diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6883 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 571521,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9956257075418056e-05,
+      "loss": 5.0842,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99125141508361e-05,
+      "loss": 3.2022,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9868771226254155e-05,
+      "loss": 2.772,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982502830167221e-05,
+      "loss": 2.503,
+      "step": 2000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.978128537709026e-05,
+      "loss": 2.3406,
+      "step": 2500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.973754245250831e-05,
+      "loss": 2.1903,
+      "step": 3000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969379952792636e-05,
+      "loss": 2.1055,
+      "step": 3500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9650056603344415e-05,
+      "loss": 2.0064,
+      "step": 4000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.960631367876246e-05,
+      "loss": 1.9395,
+      "step": 4500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9562570754180515e-05,
+      "loss": 1.9027,
+      "step": 5000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.951882782959856e-05,
+      "loss": 1.8424,
+      "step": 5500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9475084905016614e-05,
+      "loss": 1.7931,
+      "step": 6000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.943134198043467e-05,
+      "loss": 1.7543,
+      "step": 6500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938759905585272e-05,
+      "loss": 1.7311,
+      "step": 7000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9343856131270774e-05,
+      "loss": 1.6982,
+      "step": 7500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930011320668882e-05,
+      "loss": 1.6743,
+      "step": 8000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9256370282106874e-05,
+      "loss": 1.6448,
+      "step": 8500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.921262735752492e-05,
+      "loss": 1.6228,
+      "step": 9000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9168884432942974e-05,
+      "loss": 1.5979,
+      "step": 9500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.912514150836102e-05,
+      "loss": 1.5838,
+      "step": 10000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.908139858377907e-05,
+      "loss": 1.573,
+      "step": 10500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9037655659197127e-05,
+      "loss": 1.5632,
+      "step": 11000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.899391273461518e-05,
+      "loss": 1.5464,
+      "step": 11500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.895016981003323e-05,
+      "loss": 1.5251,
+      "step": 12000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890642688545128e-05,
+      "loss": 1.5211,
+      "step": 12500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.886268396086933e-05,
+      "loss": 1.4994,
+      "step": 13000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8818941036287386e-05,
+      "loss": 1.4825,
+      "step": 13500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.877519811170543e-05,
+      "loss": 1.48,
+      "step": 14000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8731455187123486e-05,
+      "loss": 1.4764,
+      "step": 14500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.868771226254153e-05,
+      "loss": 1.4416,
+      "step": 15000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8643969337959586e-05,
+      "loss": 1.4412,
+      "step": 15500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.860022641337764e-05,
+      "loss": 1.4341,
+      "step": 16000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.855648348879569e-05,
+      "loss": 1.4348,
+      "step": 16500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8512740564213745e-05,
+      "loss": 1.4227,
+      "step": 17000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.846899763963179e-05,
+      "loss": 1.4145,
+      "step": 17500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8425254715049845e-05,
+      "loss": 1.4117,
+      "step": 18000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.838151179046789e-05,
+      "loss": 1.389,
+      "step": 18500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8337768865885945e-05,
+      "loss": 1.3928,
+      "step": 19000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.829402594130399e-05,
+      "loss": 1.3828,
+      "step": 19500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8250283016722044e-05,
+      "loss": 1.3723,
+      "step": 20000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8206540092140105e-05,
+      "loss": 1.3652,
+      "step": 20500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.816279716755815e-05,
+      "loss": 1.359,
+      "step": 21000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8119054242976204e-05,
+      "loss": 1.3596,
+      "step": 21500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.807531131839425e-05,
+      "loss": 1.3405,
+      "step": 22000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8031568393812304e-05,
+      "loss": 1.3363,
+      "step": 22500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.798782546923035e-05,
+      "loss": 1.3479,
+      "step": 23000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7944082544648404e-05,
+      "loss": 1.3285,
+      "step": 23500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790033962006646e-05,
+      "loss": 1.3237,
+      "step": 24000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.785659669548451e-05,
+      "loss": 1.3186,
+      "step": 24500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7812853770902563e-05,
+      "loss": 1.3124,
+      "step": 25000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.776911084632061e-05,
+      "loss": 1.3197,
+      "step": 25500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772536792173866e-05,
+      "loss": 1.3098,
+      "step": 26000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768162499715671e-05,
+      "loss": 1.3107,
+      "step": 26500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763788207257476e-05,
+      "loss": 1.3026,
+      "step": 27000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7594139147992816e-05,
+      "loss": 1.2984,
+      "step": 27500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755039622341086e-05,
+      "loss": 1.2931,
+      "step": 28000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7506653298828916e-05,
+      "loss": 1.2904,
+      "step": 28500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746291037424697e-05,
+      "loss": 1.2794,
+      "step": 29000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.741916744966502e-05,
+      "loss": 1.2796,
+      "step": 29500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.737542452508307e-05,
+      "loss": 1.2809,
+      "step": 30000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.733168160050112e-05,
+      "loss": 1.2768,
+      "step": 30500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7287938675919175e-05,
+      "loss": 1.2627,
+      "step": 31000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.724419575133722e-05,
+      "loss": 1.2562,
+      "step": 31500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7200452826755275e-05,
+      "loss": 1.2607,
+      "step": 32000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.715670990217332e-05,
+      "loss": 1.2509,
+      "step": 32500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7112966977591375e-05,
+      "loss": 1.2463,
+      "step": 33000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706922405300943e-05,
+      "loss": 1.2486,
+      "step": 33500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.702548112842748e-05,
+      "loss": 1.2422,
+      "step": 34000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6981738203845535e-05,
+      "loss": 1.2408,
+      "step": 34500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.693799527926358e-05,
+      "loss": 1.2355,
+      "step": 35000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6894252354681634e-05,
+      "loss": 1.2319,
+      "step": 35500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.685050943009968e-05,
+      "loss": 1.2388,
+      "step": 36000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6806766505517734e-05,
+      "loss": 1.2364,
+      "step": 36500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.676302358093579e-05,
+      "loss": 1.2296,
+      "step": 37000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6719280656353834e-05,
+      "loss": 1.2273,
+      "step": 37500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.667553773177189e-05,
+      "loss": 1.2235,
+      "step": 38000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.663179480718994e-05,
+      "loss": 1.2189,
+      "step": 38500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6588051882607994e-05,
+      "loss": 1.2125,
+      "step": 39000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.654430895802604e-05,
+      "loss": 1.2097,
+      "step": 39500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.650056603344409e-05,
+      "loss": 1.2128,
+      "step": 40000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6456823108862147e-05,
+      "loss": 1.2101,
+      "step": 40500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.641308018428019e-05,
+      "loss": 1.215,
+      "step": 41000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6369337259698246e-05,
+      "loss": 1.1992,
+      "step": 41500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.632559433511629e-05,
+      "loss": 1.1992,
+      "step": 42000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6281851410534346e-05,
+      "loss": 1.1957,
+      "step": 42500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.62381084859524e-05,
+      "loss": 1.1959,
+      "step": 43000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.619436556137045e-05,
+      "loss": 1.1935,
+      "step": 43500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6150622636788506e-05,
+      "loss": 1.1899,
+      "step": 44000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.610687971220655e-05,
+      "loss": 1.1975,
+      "step": 44500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6063136787624606e-05,
+      "loss": 1.1805,
+      "step": 45000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.601939386304265e-05,
+      "loss": 1.1896,
+      "step": 45500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5975650938460705e-05,
+      "loss": 1.1796,
+      "step": 46000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.593190801387875e-05,
+      "loss": 1.1755,
+      "step": 46500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.588816508929681e-05,
+      "loss": 1.1836,
+      "step": 47000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5844422164714865e-05,
+      "loss": 1.1733,
+      "step": 47500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.580067924013291e-05,
+      "loss": 1.1846,
+      "step": 48000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5756936315550965e-05,
+      "loss": 1.1714,
+      "step": 48500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.571319339096901e-05,
+      "loss": 1.1594,
+      "step": 49000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5669450466387064e-05,
+      "loss": 1.1824,
+      "step": 49500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.562570754180511e-05,
+      "loss": 1.1728,
+      "step": 50000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5581964617223164e-05,
+      "loss": 1.1635,
+      "step": 50500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553822169264122e-05,
+      "loss": 1.1688,
+      "step": 51000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.549447876805927e-05,
+      "loss": 1.1632,
+      "step": 51500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5450735843477324e-05,
+      "loss": 1.1604,
+      "step": 52000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.540699291889537e-05,
+      "loss": 1.1567,
+      "step": 52500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5363249994313424e-05,
+      "loss": 1.1599,
+      "step": 53000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.531950706973147e-05,
+      "loss": 1.1477,
+      "step": 53500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5275764145149523e-05,
+      "loss": 1.1527,
+      "step": 54000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.523202122056758e-05,
+      "loss": 1.1545,
+      "step": 54500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.518827829598562e-05,
+      "loss": 1.1526,
+      "step": 55000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5144535371403676e-05,
+      "loss": 1.1527,
+      "step": 55500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.510079244682173e-05,
+      "loss": 1.1484,
+      "step": 56000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.505704952223978e-05,
+      "loss": 1.1385,
+      "step": 56500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.501330659765783e-05,
+      "loss": 1.1341,
+      "step": 57000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.496956367307588e-05,
+      "loss": 1.1435,
+      "step": 57500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4925820748493936e-05,
+      "loss": 1.1392,
+      "step": 58000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.488207782391198e-05,
+      "loss": 1.1442,
+      "step": 58500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4838334899330036e-05,
+      "loss": 1.1322,
+      "step": 59000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.479459197474808e-05,
+      "loss": 1.1313,
