200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.990895746492762e-05,
+      "loss": 2.1805,
+      "step": 3250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.990755681054189e-05,
+      "loss": 2.1821,
+      "step": 3300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.990615615615616e-05,
+      "loss": 2.1632,
+      "step": 3350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9904755501770426e-05,
+      "loss": 2.1328,
+      "step": 3400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.99033548473847e-05,
+      "loss": 2.1339,
+      "step": 3450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9901954192998976e-05,
+      "loss": 2.0954,
+      "step": 3500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.990055353861324e-05,
+      "loss": 2.1149,
+      "step": 3550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.989915288422751e-05,
+      "loss": 2.0974,
+      "step": 3600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9897752229841784e-05,
+      "loss": 2.0959,
+      "step": 3650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9896351575456056e-05,
+      "loss": 2.0753,
+      "step": 3700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.989495092107033e-05,
+      "loss": 2.0299,
+      "step": 3750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.98935502666846e-05,
+      "loss": 2.0623,
+      "step": 3800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9892149612298864e-05,
+      "loss": 2.0643,
+      "step": 3850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.989074895791314e-05,
+      "loss": 2.0094,
+      "step": 3900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9889348303527414e-05,
+      "loss": 2.0117,
+      "step": 3950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.988794764914168e-05,
+      "loss": 2.0083,
+      "step": 4000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.988654699475595e-05,
+      "loss": 1.9975,
+      "step": 4050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.988514634037023e-05,
+      "loss": 1.9871,
+      "step": 4100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9883745685984494e-05,
+      "loss": 1.9791,
+      "step": 4150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9882345031598766e-05,
+      "loss": 1.954,
+      "step": 4200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.988094437721304e-05,
+      "loss": 1.9414,
+      "step": 4250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.987954372282731e-05,
+      "loss": 1.9618,
+      "step": 4300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.987814306844158e-05,
+      "loss": 1.9492,
+      "step": 4350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.987674241405585e-05,
+      "loss": 1.9234,
+      "step": 4400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.987534175967012e-05,
+      "loss": 1.95,
+      "step": 4450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.987394110528439e-05,
+      "loss": 1.91,
+      "step": 4500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.987254045089867e-05,
+      "loss": 1.895,
+      "step": 4550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.987113979651293e-05,
+      "loss": 1.9085,
+      "step": 4600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9869739142127204e-05,
+      "loss": 1.9381,
+      "step": 4650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9868338487741475e-05,
+      "loss": 1.8904,
+      "step": 4700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.986693783335575e-05,
+      "loss": 1.8767,
+      "step": 4750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.986553717897002e-05,
+      "loss": 1.89,
+      "step": 4800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.986413652458429e-05,
+      "loss": 1.8853,
+      "step": 4850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9862735870198555e-05,
+      "loss": 1.8423,
+      "step": 4900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.986133521581283e-05,
+      "loss": 1.8952,
+      "step": 4950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9859934561427105e-05,
+      "loss": 1.8386,
+      "step": 5000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.985853390704137e-05,
+      "loss": 1.8445,
+      "step": 5050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.985713325265564e-05,
+      "loss": 1.8287,
+      "step": 5100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.985573259826991e-05,
+      "loss": 1.8311,
+      "step": 5150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9854331943884185e-05,
+      "loss": 1.84,
+      "step": 5200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9852931289498456e-05,
+      "loss": 1.8407,
+      "step": 5250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.985153063511273e-05,
+      "loss": 1.8136,
+      "step": 5300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9850129980727e-05,
+      "loss": 1.8238,
+      "step": 5350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.984872932634127e-05,
+      "loss": 1.8085,
+      "step": 5400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.984732867195554e-05,
+      "loss": 1.8123,
+      "step": 5450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.984592801756981e-05,
+      "loss": 1.7892,
+      "step": 5500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.984452736318408e-05,
+      "loss": 1.7986,
+      "step": 5550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.984312670879836e-05,
+      "loss": 1.7938,
+      "step": 5600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.984172605441262e-05,
+      "loss": 1.7872,
+      "step": 5650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9840325400026894e-05,
+      "loss": 1.78,
+      "step": 5700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9838924745641166e-05,
+      "loss": 1.7795,
+      "step": 5750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.983752409125544e-05,
+      "loss": 1.7643,
+      "step": 5800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.983612343686971e-05,
+      "loss": 1.7853,
+      "step": 5850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.983472278248398e-05,
+      "loss": 1.7302,
+      "step": 5900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9833322128098246e-05,
+      "loss": 1.7454,
+      "step": 5950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9831921473712524e-05,
+      "loss": 1.7523,
+      "step": 6000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9830520819326796e-05,
+      "loss": 1.7306,
+      "step": 6050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.982912016494106e-05,
+      "loss": 1.7585,
+      "step": 6100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.982771951055533e-05,
+      "loss": 1.7475,
+      "step": 6150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9826318856169604e-05,
+      "loss": 1.7738,
+      "step": 6200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9824918201783876e-05,
+      "loss": 1.7194,
+      "step": 6250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.982351754739815e-05,
+      "loss": 1.7241,
+      "step": 6300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.982211689301242e-05,
+      "loss": 1.7459,
+      "step": 6350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.982071623862669e-05,
+      "loss": 1.745,
+      "step": 6400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.981931558424096e-05,
+      "loss": 1.709,
+      "step": 6450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9817914929855234e-05,
+      "loss": 1.7029,
+      "step": 6500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.98165142754695e-05,
+      "loss": 1.7215,
+      "step": 6550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.981511362108378e-05,
+      "loss": 1.7268,
+      "step": 6600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.981371296669804e-05,
+      "loss": 1.7104,
+      "step": 6650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9812312312312313e-05,
+      "loss": 1.7045,
+      "step": 6700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9810911657926585e-05,
+      "loss": 1.7202,
+      "step": 6750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.980951100354086e-05,
+      "loss": 1.7227,
+      "step": 6800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.980811034915513e-05,
+      "loss": 1.6956,
+      "step": 6850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.98067096947694e-05,
+      "loss": 1.6987,
+      "step": 6900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.980530904038367e-05,
+      "loss": 1.7068,
+      "step": 6950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9803908385997936e-05,
+      "loss": 1.6831,
+      "step": 7000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9802507731612215e-05,
+      "loss": 1.6836,
+      "step": 7050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9801107077226487e-05,
+      "loss": 1.685,
+      "step": 7100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.979970642284075e-05,
+      "loss": 1.6383,
+      "step": 7150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.979830576845502e-05,
+      "loss": 1.6637,
+      "step": 7200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9796905114069295e-05,
+      "loss": 1.653,
+      "step": 7250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9795504459683566e-05,
+      "loss": 1.6456,
+      "step": 7300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.979410380529784e-05,
+      "loss": 1.6906,
+      "step": 7350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.979270315091211e-05,
+      "loss": 1.681,
+      "step": 7400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.979130249652638e-05,
+      "loss": 1.6843,
+      "step": 7450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.978990184214065e-05,
+      "loss": 1.6591,
+      "step": 7500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9788501187754924e-05,
+      "loss": 1.6505,
+      "step": 7550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.978710053336919e-05,
+      "loss": 1.6683,
+      "step": 7600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.978569987898347e-05,
+      "loss": 1.6337,
+      "step": 7650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978429922459773e-05,
+      "loss": 1.6364,
+      "step": 7700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9782898570212004e-05,
+      "loss": 1.6822,
+      "step": 7750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9781497915826276e-05,
+      "loss": 1.6316,
+      "step": 7800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978009726144055e-05,
+      "loss": 1.6467,
+      "step": 7850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.977869660705482e-05,
+      "loss": 1.6441,
+      "step": 7900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977729595266909e-05,
+      "loss": 1.62,
+      "step": 7950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977589529828336e-05,
+      "loss": 1.6345,
+      "step": 8000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977449464389763e-05,
+      "loss": 1.6474,
+      "step": 8050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9773093989511906e-05,
+      "loss": 1.6027,
+      "step": 8100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.977169333512617e-05,
+      "loss": 1.6124,
+      "step": 8150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.977029268074044e-05,
+      "loss": 1.641,
+      "step": 8200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9768892026354714e-05,
+      "loss": 1.6292,
+      "step": 8250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9767491371968985e-05,
+      "loss": 1.6153,
+      "step": 8300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.976609071758326e-05,
+      "loss": 1.6272,
+      "step": 8350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.976469006319753e-05,
+      "loss": 1.616,
+      "step": 8400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.97632894088118e-05,
+      "loss": 1.6062,
+      "step": 8450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.976188875442607e-05,
+      "loss": 1.5941,
+      "step": 8500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9760488100040344e-05,
+      "loss": 1.611,
+      "step": 8550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.975908744565461e-05,
+      "loss": 1.6176,
+      "step": 8600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.975768679126888e-05,
+      "loss": 1.5742,
+      "step": 8650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.975628613688316e-05,
+      "loss": 1.5945,
+      "step": 8700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9754885482497423e-05,
+      "loss": 1.5949,
+      "step": 8750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9753484828111695e-05,
+      "loss": 1.6093,
+      "step": 8800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.975208417372597e-05,
+      "loss": 1.6421,
+      "step": 8850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.975068351934024e-05,
+      "loss": 1.5867,
+      "step": 8900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.974928286495451e-05,
+      "loss": 1.5893,
+      "step": 8950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.974788221056878e-05,
+      "loss": 1.6043,
+      "step": 9000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.974648155618305e-05,
+      "loss": 1.577,
+      "step": 9050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.974508090179732e-05,
+      "loss": 1.5824,
+      "step": 9100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9743680247411596e-05,
+      "loss": 1.5603,
+      "step": 9150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.974227959302586e-05,
+      "loss": 1.5701,
+      "step": 9200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.974087893864013e-05,
+      "loss": 1.5578,
+      "step": 9250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9739478284254405e-05,
+      "loss": 1.5759,
+      "step": 9300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9738077629868676e-05,
+      "loss": 1.5798,
+      "step": 9350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.973667697548295e-05,
+      "loss": 1.5541,
+      "step": 9400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.973527632109722e-05,
+      "loss": 1.5648,
+      "step": 9450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.973387566671149e-05,
+      "loss": 1.5453,
+      "step": 9500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.973247501232576e-05,
+      "loss": 1.5376,
+      "step": 9550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9731074357940034e-05,
+      "loss": 1.5535,
+      "step": 9600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.97296737035543e-05,
+      "loss": 1.5799,
+      "step": 9650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.972827304916857e-05,
+      "loss": 1.5483,
+      "step": 9700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.972687239478285e-05,
+      "loss": 1.5823,
+      "step": 9750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9725471740397114e-05,
+      "loss": 1.5499,
+      "step": 9800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9724071086011386e-05,
+      "loss": 1.5751,
+      "step": 9850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.972267043162566e-05,
+      "loss": 1.5588,
+      "step": 9900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.972126977723993e-05,
+      "loss": 1.5987,
+      "step": 9950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.97198691228542e-05,
+      "loss": 1.5269,
+      "step": 10000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.971846846846847e-05,
+      "loss": 1.5497,
+      "step": 10050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.971706781408274e-05,
+      "loss": 1.5329,
+      "step": 10100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.971566715969701e-05,
+      "loss": 1.5383,
+      "step": 10150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.971426650531129e-05,
+      "loss": 1.5511,
+      "step": 10200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.971286585092555e-05,
+      "loss": 1.5802,
+      "step": 10250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9711465196539824e-05,
+      "loss": 1.5579,
+      "step": 10300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9710064542154095e-05,
+      "loss": 1.553,
+      "step": 10350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.970866388776837e-05,
+      "loss": 1.531,
+      "step": 10400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.970726323338264e-05,
+      "loss": 1.5311,
+      "step": 10450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.970586257899691e-05,
+      "loss": 1.5493,
+      "step": 10500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.970446192461118e-05,
+      "loss": 1.5232,
+      "step": 10550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9703061270225454e-05,
+      "loss": 1.5012,
+      "step": 10600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9701660615839725e-05,
+      "loss": 1.5243,
+      "step": 10650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.970025996145399e-05,
+      "loss": 1.524,
+      "step": 10700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.969885930706826e-05,
+      "loss": 1.5345,
+      "step": 10750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.969745865268254e-05,
+      "loss": 1.5402,
+      "step": 10800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9696057998296805e-05,
+      "loss": 1.5259,
+      "step": 10850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.969465734391108e-05,
+      "loss": 1.5375,
+      "step": 10900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.969325668952535e-05,
+      "loss": 1.5078,
+      "step": 10950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.969185603513962e-05,
+      "loss": 1.5454,
+      "step": 11000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.969045538075389e-05,
+      "loss": 1.5347,
+      "step": 11050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.968905472636816e-05,
+      "loss": 1.5035,
+      "step": 11100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.968765407198243e-05,
+      "loss": 1.5242,
+      "step": 11150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.96862534175967e-05,
+      "loss": 1.5151,
+      "step": 11200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.968485276321098e-05,
+      "loss": 1.5183,
+      "step": 11250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.968345210882524e-05,
+      "loss": 1.5342,
+      "step": 11300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9682051454439515e-05,
+      "loss": 1.5023,
+      "step": 11350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9680650800053786e-05,
+      "loss": 1.5185,
+      "step": 11400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.967925014566806e-05,
+      "loss": 1.5249,
+      "step": 11450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.967784949128233e-05,
+      "loss": 1.5236,
+      "step": 11500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.96764488368966e-05,
+      "loss": 1.5185,
+      "step": 11550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9675048182510866e-05,
+      "loss": 1.5063,
+      "step": 11600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9673647528125144e-05,
+      "loss": 1.5004,
+      "step": 11650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9672246873739416e-05,
+      "loss": 1.5408,
+      "step": 11700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.967084621935368e-05,
+      "loss": 1.4791,
+      "step": 11750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.966944556496795e-05,
+      "loss": 1.5187,
+      "step": 11800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.966804491058223e-05,
+      "loss": 1.5252,
+      "step": 11850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9666644256196496e-05,
+      "loss": 1.4862,
+      "step": 11900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.966524360181077e-05,
+      "loss": 1.4958,
+      "step": 11950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.966384294742504e-05,
+      "loss": 1.4896,
+      "step": 12000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.966244229303931e-05,
+      "loss": 1.468,
+      "step": 12050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.966104163865358e-05,
+      "loss": 1.4898,
+      "step": 12100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9659640984267854e-05,
+      "loss": 1.5018,
+      "step": 12150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.965824032988212e-05,
+      "loss": 1.5111,
+      "step": 12200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.965683967549639e-05,
+      "loss": 1.4865,
+      "step": 12250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.965543902111067e-05,
+      "loss": 1.4759,
+      "step": 12300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9654038366724934e-05,
+      "loss": 1.4959,
+      "step": 12350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9652637712339205e-05,
+      "loss": 1.5015,
+      "step": 12400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.965123705795348e-05,
+      "loss": 1.4925,
+      "step": 12450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.964983640356775e-05,
+      "loss": 1.4625,
+      "step": 12500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.964843574918202e-05,
+      "loss": 1.4867,
+      "step": 12550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.964703509479629e-05,
+      "loss": 1.4986,
+      "step": 12600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.964563444041056e-05,
+      "loss": 1.4704,
+      "step": 12650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9644233786024835e-05,
+      "loss": 1.5,
+      "step": 12700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.964283313163911e-05,
+      "loss": 1.4678,
+      "step": 12750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.964143247725337e-05,
+      "loss": 1.4832,
+      "step": 12800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.964003182286764e-05,
+      "loss": 1.4965,
+      "step": 12850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.963863116848192e-05,
+      "loss": 1.4448,
+      "step": 12900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9637230514096187e-05,
+      "loss": 1.4972,
+      "step": 12950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.963582985971046e-05,
+      "loss": 1.4942,
+      "step": 13000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.963442920532473e-05,
+      "loss": 1.4511,
+      "step": 13050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9633028550939e-05,
+      "loss": 1.4861,
+      "step": 13100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.963162789655327e-05,
+      "loss": 1.4808,
+      "step": 13150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9630227242167545e-05,
+      "loss": 1.458,
+      "step": 13200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.962882658778181e-05,
+      "loss": 1.4674,
+      "step": 13250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.962742593339608e-05,
+      "loss": 1.4772,
+      "step": 13300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.962602527901036e-05,
+      "loss": 1.4613,
+      "step": 13350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9624624624624625e-05,
+      "loss": 1.4683,
+      "step": 13400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.9623223970238896e-05,
+      "loss": 1.4877,
+      "step": 13450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.962182331585317e-05,
+      "loss": 1.448,
+      "step": 13500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.962042266146744e-05,
+      "loss": 1.471,
+      "step": 13550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.961902200708171e-05,
+      "loss": 1.4657,
+      "step": 13600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.961762135269598e-05,
+      "loss": 1.4699,
+      "step": 13650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.961622069831025e-05,
+      "loss": 1.4696,
+      "step": 13700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9614820043924526e-05,
+      "loss": 1.4938,
+      "step": 13750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.96134193895388e-05,
+      "loss": 1.448,
+      "step": 13800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.961201873515306e-05,
+      "loss": 1.4607,
+      "step": 13850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.9610618080767334e-05,
+      "loss": 1.4393,
+      "step": 13900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.960921742638161e-05,
+      "loss": 1.4952,
+      "step": 13950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.960781677199588e-05,
+      "loss": 1.4469,
+      "step": 14000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.960641611761015e-05,
+      "loss": 1.4689,
+      "step": 14050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.960501546322442e-05,
+      "loss": 1.4632,
+      "step": 14100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.960361480883869e-05,
+      "loss": 1.4726,
+      "step": 14150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9602214154452964e-05,
+      "loss": 1.4663,