+      "step": 59500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4750849050166135e-05,
+      "loss": 1.1354,
+      "step": 60000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.470710612558419e-05,
+      "loss": 1.1421,
+      "step": 60500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.466336320100224e-05,
+      "loss": 1.1385,
+      "step": 61000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4619620276420295e-05,
+      "loss": 1.1394,
+      "step": 61500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.457587735183834e-05,
+      "loss": 1.1251,
+      "step": 62000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4532134427256395e-05,
+      "loss": 1.1233,
+      "step": 62500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.448839150267444e-05,
+      "loss": 1.1182,
+      "step": 63000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4444648578092495e-05,
+      "loss": 1.1261,
+      "step": 63500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.440090565351055e-05,
+      "loss": 1.1261,
+      "step": 64000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4357162728928594e-05,
+      "loss": 1.1266,
+      "step": 64500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.431341980434665e-05,
+      "loss": 1.1068,
+      "step": 65000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.42696768797647e-05,
+      "loss": 1.1254,
+      "step": 65500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4225933955182754e-05,
+      "loss": 1.1076,
+      "step": 66000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.41821910306008e-05,
+      "loss": 1.1157,
+      "step": 66500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4138448106018854e-05,
+      "loss": 1.1159,
+      "step": 67000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.409470518143691e-05,
+      "loss": 1.1159,
+      "step": 67500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4050962256854954e-05,
+      "loss": 1.1041,
+      "step": 68000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.400721933227301e-05,
+      "loss": 1.1021,
+      "step": 68500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.396347640769105e-05,
+      "loss": 1.1095,
+      "step": 69000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.391973348310911e-05,
+      "loss": 1.104,
+      "step": 69500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.387599055852716e-05,
+      "loss": 1.117,
+      "step": 70000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.383224763394521e-05,
+      "loss": 1.1044,
+      "step": 70500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3788504709363266e-05,
+      "loss": 1.1054,
+      "step": 71000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.374476178478131e-05,
+      "loss": 1.0963,
+      "step": 71500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3701018860199366e-05,
+      "loss": 1.098,
+      "step": 72000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.365727593561741e-05,
+      "loss": 1.1046,
+      "step": 72500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3613533011035466e-05,
+      "loss": 1.0942,
+      "step": 73000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.356979008645352e-05,
+      "loss": 1.0921,
+      "step": 73500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.352604716187157e-05,
+      "loss": 1.1,
+      "step": 74000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3482304237289626e-05,
+      "loss": 1.0981,
+      "step": 74500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.343856131270767e-05,
+      "loss": 1.0947,
+      "step": 75000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3394818388125725e-05,
+      "loss": 1.0902,
+      "step": 75500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.335107546354377e-05,
+      "loss": 1.0892,
+      "step": 76000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3307332538961825e-05,
+      "loss": 1.0959,
+      "step": 76500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.326358961437987e-05,
+      "loss": 1.0793,
+      "step": 77000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3219846689797925e-05,
+      "loss": 1.0889,
+      "step": 77500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.317610376521598e-05,
+      "loss": 1.0903,
+      "step": 78000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.313236084063403e-05,
+      "loss": 1.0773,
+      "step": 78500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3088617916052085e-05,
+      "loss": 1.0856,
+      "step": 79000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.304487499147013e-05,
+      "loss": 1.0876,
+      "step": 79500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3001132066888184e-05,
+      "loss": 1.0767,
+      "step": 80000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.295738914230623e-05,
+      "loss": 1.0776,
+      "step": 80500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2913646217724284e-05,
+      "loss": 1.0822,
+      "step": 81000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.286990329314234e-05,
+      "loss": 1.0842,
+      "step": 81500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2826160368560384e-05,
+      "loss": 1.0712,
+      "step": 82000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.278241744397844e-05,
+      "loss": 1.0784,
+      "step": 82500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.273867451939649e-05,
+      "loss": 1.0708,
+      "step": 83000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2694931594814543e-05,
+      "loss": 1.0788,
+      "step": 83500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.26511886702326e-05,
+      "loss": 1.0708,
+      "step": 84000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.260744574565064e-05,
+      "loss": 1.0669,
+      "step": 84500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2563702821068696e-05,
+      "loss": 1.0752,
+      "step": 85000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.251995989648674e-05,
+      "loss": 1.0697,
+      "step": 85500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2476216971904796e-05,
+      "loss": 1.0724,
+      "step": 86000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.243247404732284e-05,
+      "loss": 1.0684,
+      "step": 86500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2388731122740896e-05,
+      "loss": 1.0609,
+      "step": 87000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.234498819815895e-05,
+      "loss": 1.0639,
+      "step": 87500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2301245273577e-05,
+      "loss": 1.0662,
+      "step": 88000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2257502348995056e-05,
+      "loss": 1.0604,
+      "step": 88500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.22137594244131e-05,
+      "loss": 1.059,
+      "step": 89000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2170016499831155e-05,
+      "loss": 1.0628,
+      "step": 89500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.21262735752492e-05,
+      "loss": 1.0674,
+      "step": 90000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2082530650667255e-05,
+      "loss": 1.0577,
+      "step": 90500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.203878772608531e-05,
+      "loss": 1.0672,
+      "step": 91000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1995044801503355e-05,
+      "loss": 1.0518,
+      "step": 91500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1951301876921415e-05,
+      "loss": 1.0598,
+      "step": 92000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.190755895233946e-05,
+      "loss": 1.0618,
+      "step": 92500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1863816027757515e-05,
+      "loss": 1.0577,
+      "step": 93000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.182007310317556e-05,
+      "loss": 1.0391,
+      "step": 93500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1776330178593614e-05,
+      "loss": 1.0532,
+      "step": 94000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.173258725401167e-05,
+      "loss": 1.056,
+      "step": 94500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1688844329429714e-05,
+      "loss": 1.0551,
+      "step": 95000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.164510140484777e-05,
+      "loss": 1.0645,
+      "step": 95500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.160135848026582e-05,
+      "loss": 1.0493,
+      "step": 96000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1557615555683874e-05,
+      "loss": 1.0474,
+      "step": 96500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.151387263110192e-05,
+      "loss": 1.0496,
+      "step": 97000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1470129706519974e-05,
+      "loss": 1.0448,
+      "step": 97500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.142638678193803e-05,
+      "loss": 1.0435,
+      "step": 98000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.138264385735607e-05,
+      "loss": 1.0425,
+      "step": 98500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1338900932774127e-05,
+      "loss": 1.0473,
+      "step": 99000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.129515800819217e-05,
+      "loss": 1.045,
+      "step": 99500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1251415083610226e-05,
+      "loss": 1.0458,
+      "step": 100000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.120767215902828e-05,
+      "loss": 1.0434,
+      "step": 100500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.116392923444633e-05,
+      "loss": 1.0472,
+      "step": 101000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1120186309864386e-05,
+      "loss": 1.0413,
+      "step": 101500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.107644338528243e-05,
+      "loss": 1.0384,
+      "step": 102000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1032700460700486e-05,
+      "loss": 1.0507,
+      "step": 102500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.098895753611853e-05,
+      "loss": 1.0463,
+      "step": 103000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0945214611536586e-05,
+      "loss": 1.0369,
+      "step": 103500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.090147168695463e-05,
+      "loss": 1.0307,
+      "step": 104000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0857728762372685e-05,
+      "loss": 1.0376,
+      "step": 104500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.081398583779074e-05,
+      "loss": 1.0388,
+      "step": 105000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.077024291320879e-05,
+      "loss": 1.0377,
+      "step": 105500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0726499988626845e-05,
+      "loss": 1.0457,
+      "step": 106000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.068275706404489e-05,
+      "loss": 1.0385,
+      "step": 106500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0639014139462945e-05,
+      "loss": 1.0309,
+      "step": 107000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0595271214881e-05,
+      "loss": 1.0304,
+      "step": 107500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0551528290299045e-05,
+      "loss": 1.0435,
+      "step": 108000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05077853657171e-05,
+      "loss": 1.0343,
+      "step": 108500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0464042441135144e-05,
+      "loss": 1.0269,
+      "step": 109000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.04202995165532e-05,
+      "loss": 1.0273,
+      "step": 109500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.037655659197125e-05,
+      "loss": 1.0279,
+      "step": 110000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0332813667389304e-05,
+      "loss": 1.0382,
+      "step": 110500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028907074280736e-05,
+      "loss": 1.0276,
+      "step": 111000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0245327818225404e-05,
+      "loss": 1.022,
+      "step": 111500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020158489364346e-05,
+      "loss": 1.0252,
+      "step": 112000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0157841969061503e-05,
+      "loss": 1.0208,
+      "step": 112500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.011409904447956e-05,
+      "loss": 1.0257,
+      "step": 113000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.00703561198976e-05,
+      "loss": 1.0258,
+      "step": 113500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0026613195315656e-05,
+      "loss": 1.0382,
+      "step": 114000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9982870270733716e-05,
+      "loss": 1.0222,
+      "step": 114500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.993912734615176e-05,
+      "loss": 1.0119,
+      "step": 115000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9895384421569816e-05,
+      "loss": 1.0255,
+      "step": 115500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.985164149698786e-05,
+      "loss": 1.0346,
+      "step": 116000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9807898572405916e-05,
+      "loss": 1.0212,
+      "step": 116500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.976415564782396e-05,
+      "loss": 1.0277,
+      "step": 117000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9720412723242016e-05,
+      "loss": 1.0188,
+      "step": 117500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.967666979866007e-05,
+      "loss": 1.0278,
+      "step": 118000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.963292687407812e-05,
+      "loss": 1.0223,
+      "step": 118500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9589183949496175e-05,
+      "loss": 1.017,