+      "step": 14200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.9600813500067236e-05,
+      "loss": 1.4647,
+      "step": 14250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.95994128456815e-05,
+      "loss": 1.4282,
+      "step": 14300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.959801219129578e-05,
+      "loss": 1.4628,
+      "step": 14350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.959661153691005e-05,
+      "loss": 1.4443,
+      "step": 14400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.9595210882524315e-05,
+      "loss": 1.4242,
+      "step": 14450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.959381022813859e-05,
+      "loss": 1.4404,
+      "step": 14500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.959240957375286e-05,
+      "loss": 1.4191,
+      "step": 14550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.959100891936713e-05,
+      "loss": 1.4304,
+      "step": 14600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.95896082649814e-05,
+      "loss": 1.4385,
+      "step": 14650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.9588207610595674e-05,
+      "loss": 1.4845,
+      "step": 14700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.958680695620994e-05,
+      "loss": 1.4561,
+      "step": 14750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.958540630182422e-05,
+      "loss": 1.4174,
+      "step": 14800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.958400564743849e-05,
+      "loss": 1.4426,
+      "step": 14850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.958260499305275e-05,
+      "loss": 1.4449,
+      "step": 14900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9581204338667025e-05,
+      "loss": 1.4493,
+      "step": 14950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.95798036842813e-05,
+      "loss": 1.4344,
+      "step": 15000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.957840302989557e-05,
+      "loss": 1.4331,
+      "step": 15050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.957700237550984e-05,
+      "loss": 1.4231,
+      "step": 15100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.957560172112411e-05,
+      "loss": 1.4466,
+      "step": 15150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.957420106673838e-05,
+      "loss": 1.4208,
+      "step": 15200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9572800412352655e-05,
+      "loss": 1.4481,
+      "step": 15250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9571399757966926e-05,
+      "loss": 1.4266,
+      "step": 15300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.956999910358119e-05,
+      "loss": 1.4182,
+      "step": 15350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.956859844919547e-05,
+      "loss": 1.439,
+      "step": 15400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.956719779480974e-05,
+      "loss": 1.4602,
+      "step": 15450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.9565797140424006e-05,
+      "loss": 1.4321,
+      "step": 15500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.956439648603828e-05,
+      "loss": 1.4211,
+      "step": 15550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.956299583165255e-05,
+      "loss": 1.4529,
+      "step": 15600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.956159517726682e-05,
+      "loss": 1.4812,
+      "step": 15650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.956019452288109e-05,
+      "loss": 1.429,
+      "step": 15700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.9558793868495364e-05,
+      "loss": 1.4248,
+      "step": 15750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.955739321410963e-05,
+      "loss": 1.4457,
+      "step": 15800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.955599255972391e-05,
+      "loss": 1.4423,
+      "step": 15850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.955459190533818e-05,
+      "loss": 1.462,
+      "step": 15900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.9553191250952444e-05,
+      "loss": 1.4297,
+      "step": 15950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9551790596566716e-05,
+      "loss": 1.4194,
+      "step": 16000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.9550389942180994e-05,
+      "loss": 1.4264,
+      "step": 16050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.954898928779526e-05,
+      "loss": 1.4297,
+      "step": 16100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.954758863340953e-05,
+      "loss": 1.4422,
+      "step": 16150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.95461879790238e-05,
+      "loss": 1.427,
+      "step": 16200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9544787324638074e-05,
+      "loss": 1.4065,
+      "step": 16250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9543386670252346e-05,
+      "loss": 1.4456,
+      "step": 16300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.954198601586662e-05,
+      "loss": 1.4264,
+      "step": 16350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.954058536148088e-05,
+      "loss": 1.4434,
+      "step": 16400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.953918470709516e-05,
+      "loss": 1.4155,
+      "step": 16450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.953778405270943e-05,
+      "loss": 1.4198,
+      "step": 16500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.95363833983237e-05,
+      "loss": 1.4149,
+      "step": 16550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.953498274393797e-05,
+      "loss": 1.4168,
+      "step": 16600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.953358208955224e-05,
+      "loss": 1.4074,
+      "step": 16650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.953218143516651e-05,
+      "loss": 1.4294,
+      "step": 16700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9530780780780783e-05,
+      "loss": 1.434,
+      "step": 16750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9529380126395055e-05,
+      "loss": 1.4247,
+      "step": 16800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.952797947200932e-05,
+      "loss": 1.4213,
+      "step": 16850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.95265788176236e-05,
+      "loss": 1.4084,
+      "step": 16900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.952517816323787e-05,
+      "loss": 1.4367,
+      "step": 16950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9523777508852135e-05,
+      "loss": 1.3964,
+      "step": 17000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9522376854466406e-05,
+      "loss": 1.4384,
+      "step": 17050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9520976200080685e-05,
+      "loss": 1.3951,
+      "step": 17100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.951957554569495e-05,
+      "loss": 1.4163,
+      "step": 17150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.951817489130922e-05,
+      "loss": 1.4204,
+      "step": 17200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.951677423692349e-05,
+      "loss": 1.3999,
+      "step": 17250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9515373582537765e-05,
+      "loss": 1.3874,
+      "step": 17300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.9513972928152036e-05,
+      "loss": 1.3935,
+      "step": 17350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.951257227376631e-05,
+      "loss": 1.3938,
+      "step": 17400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.951117161938057e-05,
+      "loss": 1.4439,
+      "step": 17450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.950977096499485e-05,
+      "loss": 1.4088,
+      "step": 17500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.950837031060912e-05,
+      "loss": 1.4006,
+      "step": 17550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.950696965622339e-05,
+      "loss": 1.4287,
+      "step": 17600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.950556900183766e-05,
+      "loss": 1.3972,
+      "step": 17650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.950416834745194e-05,
+      "loss": 1.3971,
+      "step": 17700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.95027676930662e-05,
+      "loss": 1.4028,
+      "step": 17750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9501367038680474e-05,
+      "loss": 1.4041,
+      "step": 17800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9499966384294746e-05,
+      "loss": 1.3915,
+      "step": 17850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.949856572990901e-05,
+      "loss": 1.4005,
+      "step": 17900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.949716507552329e-05,
+      "loss": 1.4176,
+      "step": 17950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.949576442113756e-05,
+      "loss": 1.3866,
+      "step": 18000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9494363766751826e-05,
+      "loss": 1.3905,
+      "step": 18050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.94929631123661e-05,
+      "loss": 1.3843,
+      "step": 18100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.9491562457980376e-05,
+      "loss": 1.4094,
+      "step": 18150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.949016180359464e-05,
+      "loss": 1.3784,
+      "step": 18200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.948876114920891e-05,
+      "loss": 1.4126,
+      "step": 18250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9487360494823184e-05,
+      "loss": 1.3982,
+      "step": 18300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9485959840437455e-05,
+      "loss": 1.3894,
+      "step": 18350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.948455918605173e-05,
+      "loss": 1.4069,
+      "step": 18400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9483158531666e-05,
+      "loss": 1.3994,
+      "step": 18450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9481757877280264e-05,
+      "loss": 1.3741,
+      "step": 18500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.948035722289454e-05,
+      "loss": 1.3939,
+      "step": 18550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.9478956568508814e-05,
+      "loss": 1.3632,
+      "step": 18600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.947755591412308e-05,
+      "loss": 1.3861,
+      "step": 18650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.947615525973735e-05,
+      "loss": 1.4011,
+      "step": 18700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.947475460535163e-05,
+      "loss": 1.421,
+      "step": 18750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9473353950965893e-05,
+      "loss": 1.3762,
+      "step": 18800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.9471953296580165e-05,
+      "loss": 1.378,
+      "step": 18850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.947055264219444e-05,
+      "loss": 1.3635,
+      "step": 18900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.94691519878087e-05,
+      "loss": 1.3582,
+      "step": 18950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.946775133342298e-05,
+      "loss": 1.3744,
+      "step": 19000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.946635067903725e-05,
+      "loss": 1.3787,
+      "step": 19050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9464950024651516e-05,
+      "loss": 1.3899,
+      "step": 19100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.946354937026579e-05,
+      "loss": 1.3576,
+      "step": 19150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9462148715880066e-05,
+      "loss": 1.369,
+      "step": 19200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.946074806149433e-05,
+      "loss": 1.3648,
+      "step": 19250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.94593474071086e-05,
+      "loss": 1.3824,
+      "step": 19300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9457946752722875e-05,
+      "loss": 1.3722,
+      "step": 19350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.9456546098337146e-05,
+      "loss": 1.3762,
+      "step": 19400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.945514544395142e-05,
+      "loss": 1.3751,
+      "step": 19450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.945374478956569e-05,
+      "loss": 1.408,
+      "step": 19500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9452344135179954e-05,
+      "loss": 1.3693,
+      "step": 19550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.945094348079423e-05,
+      "loss": 1.357,
+      "step": 19600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.9449542826408504e-05,
+      "loss": 1.3611,
+      "step": 19650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.944814217202277e-05,
+      "loss": 1.3758,
+      "step": 19700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.944674151763704e-05,
+      "loss": 1.3733,
+      "step": 19750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.944534086325132e-05,
+      "loss": 1.3941,
+      "step": 19800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9443940208865584e-05,
+      "loss": 1.3382,
+      "step": 19850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.9442539554479856e-05,
+      "loss": 1.3743,
+      "step": 19900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.944113890009413e-05,
+      "loss": 1.4127,
+      "step": 19950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.943973824570839e-05,
+      "loss": 1.3587,
+      "step": 20000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.943833759132267e-05,
+      "loss": 1.3634,
+      "step": 20050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.943693693693694e-05,
+      "loss": 1.3624,
+      "step": 20100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.943553628255121e-05,
+      "loss": 1.3669,
+      "step": 20150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.943413562816548e-05,
+      "loss": 1.385,
+      "step": 20200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.943273497377976e-05,
+      "loss": 1.3645,
+      "step": 20250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.943133431939402e-05,
+      "loss": 1.3695,
+      "step": 20300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9429933665008294e-05,
+      "loss": 1.339,
+      "step": 20350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.9428533010622565e-05,
+      "loss": 1.3919,
+      "step": 20400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.942713235623684e-05,
+      "loss": 1.3446,
+      "step": 20450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.942573170185111e-05,
+      "loss": 1.3645,
+      "step": 20500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.942433104746538e-05,
+      "loss": 1.3542,
+      "step": 20550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.9422930393079645e-05,
+      "loss": 1.3525,
+      "step": 20600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9421529738693924e-05,
+      "loss": 1.3822,
+      "step": 20650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9420129084308195e-05,
+      "loss": 1.3757,
+      "step": 20700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.941872842992246e-05,
+      "loss": 1.3809,
+      "step": 20750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.941732777553673e-05,
+      "loss": 1.3601,
+      "step": 20800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.9415927121151e-05,
+      "loss": 1.3926,
+      "step": 20850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.9414526466765275e-05,
+      "loss": 1.3775,
+      "step": 20900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.941312581237955e-05,
+      "loss": 1.379,
+      "step": 20950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.941172515799382e-05,
+      "loss": 1.376,
+      "step": 21000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.941032450360808e-05,
+      "loss": 1.3769,
+      "step": 21050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.940892384922236e-05,
+      "loss": 1.3567,
+      "step": 21100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.940752319483663e-05,
+      "loss": 1.363,
+      "step": 21150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.94061225404509e-05,
+      "loss": 1.3813,
+      "step": 21200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.940472188606517e-05,
+      "loss": 1.3464,
+      "step": 21250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.940332123167945e-05,
+      "loss": 1.3706,
+      "step": 21300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.940192057729371e-05,
+      "loss": 1.3981,
+      "step": 21350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9400519922907985e-05,
+      "loss": 1.3589,
+      "step": 21400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9399119268522256e-05,
+      "loss": 1.3631,
+      "step": 21450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.939771861413653e-05,
+      "loss": 1.3687,
+      "step": 21500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.93963179597508e-05,
+      "loss": 1.3863,
+      "step": 21550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.939491730536507e-05,
+      "loss": 1.3843,
+      "step": 21600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9393516650979336e-05,
+      "loss": 1.3488,
+      "step": 21650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9392115996593614e-05,
+      "loss": 1.3388,
+      "step": 21700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9390715342207886e-05,
+      "loss": 1.3669,
+      "step": 21750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.938931468782215e-05,
+      "loss": 1.3672,
+      "step": 21800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.938791403343642e-05,
+      "loss": 1.358,
+      "step": 21850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9386513379050694e-05,
+      "loss": 1.3578,
+      "step": 21900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9385112724664966e-05,
+      "loss": 1.3651,
+      "step": 21950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.938371207027924e-05,
+      "loss": 1.3612,
+      "step": 22000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.938231141589351e-05,
+      "loss": 1.3578,
+      "step": 22050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.938091076150778e-05,
+      "loss": 1.3665,
+      "step": 22100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.937951010712205e-05,
+      "loss": 1.338,
+      "step": 22150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9378109452736324e-05,
+      "loss": 1.3644,
+      "step": 22200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.937670879835059e-05,
+      "loss": 1.3763,
+      "step": 22250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.937530814396486e-05,
+      "loss": 1.367,
+      "step": 22300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.937390748957913e-05,
+      "loss": 1.3693,
+      "step": 22350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9372506835193404e-05,
+      "loss": 1.3245,
+      "step": 22400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.9371106180807675e-05,
+      "loss": 1.3796,
+      "step": 22450
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.936970552642195e-05,
+      "loss": 1.3595,
+      "step": 22500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.936830487203622e-05,
+      "loss": 1.3561,
+      "step": 22550
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.936690421765049e-05,
+      "loss": 1.3485,
+      "step": 22600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.936550356326476e-05,
+      "loss": 1.3683,
+      "step": 22650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.936410290887903e-05,
+      "loss": 1.3529,
+      "step": 22700
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.9362702254493305e-05,
+      "loss": 1.3511,
+      "step": 22750
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.936130160010758e-05,
+      "loss": 1.3567,
+      "step": 22800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.935990094572184e-05,
+      "loss": 1.3715,
+      "step": 22850
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.935850029133611e-05,
+      "loss": 1.3675,
+      "step": 22900
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.9357099636950385e-05,
+      "loss": 1.3675,
+      "step": 22950
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.9355698982564657e-05,
+      "loss": 1.3236,
+      "step": 23000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.935429832817893e-05,
+      "loss": 1.3666,
+      "step": 23050
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.93528976737932e-05,
+      "loss": 1.3429,
+      "step": 23100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.935149701940747e-05,
+      "loss": 1.3192,
+      "step": 23150
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.935009636502174e-05,
+      "loss": 1.3289,
+      "step": 23200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.9348695710636015e-05,
+      "loss": 1.3412,
+      "step": 23250
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.934729505625028e-05,
+      "loss": 1.3848,
+      "step": 23300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.934589440186455e-05,
+      "loss": 1.3172,
+      "step": 23350
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.934449374747882e-05,
+      "loss": 1.3364,
+      "step": 23400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.9343093093093095e-05,
+      "loss": 1.3367,
+      "step": 23450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.9341692438707366e-05,
+      "loss": 1.3251,
+      "step": 23500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.934029178432164e-05,
+      "loss": 1.3753,
+      "step": 23550
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.933889112993591e-05,
+      "loss": 1.35,
+      "step": 23600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.933749047555018e-05,
+      "loss": 1.3284,
+      "step": 23650
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.933608982116445e-05,
+      "loss": 1.363,
+      "step": 23700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.933468916677872e-05,
+      "loss": 1.3063,
+      "step": 23750
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.9333288512392996e-05,
+      "loss": 1.3346,
+      "step": 23800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.933188785800726e-05,
+      "loss": 1.3458,
+      "step": 23850
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.933048720362153e-05,
+      "loss": 1.3224,
+      "step": 23900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.9329086549235804e-05,
+      "loss": 1.3349,
+      "step": 23950
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.9327685894850076e-05,
+      "loss": 1.3494,
+      "step": 24000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.932628524046435e-05,
+      "loss": 1.3102,
+      "step": 24050
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.932488458607862e-05,
+      "loss": 1.2964,
+      "step": 24100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.932348393169289e-05,
+      "loss": 1.341,
+      "step": 24150
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.932208327730716e-05,
+      "loss": 1.3287,
+      "step": 24200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.9320682622921434e-05,
+      "loss": 1.3267,
+      "step": 24250
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.93192819685357e-05,
+      "loss": 1.3072,
+      "step": 24300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.931788131414997e-05,
+      "loss": 1.315,
+      "step": 24350
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.931648065976424e-05,
+      "loss": 1.3177,
+      "step": 24400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.9315080005378514e-05,
+      "loss": 1.3313,
+      "step": 24450
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.9313679350992785e-05,
+      "loss": 1.3679,
+      "step": 24500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.931227869660706e-05,
+      "loss": 1.3505,
+      "step": 24550
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.931087804222133e-05,
+      "loss": 1.324,
+      "step": 24600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.93094773878356e-05,
+      "loss": 1.3178,
+      "step": 24650
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.930807673344987e-05,
+      "loss": 1.3465,
+      "step": 24700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.9306676079064143e-05,
+      "loss": 1.3618,
+      "step": 24750
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.930527542467841e-05,
+      "loss": 1.3517,
+      "step": 24800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.930387477029269e-05,
+      "loss": 1.3166,
+      "step": 24850
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.930247411590695e-05,
+      "loss": 1.329,
+      "step": 24900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.930107346152122e-05,
+      "loss": 1.3431,
+      "step": 24950
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.9299672807135495e-05,
+      "loss": 1.3152,
+      "step": 25000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.9298272152749767e-05,
+      "loss": 1.3264,
+      "step": 25050
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.929687149836404e-05,
+      "loss": 1.3168,