+      "step": 119000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954544102491422e-05,
+      "loss": 1.016,
+      "step": 119500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9501698100332275e-05,
+      "loss": 1.0154,
+      "step": 120000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.945795517575032e-05,
+      "loss": 1.0134,
+      "step": 120500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9414212251168375e-05,
+      "loss": 1.0099,
+      "step": 121000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.937046932658643e-05,
+      "loss": 1.0132,
+      "step": 121500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9326726402004475e-05,
+      "loss": 1.0231,
+      "step": 122000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.928298347742253e-05,
+      "loss": 1.0193,
+      "step": 122500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.923924055284058e-05,
+      "loss": 1.0136,
+      "step": 123000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9195497628258634e-05,
+      "loss": 1.0189,
+      "step": 123500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.915175470367668e-05,
+      "loss": 1.0088,
+      "step": 124000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9108011779094734e-05,
+      "loss": 1.017,
+      "step": 124500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.906426885451279e-05,
+      "loss": 1.0128,
+      "step": 125000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9020525929930834e-05,
+      "loss": 1.0063,
+      "step": 125500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.897678300534889e-05,
+      "loss": 1.0091,
+      "step": 126000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8933040080766934e-05,
+      "loss": 1.0133,
+      "step": 126500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.888929715618499e-05,
+      "loss": 1.0127,
+      "step": 127000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.884555423160304e-05,
+      "loss": 1.0123,
+      "step": 127500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.880181130702109e-05,
+      "loss": 1.0134,
+      "step": 128000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8758068382439147e-05,
+      "loss": 1.0029,
+      "step": 128500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871432545785719e-05,
+      "loss": 1.0109,
+      "step": 129000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8670582533275246e-05,
+      "loss": 1.0069,
+      "step": 129500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.862683960869329e-05,
+      "loss": 1.0,
+      "step": 130000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8583096684111346e-05,
+      "loss": 1.0108,
+      "step": 130500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.853935375952939e-05,
+      "loss": 1.0087,
+      "step": 131000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8495610834947446e-05,
+      "loss": 0.9997,
+      "step": 131500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.84518679103655e-05,
+      "loss": 1.0022,
+      "step": 132000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.840812498578355e-05,
+      "loss": 0.9991,
+      "step": 132500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8364382061201606e-05,
+      "loss": 1.004,
+      "step": 133000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.832063913661965e-05,
+      "loss": 1.0063,
+      "step": 133500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8276896212037705e-05,
+      "loss": 0.9984,
+      "step": 134000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823315328745576e-05,
+      "loss": 1.0,
+      "step": 134500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8189410362873805e-05,
+      "loss": 1.0013,
+      "step": 135000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814566743829186e-05,
+      "loss": 0.9971,
+      "step": 135500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8101924513709905e-05,
+      "loss": 0.9998,
+      "step": 136000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805818158912796e-05,
+      "loss": 0.9941,
+      "step": 136500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801443866454601e-05,
+      "loss": 0.9899,
+      "step": 137000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7970695739964065e-05,
+      "loss": 0.9917,
+      "step": 137500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.792695281538212e-05,
+      "loss": 0.9908,
+      "step": 138000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7883209890800164e-05,
+      "loss": 1.0001,
+      "step": 138500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.783946696621822e-05,
+      "loss": 0.9951,
+      "step": 139000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7795724041636264e-05,
+      "loss": 0.9977,
+      "step": 139500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.775198111705432e-05,
+      "loss": 0.9982,
+      "step": 140000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.770823819247237e-05,
+      "loss": 0.9842,
+      "step": 140500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7664495267890424e-05,
+      "loss": 0.9843,
+      "step": 141000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.762075234330848e-05,
+      "loss": 0.9834,
+      "step": 141500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7577009418726523e-05,
+      "loss": 0.9998,
+      "step": 142000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.753326649414458e-05,
+      "loss": 0.9936,
+      "step": 142500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.748952356956262e-05,
+      "loss": 0.998,
+      "step": 143000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7445780644980676e-05,
+      "loss": 0.9969,
+      "step": 143500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.740203772039872e-05,
+      "loss": 0.9963,
+      "step": 144000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7358294795816776e-05,
+      "loss": 0.9885,
+      "step": 144500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.731455187123483e-05,
+      "loss": 0.9767,
+      "step": 145000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.727080894665288e-05,
+      "loss": 0.9867,
+      "step": 145500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7227066022070936e-05,
+      "loss": 0.9959,
+      "step": 146000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718332309748898e-05,
+      "loss": 0.9926,
+      "step": 146500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7139580172907036e-05,
+      "loss": 0.9948,
+      "step": 147000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.709583724832508e-05,
+      "loss": 0.9887,
+      "step": 147500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7052094323743135e-05,
+      "loss": 0.9894,
+      "step": 148000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700835139916119e-05,
+      "loss": 0.9862,
+      "step": 148500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6964608474579235e-05,
+      "loss": 0.9833,
+      "step": 149000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.692086554999729e-05,
+      "loss": 0.9861,
+      "step": 149500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.687712262541534e-05,
+      "loss": 0.987,
+      "step": 150000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6833379700833395e-05,
+      "loss": 0.9814,
+      "step": 150500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.678963677625144e-05,
+      "loss": 0.9857,
+      "step": 151000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6745893851669495e-05,
+      "loss": 0.9823,
+      "step": 151500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670215092708755e-05,
+      "loss": 0.9819,
+      "step": 152000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6658408002505594e-05,
+      "loss": 0.9973,
+      "step": 152500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.661466507792365e-05,
+      "loss": 0.9829,
+      "step": 153000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6570922153341694e-05,
+      "loss": 0.9864,
+      "step": 153500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.652717922875975e-05,
+      "loss": 0.9776,
+      "step": 154000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.64834363041778e-05,
+      "loss": 0.9825,
+      "step": 154500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6439693379595854e-05,
+      "loss": 0.9831,
+      "step": 155000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639595045501391e-05,
+      "loss": 0.9872,
+      "step": 155500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6352207530431954e-05,
+      "loss": 0.9818,
+      "step": 156000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.630846460585001e-05,
+      "loss": 0.973,
+      "step": 156500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.626472168126805e-05,
+      "loss": 0.9799,
+      "step": 157000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6220978756686107e-05,
+      "loss": 0.974,
+      "step": 157500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.617723583210416e-05,
+      "loss": 0.9769,
+      "step": 158000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6133492907522206e-05,
+      "loss": 0.9738,
+      "step": 158500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6089749982940266e-05,
+      "loss": 0.97,
+      "step": 159000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604600705835831e-05,
+      "loss": 0.9714,
+      "step": 159500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6002264133776366e-05,
+      "loss": 0.9758,
+      "step": 160000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.595852120919441e-05,
+      "loss": 0.9769,
+      "step": 160500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5914778284612466e-05,
+      "loss": 0.975,
+      "step": 161000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587103536003052e-05,
+      "loss": 0.9711,
+      "step": 161500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5827292435448566e-05,
+      "loss": 0.9669,
+      "step": 162000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.578354951086662e-05,
+      "loss": 0.9657,
+      "step": 162500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573980658628467e-05,
+      "loss": 0.9671,
+      "step": 163000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5696063661702725e-05,
+      "loss": 0.9737,
+      "step": 163500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.565232073712077e-05,
+      "loss": 0.9752,
+      "step": 164000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5608577812538825e-05,
+      "loss": 0.9705,
+      "step": 164500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556483488795688e-05,
+      "loss": 0.9768,
+      "step": 165000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5521091963374925e-05,
+      "loss": 0.9682,
+      "step": 165500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.547734903879298e-05,
+      "loss": 0.9684,
+      "step": 166000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5433606114211025e-05,
+      "loss": 0.969,
+      "step": 166500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538986318962908e-05,
+      "loss": 0.9683,
+      "step": 167000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534612026504713e-05,
+      "loss": 0.9636,
+      "step": 167500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5302377340465184e-05,
+      "loss": 0.9517,
+      "step": 168000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.525863441588324e-05,
+      "loss": 0.9683,
+      "step": 168500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5214891491301284e-05,
+      "loss": 0.9741,
+      "step": 169000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.517114856671934e-05,
+      "loss": 0.9611,
+      "step": 169500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5127405642137384e-05,
+      "loss": 0.9703,
+      "step": 170000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.508366271755544e-05,
+      "loss": 0.9698,
+      "step": 170500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5039919792973483e-05,
+      "loss": 0.9631,
+      "step": 171000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.499617686839154e-05,
+      "loss": 0.9657,
+      "step": 171500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.495243394380959e-05,
+      "loss": 0.9646,
+      "step": 172000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.490869101922764e-05,
+      "loss": 0.9662,
+      "step": 172500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4864948094645696e-05,
+      "loss": 0.9672,
+      "step": 173000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.482120517006374e-05,
+      "loss": 0.9601,
+      "step": 173500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4777462245481796e-05,
+      "loss": 0.9628,
+      "step": 174000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473371932089984e-05,
+      "loss": 0.9647,
+      "step": 174500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4689976396317896e-05,
+      "loss": 0.9619,
+      "step": 175000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.464623347173595e-05,
+      "loss": 0.9633,
+      "step": 175500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4602490547153996e-05,
+      "loss": 0.9666,
+      "step": 176000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455874762257205e-05,
+      "loss": 0.9527,
+      "step": 176500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45150046979901e-05,
+      "loss": 0.963,
+      "step": 177000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4471261773408155e-05,
+      "loss": 0.957,
+      "step": 177500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.44275188488262e-05,
+      "loss": 0.963,
+      "step": 178000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4383775924244255e-05,
+      "loss": 0.9636,
+      "step": 178500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.434003299966231e-05,
+      "loss": 0.9626,
+      "step": 179000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4296290075080355e-05,
+      "loss": 0.9544,
+      "step": 179500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.425254715049841e-05,
+      "loss": 0.9584,
+      "step": 180000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4208804225916455e-05,
+      "loss": 0.9591,
+      "step": 180500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.416506130133451e-05,
+      "loss": 0.9513,
+      "step": 181000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.412131837675257e-05,
+      "loss": 0.9466,
+      "step": 181500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4077575452170614e-05,
+      "loss": 0.9601,
+      "step": 182000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.403383252758867e-05,
+      "loss": 0.9593,
+      "step": 182500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3990089603006714e-05,
+      "loss": 0.9594,
+      "step": 183000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394634667842477e-05,
+      "loss": 0.9533,
+      "step": 183500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3902603753842814e-05,
+      "loss": 0.9569,
+      "step": 184000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.385886082926087e-05,
+      "loss": 0.9579,
+      "step": 184500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.381511790467892e-05,
+      "loss": 0.9514,
+      "step": 185000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3771374980096974e-05,
+      "loss": 0.9527,
+      "step": 185500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.372763205551503e-05,
+      "loss": 0.9497,
+      "step": 186000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.368388913093307e-05,
+      "loss": 0.9412,
+      "step": 186500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.364014620635113e-05,
+      "loss": 0.9494,
+      "step": 187000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.359640328176917e-05,
+      "loss": 0.9415,
+      "step": 187500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3552660357187226e-05,
+      "loss": 0.9566,
+      "step": 188000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.350891743260528e-05,
+      "loss": 0.964,
+      "step": 188500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3465174508023326e-05,
+      "loss": 0.9502,
+      "step": 189000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.342143158344138e-05,
+      "loss": 0.9452,
+      "step": 189500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.337768865885943e-05,
+      "loss": 0.9406,
+      "step": 190000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3333945734277486e-05,
+      "loss": 0.9427,
+      "step": 190500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.329020280969553e-05,
+      "loss": 0.9484,
+      "step": 191000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3246459885113586e-05,
+      "loss": 0.9478,
+      "step": 191500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.320271696053164e-05,
+      "loss": 0.9488,
+      "step": 192000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3158974035949685e-05,
+      "loss": 0.945,
+      "step": 192500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.311523111136774e-05,
+      "loss": 0.9467,
+      "step": 193000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3071488186785785e-05,
+      "loss": 0.9436,
+      "step": 193500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.302774526220384e-05,
+      "loss": 0.952,
+      "step": 194000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.298400233762189e-05,
+      "loss": 0.955,
+      "step": 194500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2940259413039945e-05,
+      "loss": 0.9507,
+      "step": 195000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2896516488458e-05,
+      "loss": 0.9454,
+      "step": 195500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2852773563876045e-05,
+      "loss": 0.9452,
+      "step": 196000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.28090306392941e-05,
+      "loss": 0.9417,
+      "step": 196500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2765287714712144e-05,
+      "loss": 0.9427,
+      "step": 197000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.27215447901302e-05,
+      "loss": 0.9443,
+      "step": 197500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2677801865548244e-05,
+      "loss": 0.9414,
+      "step": 198000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.26340589409663e-05,
+      "loss": 0.9412,
+      "step": 198500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.259031601638435e-05,
+      "loss": 0.941,
+      "step": 199000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2546573091802404e-05,
+      "loss": 0.9386,
+      "step": 199500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.250283016722046e-05,
+      "loss": 0.945,
+      "step": 200000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2459087242638503e-05,
+      "loss": 0.936,
+      "step": 200500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.241534431805656e-05,
+      "loss": 0.9411,
+      "step": 201000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.23716013934746e-05,
+      "loss": 0.9303,
+      "step": 201500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2327858468892656e-05,
+      "loss": 0.9363,
+      "step": 202000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.228411554431071e-05,
+      "loss": 0.9218,
+      "step": 202500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2240372619728756e-05,
+      "loss": 0.9431,
+      "step": 203000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.219662969514681e-05,
+      "loss": 0.9437,
+      "step": 203500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.215288677056486e-05,
+      "loss": 0.9304,
+      "step": 204000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2109143845982916e-05,
+      "loss": 0.9485,
+      "step": 204500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.206540092140097e-05,
+      "loss": 0.941,
+      "step": 205000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2021657996819016e-05,
+      "loss": 0.9384,
+      "step": 205500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.197791507223707e-05,
+      "loss": 0.9423,
+      "step": 206000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.1934172147655115e-05,
+      "loss": 0.9406,
+      "step": 206500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.189042922307317e-05,
+      "loss": 0.9345,
+      "step": 207000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1846686298491215e-05,
+      "loss": 0.9403,
+      "step": 207500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1802943373909275e-05,
+      "loss": 0.9406,
+      "step": 208000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.175920044932733e-05,
+      "loss": 0.9395,
+      "step": 208500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1715457524745375e-05,
+      "loss": 0.9338,
+      "step": 209000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.167171460016343e-05,
+      "loss": 0.9299,
+      "step": 209500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1627971675581475e-05,
+      "loss": 0.9329,
+      "step": 210000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.158422875099953e-05,
+      "loss": 0.9317,
+      "step": 210500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1540485826417574e-05,
+      "loss": 0.9372,
+      "step": 211000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.149674290183563e-05,
+      "loss": 0.9328,
+      "step": 211500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.145299997725368e-05,
+      "loss": 0.9345,
+      "step": 212000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1409257052671734e-05,
+      "loss": 0.9358,
+      "step": 212500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.136551412808979e-05,
+      "loss": 0.9345,
+      "step": 213000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1321771203507834e-05,
+      "loss": 0.9223,
+      "step": 213500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.127802827892589e-05,
+      "loss": 0.9361,
+      "step": 214000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1234285354343934e-05,
+      "loss": 0.933,
+      "step": 214500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.119054242976199e-05,
+      "loss": 0.9349,
+      "step": 215000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.114679950518004e-05,
+      "loss": 0.9336,
+      "step": 215500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1103056580598087e-05,
+      "loss": 0.9267,
+      "step": 216000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.105931365601614e-05,
+      "loss": 0.9306,
+      "step": 216500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.101557073143419e-05,
+      "loss": 0.9295,
+      "step": 217000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.0971827806852246e-05,
+      "loss": 0.9341,
+      "step": 217500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.092808488227029e-05,
+      "loss": 0.9355,
+      "step": 218000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0884341957688346e-05,
+      "loss": 0.9207,
+      "step": 218500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.08405990331064e-05,
+      "loss": 0.925,
+      "step": 219000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0796856108524446e-05,
+      "loss": 0.9273,
+      "step": 219500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.07531131839425e-05,
+      "loss": 0.9292,
+      "step": 220000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0709370259360546e-05,
+      "loss": 0.9266,
+      "step": 220500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.06656273347786e-05,
+      "loss": 0.9311,
+      "step": 221000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.062188441019665e-05,
+      "loss": 0.9332,
+      "step": 221500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0578141485614705e-05,
+      "loss": 0.9362,
+      "step": 222000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.053439856103276e-05,
+      "loss": 0.9212,
+      "step": 222500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0490655636450805e-05,
+      "loss": 0.927,
+      "step": 223000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.044691271186886e-05,
+      "loss": 0.9247,
+      "step": 223500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0403169787286905e-05,
+      "loss": 0.9227,
+      "step": 224000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0359426862704958e-05,
+      "loss": 0.9303,
+      "step": 224500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0315683938123008e-05,
+      "loss": 0.9291,
+      "step": 225000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.027194101354106e-05,
+      "loss": 0.9285,
+      "step": 225500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0228198088959114e-05,
+      "loss": 0.9244,
+      "step": 226000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.018445516437716e-05,
+      "loss": 0.922,
+      "step": 226500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0140712239795214e-05,
+      "loss": 0.9282,
+      "step": 227000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0096969315213264e-05,
+      "loss": 0.92,
+      "step": 227500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0053226390631317e-05,
+      "loss": 0.9246,
+      "step": 228000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.000948346604937e-05,
+      "loss": 0.9303,
+      "step": 228500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9965740541467417e-05,
+      "loss": 0.9259,
+      "step": 229000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9921997616885474e-05,
+      "loss": 0.9201,
+      "step": 229500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.987825469230352e-05,
+      "loss": 0.9246,
+      "step": 230000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9834511767721573e-05,
+      "loss": 0.9209,
+      "step": 230500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.979076884313962e-05,
+      "loss": 0.9242,
+      "step": 231000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9747025918557677e-05,
+      "loss": 0.916,
+      "step": 231500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.970328299397573e-05,
+      "loss": 0.9233,
+      "step": 232000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9659540069393776e-05,
+      "loss": 0.9153,
+      "step": 232500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.961579714481183e-05,
+      "loss": 0.9244,
+      "step": 233000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.957205422022988e-05,
+      "loss": 0.9212,
+      "step": 233500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9528311295647933e-05,
+      "loss": 0.9225,
+      "step": 234000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.948456837106598e-05,
+      "loss": 0.9136,
+      "step": 234500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9440825446484032e-05,
+      "loss": 0.9268,
+      "step": 235000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9397082521902086e-05,
+      "loss": 0.9245,
+      "step": 235500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9353339597320135e-05,
+      "loss": 0.9256,
+      "step": 236000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.930959667273819e-05,
+      "loss": 0.9071,
+      "step": 236500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9265853748156235e-05,
+      "loss": 0.9128,
+      "step": 237000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.922211082357429e-05,
+      "loss": 0.9177,
+      "step": 237500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.917836789899234e-05,