+      "step": 25100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.929547084397831e-05,
+      "loss": 1.3626,
+      "step": 25150
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.929407018959258e-05,
+      "loss": 1.3371,
+      "step": 25200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.929266953520685e-05,
+      "loss": 1.3076,
+      "step": 25250
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.9291268880821125e-05,
+      "loss": 1.3047,
+      "step": 25300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.928986822643539e-05,
+      "loss": 1.3129,
+      "step": 25350
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.928846757204966e-05,
+      "loss": 1.3037,
+      "step": 25400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.928706691766394e-05,
+      "loss": 1.3246,
+      "step": 25450
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.9285666263278204e-05,
+      "loss": 1.3469,
+      "step": 25500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.9284265608892476e-05,
+      "loss": 1.3285,
+      "step": 25550
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.928286495450675e-05,
+      "loss": 1.3412,
+      "step": 25600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.928146430012102e-05,
+      "loss": 1.3245,
+      "step": 25650
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.928006364573529e-05,
+      "loss": 1.3652,
+      "step": 25700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.927866299134956e-05,
+      "loss": 1.3307,
+      "step": 25750
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.927726233696383e-05,
+      "loss": 1.3644,
+      "step": 25800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.92758616825781e-05,
+      "loss": 1.3049,
+      "step": 25850
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.927446102819238e-05,
+      "loss": 1.3461,
+      "step": 25900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.927306037380664e-05,
+      "loss": 1.3252,
+      "step": 25950
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.9271659719420914e-05,
+      "loss": 1.3157,
+      "step": 26000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.9270259065035186e-05,
+      "loss": 1.3196,
+      "step": 26050
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.926885841064946e-05,
+      "loss": 1.332,
+      "step": 26100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.926745775626373e-05,
+      "loss": 1.3552,
+      "step": 26150
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.9266057101878e-05,
+      "loss": 1.3087,
+      "step": 26200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.926465644749227e-05,
+      "loss": 1.3369,
+      "step": 26250
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.9263255793106544e-05,
+      "loss": 1.3391,
+      "step": 26300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.9261855138720815e-05,
+      "loss": 1.3279,
+      "step": 26350
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.926045448433508e-05,
+      "loss": 1.3224,
+      "step": 26400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.925905382994935e-05,
+      "loss": 1.3283,
+      "step": 26450
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.925765317556363e-05,
+      "loss": 1.3316,
+      "step": 26500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.9256252521177895e-05,
+      "loss": 1.307,
+      "step": 26550
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.925485186679217e-05,
+      "loss": 1.3205,
+      "step": 26600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.925345121240644e-05,
+      "loss": 1.326,
+      "step": 26650
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.925205055802071e-05,
+      "loss": 1.3374,
+      "step": 26700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.925064990363498e-05,
+      "loss": 1.331,
+      "step": 26750
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.9249249249249253e-05,
+      "loss": 1.3382,
+      "step": 26800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.924784859486352e-05,
+      "loss": 1.3381,
+      "step": 26850
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.924644794047779e-05,
+      "loss": 1.3337,
+      "step": 26900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.924504728609207e-05,
+      "loss": 1.321,
+      "step": 26950
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.924364663170633e-05,
+      "loss": 1.3078,
+      "step": 27000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.9242245977320605e-05,
+      "loss": 1.3166,
+      "step": 27050
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.9240845322934876e-05,
+      "loss": 1.3202,
+      "step": 27100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.923944466854915e-05,
+      "loss": 1.2926,
+      "step": 27150
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.923804401416342e-05,
+      "loss": 1.3176,
+      "step": 27200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.923664335977769e-05,
+      "loss": 1.2848,
+      "step": 27250
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.9235242705391956e-05,
+      "loss": 1.2999,
+      "step": 27300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.9233842051006235e-05,
+      "loss": 1.3205,
+      "step": 27350
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.9232441396620506e-05,
+      "loss": 1.2959,
+      "step": 27400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.923104074223477e-05,
+      "loss": 1.3133,
+      "step": 27450
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.922964008784904e-05,
+      "loss": 1.3072,
+      "step": 27500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.922823943346332e-05,
+      "loss": 1.2885,
+      "step": 27550
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.9226838779077586e-05,
+      "loss": 1.3363,
+      "step": 27600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.922543812469186e-05,
+      "loss": 1.3299,
+      "step": 27650
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.922403747030613e-05,
+      "loss": 1.2851,
+      "step": 27700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.92226368159204e-05,
+      "loss": 1.2899,
+      "step": 27750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.922123616153467e-05,
+      "loss": 1.2967,
+      "step": 27800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.9219835507148944e-05,
+      "loss": 1.3213,
+      "step": 27850
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.921843485276321e-05,
+      "loss": 1.2822,
+      "step": 27900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.921703419837748e-05,
+      "loss": 1.2979,
+      "step": 27950
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.921563354399176e-05,
+      "loss": 1.3157,
+      "step": 28000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.9214232889606024e-05,
+      "loss": 1.3302,
+      "step": 28050
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.9212832235220296e-05,
+      "loss": 1.3277,
+      "step": 28100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.921143158083457e-05,
+      "loss": 1.3079,
+      "step": 28150
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.921003092644884e-05,
+      "loss": 1.3159,
+      "step": 28200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.920863027206311e-05,
+      "loss": 1.2898,
+      "step": 28250
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.920722961767738e-05,
+      "loss": 1.2955,
+      "step": 28300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.920582896329165e-05,
+      "loss": 1.328,
+      "step": 28350
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.9204428308905925e-05,
+      "loss": 1.3313,
+      "step": 28400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.92030276545202e-05,
+      "loss": 1.3039,
+      "step": 28450
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.920162700013446e-05,
+      "loss": 1.3175,
+      "step": 28500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.9200226345748734e-05,
+      "loss": 1.3283,
+      "step": 28550
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.919882569136301e-05,
+      "loss": 1.3091,
+      "step": 28600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.919742503697728e-05,
+      "loss": 1.3048,
+      "step": 28650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.919602438259155e-05,
+      "loss": 1.2843,
+      "step": 28700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.919462372820582e-05,
+      "loss": 1.3218,
+      "step": 28750
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.919322307382009e-05,
+      "loss": 1.3225,
+      "step": 28800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.9191822419434363e-05,
+      "loss": 1.2906,
+      "step": 28850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.9190421765048635e-05,
+      "loss": 1.3013,
+      "step": 28900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.91890211106629e-05,
+      "loss": 1.3285,
+      "step": 28950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.918762045627717e-05,
+      "loss": 1.2895,
+      "step": 29000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.918621980189145e-05,
+      "loss": 1.2707,
+      "step": 29050
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.9184819147505715e-05,
+      "loss": 1.3069,
+      "step": 29100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.9183418493119986e-05,
+      "loss": 1.3097,
+      "step": 29150
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.918201783873426e-05,
+      "loss": 1.3229,
+      "step": 29200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.918061718434853e-05,
+      "loss": 1.3097,
+      "step": 29250
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.91792165299628e-05,
+      "loss": 1.3154,
+      "step": 29300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.917781587557707e-05,
+      "loss": 1.3012,
+      "step": 29350
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.917641522119134e-05,
+      "loss": 1.2962,
+      "step": 29400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.9175014566805616e-05,
+      "loss": 1.3123,
+      "step": 29450
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.917361391241989e-05,
+      "loss": 1.2856,
+      "step": 29500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.917221325803415e-05,
+      "loss": 1.3,
+      "step": 29550
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.9170812603648424e-05,
+      "loss": 1.2847,
+      "step": 29600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.91694119492627e-05,
+      "loss": 1.3098,
+      "step": 29650
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.916801129487697e-05,
+      "loss": 1.3073,
+      "step": 29700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.916661064049124e-05,
+      "loss": 1.332,
+      "step": 29750
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.916520998610551e-05,
+      "loss": 1.3081,
+      "step": 29800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.916380933171978e-05,
+      "loss": 1.2914,
+      "step": 29850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.9162408677334054e-05,
+      "loss": 1.2642,
+      "step": 29900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.9161008022948326e-05,
+      "loss": 1.2792,
+      "step": 29950
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.915960736856259e-05,
+      "loss": 1.3048,
+      "step": 30000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.915820671417686e-05,
+      "loss": 1.2826,
+      "step": 30050
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.915680605979114e-05,
+      "loss": 1.2759,
+      "step": 30100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9155405405405406e-05,
+      "loss": 1.2902,
+      "step": 30150
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.915400475101968e-05,
+      "loss": 1.3175,
+      "step": 30200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.915260409663395e-05,
+      "loss": 1.2784,
+      "step": 30250
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.915120344224822e-05,
+      "loss": 1.2925,
+      "step": 30300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.914980278786249e-05,
+      "loss": 1.3032,
+      "step": 30350
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9148402133476764e-05,
+      "loss": 1.2928,
+      "step": 30400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.914700147909103e-05,
+      "loss": 1.3201,
+      "step": 30450
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.914560082470531e-05,
+      "loss": 1.2744,
+      "step": 30500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.914420017031958e-05,
+      "loss": 1.2872,
+      "step": 30550
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.9142799515933844e-05,
+      "loss": 1.2961,
+      "step": 30600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.9141398861548115e-05,
+      "loss": 1.3028,
+      "step": 30650
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.9139998207162394e-05,
+      "loss": 1.2775,
+      "step": 30700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.913859755277666e-05,
+      "loss": 1.2889,
+      "step": 30750
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.913719689839093e-05,
+      "loss": 1.2834,
+      "step": 30800
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.91357962440052e-05,
+      "loss": 1.2941,
+      "step": 30850
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.913439558961947e-05,
+      "loss": 1.3068,
+      "step": 30900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.9132994935233745e-05,
+      "loss": 1.3122,
+      "step": 30950
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.913159428084802e-05,
+      "loss": 1.3007,
+      "step": 31000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.913019362646228e-05,
+      "loss": 1.2985,
+      "step": 31050
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.912879297207655e-05,
+      "loss": 1.296,
+      "step": 31100
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.912739231769083e-05,
+      "loss": 1.2782,
+      "step": 31150
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.9125991663305096e-05,
+      "loss": 1.3056,
+      "step": 31200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.912459100891937e-05,
+      "loss": 1.2807,
+      "step": 31250
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.912319035453364e-05,
+      "loss": 1.3004,
+      "step": 31300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.912178970014791e-05,
+      "loss": 1.2904,
+      "step": 31350
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.912038904576218e-05,
+      "loss": 1.2898,
+      "step": 31400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.9118988391376455e-05,
+      "loss": 1.2742,
+      "step": 31450
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.911758773699072e-05,
+      "loss": 1.3027,
+      "step": 31500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.9116187082605e-05,
+      "loss": 1.3086,
+      "step": 31550
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.911478642821927e-05,
+      "loss": 1.2734,
+      "step": 31600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.9113385773833534e-05,
+      "loss": 1.2884,
+      "step": 31650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.9111985119447806e-05,
+      "loss": 1.3396,
+      "step": 31700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.9110584465062084e-05,
+      "loss": 1.3018,
+      "step": 31750
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.910918381067635e-05,
+      "loss": 1.2828,
+      "step": 31800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.910778315629062e-05,
+      "loss": 1.3015,
+      "step": 31850
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.910638250190489e-05,
+      "loss": 1.3118,
+      "step": 31900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.9104981847519164e-05,
+      "loss": 1.288,
+      "step": 31950
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.9103581193133436e-05,
+      "loss": 1.288,
+      "step": 32000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.910218053874771e-05,
+      "loss": 1.2914,
+      "step": 32050
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.910077988436197e-05,
+      "loss": 1.2784,
+      "step": 32100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.9099379229976244e-05,
+      "loss": 1.2836,
+      "step": 32150
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.909797857559052e-05,
+      "loss": 1.2589,
+      "step": 32200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.909657792120479e-05,
+      "loss": 1.3241,
+      "step": 32250
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.909517726681906e-05,
+      "loss": 1.2986,
+      "step": 32300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.909377661243333e-05,
+      "loss": 1.2801,
+      "step": 32350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.90923759580476e-05,
+      "loss": 1.2766,
+      "step": 32400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.9090975303661874e-05,
+      "loss": 1.313,
+      "step": 32450
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.9089574649276145e-05,
+      "loss": 1.2746,
+      "step": 32500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.908817399489041e-05,
+      "loss": 1.3009,
+      "step": 32550
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.908677334050469e-05,
+      "loss": 1.2809,
+      "step": 32600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.908537268611896e-05,
+      "loss": 1.2728,
+      "step": 32650
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.9083972031733225e-05,
+      "loss": 1.2999,
+      "step": 32700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.90825713773475e-05,
+      "loss": 1.2662,
+      "step": 32750
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.9081170722961775e-05,
+      "loss": 1.2843,
+      "step": 32800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.907977006857604e-05,
+      "loss": 1.2862,
+      "step": 32850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.907836941419031e-05,
+      "loss": 1.2926,
+      "step": 32900
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.907696875980458e-05,
+      "loss": 1.2801,
+      "step": 32950
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9075568105418855e-05,
+      "loss": 1.3205,
+      "step": 33000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.9074167451033127e-05,
+      "loss": 1.2809,
+      "step": 33050
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.90727667966474e-05,
+      "loss": 1.2742,
+      "step": 33100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.907136614226166e-05,
+      "loss": 1.2854,
+      "step": 33150
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.906996548787594e-05,
+      "loss": 1.2648,
+      "step": 33200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.906856483349021e-05,
+      "loss": 1.3033,
+      "step": 33250
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.906716417910448e-05,
+      "loss": 1.2548,
+      "step": 33300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.906576352471875e-05,
+      "loss": 1.3158,
+      "step": 33350
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.906436287033302e-05,
+      "loss": 1.3022,
+      "step": 33400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.906296221594729e-05,
+      "loss": 1.2948,
+      "step": 33450
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.9061561561561565e-05,
+      "loss": 1.2916,
+      "step": 33500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.9060160907175836e-05,
+      "loss": 1.285,
+      "step": 33550
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.90587602527901e-05,
+      "loss": 1.2831,
+      "step": 33600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.905735959840438e-05,
+      "loss": 1.2866,
+      "step": 33650
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.905595894401865e-05,
+      "loss": 1.266,
+      "step": 33700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.9054558289632916e-05,
+      "loss": 1.2396,
+      "step": 33750
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.905315763524719e-05,
+      "loss": 1.2913,
+      "step": 33800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.9051756980861466e-05,
+      "loss": 1.321,
+      "step": 33850
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.905035632647573e-05,
+      "loss": 1.2768,
+      "step": 33900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.904895567209e-05,
+      "loss": 1.2767,
+      "step": 33950
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.9047555017704274e-05,
+      "loss": 1.2707,
+      "step": 34000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.9046154363318546e-05,
+      "loss": 1.2705,
+      "step": 34050
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.904475370893282e-05,
+      "loss": 1.3089,
+      "step": 34100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.904335305454709e-05,
+      "loss": 1.2597,
+      "step": 34150
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.9041952400161354e-05,
+      "loss": 1.2656,
+      "step": 34200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.904055174577563e-05,
+      "loss": 1.2772,
+      "step": 34250
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.9039151091389904e-05,
+      "loss": 1.2838,
+      "step": 34300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.903775043700417e-05,
+      "loss": 1.2743,
+      "step": 34350
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.903634978261844e-05,
+      "loss": 1.2745,
+      "step": 34400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.903494912823271e-05,
+      "loss": 1.2955,
+      "step": 34450
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.9033548473846984e-05,
+      "loss": 1.2695,
+      "step": 34500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.9032147819461255e-05,
+      "loss": 1.2664,
+      "step": 34550
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.903074716507553e-05,
+      "loss": 1.3064,
+      "step": 34600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.902934651068979e-05,
+      "loss": 1.275,
+      "step": 34650
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.902794585630407e-05,
+      "loss": 1.2841,
+      "step": 34700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.902654520191834e-05,
+      "loss": 1.2625,
+      "step": 34750
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.902514454753261e-05,
+      "loss": 1.287,
+      "step": 34800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.902374389314688e-05,
+      "loss": 1.2852,
+      "step": 34850
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.902234323876116e-05,
+      "loss": 1.2593,
+      "step": 34900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.902094258437542e-05,
+      "loss": 1.2636,
+      "step": 34950
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.901954192998969e-05,
+      "loss": 1.2938,
+      "step": 35000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.9018141275603965e-05,
+      "loss": 1.2508,
+      "step": 35050
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.9016740621218237e-05,
+      "loss": 1.3109,
+      "step": 35100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.901533996683251e-05,
+      "loss": 1.2902,
+      "step": 35150
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.901393931244678e-05,
+      "loss": 1.2899,
+      "step": 35200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.9012538658061045e-05,
+      "loss": 1.2717,
+      "step": 35250
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.901113800367532e-05,
+      "loss": 1.2707,
+      "step": 35300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.9009737349289595e-05,
+      "loss": 1.3071,
+      "step": 35350
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.900833669490386e-05,
+      "loss": 1.279,
+      "step": 35400
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.900693604051813e-05,
+      "loss": 1.269,
+      "step": 35450
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.90055353861324e-05,
+      "loss": 1.2757,
+      "step": 35500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.9004134731746674e-05,
+      "loss": 1.2808,
+      "step": 35550
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.9002734077360946e-05,
+      "loss": 1.2825,
+      "step": 35600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.900133342297522e-05,
+      "loss": 1.2902,
+      "step": 35650
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.899993276858948e-05,
+      "loss": 1.2926,
+      "step": 35700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.899853211420376e-05,
+      "loss": 1.2941,
+      "step": 35750
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.899713145981803e-05,
+      "loss": 1.2875,
+      "step": 35800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.89957308054323e-05,
+      "loss": 1.27,
+      "step": 35850
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.899433015104657e-05,
+      "loss": 1.2614,
+      "step": 35900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.899292949666085e-05,
+      "loss": 1.2542,
+      "step": 35950
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.899152884227511e-05,
+      "loss": 1.2709,
+      "step": 36000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.8990128187889384e-05,
+      "loss": 1.2664,
+      "step": 36050
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.8988727533503656e-05,
+      "loss": 1.2685,
+      "step": 36100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.898732687911793e-05,
+      "loss": 1.2621,
+      "step": 36150
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.89859262247322e-05,
+      "loss": 1.2873,
+      "step": 36200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.898452557034647e-05,
+      "loss": 1.2842,
+      "step": 36250
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.8983124915960735e-05,
+      "loss": 1.2622,
+      "step": 36300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.8981724261575014e-05,
+      "loss": 1.2695,
+      "step": 36350
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.8980323607189285e-05,
+      "loss": 1.2492,
+      "step": 36400
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.897892295280355e-05,
+      "loss": 1.2736,
+      "step": 36450
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.897752229841782e-05,
+      "loss": 1.2612,
+      "step": 36500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.8976121644032094e-05,
+      "loss": 1.2644,
+      "step": 36550
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.8974720989646365e-05,
+      "loss": 1.2708,
+      "step": 36600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.897332033526064e-05,
+      "loss": 1.2738,
+      "step": 36650
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.897191968087491e-05,
+      "loss": 1.26,
+      "step": 36700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.897051902648917e-05,
+      "loss": 1.2805,
+      "step": 36750
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.896911837210345e-05,
+      "loss": 1.2871,
+      "step": 36800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.8967717717717723e-05,
+      "loss": 1.284,
+      "step": 36850
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.896631706333199e-05,
+      "loss": 1.2666,
+      "step": 36900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.896491640894626e-05,
+      "loss": 1.261,
+      "step": 36950
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.896351575456054e-05,
+      "loss": 1.2892,
+      "step": 37000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.89621151001748e-05,
+      "loss": 1.2672,
+      "step": 37050
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.8960714445789075e-05,
+      "loss": 1.2791,
+      "step": 37100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.8959313791403346e-05,
+      "loss": 1.283,
+      "step": 37150
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.895791313701762e-05,
+      "loss": 1.2304,
+      "step": 37200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.895651248263189e-05,
+      "loss": 1.2807,
+      "step": 37250
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.895511182824616e-05,
+      "loss": 1.2712,
+      "step": 37300
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.8953711173860426e-05,
+      "loss": 1.2533,
+      "step": 37350
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8952310519474705e-05,
+      "loss": 1.2567,
+      "step": 37400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8950909865088976e-05,
+      "loss": 1.259,
+      "step": 37450
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.894950921070324e-05,
+      "loss": 1.2832,
+      "step": 37500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.894810855631751e-05,
+      "loss": 1.2656,
+      "step": 37550
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.8946707901931784e-05,
+      "loss": 1.2681,
+      "step": 37600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.8945307247546056e-05,
+      "loss": 1.2587,
+      "step": 37650
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.894390659316033e-05,
+      "loss": 1.263,
+      "step": 37700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.89425059387746e-05,
+      "loss": 1.2906,
+      "step": 37750
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.8941105284388864e-05,
+      "loss": 1.241,
+      "step": 37800
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.893970463000314e-05,
+      "loss": 1.2742,
+      "step": 37850
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.8938303975617414e-05,
+      "loss": 1.2652,
+      "step": 37900
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.893690332123168e-05,
+      "loss": 1.2572,
+      "step": 37950
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.893550266684595e-05,
+      "loss": 1.2905,
+      "step": 38000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.893410201246022e-05,
+      "loss": 1.2526,
+      "step": 38050
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.8932701358074494e-05,
+      "loss": 1.2871,
+      "step": 38100
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.8931300703688766e-05,
+      "loss": 1.2729,
+      "step": 38150
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.892990004930304e-05,
+      "loss": 1.2516,
+      "step": 38200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.892849939491731e-05,
+      "loss": 1.2629,
+      "step": 38250
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.892709874053158e-05,
+      "loss": 1.253,
+      "step": 38300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.892569808614585e-05,
+      "loss": 1.2896,
+      "step": 38350
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.892429743176012e-05,
+      "loss": 1.2799,
+      "step": 38400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.8922896777374395e-05,
+      "loss": 1.2734,
+      "step": 38450
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.892149612298867e-05,
+      "loss": 1.2748,
+      "step": 38500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.892009546860293e-05,
+      "loss": 1.2544,
+      "step": 38550
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.8918694814217204e-05,
+      "loss": 1.2698,
+      "step": 38600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.8917294159831475e-05,
+      "loss": 1.2794,
+      "step": 38650
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.891589350544575e-05,
+      "loss": 1.2711,
+      "step": 38700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.891449285106002e-05,
+      "loss": 1.2647,
+      "step": 38750
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.891309219667429e-05,
+      "loss": 1.2614,
+      "step": 38800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.8911691542288555e-05,
+      "loss": 1.271,
+      "step": 38850
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.891029088790283e-05,
+      "loss": 1.2831,
+      "step": 38900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.8908890233517105e-05,
+      "loss": 1.2748,
+      "step": 38950
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.890748957913137e-05,
+      "loss": 1.2796,
+      "step": 39000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.890608892474564e-05,
+      "loss": 1.279,
+      "step": 39050
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.890468827035991e-05,
+      "loss": 1.2432,
+      "step": 39100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.8903287615974185e-05,
+      "loss": 1.2755,
+      "step": 39150
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.8901886961588456e-05,
+      "loss": 1.2341,
+      "step": 39200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.890048630720273e-05,
+      "loss": 1.2592,
+      "step": 39250
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.8899085652817e-05,
+      "loss": 1.2487,
+      "step": 39300
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.889768499843127e-05,
+      "loss": 1.2737,
+      "step": 39350
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.889628434404554e-05,
+      "loss": 1.2582,
+      "step": 39400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.889488368965981e-05,
+      "loss": 1.2543,
+      "step": 39450
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.8893483035274086e-05,
+      "loss": 1.2669,
+      "step": 39500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.889208238088835e-05,
+      "loss": 1.2544,
+      "step": 39550
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.889068172650262e-05,
+      "loss": 1.2764,
+      "step": 39600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.8889281072116894e-05,
+      "loss": 1.2518,
+      "step": 39650
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.8887880417731166e-05,
+      "loss": 1.2614,
+      "step": 39700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.888647976334544e-05,
+      "loss": 1.2271,
+      "step": 39750
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.888507910895971e-05,
+      "loss": 1.2771,
+      "step": 39800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.888367845457398e-05,
+      "loss": 1.2457,
+      "step": 39850
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.8882277800188246e-05,
+      "loss": 1.2609,
+      "step": 39900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.8880877145802524e-05,
+      "loss": 1.2579,
+      "step": 39950
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.887947649141679e-05,
+      "loss": 1.2853,
+      "step": 40000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.887807583703106e-05,
+      "loss": 1.2496,
+      "step": 40050
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.887667518264533e-05,
+      "loss": 1.2633,
+      "step": 40100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.8875274528259604e-05,
+      "loss": 1.2668,
+      "step": 40150
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.8873873873873876e-05,
+      "loss": 1.2649,
+      "step": 40200
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.887247321948815e-05,
+      "loss": 1.2554,
+      "step": 40250
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.887107256510242e-05,
+      "loss": 1.2577,
+      "step": 40300
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.886967191071669e-05,
+      "loss": 1.2373,
+      "step": 40350
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.886827125633096e-05,
+      "loss": 1.2435,
+      "step": 40400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.8866870601945234e-05,
+      "loss": 1.2559,
+      "step": 40450
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.88654699475595e-05,
+      "loss": 1.2583,
+      "step": 40500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.886406929317378e-05,
+      "loss": 1.2645,
+      "step": 40550
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.886266863878804e-05,
+      "loss": 1.274,
+      "step": 40600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.8861267984402314e-05,
+      "loss": 1.2543,
+      "step": 40650
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.8859867330016585e-05,
+      "loss": 1.2648,
+      "step": 40700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.885846667563086e-05,
+      "loss": 1.248,
+      "step": 40750
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.885706602124513e-05,
+      "loss": 1.2658,
+      "step": 40800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.88556653668594e-05,
+      "loss": 1.2763,
+      "step": 40850
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.885426471247367e-05,
+      "loss": 1.238,
+      "step": 40900
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.885286405808794e-05,
+      "loss": 1.2763,
+      "step": 40950
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.8851463403702215e-05,
+      "loss": 1.2537,
+      "step": 41000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.885006274931648e-05,
+      "loss": 1.2257,
+      "step": 41050
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.884866209493075e-05,
+      "loss": 1.2472,
+      "step": 41100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.884726144054502e-05,
+      "loss": 1.271,
+      "step": 41150
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.8845860786159295e-05,
+      "loss": 1.2635,
+      "step": 41200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.8844460131773566e-05,
+      "loss": 1.233,
+      "step": 41250
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.884305947738784e-05,
+      "loss": 1.2269,
+      "step": 41300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.884165882300211e-05,
+      "loss": 1.2529,
+      "step": 41350
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.884025816861638e-05,
+      "loss": 1.2595,
+      "step": 41400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.883885751423065e-05,
+      "loss": 1.23,
+      "step": 41450
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.883745685984492e-05,
+      "loss": 1.2752,
+      "step": 41500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.883605620545919e-05,
+      "loss": 1.2482,
+      "step": 41550
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.883465555107347e-05,
+      "loss": 1.2675,
+      "step": 41600
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.883325489668773e-05,
+      "loss": 1.2196,
+      "step": 41650
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.8831854242302004e-05,
+      "loss": 1.2275,
+      "step": 41700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.8830453587916276e-05,
+      "loss": 1.2709,
+      "step": 41750
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.882905293353055e-05,
+      "loss": 1.2194,
+      "step": 41800
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.882765227914482e-05,
+      "loss": 1.2408,
+      "step": 41850
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.882625162475909e-05,
+      "loss": 1.2282,
+      "step": 41900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.882485097037336e-05,
+      "loss": 1.2642,
+      "step": 41950
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.8823450315987634e-05,
+      "loss": 1.225,
+      "step": 42000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.8822049661601906e-05,
+      "loss": 1.2541,
+      "step": 42050
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.882064900721617e-05,
+      "loss": 1.2723,
+      "step": 42100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.881924835283044e-05,
+      "loss": 1.233,
+      "step": 42150
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8817847698444714e-05,
+      "loss": 1.2642,
+      "step": 42200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.8816447044058986e-05,
+      "loss": 1.2538,
+      "step": 42250
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.881504638967326e-05,
+      "loss": 1.2425,
+      "step": 42300
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.881364573528753e-05,
+      "loss": 1.2449,
+      "step": 42350
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.88122450809018e-05,
+      "loss": 1.2382,
+      "step": 42400
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.881084442651607e-05,
+      "loss": 1.1989,
+      "step": 42450
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.8809443772130344e-05,
+      "loss": 1.2473,
+      "step": 42500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.880804311774461e-05,
+      "loss": 1.2382,
+      "step": 42550
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.880664246335888e-05,
+      "loss": 1.2732,
+      "step": 42600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.880524180897316e-05,
+      "loss": 1.2389,
+      "step": 42650
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.8803841154587423e-05,
+      "loss": 1.2669,
+      "step": 42700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.8802440500201695e-05,
+      "loss": 1.2414,
+      "step": 42750
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.880103984581597e-05,
+      "loss": 1.2243,
+      "step": 42800
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.879963919143024e-05,
+      "loss": 1.2497,
+      "step": 42850
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.879823853704451e-05,
+      "loss": 1.2445,
+      "step": 42900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.879683788265878e-05,
+      "loss": 1.2558,
+      "step": 42950
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.8795437228273047e-05,
+      "loss": 1.2713,
+      "step": 43000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.8794036573887325e-05,
+      "loss": 1.2751,
+      "step": 43050
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.8792635919501597e-05,
+      "loss": 1.2432,
+      "step": 43100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.879123526511586e-05,
+      "loss": 1.2291,
+      "step": 43150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.878983461073013e-05,
+      "loss": 1.2433,
+      "step": 43200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.8788433956344405e-05,
+      "loss": 1.2496,
+      "step": 43250
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.8787033301958676e-05,
+      "loss": 1.256,
+      "step": 43300
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.878563264757295e-05,
+      "loss": 1.2297,
+      "step": 43350
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.878423199318722e-05,
+      "loss": 1.2287,
+      "step": 43400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.878283133880149e-05,
+      "loss": 1.2465,
+      "step": 43450
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.878143068441576e-05,
+      "loss": 1.2325,
+      "step": 43500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.8780030030030034e-05,
+      "loss": 1.2047,
+      "step": 43550
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.87786293756443e-05,
+      "loss": 1.2457,
+      "step": 43600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.877722872125857e-05,
+      "loss": 1.2407,
+      "step": 43650
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.877582806687285e-05,
+      "loss": 1.2483,
+      "step": 43700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.8774427412487114e-05,
+      "loss": 1.2554,
+      "step": 43750
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.8773026758101386e-05,
+      "loss": 1.2362,
+      "step": 43800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.877162610371566e-05,
+      "loss": 1.2089,
+      "step": 43850
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.877022544932993e-05,
+      "loss": 1.2505,
+      "step": 43900
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.87688247949442e-05,
+      "loss": 1.2423,
+      "step": 43950
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.876742414055847e-05,
+      "loss": 1.2723,
+      "step": 44000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.876602348617274e-05,
+      "loss": 1.2431,
+      "step": 44050
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.8764622831787016e-05,
+      "loss": 1.2646,
+      "step": 44100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.876322217740129e-05,
+      "loss": 1.23,
+      "step": 44150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.876182152301555e-05,
+      "loss": 1.2269,
+      "step": 44200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.8760420868629824e-05,
+      "loss": 1.2469,
+      "step": 44250
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.87590202142441e-05,
+      "loss": 1.2559,
+      "step": 44300
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.875761955985837e-05,
+      "loss": 1.2738,
+      "step": 44350
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.875621890547264e-05,
+      "loss": 1.2561,
+      "step": 44400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.875481825108691e-05,
+      "loss": 1.2353,
+      "step": 44450
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.8753417596701175e-05,
+      "loss": 1.2301,
+      "step": 44500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.8752016942315454e-05,
+      "loss": 1.2695,
+      "step": 44550
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.8750616287929725e-05,
+      "loss": 1.2462,
+      "step": 44600
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.874921563354399e-05,
+      "loss": 1.2627,
+      "step": 44650
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.874781497915826e-05,
+      "loss": 1.2441,
+      "step": 44700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.874641432477254e-05,
+      "loss": 1.2372,
+      "step": 44750
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.8745013670386805e-05,
+      "loss": 1.2305,
+      "step": 44800
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.874361301600108e-05,
+      "loss": 1.2258,
+      "step": 44850
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.874221236161535e-05,
+      "loss": 1.2477,
+      "step": 44900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.874081170722962e-05,
+      "loss": 1.2799,
+      "step": 44950
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.873941105284389e-05,
+      "loss": 1.2465,
+      "step": 45000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.873801039845816e-05,
+      "loss": 1.2296,
+      "step": 45050
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.873660974407243e-05,
+      "loss": 1.2494,
+      "step": 45100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.8735209089686707e-05,
+      "loss": 1.2313,
+      "step": 45150
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.873380843530098e-05,
+      "loss": 1.2327,
+      "step": 45200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.873240778091524e-05,
+      "loss": 1.2189,
+      "step": 45250
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.8731007126529515e-05,
+      "loss": 1.237,
+      "step": 45300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.872960647214379e-05,
+      "loss": 1.2046,
+      "step": 45350
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.872820581775806e-05,
+      "loss": 1.2447,
+      "step": 45400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.872680516337233e-05,
+      "loss": 1.2393,
+      "step": 45450
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.87254045089866e-05,
+      "loss": 1.2266,
+      "step": 45500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8724003854600866e-05,
+      "loss": 1.2423,
+      "step": 45550
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8722603200215144e-05,
+      "loss": 1.2575,
+      "step": 45600
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8721202545829416e-05,
+      "loss": 1.2202,
+      "step": 45650
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.871980189144368e-05,
+      "loss": 1.2145,
+      "step": 45700
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.871840123705795e-05,
+      "loss": 1.2246,
+      "step": 45750
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.871700058267223e-05,
+      "loss": 1.2407,
+      "step": 45800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.8715599928286496e-05,
+      "loss": 1.2289,
+      "step": 45850
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.871419927390077e-05,
+      "loss": 1.261,
+      "step": 45900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.871279861951504e-05,
+      "loss": 1.203,
+      "step": 45950
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.871139796512931e-05,
+      "loss": 1.2226,
+      "step": 46000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.870999731074358e-05,
+      "loss": 1.2498,
+      "step": 46050
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.8708596656357854e-05,
+      "loss": 1.2004,
+      "step": 46100
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.870719600197212e-05,
+      "loss": 1.2399,
+      "step": 46150
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.87057953475864e-05,
+      "loss": 1.2406,
+      "step": 46200
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.870439469320067e-05,
+      "loss": 1.2323,
+      "step": 46250
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.8702994038814934e-05,
+      "loss": 1.2299,
+      "step": 46300
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.8701593384429205e-05,
+      "loss": 1.2521,
+      "step": 46350
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.8700192730043484e-05,
+      "loss": 1.2284,
+      "step": 46400
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.869879207565775e-05,
+      "loss": 1.2275,
+      "step": 46450
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.869739142127202e-05,
+      "loss": 1.2482,
+      "step": 46500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.869599076688629e-05,
+      "loss": 1.2348,
+      "step": 46550
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.869459011250056e-05,
+      "loss": 1.2323,
+      "step": 46600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.8693189458114835e-05,
+      "loss": 1.2286,
+      "step": 46650
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.869178880372911e-05,
+      "loss": 1.2215,
+      "step": 46700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.869038814934337e-05,
+      "loss": 1.2196,
+      "step": 46750
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.868898749495764e-05,