+      "loss": 0.9146,
+      "step": 238000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.913462497441039e-05,
+      "loss": 0.9258,
+      "step": 238500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9090882049828445e-05,
+      "loss": 0.9175,
+      "step": 239000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.904713912524649e-05,
+      "loss": 0.9231,
+      "step": 239500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9003396200664545e-05,
+      "loss": 0.9188,
+      "step": 240000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8959653276082594e-05,
+      "loss": 0.9251,
+      "step": 240500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8915910351500648e-05,
+      "loss": 0.912,
+      "step": 241000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8872167426918694e-05,
+      "loss": 0.9169,
+      "step": 241500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8828424502336747e-05,
+      "loss": 0.9219,
+      "step": 242000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.87846815777548e-05,
+      "loss": 0.9263,
+      "step": 242500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.874093865317285e-05,
+      "loss": 0.9106,
+      "step": 243000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8697195728590904e-05,
+      "loss": 0.9096,
+      "step": 243500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.865345280400895e-05,
+      "loss": 0.9127,
+      "step": 244000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8609709879427004e-05,
+      "loss": 0.9255,
+      "step": 244500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8565966954845053e-05,
+      "loss": 0.9191,
+      "step": 245000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8522224030263107e-05,
+      "loss": 0.9222,
+      "step": 245500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.847848110568116e-05,
+      "loss": 0.9153,
+      "step": 246000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8434738181099206e-05,
+      "loss": 0.9129,
+      "step": 246500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.839099525651726e-05,
+      "loss": 0.9101,
+      "step": 247000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.834725233193531e-05,
+      "loss": 0.9109,
+      "step": 247500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8303509407353363e-05,
+      "loss": 0.9116,
+      "step": 248000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.825976648277141e-05,
+      "loss": 0.9156,
+      "step": 248500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8216023558189462e-05,
+      "loss": 0.9154,
+      "step": 249000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8172280633607516e-05,
+      "loss": 0.9124,
+      "step": 249500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8128537709025566e-05,
+      "loss": 0.9136,
+      "step": 250000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.808479478444362e-05,
+      "loss": 0.9161,
+      "step": 250500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8041051859861665e-05,
+      "loss": 0.9217,
+      "step": 251000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.799730893527972e-05,
+      "loss": 0.9175,
+      "step": 251500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7953566010697775e-05,
+      "loss": 0.9093,
+      "step": 252000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7909823086115822e-05,
+      "loss": 0.9128,
+      "step": 252500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7866080161533875e-05,
+      "loss": 0.9141,
+      "step": 253000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.782233723695192e-05,
+      "loss": 0.92,
+      "step": 253500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7778594312369978e-05,
+      "loss": 0.9134,
+      "step": 254000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7734851387788025e-05,
+      "loss": 0.9142,
+      "step": 254500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7691108463206078e-05,
+      "loss": 0.9059,
+      "step": 255000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.764736553862413e-05,
+      "loss": 0.9113,
+      "step": 255500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.760362261404218e-05,
+      "loss": 0.9042,
+      "step": 256000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7559879689460234e-05,
+      "loss": 0.9118,
+      "step": 256500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.751613676487828e-05,
+      "loss": 0.9101,
+      "step": 257000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7472393840296334e-05,
+      "loss": 0.9068,
+      "step": 257500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7428650915714384e-05,
+      "loss": 0.9,
+      "step": 258000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7384907991132437e-05,
+      "loss": 0.9139,
+      "step": 258500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.734116506655049e-05,
+      "loss": 0.914,
+      "step": 259000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7297422141968537e-05,
+      "loss": 0.9036,
+      "step": 259500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.725367921738659e-05,
+      "loss": 0.8963,
+      "step": 260000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.720993629280464e-05,
+      "loss": 0.9092,
+      "step": 260500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7166193368222693e-05,
+      "loss": 0.9088,
+      "step": 261000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.712245044364074e-05,
+      "loss": 0.9062,
+      "step": 261500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7078707519058793e-05,
+      "loss": 0.9079,
+      "step": 262000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7034964594476846e-05,
+      "loss": 0.9013,
+      "step": 262500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6991221669894896e-05,
+      "loss": 0.91,
+      "step": 263000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.694747874531295e-05,
+      "loss": 0.9053,
+      "step": 263500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6903735820730996e-05,
+      "loss": 0.9064,
+      "step": 264000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.685999289614905e-05,
+      "loss": 0.9046,
+      "step": 264500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.68162499715671e-05,
+      "loss": 0.9112,
+      "step": 265000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6772507046985152e-05,
+      "loss": 0.9095,
+      "step": 265500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6728764122403205e-05,
+      "loss": 0.9012,
+      "step": 266000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6685021197821252e-05,
+      "loss": 0.9013,
+      "step": 266500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6641278273239305e-05,
+      "loss": 0.9079,
+      "step": 267000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6597535348657355e-05,
+      "loss": 0.9006,
+      "step": 267500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6553792424075408e-05,
+      "loss": 0.9007,
+      "step": 268000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6510049499493455e-05,
+      "loss": 0.9021,
+      "step": 268500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6466306574911508e-05,
+      "loss": 0.9024,
+      "step": 269000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.642256365032956e-05,
+      "loss": 0.9,
+      "step": 269500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.637882072574761e-05,
+      "loss": 0.9006,
+      "step": 270000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6335077801165664e-05,
+      "loss": 0.9132,
+      "step": 270500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.629133487658371e-05,
+      "loss": 0.9029,
+      "step": 271000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6247591952001764e-05,
+      "loss": 0.9044,
+      "step": 271500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6203849027419814e-05,
+      "loss": 0.9002,
+      "step": 272000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6160106102837867e-05,
+      "loss": 0.9009,
+      "step": 272500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.611636317825592e-05,
+      "loss": 0.892,
+      "step": 273000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6072620253673967e-05,
+      "loss": 0.9094,
+      "step": 273500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.602887732909202e-05,
+      "loss": 0.9029,
+      "step": 274000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.598513440451007e-05,
+      "loss": 0.9008,
+      "step": 274500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5941391479928123e-05,
+      "loss": 0.903,
+      "step": 275000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5897648555346177e-05,
+      "loss": 0.8976,
+      "step": 275500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5853905630764223e-05,
+      "loss": 0.8995,
+      "step": 276000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.581016270618228e-05,
+      "loss": 0.9017,
+      "step": 276500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5766419781600326e-05,
+      "loss": 0.8977,
+      "step": 277000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.572267685701838e-05,
+      "loss": 0.9115,
+      "step": 277500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5678933932436426e-05,
+      "loss": 0.8982,
+      "step": 278000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5635191007854482e-05,
+      "loss": 0.9043,
+      "step": 278500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5591448083272536e-05,
+      "loss": 0.901,
+      "step": 279000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5547705158690582e-05,
+      "loss": 0.8973,
+      "step": 279500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5503962234108635e-05,
+      "loss": 0.896,
+      "step": 280000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5460219309526685e-05,
+      "loss": 0.8921,
+      "step": 280500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.541647638494474e-05,
+      "loss": 0.8933,
+      "step": 281000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5372733460362785e-05,
+      "loss": 0.8947,
+      "step": 281500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.532899053578084e-05,
+      "loss": 0.8948,
+      "step": 282000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.528524761119889e-05,
+      "loss": 0.8995,
+      "step": 282500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.524150468661694e-05,
+      "loss": 0.8924,
+      "step": 283000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5197761762034995e-05,
+      "loss": 0.8966,
+      "step": 283500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.515401883745304e-05,
+      "loss": 0.8905,
+      "step": 284000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5110275912871094e-05,
+      "loss": 0.8923,
+      "step": 284500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5066532988289144e-05,
+      "loss": 0.8954,
+      "step": 285000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5022790063707198e-05,
+      "loss": 0.8979,
+      "step": 285500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.4979047139125247e-05,
+      "loss": 0.9015,
+      "step": 286000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.4935304214543297e-05,
+      "loss": 0.8922,
+      "step": 286500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4891561289961347e-05,
+      "loss": 0.8916,
+      "step": 287000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4847818365379404e-05,
+      "loss": 0.8892,
+      "step": 287500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4804075440797454e-05,
+      "loss": 0.8972,
+      "step": 288000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4760332516215504e-05,
+      "loss": 0.8991,
+      "step": 288500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4716589591633553e-05,
+      "loss": 0.8923,
+      "step": 289000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4672846667051607e-05,
+      "loss": 0.8889,
+      "step": 289500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4629103742469657e-05,
+      "loss": 0.8968,
+      "step": 290000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4585360817887706e-05,
+      "loss": 0.9009,
+      "step": 290500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.454161789330576e-05,
+      "loss": 0.8931,
+      "step": 291000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.449787496872381e-05,
+      "loss": 0.8886,
+      "step": 291500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4454132044141863e-05,
+      "loss": 0.8802,
+      "step": 292000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4410389119559913e-05,
+      "loss": 0.8892,
+      "step": 292500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4366646194977962e-05,
+      "loss": 0.8948,
+      "step": 293000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4322903270396012e-05,
+      "loss": 0.893,
+      "step": 293500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4279160345814066e-05,
+      "loss": 0.8882,
+      "step": 294000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.423541742123212e-05,
+      "loss": 0.896,
+      "step": 294500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.419167449665017e-05,
+      "loss": 0.891,
+      "step": 295000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.414793157206822e-05,
+      "loss": 0.8908,
+      "step": 295500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.410418864748627e-05,
+      "loss": 0.8884,
+      "step": 296000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4060445722904322e-05,
+      "loss": 0.8988,
+      "step": 296500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.401670279832237e-05,