+      "loss": 1.2544,
+      "step": 46800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.868758684057192e-05,
+      "loss": 1.2101,
+      "step": 46850
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.868618618618619e-05,
+      "loss": 1.2215,
+      "step": 46900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.868478553180046e-05,
+      "loss": 1.2258,
+      "step": 46950
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.868338487741473e-05,
+      "loss": 1.2388,
+      "step": 47000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.8681984223029e-05,
+      "loss": 1.2367,
+      "step": 47050
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.868058356864327e-05,
+      "loss": 1.2129,
+      "step": 47100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.8679182914257545e-05,
+      "loss": 1.2243,
+      "step": 47150
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.867778225987181e-05,
+      "loss": 1.229,
+      "step": 47200
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.867638160548609e-05,
+      "loss": 1.21,
+      "step": 47250
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.867498095110036e-05,
+      "loss": 1.2209,
+      "step": 47300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.8673580296714625e-05,
+      "loss": 1.2399,
+      "step": 47350
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.8672179642328896e-05,
+      "loss": 1.2307,
+      "step": 47400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.8670778987943175e-05,
+      "loss": 1.2654,
+      "step": 47450
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.866937833355744e-05,
+      "loss": 1.2266,
+      "step": 47500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.866797767917171e-05,
+      "loss": 1.1997,
+      "step": 47550
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.866657702478598e-05,
+      "loss": 1.234,
+      "step": 47600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.866517637040025e-05,
+      "loss": 1.2297,
+      "step": 47650
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.8663775716014526e-05,
+      "loss": 1.2273,
+      "step": 47700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.86623750616288e-05,
+      "loss": 1.2497,
+      "step": 47750
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.866097440724306e-05,
+      "loss": 1.2065,
+      "step": 47800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.8659573752857334e-05,
+      "loss": 1.2506,
+      "step": 47850
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.865817309847161e-05,
+      "loss": 1.2257,
+      "step": 47900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.865677244408588e-05,
+      "loss": 1.1959,
+      "step": 47950
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.865537178970015e-05,
+      "loss": 1.2571,
+      "step": 48000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.865397113531442e-05,
+      "loss": 1.2162,
+      "step": 48050
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.865257048092869e-05,
+      "loss": 1.2363,
+      "step": 48100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.8651169826542964e-05,
+      "loss": 1.2163,
+      "step": 48150
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.8649769172157236e-05,
+      "loss": 1.2321,
+      "step": 48200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.86483685177715e-05,
+      "loss": 1.2351,
+      "step": 48250
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.864696786338578e-05,
+      "loss": 1.2446,
+      "step": 48300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.864556720900005e-05,
+      "loss": 1.2382,
+      "step": 48350
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.8644166554614315e-05,
+      "loss": 1.2366,
+      "step": 48400
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.864276590022859e-05,
+      "loss": 1.231,
+      "step": 48450
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.8641365245842865e-05,
+      "loss": 1.1926,
+      "step": 48500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.863996459145713e-05,
+      "loss": 1.2306,
+      "step": 48550
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.86385639370714e-05,
+      "loss": 1.2361,
+      "step": 48600
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.8637163282685674e-05,
+      "loss": 1.2208,
+      "step": 48650
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.8635762628299945e-05,
+      "loss": 1.2187,
+      "step": 48700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.863436197391422e-05,
+      "loss": 1.2268,
+      "step": 48750
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.863296131952849e-05,
+      "loss": 1.2174,
+      "step": 48800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.863156066514275e-05,
+      "loss": 1.225,
+      "step": 48850
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.8630160010757025e-05,
+      "loss": 1.2393,
+      "step": 48900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.86287593563713e-05,
+      "loss": 1.2388,
+      "step": 48950
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.862735870198557e-05,
+      "loss": 1.2231,
+      "step": 49000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.862595804759984e-05,
+      "loss": 1.2132,
+      "step": 49050
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.862455739321411e-05,
+      "loss": 1.229,
+      "step": 49100
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.862315673882838e-05,
+      "loss": 1.2409,
+      "step": 49150
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.8621756084442655e-05,
+      "loss": 1.2225,
+      "step": 49200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.8620355430056926e-05,
+      "loss": 1.2442,
+      "step": 49250
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.861895477567119e-05,
+      "loss": 1.2242,
+      "step": 49300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.861755412128547e-05,
+      "loss": 1.2319,
+      "step": 49350
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.861615346689974e-05,
+      "loss": 1.2594,
+      "step": 49400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.8614752812514006e-05,
+      "loss": 1.2201,
+      "step": 49450
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.861335215812828e-05,
+      "loss": 1.205,
+      "step": 49500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.8611951503742556e-05,
+      "loss": 1.2434,
+      "step": 49550
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.861055084935682e-05,
+      "loss": 1.2036,
+      "step": 49600
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.860915019497109e-05,
+      "loss": 1.2229,
+      "step": 49650
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.8607749540585364e-05,
+      "loss": 1.215,
+      "step": 49700
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.8606348886199636e-05,
+      "loss": 1.2174,
+      "step": 49750
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.860494823181391e-05,
+      "loss": 1.2042,
+      "step": 49800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.860354757742818e-05,
+      "loss": 1.2096,
+      "step": 49850
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.8602146923042444e-05,
+      "loss": 1.2342,
+      "step": 49900
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.8600746268656716e-05,
+      "loss": 1.2375,
+      "step": 49950
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.8599345614270994e-05,
+      "loss": 1.2186,
+      "step": 50000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.859794495988526e-05,
+      "loss": 1.2533,
+      "step": 50050
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.859654430549953e-05,
+      "loss": 1.2225,
+      "step": 50100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.85951436511138e-05,
+      "loss": 1.2329,
+      "step": 50150
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.8593742996728074e-05,
+      "loss": 1.1887,
+      "step": 50200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.8592342342342346e-05,
+      "loss": 1.2638,
+      "step": 50250
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.859094168795662e-05,
+      "loss": 1.2038,
+      "step": 50300
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.858954103357088e-05,
+      "loss": 1.2328,
+      "step": 50350
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.858814037918516e-05,
+      "loss": 1.2415,
+      "step": 50400
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.858673972479943e-05,
+      "loss": 1.2019,
+      "step": 50450
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.85853390704137e-05,
+      "loss": 1.223,
+      "step": 50500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.858393841602797e-05,
+      "loss": 1.2338,
+      "step": 50550
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.858253776164225e-05,
+      "loss": 1.2546,
+      "step": 50600
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.858113710725651e-05,
+      "loss": 1.2223,
+      "step": 50650
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8579736452870784e-05,
+      "loss": 1.2409,
+      "step": 50700
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.8578335798485055e-05,
+      "loss": 1.2193,
+      "step": 50750
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.857693514409933e-05,
+      "loss": 1.2109,
+      "step": 50800
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.85755344897136e-05,
+      "loss": 1.2168,
+      "step": 50850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.857413383532787e-05,
+      "loss": 1.2266,
+      "step": 50900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8572733180942135e-05,
+      "loss": 1.2056,
+      "step": 50950
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8571332526556407e-05,
+      "loss": 1.2467,
+      "step": 51000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8569931872170685e-05,
+      "loss": 1.2152,
+      "step": 51050
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.856853121778495e-05,
+      "loss": 1.2269,
+      "step": 51100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.856713056339922e-05,
+      "loss": 1.2012,
+      "step": 51150
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.856572990901349e-05,
+      "loss": 1.2035,
+      "step": 51200
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.8564329254627765e-05,
+      "loss": 1.2232,
+      "step": 51250
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.8562928600242036e-05,
+      "loss": 1.2268,
+      "step": 51300
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.856152794585631e-05,
+      "loss": 1.1912,
+      "step": 51350
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.856012729147057e-05,
+      "loss": 1.1985,
+      "step": 51400
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.855872663708485e-05,
+      "loss": 1.2232,
+      "step": 51450
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.855732598269912e-05,
+      "loss": 1.2355,
+      "step": 51500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.855592532831339e-05,
+      "loss": 1.2261,
+      "step": 51550
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.855452467392766e-05,
+      "loss": 1.2322,
+      "step": 51600
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.855312401954194e-05,
+      "loss": 1.2263,
+      "step": 51650
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.85517233651562e-05,
+      "loss": 1.1849,
+      "step": 51700
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.8550322710770474e-05,
+      "loss": 1.2083,
+      "step": 51750
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.8548922056384746e-05,
+      "loss": 1.2184,
+      "step": 51800
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.854752140199902e-05,
+      "loss": 1.2076,
+      "step": 51850
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.854612074761329e-05,
+      "loss": 1.2223,
+      "step": 51900
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.854472009322756e-05,
+      "loss": 1.2183,
+      "step": 51950
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.8543319438841826e-05,
+      "loss": 1.2058,
+      "step": 52000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.8541918784456104e-05,
+      "loss": 1.2209,
+      "step": 52050
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8540518130070376e-05,
+      "loss": 1.1996,
+      "step": 52100
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.853911747568464e-05,
+      "loss": 1.216,
+      "step": 52150
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.853771682129891e-05,
+      "loss": 1.2479,
+      "step": 52200
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8536316166913184e-05,
+      "loss": 1.2443,
+      "step": 52250
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.8534915512527456e-05,
+      "loss": 1.2272,
+      "step": 52300
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.853351485814173e-05,
+      "loss": 1.2185,
+      "step": 52350
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.8532114203756e-05,
+      "loss": 1.2367,
+      "step": 52400
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.8530713549370264e-05,
+      "loss": 1.2073,
+      "step": 52450
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.852931289498454e-05,
+      "loss": 1.2098,
+      "step": 52500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.8527912240598814e-05,
+      "loss": 1.2523,
+      "step": 52550
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.852651158621308e-05,
+      "loss": 1.2031,
+      "step": 52600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.852511093182735e-05,
+      "loss": 1.2386,
+      "step": 52650
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.852371027744163e-05,
+      "loss": 1.22,
+      "step": 52700
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.8522309623055893e-05,
+      "loss": 1.199,
+      "step": 52750
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.8520908968670165e-05,
+      "loss": 1.2325,
+      "step": 52800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.851950831428444e-05,
+      "loss": 1.2192,
+      "step": 52850
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.851810765989871e-05,
+      "loss": 1.2283,
+      "step": 52900
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.851670700551298e-05,
+      "loss": 1.2121,
+      "step": 52950
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.851530635112725e-05,
+      "loss": 1.2178,
+      "step": 53000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.8513905696741516e-05,
+      "loss": 1.2375,
+      "step": 53050
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.8512505042355795e-05,
+      "loss": 1.2134,
+      "step": 53100
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.8511104387970067e-05,
+      "loss": 1.2103,
+      "step": 53150
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.850970373358433e-05,
+      "loss": 1.2134,
+      "step": 53200
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.85083030791986e-05,
+      "loss": 1.2352,
+      "step": 53250
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.8506902424812875e-05,
+      "loss": 1.2261,
+      "step": 53300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.8505501770427146e-05,
+      "loss": 1.2092,
+      "step": 53350
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.850410111604142e-05,
+      "loss": 1.1907,
+      "step": 53400
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.850270046165569e-05,
+      "loss": 1.2244,
+      "step": 53450
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.8501299807269954e-05,
+      "loss": 1.2199,
+      "step": 53500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.849989915288423e-05,
+      "loss": 1.2204,
+      "step": 53550
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.8498498498498504e-05,
+      "loss": 1.2152,
+      "step": 53600
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.849709784411277e-05,
+      "loss": 1.2411,
+      "step": 53650
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.849569718972704e-05,
+      "loss": 1.1775,
+      "step": 53700
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.849429653534131e-05,
+      "loss": 1.2329,
+      "step": 53750
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.8492895880955584e-05,
+      "loss": 1.1837,
+      "step": 53800
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.8491495226569856e-05,
+      "loss": 1.2036,
+      "step": 53850
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.849009457218413e-05,
+      "loss": 1.2453,
+      "step": 53900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.84886939177984e-05,
+      "loss": 1.1989,
+      "step": 53950
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.848729326341267e-05,
+      "loss": 1.2009,
+      "step": 54000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.848589260902694e-05,
+      "loss": 1.2301,
+      "step": 54050
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.848449195464121e-05,
+      "loss": 1.1844,
+      "step": 54100
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.8483091300255486e-05,
+      "loss": 1.2068,
+      "step": 54150
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.848169064586976e-05,
+      "loss": 1.2124,
+      "step": 54200
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.848028999148402e-05,
+      "loss": 1.2168,
+      "step": 54250
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.8478889337098294e-05,
+      "loss": 1.221,
+      "step": 54300
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.8477488682712565e-05,
+      "loss": 1.2268,
+      "step": 54350
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.847608802832684e-05,
+      "loss": 1.2148,
+      "step": 54400
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.847468737394111e-05,
+      "loss": 1.229,
+      "step": 54450
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.847328671955538e-05,
+      "loss": 1.2287,
+      "step": 54500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.8471886065169645e-05,
+      "loss": 1.2305,
+      "step": 54550
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.8470485410783924e-05,
+      "loss": 1.2139,
+      "step": 54600
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.8469084756398195e-05,
+      "loss": 1.2446,
+      "step": 54650
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.846768410201246e-05,
+      "loss": 1.2166,
+      "step": 54700
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.846628344762673e-05,
+      "loss": 1.2111,
+      "step": 54750
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.8464882793241003e-05,
+      "loss": 1.2257,
+      "step": 54800
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.8463482138855275e-05,
+      "loss": 1.2241,
+      "step": 54850
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.846208148446955e-05,
+      "loss": 1.214,
+      "step": 54900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.846068083008382e-05,
+      "loss": 1.2435,
+      "step": 54950
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.845928017569809e-05,
+      "loss": 1.2214,
+      "step": 55000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.845787952131236e-05,
+      "loss": 1.2371,
+      "step": 55050
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.845647886692663e-05,
+      "loss": 1.2145,
+      "step": 55100
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.84550782125409e-05,
+      "loss": 1.2035,
+      "step": 55150
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.8453677558155176e-05,
+      "loss": 1.2216,
+      "step": 55200
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.845227690376944e-05,
+      "loss": 1.2122,
+      "step": 55250
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.845087624938371e-05,
+      "loss": 1.235,
+      "step": 55300
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.8449475594997985e-05,
+      "loss": 1.2226,
+      "step": 55350
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.8448074940612256e-05,
+      "loss": 1.1771,
+      "step": 55400
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.844667428622653e-05,
+      "loss": 1.1974,
+      "step": 55450
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.84452736318408e-05,
+      "loss": 1.1854,
+      "step": 55500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.844387297745507e-05,
+      "loss": 1.1791,
+      "step": 55550
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.8442472323069336e-05,
+      "loss": 1.2525,
+      "step": 55600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.8441071668683614e-05,
+      "loss": 1.1801,
+      "step": 55650
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.843967101429788e-05,
+      "loss": 1.2178,
+      "step": 55700
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.843827035991215e-05,
+      "loss": 1.2229,
+      "step": 55750
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.843686970552642e-05,
+      "loss": 1.2264,
+      "step": 55800
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.8435469051140694e-05,
+      "loss": 1.2389,
+      "step": 55850
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.8434068396754966e-05,
+      "loss": 1.2246,
+      "step": 55900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.843266774236924e-05,
+      "loss": 1.2117,
+      "step": 55950
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.843126708798351e-05,
+      "loss": 1.1933,
+      "step": 56000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.842986643359778e-05,
+      "loss": 1.2031,
+      "step": 56050
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.842846577921205e-05,
+      "loss": 1.1818,
+      "step": 56100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.8427065124826324e-05,
+      "loss": 1.2084,
+      "step": 56150
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.842566447044059e-05,
+      "loss": 1.2158,
+      "step": 56200
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.842426381605487e-05,
+      "loss": 1.2176,
+      "step": 56250
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.842286316166913e-05,
+      "loss": 1.2081,
+      "step": 56300
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.8421462507283404e-05,
+      "loss": 1.2115,
+      "step": 56350
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.8420061852897675e-05,
+      "loss": 1.2184,
+      "step": 56400
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.841866119851195e-05,
+      "loss": 1.2222,
+      "step": 56450
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.841726054412622e-05,
+      "loss": 1.1938,
+      "step": 56500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.841585988974049e-05,
+      "loss": 1.1925,
+      "step": 56550
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.841445923535476e-05,
+      "loss": 1.2015,
+      "step": 56600
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.841305858096903e-05,
+      "loss": 1.2042,
+      "step": 56650
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.8411657926583305e-05,
+      "loss": 1.2039,
+      "step": 56700
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.841025727219757e-05,
+      "loss": 1.2061,
+      "step": 56750
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.840885661781184e-05,
+      "loss": 1.1724,
+      "step": 56800
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.840745596342611e-05,
+      "loss": 1.2074,
+      "step": 56850
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.8406055309040385e-05,
+      "loss": 1.2245,
+      "step": 56900
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.840465465465466e-05,
+      "loss": 1.2051,
+      "step": 56950
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.840325400026893e-05,
+      "loss": 1.2219,
+      "step": 57000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.84018533458832e-05,
+      "loss": 1.1998,
+      "step": 57050
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.840045269149747e-05,
+      "loss": 1.2152,
+      "step": 57100
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.839905203711174e-05,
+      "loss": 1.2081,
+      "step": 57150
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.839765138272601e-05,
+      "loss": 1.2091,
+      "step": 57200
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.839625072834028e-05,
+      "loss": 1.2206,
+      "step": 57250
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.839485007395456e-05,
+      "loss": 1.2193,
+      "step": 57300
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.839344941956882e-05,
+      "loss": 1.2062,
+      "step": 57350
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.8392048765183095e-05,
+      "loss": 1.2025,
+      "step": 57400
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.8390648110797366e-05,
+      "loss": 1.2136,
+      "step": 57450
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.838924745641164e-05,
+      "loss": 1.227,
+      "step": 57500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.838784680202591e-05,
+      "loss": 1.1956,
+      "step": 57550
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.838644614764018e-05,
+      "loss": 1.2036,
+      "step": 57600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.838504549325445e-05,
+      "loss": 1.2001,
+      "step": 57650
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.838364483886872e-05,
+      "loss": 1.1784,
+      "step": 57700
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.8382244184482996e-05,
+      "loss": 1.2165,
+      "step": 57750
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.838084353009726e-05,
+      "loss": 1.1924,
+      "step": 57800
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.837944287571153e-05,
+      "loss": 1.214,
+      "step": 57850
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.8378042221325804e-05,
+      "loss": 1.19,
+      "step": 57900
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.8376641566940076e-05,
+      "loss": 1.2157,
+      "step": 57950
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.837524091255435e-05,
+      "loss": 1.2004,
+      "step": 58000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.837384025816862e-05,
+      "loss": 1.1887,
+      "step": 58050
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.837243960378289e-05,
+      "loss": 1.1884,
+      "step": 58100
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.837103894939716e-05,
+      "loss": 1.1989,
+      "step": 58150
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.8369638295011434e-05,
+      "loss": 1.2413,
+      "step": 58200
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.83682376406257e-05,
+      "loss": 1.1958,
+      "step": 58250
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.836683698623997e-05,
+      "loss": 1.2177,
+      "step": 58300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.836543633185425e-05,
+      "loss": 1.2274,
+      "step": 58350
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.8364035677468514e-05,
+      "loss": 1.2218,
+      "step": 58400
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.8362635023082785e-05,
+      "loss": 1.2179,
+      "step": 58450
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.836123436869706e-05,
+      "loss": 1.1932,
+      "step": 58500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.835983371431133e-05,
+      "loss": 1.2312,
+      "step": 58550
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.83584330599256e-05,
+      "loss": 1.207,
+      "step": 58600
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.835703240553987e-05,
+      "loss": 1.1956,
+      "step": 58650
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.835563175115414e-05,
+      "loss": 1.1989,
+      "step": 58700
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.835423109676841e-05,
+      "loss": 1.2206,
+      "step": 58750
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.835283044238269e-05,
+      "loss": 1.1979,
+      "step": 58800
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.835142978799695e-05,
+      "loss": 1.2383,
+      "step": 58850
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.835002913361122e-05,
+      "loss": 1.1947,
+      "step": 58900
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.8348628479225495e-05,
+      "loss": 1.1791,
+      "step": 58950
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.8347227824839767e-05,
+      "loss": 1.208,
+      "step": 59000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.834582717045404e-05,
+      "loss": 1.2088,
+      "step": 59050
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.834442651606831e-05,
+      "loss": 1.1862,
+      "step": 59100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.834302586168258e-05,
+      "loss": 1.1975,
+      "step": 59150
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.834162520729685e-05,
+      "loss": 1.2037,
+      "step": 59200
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8340224552911125e-05,
+      "loss": 1.1905,
+      "step": 59250
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.833882389852539e-05,
+      "loss": 1.22,
+      "step": 59300
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.833742324413966e-05,
+      "loss": 1.1894,
+      "step": 59350
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.833602258975394e-05,
+      "loss": 1.1658,
+      "step": 59400
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8334621935368205e-05,
+      "loss": 1.2109,
+      "step": 59450
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.8333221280982476e-05,
+      "loss": 1.1792,
+      "step": 59500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.833182062659675e-05,
+      "loss": 1.2071,
+      "step": 59550
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.833041997221102e-05,
+      "loss": 1.212,
+      "step": 59600
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.832901931782529e-05,
+      "loss": 1.2113,
+      "step": 59650
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.832761866343956e-05,
+      "loss": 1.2427,
+      "step": 59700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.832621800905383e-05,
+      "loss": 1.2051,
+      "step": 59750
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.8324817354668106e-05,
+      "loss": 1.2013,
+      "step": 59800
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.832341670028238e-05,
+      "loss": 1.1794,
+      "step": 59850
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.832201604589664e-05,
+      "loss": 1.1997,
+      "step": 59900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.8320615391510914e-05,
+      "loss": 1.2033,
+      "step": 59950
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.8319214737125186e-05,
+      "loss": 1.2245,
+      "step": 60000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.831781408273946e-05,
+      "loss": 1.2218,
+      "step": 60050
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.831641342835373e-05,
+      "loss": 1.1802,
+      "step": 60100
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.8315012773968e-05,
+      "loss": 1.1845,
+      "step": 60150
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.8313612119582266e-05,
+      "loss": 1.2241,
+      "step": 60200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.8312211465196544e-05,
+      "loss": 1.2201,
+      "step": 60250
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.8310810810810816e-05,
+      "loss": 1.1924,
+      "step": 60300
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.830941015642508e-05,
+      "loss": 1.208,
+      "step": 60350
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.830800950203935e-05,
+      "loss": 1.1838,
+      "step": 60400
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.830660884765363e-05,
+      "loss": 1.1818,
+      "step": 60450
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8305208193267895e-05,
+      "loss": 1.1835,
+      "step": 60500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.830380753888217e-05,
+      "loss": 1.2058,
+      "step": 60550
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.830240688449644e-05,
+      "loss": 1.184,
+      "step": 60600
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.830100623011071e-05,
+      "loss": 1.2118,
+      "step": 60650
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.829960557572498e-05,
+      "loss": 1.2301,
+      "step": 60700
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.8298204921339253e-05,
+      "loss": 1.1792,
+      "step": 60750
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.829680426695352e-05,
+      "loss": 1.2141,
+      "step": 60800
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.82954036125678e-05,
+      "loss": 1.2089,
+      "step": 60850
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.829400295818207e-05,
+      "loss": 1.2261,
+      "step": 60900
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.829260230379633e-05,
+      "loss": 1.2155,
+      "step": 60950
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.8291201649410605e-05,
+      "loss": 1.201,
+      "step": 61000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.8289800995024877e-05,
+      "loss": 1.1915,
+      "step": 61050
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.828840034063915e-05,
+      "loss": 1.1811,
+      "step": 61100
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.828699968625342e-05,
+      "loss": 1.2219,
+      "step": 61150
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.828559903186769e-05,
+      "loss": 1.1838,
+      "step": 61200
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.8284198377481956e-05,
+      "loss": 1.1888,
+      "step": 61250
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.8282797723096235e-05,
+      "loss": 1.2009,
+      "step": 61300
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.8281397068710506e-05,
+      "loss": 1.2059,
+      "step": 61350
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.827999641432477e-05,
+      "loss": 1.2007,
+      "step": 61400
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.827859575993904e-05,
+      "loss": 1.2421,
+      "step": 61450
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.827719510555332e-05,
+      "loss": 1.1792,
+      "step": 61500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.8275794451167586e-05,
+      "loss": 1.2004,
+      "step": 61550
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.827439379678186e-05,
+      "loss": 1.2036,
+      "step": 61600
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.827299314239613e-05,
+      "loss": 1.2353,
+      "step": 61650
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.82715924880104e-05,
+      "loss": 1.1929,
+      "step": 61700
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.827019183362467e-05,
+      "loss": 1.1829,
+      "step": 61750
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.8268791179238944e-05,
+      "loss": 1.2093,
+      "step": 61800
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.826739052485321e-05,
+      "loss": 1.1955,
+      "step": 61850
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.826598987046749e-05,
+      "loss": 1.2073,
+      "step": 61900
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.826458921608176e-05,
+      "loss": 1.1858,
+      "step": 61950
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.8263188561696024e-05,
+      "loss": 1.2025,
+      "step": 62000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.8261787907310296e-05,
+      "loss": 1.2035,
+      "step": 62050
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.826038725292457e-05,
+      "loss": 1.1939,
+      "step": 62100
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.825898659853884e-05,
+      "loss": 1.187,
+      "step": 62150
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.825758594415311e-05,
+      "loss": 1.1798,
+      "step": 62200
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.825618528976738e-05,
+      "loss": 1.1662,
+      "step": 62250
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.825478463538165e-05,
+      "loss": 1.2151,
+      "step": 62300
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.8253383980995926e-05,
+      "loss": 1.1957,
+      "step": 62350
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.82519833266102e-05,
+      "loss": 1.1992,
+      "step": 62400
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.825058267222446e-05,
+      "loss": 1.184,
+      "step": 62450
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.8249182017838734e-05,
+      "loss": 1.2138,
+      "step": 62500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.824778136345301e-05,
+      "loss": 1.2103,
+      "step": 62550
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.824638070906728e-05,
+      "loss": 1.2085,
+      "step": 62600
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.824498005468155e-05,
+      "loss": 1.1934,
+      "step": 62650
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.824357940029582e-05,
+      "loss": 1.1775,
+      "step": 62700
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.824217874591009e-05,
+      "loss": 1.194,
+      "step": 62750
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.8240778091524363e-05,
+      "loss": 1.1881,
+      "step": 62800
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.8239377437138635e-05,
+      "loss": 1.2089,
+      "step": 62850
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.82379767827529e-05,
+      "loss": 1.2112,
+      "step": 62900
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.823657612836718e-05,
+      "loss": 1.1858,
+      "step": 62950
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.823517547398145e-05,
+      "loss": 1.2052,
+      "step": 63000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.8233774819595715e-05,
+      "loss": 1.1938,
+      "step": 63050
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.8232374165209986e-05,
+      "loss": 1.1854,
+      "step": 63100
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.823097351082426e-05,
+      "loss": 1.1946,
+      "step": 63150
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.822957285643853e-05,
+      "loss": 1.1879,
+      "step": 63200
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.82281722020528e-05,
+      "loss": 1.1887,
+      "step": 63250
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.822677154766707e-05,
+      "loss": 1.2325,
+      "step": 63300
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.822537089328134e-05,
+      "loss": 1.2207,
+      "step": 63350
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.8223970238895616e-05,
+      "loss": 1.1916,
+      "step": 63400
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.822256958450989e-05,
+      "loss": 1.1968,
+      "step": 63450
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.822116893012415e-05,
+      "loss": 1.2071,
+      "step": 63500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.8219768275738424e-05,
+      "loss": 1.1849,
+      "step": 63550
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.82183676213527e-05,
+      "loss": 1.1992,
+      "step": 63600
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.821696696696697e-05,
+      "loss": 1.1895,
+      "step": 63650
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.821556631258124e-05,
+      "loss": 1.2329,
+      "step": 63700
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.821416565819551e-05,
+      "loss": 1.1962,
+      "step": 63750
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.821276500380978e-05,
+      "loss": 1.2112,
+      "step": 63800
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.8211364349424054e-05,
+      "loss": 1.204,
+      "step": 63850
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.8209963695038326e-05,
+      "loss": 1.1842,
+      "step": 63900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.820856304065259e-05,
+      "loss": 1.1746,
+      "step": 63950
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.820716238626687e-05,
+      "loss": 1.2031,
+      "step": 64000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.820576173188114e-05,
+      "loss": 1.1804,
+      "step": 64050
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.8204361077495406e-05,
+      "loss": 1.2164,
+      "step": 64100
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.820296042310968e-05,
+      "loss": 1.2014,
+      "step": 64150
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.8201559768723956e-05,
+      "loss": 1.1894,
+      "step": 64200
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.820015911433822e-05,
+      "loss": 1.1824,
+      "step": 64250
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.819875845995249e-05,
+      "loss": 1.1798,
+      "step": 64300
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.8197357805566764e-05,
+      "loss": 1.2037,
+      "step": 64350
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.819595715118103e-05,
+      "loss": 1.2034,
+      "step": 64400
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.819455649679531e-05,
+      "loss": 1.2143,
+      "step": 64450
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.819315584240958e-05,
+      "loss": 1.1951,
+      "step": 64500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.8191755188023844e-05,
+      "loss": 1.2028,
+      "step": 64550
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.8190354533638115e-05,
+      "loss": 1.2084,
+      "step": 64600
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.8188953879252394e-05,
+      "loss": 1.1917,
+      "step": 64650
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.818755322486666e-05,
+      "loss": 1.1943,
+      "step": 64700
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.818615257048093e-05,
+      "loss": 1.1768,
+      "step": 64750
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.81847519160952e-05,
+      "loss": 1.2175,
+      "step": 64800
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8183351261709473e-05,
+      "loss": 1.2022,
+      "step": 64850
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8181950607323745e-05,
+      "loss": 1.1776,
+      "step": 64900
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.818054995293802e-05,
+      "loss": 1.1815,
+      "step": 64950
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.817914929855228e-05,
+      "loss": 1.2024,
+      "step": 65000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.817774864416656e-05,
+      "loss": 1.19,
+      "step": 65050
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.817634798978083e-05,
+      "loss": 1.1711,
+      "step": 65100
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.8174947335395096e-05,
+      "loss": 1.205,
+      "step": 65150
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.817354668100937e-05,
+      "loss": 1.2095,
+      "step": 65200
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.8172146026623646e-05,
+      "loss": 1.2057,
+      "step": 65250
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.817074537223791e-05,
+      "loss": 1.2161,
+      "step": 65300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.816934471785218e-05,
+      "loss": 1.2183,
+      "step": 65350
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.8167944063466455e-05,
+      "loss": 1.1944,
+      "step": 65400
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.816654340908072e-05,
+      "loss": 1.1997,
+      "step": 65450
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.8165142754695e-05,
+      "loss": 1.2016,
+      "step": 65500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.816374210030927e-05,
+      "loss": 1.2132,
+      "step": 65550
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.8162341445923534e-05,
+      "loss": 1.1729,
+      "step": 65600
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.8160940791537806e-05,
+      "loss": 1.1957,
+      "step": 65650
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.8159540137152084e-05,
+      "loss": 1.2012,
+      "step": 65700
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.815813948276635e-05,
+      "loss": 1.1959,
+      "step": 65750
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.815673882838062e-05,
+      "loss": 1.2101,
+      "step": 65800
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.815533817399489e-05,
+      "loss": 1.1908,
+      "step": 65850
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.8153937519609164e-05,
+      "loss": 1.1917,
+      "step": 65900
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.8152536865223436e-05,
+      "loss": 1.1876,
+      "step": 65950
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.815113621083771e-05,
+      "loss": 1.185,
+      "step": 66000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.814973555645197e-05,
+      "loss": 1.1991,
+      "step": 66050
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.814833490206625e-05,
+      "loss": 1.1658,
+      "step": 66100
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.814693424768052e-05,
+      "loss": 1.1782,
+      "step": 66150
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.814553359329479e-05,
+      "loss": 1.1902,
+      "step": 66200
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.814413293890906e-05,
+      "loss": 1.2057,
+      "step": 66250
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.814273228452334e-05,
+      "loss": 1.19,
+      "step": 66300
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.81413316301376e-05,
+      "loss": 1.2044,
+      "step": 66350
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.8139930975751874e-05,
+      "loss": 1.1954,
+      "step": 66400
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.8138530321366145e-05,
+      "loss": 1.1804,
+      "step": 66450
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.813712966698041e-05,
+      "loss": 1.2178,
+      "step": 66500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.813572901259469e-05,
+      "loss": 1.2109,
+      "step": 66550
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.813432835820896e-05,
+      "loss": 1.1651,
+      "step": 66600
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.8132927703823225e-05,
+      "loss": 1.2187,
+      "step": 66650
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.81315270494375e-05,
+      "loss": 1.2112,
+      "step": 66700
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.8130126395051775e-05,
+      "loss": 1.1817,
+      "step": 66750
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.812872574066604e-05,
+      "loss": 1.1852,
+      "step": 66800
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.812732508628031e-05,
+      "loss": 1.2272,
+      "step": 66850
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.812592443189458e-05,
+      "loss": 1.2002,
+      "step": 66900
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.8124523777508855e-05,
+      "loss": 1.1906,
+      "step": 66950
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.812312312312313e-05,
+      "loss": 1.1791,
+      "step": 67000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.81217224687374e-05,
+      "loss": 1.1799,
+      "step": 67050
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.812032181435166e-05,
+      "loss": 1.1557,
+      "step": 67100
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.811892115996594e-05,
+      "loss": 1.1781,
+      "step": 67150
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.811752050558021e-05,
+      "loss": 1.2152,
+      "step": 67200
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.811611985119448e-05,
+      "loss": 1.2041,
+      "step": 67250
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.811471919680875e-05,
+      "loss": 1.1565,
+      "step": 67300
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.811331854242303e-05,
+      "loss": 1.1857,
+      "step": 67350
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.811191788803729e-05,
+      "loss": 1.1949,
+      "step": 67400
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.8110517233651565e-05,
+      "loss": 1.1924,
+      "step": 67450
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.8109116579265836e-05,
+      "loss": 1.1577,
+      "step": 67500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.810771592488011e-05,
+      "loss": 1.1872,
+      "step": 67550
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.810631527049438e-05,
+      "loss": 1.1786,
+      "step": 67600
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.810491461610865e-05,
+      "loss": 1.2033,
+      "step": 67650
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.8103513961722916e-05,
+      "loss": 1.1505,
+      "step": 67700
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.810211330733719e-05,
+      "loss": 1.2049,
+      "step": 67750
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.8100712652951466e-05,
+      "loss": 1.172,
+      "step": 67800
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.809931199856573e-05,
+      "loss": 1.1887,
+      "step": 67850
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.809791134418e-05,
+      "loss": 1.1769,
+      "step": 67900
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.8096510689794274e-05,
+      "loss": 1.1956,
+      "step": 67950
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.8095110035408546e-05,
+      "loss": 1.1847,
+      "step": 68000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.809370938102282e-05,
+      "loss": 1.2055,
+      "step": 68050
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.809230872663709e-05,
+      "loss": 1.2131,
+      "step": 68100
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.8090908072251354e-05,
+      "loss": 1.1655,
+      "step": 68150
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.808950741786563e-05,
+      "loss": 1.2098,
+      "step": 68200
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.8088106763479904e-05,
+      "loss": 1.1924,
+      "step": 68250
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.808670610909417e-05,
+      "loss": 1.2078,
+      "step": 68300
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.808530545470844e-05,
+      "loss": 1.1757,
+      "step": 68350
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.808390480032272e-05,
+      "loss": 1.1702,
+      "step": 68400
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.8082504145936984e-05,