+      "loss": 0.8829,
+      "step": 297000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3972959873740425e-05,
+      "loss": 0.8892,
+      "step": 297500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3929216949158475e-05,
+      "loss": 0.8917,
+      "step": 298000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3885474024576528e-05,
+      "loss": 0.902,
+      "step": 298500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3841731099994578e-05,
+      "loss": 0.8899,
+      "step": 299000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3797988175412628e-05,
+      "loss": 0.8855,
+      "step": 299500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3754245250830678e-05,
+      "loss": 0.883,
+      "step": 300000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.371050232624873e-05,
+      "loss": 0.8831,
+      "step": 300500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3666759401666784e-05,
+      "loss": 0.8867,
+      "step": 301000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3623016477084834e-05,
+      "loss": 0.8907,
+      "step": 301500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3579273552502884e-05,
+      "loss": 0.8897,
+      "step": 302000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3535530627920934e-05,
+      "loss": 0.8842,
+      "step": 302500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3491787703338987e-05,
+      "loss": 0.8853,
+      "step": 303000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3448044778757037e-05,
+      "loss": 0.8922,
+      "step": 303500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3404301854175087e-05,
+      "loss": 0.8902,
+      "step": 304000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.336055892959314e-05,
+      "loss": 0.892,
+      "step": 304500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.331681600501119e-05,
+      "loss": 0.8848,
+      "step": 305000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3273073080429243e-05,
+      "loss": 0.8929,
+      "step": 305500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3229330155847293e-05,
+      "loss": 0.8909,
+      "step": 306000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3185587231265343e-05,
+      "loss": 0.9047,
+      "step": 306500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3141844306683393e-05,
+      "loss": 0.8952,
+      "step": 307000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3098101382101446e-05,
+      "loss": 0.8832,
+      "step": 307500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.30543584575195e-05,
+      "loss": 0.8892,
+      "step": 308000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.301061553293755e-05,
+      "loss": 0.8865,
+      "step": 308500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.29668726083556e-05,
+      "loss": 0.8829,
+      "step": 309000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.292312968377365e-05,
+      "loss": 0.885,
+      "step": 309500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2879386759191702e-05,
+      "loss": 0.8789,
+      "step": 310000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2835643834609752e-05,
+      "loss": 0.8861,
+      "step": 310500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2791900910027805e-05,
+      "loss": 0.8869,
+      "step": 311000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2748157985445855e-05,
+      "loss": 0.8885,
+      "step": 311500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2704415060863908e-05,
+      "loss": 0.8849,
+      "step": 312000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2660672136281958e-05,
+      "loss": 0.8824,
+      "step": 312500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2616929211700008e-05,
+      "loss": 0.8834,
+      "step": 313000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2573186287118058e-05,
+      "loss": 0.888,
+      "step": 313500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.252944336253611e-05,
+      "loss": 0.8845,
+      "step": 314000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2485700437954164e-05,
+      "loss": 0.8777,
+      "step": 314500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2441957513372214e-05,
+      "loss": 0.885,
+      "step": 315000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2398214588790264e-05,
+      "loss": 0.8849,
+      "step": 315500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2354471664208314e-05,
+      "loss": 0.888,
+      "step": 316000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2310728739626367e-05,
+      "loss": 0.8833,
+      "step": 316500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2266985815044417e-05,
+      "loss": 0.8724,
+      "step": 317000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.222324289046247e-05,
+      "loss": 0.8854,
+      "step": 317500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.217949996588052e-05,
+      "loss": 0.8858,
+      "step": 318000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.213575704129857e-05,
+      "loss": 0.8824,
+      "step": 318500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2092014116716623e-05,
+      "loss": 0.8792,
+      "step": 319000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2048271192134673e-05,
+      "loss": 0.8832,
+      "step": 319500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2004528267552723e-05,
+      "loss": 0.8812,
+      "step": 320000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1960785342970773e-05,
+      "loss": 0.8799,
+      "step": 320500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.191704241838883e-05,
+      "loss": 0.8831,
+      "step": 321000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.187329949380688e-05,
+      "loss": 0.895,
+      "step": 321500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.182955656922493e-05,
+      "loss": 0.8882,
+      "step": 322000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.178581364464298e-05,
+      "loss": 0.8905,
+      "step": 322500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1742070720061032e-05,
+      "loss": 0.887,
+      "step": 323000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1698327795479082e-05,
+      "loss": 0.8792,
+      "step": 323500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1654584870897132e-05,
+      "loss": 0.8734,
+      "step": 324000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1610841946315185e-05,
+      "loss": 0.8756,
+      "step": 324500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1567099021733235e-05,
+      "loss": 0.878,
+      "step": 325000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.152335609715129e-05,
+      "loss": 0.8806,
+      "step": 325500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.147961317256934e-05,
+      "loss": 0.8846,
+      "step": 326000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1435870247987388e-05,
+      "loss": 0.8813,
+      "step": 326500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1392127323405438e-05,
+      "loss": 0.8765,
+      "step": 327000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.134838439882349e-05,
+      "loss": 0.8765,
+      "step": 327500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1304641474241545e-05,
+      "loss": 0.8834,
+      "step": 328000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1260898549659594e-05,
+      "loss": 0.8834,
+      "step": 328500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1217155625077644e-05,
+      "loss": 0.8704,
+      "step": 329000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1173412700495694e-05,
+      "loss": 0.8741,
+      "step": 329500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1129669775913747e-05,
+      "loss": 0.8856,
+      "step": 330000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1085926851331797e-05,
+      "loss": 0.8743,
+      "step": 330500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.104218392674985e-05,
+      "loss": 0.8745,
+      "step": 331000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.09984410021679e-05,
+      "loss": 0.8841,
+      "step": 331500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.095469807758595e-05,
+      "loss": 0.8729,
+      "step": 332000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0910955153004004e-05,
+      "loss": 0.8849,
+      "step": 332500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0867212228422053e-05,
+      "loss": 0.8688,
+      "step": 333000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0823469303840103e-05,
+      "loss": 0.881,
+      "step": 333500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0779726379258153e-05,
+      "loss": 0.8702,
+      "step": 334000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.073598345467621e-05,
+      "loss": 0.873,
+      "step": 334500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.069224053009426e-05,
+      "loss": 0.875,
+      "step": 335000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.064849760551231e-05,
+      "loss": 0.877,
+      "step": 335500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.060475468093036e-05,
+      "loss": 0.8762,
+      "step": 336000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0561011756348413e-05,
+      "loss": 0.8759,
+      "step": 336500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0517268831766463e-05,
+      "loss": 0.8769,
+      "step": 337000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0473525907184512e-05,
+      "loss": 0.8723,
+      "step": 337500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0429782982602566e-05,
+      "loss": 0.877,
+      "step": 338000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0386040058020615e-05,
+      "loss": 0.8787,
+      "step": 338500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.034229713343867e-05,
+      "loss": 0.8702,
+      "step": 339000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.029855420885672e-05,
+      "loss": 0.8868,
+      "step": 339500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.025481128427477e-05,
+      "loss": 0.8742,
+      "step": 340000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.021106835969282e-05,
+      "loss": 0.8718,
+      "step": 340500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.016732543511087e-05,
+      "loss": 0.8691,
+      "step": 341000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0123582510528925e-05,
+      "loss": 0.8759,
+      "step": 341500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0079839585946975e-05,
+      "loss": 0.8678,
+      "step": 342000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0036096661365025e-05,
+      "loss": 0.875,
+      "step": 342500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9992353736783074e-05,
+      "loss": 0.8752,
+      "step": 343000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9948610812201128e-05,
+      "loss": 0.8683,
+      "step": 343500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9904867887619178e-05,
+      "loss": 0.875,
+      "step": 344000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.986112496303723e-05,
+      "loss": 0.865,
+      "step": 344500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.981738203845528e-05,
+      "loss": 0.8785,
+      "step": 345000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9773639113873334e-05,
+      "loss": 0.8675,
+      "step": 345500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9729896189291384e-05,
+      "loss": 0.8764,
+      "step": 346000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9686153264709434e-05,
+      "loss": 0.8702,
+      "step": 346500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9642410340127484e-05,
+      "loss": 0.8766,
+      "step": 347000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9598667415545537e-05,
+      "loss": 0.8647,
+      "step": 347500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.955492449096359e-05,
+      "loss": 0.8635,
+      "step": 348000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.951118156638164e-05,
+      "loss": 0.8781,
+      "step": 348500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.946743864179969e-05,
+      "loss": 0.8749,
+      "step": 349000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.942369571721774e-05,
+      "loss": 0.869,
+      "step": 349500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9379952792635793e-05,
+      "loss": 0.8686,
+      "step": 350000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9336209868053843e-05,
+      "loss": 0.8678,
+      "step": 350500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9292466943471893e-05,
+      "loss": 0.8731,
+      "step": 351000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9248724018889946e-05,
+      "loss": 0.8716,
+      "step": 351500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9204981094307996e-05,
+      "loss": 0.8737,
+      "step": 352000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.916123816972605e-05,
+      "loss": 0.8757,
+      "step": 352500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.91174952451441e-05,
+      "loss": 0.8781,
+      "step": 353000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.907375232056215e-05,
+      "loss": 0.8746,
+      "step": 353500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.90300093959802e-05,
+      "loss": 0.8686,
+      "step": 354000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8986266471398252e-05,
+      "loss": 0.8619,
+      "step": 354500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8942523546816305e-05,
+      "loss": 0.8672,
+      "step": 355000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8898780622234355e-05,
+      "loss": 0.8666,
+      "step": 355500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8855037697652405e-05,
+      "loss": 0.8717,