+      "loss": 1.1891,
+      "step": 68450
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.8081103491551255e-05,
+      "loss": 1.1555,
+      "step": 68500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.807970283716553e-05,
+      "loss": 1.1996,
+      "step": 68550
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.80783021827798e-05,
+      "loss": 1.1781,
+      "step": 68600
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.807690152839407e-05,
+      "loss": 1.1918,
+      "step": 68650
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.807550087400834e-05,
+      "loss": 1.1852,
+      "step": 68700
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.807410021962261e-05,
+      "loss": 1.1847,
+      "step": 68750
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.807269956523688e-05,
+      "loss": 1.1803,
+      "step": 68800
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.807129891085116e-05,
+      "loss": 1.1916,
+      "step": 68850
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.806989825646542e-05,
+      "loss": 1.1877,
+      "step": 68900
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.806849760207969e-05,
+      "loss": 1.1928,
+      "step": 68950
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.8067096947693965e-05,
+      "loss": 1.1607,
+      "step": 69000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.8065696293308237e-05,
+      "loss": 1.1952,
+      "step": 69050
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.806429563892251e-05,
+      "loss": 1.1779,
+      "step": 69100
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.806289498453678e-05,
+      "loss": 1.1702,
+      "step": 69150
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.8061494330151045e-05,
+      "loss": 1.1463,
+      "step": 69200
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.806009367576532e-05,
+      "loss": 1.1788,
+      "step": 69250
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.8058693021379595e-05,
+      "loss": 1.1471,
+      "step": 69300
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.805729236699386e-05,
+      "loss": 1.195,
+      "step": 69350
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.805589171260813e-05,
+      "loss": 1.1965,
+      "step": 69400
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.80544910582224e-05,
+      "loss": 1.1807,
+      "step": 69450
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.8053090403836675e-05,
+      "loss": 1.181,
+      "step": 69500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.8051689749450946e-05,
+      "loss": 1.1963,
+      "step": 69550
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.805028909506522e-05,
+      "loss": 1.2128,
+      "step": 69600
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.804888844067949e-05,
+      "loss": 1.204,
+      "step": 69650
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.804748778629376e-05,
+      "loss": 1.191,
+      "step": 69700
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.804608713190803e-05,
+      "loss": 1.1903,
+      "step": 69750
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.80446864775223e-05,
+      "loss": 1.1931,
+      "step": 69800
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.804328582313657e-05,
+      "loss": 1.1751,
+      "step": 69850
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.804188516875085e-05,
+      "loss": 1.1742,
+      "step": 69900
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.804048451436511e-05,
+      "loss": 1.2178,
+      "step": 69950
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.8039083859979384e-05,
+      "loss": 1.2015,
+      "step": 70000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.8037683205593656e-05,
+      "loss": 1.1839,
+      "step": 70050
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.803628255120793e-05,
+      "loss": 1.2029,
+      "step": 70100
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.80348818968222e-05,
+      "loss": 1.2018,
+      "step": 70150
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.803348124243647e-05,
+      "loss": 1.1907,
+      "step": 70200
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.8032080588050735e-05,
+      "loss": 1.2129,
+      "step": 70250
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.8030679933665014e-05,
+      "loss": 1.1822,
+      "step": 70300
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.8029279279279286e-05,
+      "loss": 1.1658,
+      "step": 70350
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.802787862489355e-05,
+      "loss": 1.2053,
+      "step": 70400
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.802647797050782e-05,
+      "loss": 1.1671,
+      "step": 70450
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.8025077316122094e-05,
+      "loss": 1.1622,
+      "step": 70500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.8023676661736365e-05,
+      "loss": 1.18,
+      "step": 70550
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.802227600735064e-05,
+      "loss": 1.1882,
+      "step": 70600
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.802087535296491e-05,
+      "loss": 1.1913,
+      "step": 70650
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.801947469857918e-05,
+      "loss": 1.1983,
+      "step": 70700
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.801807404419345e-05,
+      "loss": 1.1769,
+      "step": 70750
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.8016673389807723e-05,
+      "loss": 1.1798,
+      "step": 70800
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.801527273542199e-05,
+      "loss": 1.16,
+      "step": 70850
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.801387208103626e-05,
+      "loss": 1.1444,
+      "step": 70900
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.801247142665053e-05,
+      "loss": 1.1725,
+      "step": 70950
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.80110707722648e-05,
+      "loss": 1.1695,
+      "step": 71000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.8009670117879075e-05,
+      "loss": 1.1794,
+      "step": 71050
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.8008269463493347e-05,
+      "loss": 1.1908,
+      "step": 71100
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.800686880910762e-05,
+      "loss": 1.1891,
+      "step": 71150
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.800546815472189e-05,
+      "loss": 1.1919,
+      "step": 71200
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.800406750033616e-05,
+      "loss": 1.1828,
+      "step": 71250
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.8002666845950426e-05,
+      "loss": 1.1922,
+      "step": 71300
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.8001266191564705e-05,
+      "loss": 1.1989,
+      "step": 71350
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.799986553717897e-05,
+      "loss": 1.169,
+      "step": 71400
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.799846488279324e-05,
+      "loss": 1.1873,
+      "step": 71450
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.799706422840751e-05,
+      "loss": 1.2123,
+      "step": 71500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.7995663574021784e-05,
+      "loss": 1.1478,
+      "step": 71550
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.7994262919636056e-05,
+      "loss": 1.167,
+      "step": 71600
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.799286226525033e-05,
+      "loss": 1.1834,
+      "step": 71650
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.79914616108646e-05,
+      "loss": 1.1541,
+      "step": 71700
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.799006095647887e-05,
+      "loss": 1.1843,
+      "step": 71750
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.798866030209314e-05,
+      "loss": 1.1918,
+      "step": 71800
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.7987259647707414e-05,
+      "loss": 1.199,
+      "step": 71850
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.798585899332168e-05,
+      "loss": 1.186,
+      "step": 71900
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.798445833893596e-05,
+      "loss": 1.1835,
+      "step": 71950
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.798305768455022e-05,
+      "loss": 1.1883,
+      "step": 72000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.7981657030164494e-05,
+      "loss": 1.1709,
+      "step": 72050
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.7980256375778766e-05,
+      "loss": 1.1745,
+      "step": 72100
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.797885572139304e-05,
+      "loss": 1.1799,
+      "step": 72150
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.797745506700731e-05,
+      "loss": 1.1752,
+      "step": 72200
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.797605441262158e-05,
+      "loss": 1.2008,
+      "step": 72250
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.797465375823585e-05,
+      "loss": 1.1907,
+      "step": 72300
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.797325310385012e-05,
+      "loss": 1.1852,
+      "step": 72350
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.7971852449464395e-05,
+      "loss": 1.1735,
+      "step": 72400
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.797045179507866e-05,
+      "loss": 1.1996,
+      "step": 72450
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.796905114069293e-05,
+      "loss": 1.1798,
+      "step": 72500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.7967650486307204e-05,
+      "loss": 1.2018,
+      "step": 72550
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.7966249831921475e-05,
+      "loss": 1.1704,
+      "step": 72600
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.796484917753575e-05,
+      "loss": 1.1998,
+      "step": 72650
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.796344852315002e-05,
+      "loss": 1.1909,
+      "step": 72700
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.796204786876429e-05,
+      "loss": 1.168,
+      "step": 72750
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.796064721437856e-05,
+      "loss": 1.1671,
+      "step": 72800
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.7959246559992833e-05,
+      "loss": 1.1733,
+      "step": 72850
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.79578459056071e-05,
+      "loss": 1.1839,
+      "step": 72900
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.795644525122137e-05,
+      "loss": 1.1614,
+      "step": 72950
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.795504459683565e-05,
+      "loss": 1.1931,
+      "step": 73000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.795364394244991e-05,
+      "loss": 1.1983,
+      "step": 73050
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.7952243288064185e-05,
+      "loss": 1.1815,
+      "step": 73100
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.7950842633678456e-05,
+      "loss": 1.2087,
+      "step": 73150
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.794944197929273e-05,
+      "loss": 1.1793,
+      "step": 73200
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.7948041324907e-05,
+      "loss": 1.1882,
+      "step": 73250
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.794664067052127e-05,
+      "loss": 1.1631,
+      "step": 73300
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.794524001613554e-05,
+      "loss": 1.17,
+      "step": 73350
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.794383936174981e-05,
+      "loss": 1.1803,
+      "step": 73400
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.7942438707364086e-05,
+      "loss": 1.1812,
+      "step": 73450
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.794103805297835e-05,
+      "loss": 1.1729,
+      "step": 73500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.793963739859262e-05,
+      "loss": 1.1689,
+      "step": 73550
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.7938236744206894e-05,
+      "loss": 1.216,
+      "step": 73600
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.7936836089821166e-05,
+      "loss": 1.1708,
+      "step": 73650
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.793543543543544e-05,
+      "loss": 1.183,
+      "step": 73700
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.793403478104971e-05,
+      "loss": 1.1879,
+      "step": 73750
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.793263412666398e-05,
+      "loss": 1.1595,
+      "step": 73800
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.793123347227825e-05,
+      "loss": 1.1903,
+      "step": 73850
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.7929832817892524e-05,
+      "loss": 1.211,
+      "step": 73900
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.792843216350679e-05,
+      "loss": 1.1945,
+      "step": 73950
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.792703150912106e-05,
+      "loss": 1.1992,
+      "step": 74000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.792563085473534e-05,
+      "loss": 1.1741,
+      "step": 74050
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.7924230200349604e-05,
+      "loss": 1.1773,
+      "step": 74100
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.7922829545963876e-05,
+      "loss": 1.2041,
+      "step": 74150
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.792142889157815e-05,
+      "loss": 1.1684,
+      "step": 74200
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.792002823719242e-05,
+      "loss": 1.1625,
+      "step": 74250
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.791862758280669e-05,
+      "loss": 1.1374,
+      "step": 74300
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.791722692842096e-05,
+      "loss": 1.1701,
+      "step": 74350
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.791582627403523e-05,
+      "loss": 1.1549,
+      "step": 74400
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.79144256196495e-05,
+      "loss": 1.1852,
+      "step": 74450
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.791302496526378e-05,
+      "loss": 1.1694,
+      "step": 74500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.791162431087804e-05,
+      "loss": 1.1832,
+      "step": 74550
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.7910223656492314e-05,
+      "loss": 1.1613,
+      "step": 74600
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.7908823002106585e-05,
+      "loss": 1.1594,
+      "step": 74650
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.790742234772086e-05,
+      "loss": 1.1749,
+      "step": 74700
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.790602169333513e-05,
+      "loss": 1.1597,
+      "step": 74750
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.79046210389494e-05,
+      "loss": 1.1667,
+      "step": 74800
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.790322038456367e-05,
+      "loss": 1.1633,
+      "step": 74850
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.790181973017794e-05,
+      "loss": 1.1702,
+      "step": 74900
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.7900419075792215e-05,
+      "loss": 1.1697,
+      "step": 74950
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.789901842140648e-05,
+      "loss": 1.2037,
+      "step": 75000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.789761776702075e-05,
+      "loss": 1.1664,
+      "step": 75050
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.789621711263503e-05,
+      "loss": 1.1631,
+      "step": 75100
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.7894816458249295e-05,
+      "loss": 1.1686,
+      "step": 75150
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.7893415803863566e-05,
+      "loss": 1.1614,
+      "step": 75200
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.789201514947784e-05,
+      "loss": 1.1816,
+      "step": 75250
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.789061449509211e-05,
+      "loss": 1.1659,
+      "step": 75300
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.788921384070638e-05,
+      "loss": 1.1634,
+      "step": 75350
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.788781318632065e-05,
+      "loss": 1.1678,
+      "step": 75400
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.788641253193492e-05,
+      "loss": 1.1807,
+      "step": 75450
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.788501187754919e-05,
+      "loss": 1.1632,
+      "step": 75500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.788361122316347e-05,
+      "loss": 1.1619,
+      "step": 75550
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.788221056877773e-05,
+      "loss": 1.1506,
+      "step": 75600
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.7880809914392004e-05,
+      "loss": 1.1797,
+      "step": 75650
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.7879409260006276e-05,
+      "loss": 1.1735,
+      "step": 75700
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.787800860562055e-05,
+      "loss": 1.1532,
+      "step": 75750
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.787660795123482e-05,
+      "loss": 1.1821,
+      "step": 75800
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.787520729684909e-05,
+      "loss": 1.1809,
+      "step": 75850
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.7873806642463356e-05,
+      "loss": 1.1986,
+      "step": 75900
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.7872405988077634e-05,
+      "loss": 1.1636,
+      "step": 75950
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.7871005333691906e-05,
+      "loss": 1.1619,
+      "step": 76000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.786960467930617e-05,
+      "loss": 1.1842,
+      "step": 76050
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.786820402492044e-05,
+      "loss": 1.1465,
+      "step": 76100
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.786680337053472e-05,
+      "loss": 1.2114,
+      "step": 76150
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7865402716148986e-05,
+      "loss": 1.1696,
+      "step": 76200
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.786400206176326e-05,
+      "loss": 1.1876,
+      "step": 76250
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.786260140737753e-05,
+      "loss": 1.163,
+      "step": 76300
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.78612007529918e-05,
+      "loss": 1.1856,
+      "step": 76350
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.785980009860607e-05,
+      "loss": 1.2008,
+      "step": 76400
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7858399444220344e-05,
+      "loss": 1.1641,
+      "step": 76450
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.785699878983461e-05,
+      "loss": 1.1701,
+      "step": 76500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.785559813544888e-05,
+      "loss": 1.1762,
+      "step": 76550
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.785419748106316e-05,
+      "loss": 1.1818,
+      "step": 76600
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7852796826677424e-05,
+      "loss": 1.1875,
+      "step": 76650
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7851396172291695e-05,
+      "loss": 1.1868,
+      "step": 76700
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.784999551790597e-05,
+      "loss": 1.1566,
+      "step": 76750
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.784859486352024e-05,
+      "loss": 1.173,
+      "step": 76800
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.784719420913451e-05,
+      "loss": 1.1841,
+      "step": 76850
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.784579355474878e-05,
+      "loss": 1.1344,
+      "step": 76900
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.7844392900363047e-05,
+      "loss": 1.183,
+      "step": 76950
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.7842992245977325e-05,
+      "loss": 1.1518,
+      "step": 77000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.7841591591591597e-05,
+      "loss": 1.1861,
+      "step": 77050
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.784019093720586e-05,
+      "loss": 1.1881,
+      "step": 77100
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.783879028282013e-05,
+      "loss": 1.1523,
+      "step": 77150
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.783738962843441e-05,
+      "loss": 1.1709,
+      "step": 77200
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.7835988974048676e-05,
+      "loss": 1.1667,
+      "step": 77250
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.783458831966295e-05,
+      "loss": 1.2023,
+      "step": 77300
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.783318766527722e-05,
+      "loss": 1.1518,
+      "step": 77350
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.783178701089149e-05,
+      "loss": 1.1496,
+      "step": 77400
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.783038635650576e-05,
+      "loss": 1.169,
+      "step": 77450
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.7828985702120035e-05,
+      "loss": 1.1592,
+      "step": 77500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.78275850477343e-05,
+      "loss": 1.1706,
+      "step": 77550
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.782618439334857e-05,
+      "loss": 1.1922,
+      "step": 77600
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.782478373896285e-05,
+      "loss": 1.1403,
+      "step": 77650
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.7823383084577114e-05,
+      "loss": 1.1817,
+      "step": 77700
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.7821982430191386e-05,
+      "loss": 1.1655,
+      "step": 77750
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.782058177580566e-05,
+      "loss": 1.1826,
+      "step": 77800
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.781918112141993e-05,
+      "loss": 1.2345,
+      "step": 77850
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.78177804670342e-05,
+      "loss": 1.1927,
+      "step": 77900
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.781637981264847e-05,
+      "loss": 1.1525,
+      "step": 77950
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.781497915826274e-05,
+      "loss": 1.1508,
+      "step": 78000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.7813578503877016e-05,
+      "loss": 1.2115,
+      "step": 78050
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.781217784949129e-05,
+      "loss": 1.1696,
+      "step": 78100
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.781077719510555e-05,
+      "loss": 1.1726,
+      "step": 78150
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.7809376540719824e-05,
+      "loss": 1.1721,
+      "step": 78200
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.78079758863341e-05,
+      "loss": 1.1736,
+      "step": 78250
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.780657523194837e-05,
+      "loss": 1.1784,
+      "step": 78300
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.780517457756264e-05,
+      "loss": 1.14,
+      "step": 78350
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.780377392317691e-05,
+      "loss": 1.1828,
+      "step": 78400
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.780237326879118e-05,
+      "loss": 1.1638,
+      "step": 78450
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.7800972614405454e-05,
+      "loss": 1.1667,
+      "step": 78500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.7799571960019725e-05,
+      "loss": 1.154,
+      "step": 78550
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.779817130563399e-05,
+      "loss": 1.1937,
+      "step": 78600
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.779677065124827e-05,
+      "loss": 1.1851,
+      "step": 78650
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.779536999686254e-05,
+      "loss": 1.1986,
+      "step": 78700
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.7793969342476805e-05,
+      "loss": 1.1953,
+      "step": 78750
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.779256868809108e-05,
+      "loss": 1.1659,
+      "step": 78800
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.779116803370535e-05,
+      "loss": 1.1662,
+      "step": 78850
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.778976737931962e-05,
+      "loss": 1.1517,
+      "step": 78900
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.778836672493389e-05,
+      "loss": 1.1697,
+      "step": 78950
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.778696607054816e-05,
+      "loss": 1.1895,
+      "step": 79000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.778556541616243e-05,
+      "loss": 1.1425,
+      "step": 79050
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.7784164761776707e-05,
+      "loss": 1.1622,
+      "step": 79100
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.778276410739098e-05,
+      "loss": 1.1712,
+      "step": 79150
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.778136345300524e-05,
+      "loss": 1.1632,
+      "step": 79200
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.7779962798619515e-05,
+      "loss": 1.171,
+      "step": 79250
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.777856214423379e-05,
+      "loss": 1.159,
+      "step": 79300
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.777716148984806e-05,
+      "loss": 1.1655,
+      "step": 79350
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.777576083546233e-05,
+      "loss": 1.1825,
+      "step": 79400
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.77743601810766e-05,
+      "loss": 1.2058,
+      "step": 79450
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.777295952669087e-05,
+      "loss": 1.1672,
+      "step": 79500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.7771558872305145e-05,
+      "loss": 1.1774,
+      "step": 79550
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.7770158217919416e-05,
+      "loss": 1.1497,
+      "step": 79600
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.776875756353368e-05,
+      "loss": 1.1707,
+      "step": 79650
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.776735690914796e-05,
+      "loss": 1.2059,
+      "step": 79700
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.776595625476223e-05,
+      "loss": 1.1527,
+      "step": 79750
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.7764555600376496e-05,
+      "loss": 1.1573,
+      "step": 79800
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.776315494599077e-05,
+      "loss": 1.1846,
+      "step": 79850
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.776175429160504e-05,
+      "loss": 1.2171,
+      "step": 79900
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.776035363721931e-05,
+      "loss": 1.1666,
+      "step": 79950
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.775895298283358e-05,
+      "loss": 1.1736,
+      "step": 80000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.7757552328447854e-05,
+      "loss": 1.157,
+      "step": 80050
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.775615167406212e-05,
+      "loss": 1.1828,
+      "step": 80100
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.77547510196764e-05,
+      "loss": 1.1898,
+      "step": 80150
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.775335036529067e-05,
+      "loss": 1.1761,
+      "step": 80200
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.7751949710904934e-05,
+      "loss": 1.1554,
+      "step": 80250
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.7750549056519205e-05,
+      "loss": 1.1623,
+      "step": 80300
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.7749148402133484e-05,
+      "loss": 1.1596,
+      "step": 80350
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.774774774774775e-05,
+      "loss": 1.1777,
+      "step": 80400
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.774634709336202e-05,
+      "loss": 1.1608,
+      "step": 80450
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.774494643897629e-05,
+      "loss": 1.1305,
+      "step": 80500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.7743545784590564e-05,
+      "loss": 1.1636,
+      "step": 80550
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.7742145130204835e-05,
+      "loss": 1.1405,
+      "step": 80600
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.774074447581911e-05,
+      "loss": 1.164,
+      "step": 80650
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.773934382143337e-05,
+      "loss": 1.1696,
+      "step": 80700
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.773794316704765e-05,
+      "loss": 1.1624,
+      "step": 80750
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.773654251266192e-05,
+      "loss": 1.1748,
+      "step": 80800
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.773514185827619e-05,
+      "loss": 1.1486,
+      "step": 80850
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.773374120389046e-05,
+      "loss": 1.1813,
+      "step": 80900
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.773234054950473e-05,
+      "loss": 1.1725,
+      "step": 80950
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.7730939895119e-05,
+      "loss": 1.1872,
+      "step": 81000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.772953924073327e-05,
+      "loss": 1.1534,
+      "step": 81050
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.7728138586347545e-05,
+      "loss": 1.1725,
+      "step": 81100
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.772673793196181e-05,
+      "loss": 1.1668,
+      "step": 81150
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.772533727757609e-05,
+      "loss": 1.1781,
+      "step": 81200
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.772393662319036e-05,
+      "loss": 1.1684,
+      "step": 81250
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.7722535968804625e-05,
+      "loss": 1.1879,
+      "step": 81300
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.7721135314418896e-05,
+      "loss": 1.1728,
+      "step": 81350
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.7719734660033175e-05,
+      "loss": 1.1831,
+      "step": 81400
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.771833400564744e-05,
+      "loss": 1.1714,
+      "step": 81450
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.771693335126171e-05,
+      "loss": 1.1622,
+      "step": 81500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.771553269687598e-05,
+      "loss": 1.1586,
+      "step": 81550
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.7714132042490254e-05,
+      "loss": 1.1451,
+      "step": 81600
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.7712731388104526e-05,
+      "loss": 1.1707,
+      "step": 81650
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.77113307337188e-05,
+      "loss": 1.1656,
+      "step": 81700
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.770993007933306e-05,
+      "loss": 1.1915,
+      "step": 81750
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.770852942494734e-05,
+      "loss": 1.1767,
+      "step": 81800
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.770712877056161e-05,
+      "loss": 1.1862,
+      "step": 81850
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.770572811617588e-05,
+      "loss": 1.1925,
+      "step": 81900
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.770432746179015e-05,
+      "loss": 1.1957,
+      "step": 81950
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.770292680740442e-05,
+      "loss": 1.1652,
+      "step": 82000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.770152615301869e-05,
+      "loss": 1.1723,
+      "step": 82050
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.7700125498632964e-05,
+      "loss": 1.183,
+      "step": 82100
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.7698724844247236e-05,
+      "loss": 1.1691,
+      "step": 82150
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.76973241898615e-05,
+      "loss": 1.1415,
+      "step": 82200
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.769592353547578e-05,
+      "loss": 1.1928,
+      "step": 82250
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.769452288109005e-05,
+      "loss": 1.1779,
+      "step": 82300
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.7693122226704315e-05,
+      "loss": 1.155,
+      "step": 82350
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.769172157231859e-05,
+      "loss": 1.1755,
+      "step": 82400
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.7690320917932865e-05,
+      "loss": 1.1804,
+      "step": 82450
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.768892026354713e-05,
+      "loss": 1.1769,
+      "step": 82500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.76875196091614e-05,
+      "loss": 1.1553,
+      "step": 82550
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.7686118954775674e-05,
+      "loss": 1.1611,
+      "step": 82600
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.7684718300389945e-05,
+      "loss": 1.1786,
+      "step": 82650
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.768331764600422e-05,
+      "loss": 1.1768,
+      "step": 82700
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.768191699161849e-05,
+      "loss": 1.1694,
+      "step": 82750
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.768051633723275e-05,
+      "loss": 1.214,
+      "step": 82800
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.767911568284703e-05,
+      "loss": 1.1711,
+      "step": 82850
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.7677715028461303e-05,
+      "loss": 1.1475,
+      "step": 82900
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.767631437407557e-05,
+      "loss": 1.1716,
+      "step": 82950
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.767491371968984e-05,
+      "loss": 1.184,
+      "step": 83000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.767351306530412e-05,
+      "loss": 1.1876,
+      "step": 83050
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.767211241091838e-05,
+      "loss": 1.2014,
+      "step": 83100
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.7670711756532655e-05,
+      "loss": 1.1782,
+      "step": 83150
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.7669311102146926e-05,
+      "loss": 1.1751,
+      "step": 83200
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.766791044776119e-05,
+      "loss": 1.1593,
+      "step": 83250
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.766650979337547e-05,
+      "loss": 1.1758,
+      "step": 83300
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.766510913898974e-05,
+      "loss": 1.157,
+      "step": 83350
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.7663708484604006e-05,
+      "loss": 1.1296,
+      "step": 83400
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.766230783021828e-05,
+      "loss": 1.1392,
+      "step": 83450
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.7660907175832556e-05,
+      "loss": 1.145,
+      "step": 83500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.765950652144682e-05,
+      "loss": 1.1592,
+      "step": 83550
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.765810586706109e-05,
+      "loss": 1.1619,
+      "step": 83600
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.7656705212675364e-05,
+      "loss": 1.1685,
+      "step": 83650
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.7655304558289636e-05,
+      "loss": 1.1478,
+      "step": 83700
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.765390390390391e-05,
+      "loss": 1.1554,
+      "step": 83750
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.765250324951818e-05,
+      "loss": 1.1587,
+      "step": 83800
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.7651102595132444e-05,
+      "loss": 1.1696,
+      "step": 83850
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.764970194074672e-05,
+      "loss": 1.1849,
+      "step": 83900
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.7648301286360994e-05,
+      "loss": 1.1463,
+      "step": 83950
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.764690063197526e-05,
+      "loss": 1.1289,
+      "step": 84000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.764549997758953e-05,
+      "loss": 1.1503,
+      "step": 84050
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.764409932320381e-05,
+      "loss": 1.1848,
+      "step": 84100
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.7642698668818074e-05,
+      "loss": 1.167,
+      "step": 84150
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.7641298014432346e-05,
+      "loss": 1.1467,
+      "step": 84200
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.763989736004662e-05,
+      "loss": 1.1529,
+      "step": 84250
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.763849670566088e-05,
+      "loss": 1.1826,
+      "step": 84300
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.763709605127516e-05,
+      "loss": 1.1499,
+      "step": 84350
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.763569539688943e-05,
+      "loss": 1.18,
+      "step": 84400
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.76342947425037e-05,
+      "loss": 1.1758,
+      "step": 84450
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.763289408811797e-05,
+      "loss": 1.1627,
+      "step": 84500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.763149343373225e-05,
+      "loss": 1.1634,
+      "step": 84550
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.763009277934651e-05,
+      "loss": 1.1744,
+      "step": 84600
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.7628692124960784e-05,
+      "loss": 1.1388,
+      "step": 84650
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.7627291470575055e-05,
+      "loss": 1.1546,
+      "step": 84700
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.762589081618933e-05,
+      "loss": 1.1508,
+      "step": 84750
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.76244901618036e-05,
+      "loss": 1.1785,
+      "step": 84800
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.762308950741787e-05,
+      "loss": 1.184,
+      "step": 84850
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.7621688853032135e-05,
+      "loss": 1.1683,
+      "step": 84900
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.762028819864641e-05,
+      "loss": 1.155,
+      "step": 84950
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.7618887544260685e-05,
+      "loss": 1.1493,
+      "step": 85000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.761748688987495e-05,
+      "loss": 1.1736,
+      "step": 85050
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.761608623548922e-05,
+      "loss": 1.1701,
+      "step": 85100
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.761468558110349e-05,
+      "loss": 1.1473,
+      "step": 85150
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.7613284926717765e-05,
+      "loss": 1.1543,
+      "step": 85200
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.7611884272332036e-05,
+      "loss": 1.16,
+      "step": 85250
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.761048361794631e-05,
+      "loss": 1.1549,
+      "step": 85300
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.760908296356057e-05,
+      "loss": 1.1629,
+      "step": 85350
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.760768230917485e-05,
+      "loss": 1.1526,
+      "step": 85400
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.760628165478912e-05,
+      "loss": 1.1535,
+      "step": 85450
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.760488100040339e-05,
+      "loss": 1.1512,
+      "step": 85500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.760348034601766e-05,
+      "loss": 1.1895,
+      "step": 85550
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.760207969163194e-05,
+      "loss": 1.1491,
+      "step": 85600
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.76006790372462e-05,
+      "loss": 1.1818,
+      "step": 85650
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.7599278382860474e-05,
+      "loss": 1.1703,
+      "step": 85700
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.7597877728474746e-05,
+      "loss": 1.1637,
+      "step": 85750
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.759647707408902e-05,
+      "loss": 1.1561,
+      "step": 85800
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.759507641970329e-05,
+      "loss": 1.179,
+      "step": 85850
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.759367576531756e-05,
+      "loss": 1.1416,
+      "step": 85900
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.7592275110931826e-05,
+      "loss": 1.1354,
+      "step": 85950
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.7590874456546104e-05,
+      "loss": 1.1555,
+      "step": 86000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.7589473802160376e-05,
+      "loss": 1.1897,
+      "step": 86050
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.758807314777464e-05,
+      "loss": 1.159,
+      "step": 86100
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.758667249338891e-05,
+      "loss": 1.1614,
+      "step": 86150
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.7585271839003184e-05,
+      "loss": 1.177,
+      "step": 86200
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.7583871184617456e-05,
+      "loss": 1.1443,
+      "step": 86250
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.758247053023173e-05,
+      "loss": 1.1603,
+      "step": 86300
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.7581069875846e-05,
+      "loss": 1.198,
+      "step": 86350
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.757966922146027e-05,
+      "loss": 1.1655,
+      "step": 86400
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.757826856707454e-05,
+      "loss": 1.2059,
+      "step": 86450
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.7576867912688814e-05,
+      "loss": 1.1499,
+      "step": 86500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.757546725830308e-05,
+      "loss": 1.159,
+      "step": 86550
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.757406660391735e-05,
+      "loss": 1.1589,
+      "step": 86600
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.757266594953162e-05,
+      "loss": 1.1699,
+      "step": 86650
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.7571265295145894e-05,
+      "loss": 1.1401,
+      "step": 86700
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.7569864640760165e-05,
+      "loss": 1.1742,
+      "step": 86750
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.756846398637444e-05,
+      "loss": 1.1591,
+      "step": 86800
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.756706333198871e-05,
+      "loss": 1.1519,
+      "step": 86850
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.756566267760298e-05,
+      "loss": 1.1516,
+      "step": 86900
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.756426202321725e-05,
+      "loss": 1.1428,
+      "step": 86950
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.7562861368831517e-05,
+      "loss": 1.1925,
+      "step": 87000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.7561460714445795e-05,
+      "loss": 1.1404,
+      "step": 87050
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.756006006006006e-05,
+      "loss": 1.1496,
+      "step": 87100
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.755865940567433e-05,
+      "loss": 1.1784,
+      "step": 87150
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.75572587512886e-05,
+      "loss": 1.1627,
+      "step": 87200
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.7555858096902875e-05,
+      "loss": 1.1426,
+      "step": 87250
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.7554457442517146e-05,
+      "loss": 1.1499,
+      "step": 87300
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.755305678813142e-05,
+      "loss": 1.1624,
+      "step": 87350
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.755165613374569e-05,
+      "loss": 1.1788,
+      "step": 87400
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.755025547935996e-05,
+      "loss": 1.1411,
+      "step": 87450
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.754885482497423e-05,
+      "loss": 1.1643,
+      "step": 87500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.7547454170588505e-05,
+      "loss": 1.1557,
+      "step": 87550
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.754605351620277e-05,
+      "loss": 1.1751,
+      "step": 87600
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.754465286181704e-05,
+      "loss": 1.1482,
+      "step": 87650
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.754325220743131e-05,
+      "loss": 1.1461,
+      "step": 87700
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.7541851553045584e-05,
+      "loss": 1.175,
+      "step": 87750
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.7540450898659856e-05,
+      "loss": 1.1366,
+      "step": 87800
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.753905024427413e-05,
+      "loss": 1.1696,
+      "step": 87850
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.75376495898884e-05,
+      "loss": 1.1603,
+      "step": 87900
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.753624893550267e-05,
+      "loss": 1.1581,
+      "step": 87950
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.753484828111694e-05,
+      "loss": 1.1782,
+      "step": 88000
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.753344762673121e-05,
+      "loss": 1.1512,
+      "step": 88050
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.7532046972345486e-05,
+      "loss": 1.1958,
+      "step": 88100
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.753064631795975e-05,
+      "loss": 1.1588,
+      "step": 88150
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.752924566357402e-05,
+      "loss": 1.1503,
+      "step": 88200
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.7527845009188294e-05,
+      "loss": 1.1552,
+      "step": 88250
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.7526444354802566e-05,
+      "loss": 1.178,
+      "step": 88300
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.752504370041684e-05,
+      "loss": 1.1439,
+      "step": 88350
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.752364304603111e-05,
+      "loss": 1.1596,
+      "step": 88400
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.752224239164538e-05,
+      "loss": 1.1738,
+      "step": 88450
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.752084173725965e-05,
+      "loss": 1.1629,
+      "step": 88500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.7519441082873924e-05,
+      "loss": 1.1586,
+      "step": 88550
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.751804042848819e-05,
+      "loss": 1.1825,
+      "step": 88600
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.751663977410246e-05,
+      "loss": 1.1345,
+      "step": 88650
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.751523911971673e-05,
+      "loss": 1.1739,
+      "step": 88700
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.7513838465331003e-05,
+      "loss": 1.1766,
+      "step": 88750
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.7512437810945275e-05,
+      "loss": 1.1577,
+      "step": 88800
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.751103715655955e-05,
+      "loss": 1.1732,
+      "step": 88850
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.750963650217382e-05,
+      "loss": 1.1485,
+      "step": 88900
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.750823584778809e-05,
+      "loss": 1.1586,
+      "step": 88950
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.750683519340236e-05,
+      "loss": 1.1265,
+      "step": 89000
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.750543453901663e-05,
+      "loss": 1.1527,
+      "step": 89050
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.75040338846309e-05,
+      "loss": 1.1542,
+      "step": 89100
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.7502633230245177e-05,
+      "loss": 1.1522,
+      "step": 89150
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.750123257585944e-05,
+      "loss": 1.1272,
+      "step": 89200
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.749983192147371e-05,
+      "loss": 1.1496,
+      "step": 89250
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.7498431267087985e-05,
+      "loss": 1.1714,
+      "step": 89300
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.7497030612702256e-05,
+      "loss": 1.1457,
+      "step": 89350
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.749562995831653e-05,
+      "loss": 1.149,
+      "step": 89400
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.74942293039308e-05,
+      "loss": 1.1828,
+      "step": 89450
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.749282864954507e-05,
+      "loss": 1.1107,
+      "step": 89500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.749142799515934e-05,
+      "loss": 1.1306,
+      "step": 89550
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.7490027340773614e-05,
+      "loss": 1.1476,
+      "step": 89600
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.748862668638788e-05,
+      "loss": 1.1422,
+      "step": 89650
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.748722603200215e-05,
+      "loss": 1.1734,
+      "step": 89700
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.748582537761642e-05,
+      "loss": 1.1281,
+      "step": 89750
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.7484424723230694e-05,
+      "loss": 1.1658,
+      "step": 89800
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.7483024068844966e-05,
+      "loss": 1.1493,
+      "step": 89850
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.748162341445924e-05,
+      "loss": 1.1633,
+      "step": 89900
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.748022276007351e-05,
+      "loss": 1.1526,
+      "step": 89950
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.747882210568778e-05,
+      "loss": 1.1744,
+      "step": 90000
+    }
+  ],
+  "max_steps": 1784880,
+  "num_train_epochs": 80,
+  "total_flos": 1.1156563427328e+17,
+  "trial_name": null,
+  "trial_params": null
+}