+      "step": 356000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8811294773070455e-05,
+      "loss": 0.8694,
+      "step": 356500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8767551848488508e-05,
+      "loss": 0.8698,
+      "step": 357000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8723808923906558e-05,
+      "loss": 0.8635,
+      "step": 357500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.868006599932461e-05,
+      "loss": 0.8714,
+      "step": 358000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.863632307474266e-05,
+      "loss": 0.8658,
+      "step": 358500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8592580150160714e-05,
+      "loss": 0.8631,
+      "step": 359000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8548837225578764e-05,
+      "loss": 0.8599,
+      "step": 359500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8505094300996814e-05,
+      "loss": 0.8636,
+      "step": 360000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8461351376414864e-05,
+      "loss": 0.8689,
+      "step": 360500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8417608451832917e-05,
+      "loss": 0.8654,
+      "step": 361000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.837386552725097e-05,
+      "loss": 0.8588,
+      "step": 361500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.833012260266902e-05,
+      "loss": 0.8729,
+      "step": 362000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.828637967808707e-05,
+      "loss": 0.8614,
+      "step": 362500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.824263675350512e-05,
+      "loss": 0.8673,
+      "step": 363000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8198893828923173e-05,
+      "loss": 0.8617,
+      "step": 363500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8155150904341223e-05,
+      "loss": 0.8641,
+      "step": 364000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8111407979759276e-05,
+      "loss": 0.8661,
+      "step": 364500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8067665055177326e-05,
+      "loss": 0.8686,
+      "step": 365000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8023922130595376e-05,
+      "loss": 0.8639,
+      "step": 365500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.798017920601343e-05,
+      "loss": 0.8755,
+      "step": 366000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.793643628143148e-05,
+      "loss": 0.8638,
+      "step": 366500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.789269335684953e-05,
+      "loss": 0.871,
+      "step": 367000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.784895043226758e-05,
+      "loss": 0.8695,
+      "step": 367500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7805207507685636e-05,
+      "loss": 0.8613,
+      "step": 368000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7761464583103685e-05,
+      "loss": 0.8649,
+      "step": 368500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7717721658521735e-05,
+      "loss": 0.8547,
+      "step": 369000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7673978733939785e-05,
+      "loss": 0.8678,
+      "step": 369500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.763023580935784e-05,
+      "loss": 0.8696,
+      "step": 370000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7586492884775888e-05,
+      "loss": 0.8674,
+      "step": 370500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7542749960193938e-05,
+      "loss": 0.8682,
+      "step": 371000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.749900703561199e-05,
+      "loss": 0.8583,
+      "step": 371500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.745526411103004e-05,
+      "loss": 0.8672,
+      "step": 372000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7411521186448094e-05,
+      "loss": 0.8677,
+      "step": 372500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7367778261866144e-05,
+      "loss": 0.8607,
+      "step": 373000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7324035337284194e-05,
+      "loss": 0.8621,
+      "step": 373500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7280292412702244e-05,
+      "loss": 0.8674,
+      "step": 374000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7236549488120297e-05,
+      "loss": 0.8538,
+      "step": 374500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.719280656353835e-05,
+      "loss": 0.8583,
+      "step": 375000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.71490636389564e-05,
+      "loss": 0.8611,
+      "step": 375500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.710532071437445e-05,
+      "loss": 0.8761,
+      "step": 376000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.70615777897925e-05,
+      "loss": 0.8647,
+      "step": 376500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7017834865210553e-05,
+      "loss": 0.859,
+      "step": 377000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6974091940628603e-05,
+      "loss": 0.8647,
+      "step": 377500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6930349016046657e-05,
+      "loss": 0.8579,
+      "step": 378000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6886606091464706e-05,
+      "loss": 0.8639,
+      "step": 378500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.684286316688276e-05,
+      "loss": 0.8665,
+      "step": 379000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.679912024230081e-05,
+      "loss": 0.8596,
+      "step": 379500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.675537731771886e-05,
+      "loss": 0.8619,
+      "step": 380000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.671163439313691e-05,
+      "loss": 0.8534,
+      "step": 380500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6667891468554963e-05,
+      "loss": 0.8588,
+      "step": 381000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6624148543973016e-05,
+      "loss": 0.8533,
+      "step": 381500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6580405619391066e-05,
+      "loss": 0.8575,
+      "step": 382000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6536662694809116e-05,
+      "loss": 0.8621,
+      "step": 382500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6492919770227165e-05,
+      "loss": 0.8606,
+      "step": 383000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.644917684564522e-05,
+      "loss": 0.8583,
+      "step": 383500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.640543392106327e-05,
+      "loss": 0.8526,
+      "step": 384000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.636169099648132e-05,
+      "loss": 0.8559,
+      "step": 384500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.631794807189937e-05,
+      "loss": 0.8563,
+      "step": 385000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.627420514731742e-05,
+      "loss": 0.8448,
+      "step": 385500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6230462222735475e-05,
+      "loss": 0.8571,
+      "step": 386000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6186719298153525e-05,
+      "loss": 0.8517,
+      "step": 386500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6142976373571574e-05,
+      "loss": 0.8576,
+      "step": 387000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6099233448989624e-05,
+      "loss": 0.8528,
+      "step": 387500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6055490524407678e-05,
+      "loss": 0.8664,
+      "step": 388000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.601174759982573e-05,
+      "loss": 0.8547,
+      "step": 388500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.596800467524378e-05,
+      "loss": 0.8538,
+      "step": 389000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.592426175066183e-05,
+      "loss": 0.8618,
+      "step": 389500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.588051882607988e-05,
+      "loss": 0.8601,
+      "step": 390000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5836775901497934e-05,
+      "loss": 0.846,
+      "step": 390500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5793032976915984e-05,
+      "loss": 0.852,
+      "step": 391000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5749290052334037e-05,
+      "loss": 0.8413,
+      "step": 391500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5705547127752087e-05,
+      "loss": 0.8556,
+      "step": 392000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.566180420317014e-05,
+      "loss": 0.8622,
+      "step": 392500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.561806127858819e-05,
+      "loss": 0.8528,
+      "step": 393000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.557431835400624e-05,
+      "loss": 0.8561,
+      "step": 393500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.553057542942429e-05,
+      "loss": 0.8598,
+      "step": 394000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5486832504842343e-05,
+      "loss": 0.8533,
+      "step": 394500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5443089580260396e-05,
+      "loss": 0.8522,
+      "step": 395000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5399346655678446e-05,
+      "loss": 0.8559,
+      "step": 395500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5355603731096496e-05,
+      "loss": 0.8618,
+      "step": 396000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5311860806514546e-05,
+      "loss": 0.8563,
+      "step": 396500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.52681178819326e-05,
+      "loss": 0.8524,
+      "step": 397000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5224374957350649e-05,
+      "loss": 0.8506,
+      "step": 397500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5180632032768699e-05,
+      "loss": 0.8613,
+      "step": 398000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5136889108186752e-05,
+      "loss": 0.8543,
+      "step": 398500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5093146183604803e-05,
+      "loss": 0.8531,
+      "step": 399000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5049403259022853e-05,
+      "loss": 0.8462,
+      "step": 399500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5005660334440905e-05,
+      "loss": 0.8581,
+      "step": 400000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4961917409858955e-05,
+      "loss": 0.8507,
+      "step": 400500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4918174485277006e-05,
+      "loss": 0.8624,
+      "step": 401000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.487443156069506e-05,
+      "loss": 0.8547,
+      "step": 401500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4830688636113111e-05,
+      "loss": 0.8495,
+      "step": 402000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4786945711531161e-05,
+      "loss": 0.8529,
+      "step": 402500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4743202786949213e-05,
+      "loss": 0.8506,
+      "step": 403000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4699459862367262e-05,
+      "loss": 0.8531,
+      "step": 403500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4655716937785314e-05,
+      "loss": 0.8582,
+      "step": 404000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4611974013203364e-05,
+      "loss": 0.8573,
+      "step": 404500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4568231088621417e-05,
+      "loss": 0.8533,
+      "step": 405000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4524488164039469e-05,
+      "loss": 0.8589,
+      "step": 405500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4480745239457518e-05,
+      "loss": 0.8526,
+      "step": 406000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.443700231487557e-05,
+      "loss": 0.8547,
+      "step": 406500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.439325939029362e-05,
+      "loss": 0.8522,
+      "step": 407000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4349516465711671e-05,
+      "loss": 0.847,
+      "step": 407500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4305773541129721e-05,
+      "loss": 0.8542,
+      "step": 408000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4262030616547775e-05,
+      "loss": 0.8488,
+      "step": 408500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4218287691965826e-05,
+      "loss": 0.8548,
+      "step": 409000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4174544767383876e-05,
+      "loss": 0.8529,
+      "step": 409500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4130801842801928e-05,
+      "loss": 0.8419,
+      "step": 410000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4087058918219977e-05,
+      "loss": 0.8484,
+      "step": 410500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4043315993638029e-05,
+      "loss": 0.8531,
+      "step": 411000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3999573069056082e-05,
+      "loss": 0.847,
+      "step": 411500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3955830144474132e-05,
+      "loss": 0.8524,
+      "step": 412000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3912087219892184e-05,
+      "loss": 0.8525,
+      "step": 412500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3868344295310234e-05,
+      "loss": 0.8579,
+      "step": 413000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3824601370728285e-05,
+      "loss": 0.8526,
+      "step": 413500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3780858446146335e-05,
+      "loss": 0.8478,
+      "step": 414000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3737115521564387e-05,
+      "loss": 0.8586,
+      "step": 414500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.369337259698244e-05,
+      "loss": 0.8543,
+      "step": 415000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3649629672400491e-05,
+      "loss": 0.8511,
+      "step": 415500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3605886747818541e-05,
+      "loss": 0.846,
+      "step": 416000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3562143823236593e-05,
+      "loss": 0.849,
+      "step": 416500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3518400898654643e-05,
+      "loss": 0.848,
+      "step": 417000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3474657974072694e-05,
+      "loss": 0.8559,
+      "step": 417500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3430915049490744e-05,
+      "loss": 0.8517,
+      "step": 418000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3387172124908797e-05,
+      "loss": 0.838,
+      "step": 418500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3343429200326849e-05,
+      "loss": 0.8467,
+      "step": 419000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3299686275744899e-05,
+      "loss": 0.8423,
+      "step": 419500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.325594335116295e-05,
+      "loss": 0.8529,
+      "step": 420000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3212200426581e-05,
+      "loss": 0.8522,
+      "step": 420500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3168457501999052e-05,
+      "loss": 0.8405,
+      "step": 421000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3124714577417102e-05,
+      "loss": 0.8408,
+      "step": 421500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3080971652835155e-05,
+      "loss": 0.8469,
+      "step": 422000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3037228728253206e-05,
+      "loss": 0.8422,
+      "step": 422500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2993485803671256e-05,
+      "loss": 0.8474,
+      "step": 423000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2949742879089308e-05,
+      "loss": 0.8401,
+      "step": 423500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2905999954507358e-05,
+      "loss": 0.8438,
+      "step": 424000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.286225702992541e-05,
+      "loss": 0.8442,
+      "step": 424500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2818514105343463e-05,
+      "loss": 0.8529,
+      "step": 425000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2774771180761514e-05,
+      "loss": 0.8435,
+      "step": 425500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2731028256179564e-05,
+      "loss": 0.8509,
+      "step": 426000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2687285331597616e-05,
+      "loss": 0.846,
+      "step": 426500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2643542407015665e-05,
+      "loss": 0.8487,
+      "step": 427000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2599799482433717e-05,
+      "loss": 0.8331,
+      "step": 427500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2556056557851767e-05,
+      "loss": 0.8459,
+      "step": 428000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.251231363326982e-05,
+      "loss": 0.8433,
+      "step": 428500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.246857070868787e-05,
+      "loss": 0.8431,
+      "step": 429000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2424827784105921e-05,
+      "loss": 0.8452,
+      "step": 429500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2381084859523973e-05,
+      "loss": 0.8362,
+      "step": 430000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2337341934942023e-05,
+      "loss": 0.8514,
+      "step": 430500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2293599010360076e-05,
+      "loss": 0.8415,
+      "step": 431000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2249856085778126e-05,
+      "loss": 0.8491,
+      "step": 431500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2206113161196178e-05,
+      "loss": 0.8366,
+      "step": 432000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.216237023661423e-05,
+      "loss": 0.8469,
+      "step": 432500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2118627312032279e-05,
+      "loss": 0.8415,
+      "step": 433000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.207488438745033e-05,
+      "loss": 0.84,
+      "step": 433500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.203114146286838e-05,
+      "loss": 0.8367,
+      "step": 434000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1987398538286434e-05,
+      "loss": 0.8504,
+      "step": 434500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1943655613704484e-05,
+      "loss": 0.8424,
+      "step": 435000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1899912689122535e-05,
+      "loss": 0.8405,
+      "step": 435500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1856169764540587e-05,
+      "loss": 0.8361,
+      "step": 436000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1812426839958638e-05,
+      "loss": 0.8459,
+      "step": 436500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1768683915376688e-05,
+      "loss": 0.8344,
+      "step": 437000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.172494099079474e-05,
+      "loss": 0.8408,
+      "step": 437500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1681198066212791e-05,
+      "loss": 0.8475,
+      "step": 438000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1637455141630841e-05,
+      "loss": 0.8513,
+      "step": 438500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1593712217048893e-05,
+      "loss": 0.8446,
+      "step": 439000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1549969292466944e-05,
+      "loss": 0.8367,
+      "step": 439500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1506226367884996e-05,
+      "loss": 0.8479,
+      "step": 440000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1462483443303046e-05,
+      "loss": 0.8366,
+      "step": 440500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1418740518721097e-05,
+      "loss": 0.8387,
+      "step": 441000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1374997594139149e-05,
+      "loss": 0.8442,
+      "step": 441500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1331254669557199e-05,
+      "loss": 0.8406,
+      "step": 442000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.128751174497525e-05,
+      "loss": 0.8405,
+      "step": 442500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1243768820393302e-05,
+      "loss": 0.8383,
+      "step": 443000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1200025895811353e-05,
+      "loss": 0.849,
+      "step": 443500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1156282971229403e-05,
+      "loss": 0.8433,
+      "step": 444000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1112540046647456e-05,
+      "loss": 0.8465,
+      "step": 444500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1068797122065506e-05,
+      "loss": 0.8367,
+      "step": 445000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1025054197483558e-05,
+      "loss": 0.8367,
+      "step": 445500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.098131127290161e-05,
+      "loss": 0.8443,
+      "step": 446000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.093756834831966e-05,
+      "loss": 0.8406,
+      "step": 446500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.089382542373771e-05,
+      "loss": 0.8314,
+      "step": 447000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.085008249915576e-05,
+      "loss": 0.8387,
+      "step": 447500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0806339574573814e-05,
+      "loss": 0.841,
+      "step": 448000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0762596649991864e-05,
+      "loss": 0.8367,
+      "step": 448500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0718853725409915e-05,
+      "loss": 0.8469,
+      "step": 449000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0675110800827967e-05,
+      "loss": 0.8395,
+      "step": 449500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0631367876246019e-05,
+      "loss": 0.8365,
+      "step": 450000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0587624951664068e-05,
+      "loss": 0.8398,
+      "step": 450500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.054388202708212e-05,
+      "loss": 0.84,
+      "step": 451000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0500139102500171e-05,
+      "loss": 0.8371,
+      "step": 451500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0456396177918221e-05,
+      "loss": 0.836,
+      "step": 452000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0412653253336273e-05,
+      "loss": 0.8414,
+      "step": 452500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0368910328754324e-05,
+      "loss": 0.8404,
+      "step": 453000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0325167404172376e-05,
+      "loss": 0.8399,
+      "step": 453500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0281424479590426e-05,
+      "loss": 0.8297,
+      "step": 454000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.023768155500848e-05,
+      "loss": 0.8366,
+      "step": 454500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0193938630426529e-05,
+      "loss": 0.8358,
+      "step": 455000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.015019570584458e-05,
+      "loss": 0.8399,
+      "step": 455500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.010645278126263e-05,
+      "loss": 0.8428,
+      "step": 456000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0062709856680682e-05,
+      "loss": 0.8367,
+      "step": 456500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0018966932098734e-05,
+      "loss": 0.8364,
+      "step": 457000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.975224007516783e-06,
+      "loss": 0.8386,
+      "step": 457500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.931481082934837e-06,
+      "loss": 0.8383,
+      "step": 458000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.887738158352887e-06,
+      "loss": 0.8351,
+      "step": 458500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.843995233770938e-06,
+      "loss": 0.835,
+      "step": 459000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.80025230918899e-06,
+      "loss": 0.8374,
+      "step": 459500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.756509384607041e-06,
+      "loss": 0.8373,
+      "step": 460000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.712766460025091e-06,
+      "loss": 0.844,
+      "step": 460500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.669023535443143e-06,
+      "loss": 0.8387,
+      "step": 461000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.625280610861194e-06,
+      "loss": 0.8353,
+      "step": 461500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.581537686279244e-06,
+      "loss": 0.835,
+      "step": 462000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.537794761697296e-06,
+      "loss": 0.8334,
+      "step": 462500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.494051837115347e-06,
+      "loss": 0.8387,
+      "step": 463000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.450308912533399e-06,
+      "loss": 0.8443,
+      "step": 463500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.406565987951449e-06,
+      "loss": 0.8302,
+      "step": 464000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.3628230633695e-06,
+      "loss": 0.8364,
+      "step": 464500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.319080138787552e-06,
+      "loss": 0.8371,
+      "step": 465000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.275337214205603e-06,
+      "loss": 0.8382,
+      "step": 465500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.231594289623653e-06,
+      "loss": 0.8427,
+      "step": 466000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.187851365041705e-06,
+      "loss": 0.8335,
+      "step": 466500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.144108440459756e-06,
+      "loss": 0.8349,
+      "step": 467000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.100365515877806e-06,
+      "loss": 0.8272,
+      "step": 467500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.05662259129586e-06,
+      "loss": 0.8319,
+      "step": 468000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.01287966671391e-06,
+      "loss": 0.8344,
+      "step": 468500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.969136742131961e-06,
+      "loss": 0.8387,
+      "step": 469000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.925393817550012e-06,
+      "loss": 0.8378,
+      "step": 469500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.881650892968062e-06,
+      "loss": 0.8352,
+      "step": 470000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.837907968386114e-06,
+      "loss": 0.8298,
+      "step": 470500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.794165043804164e-06,
+      "loss": 0.8291,
+      "step": 471000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.750422119222217e-06,
+      "loss": 0.8323,
+      "step": 471500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.706679194640267e-06,
+      "loss": 0.8338,
+      "step": 472000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.662936270058318e-06,
+      "loss": 0.8374,
+      "step": 472500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.61919334547637e-06,
+      "loss": 0.8313,
+      "step": 473000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.575450420894422e-06,
+      "loss": 0.8288,
+      "step": 473500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.531707496312471e-06,
+      "loss": 0.8395,
+      "step": 474000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.487964571730523e-06,
+      "loss": 0.8343,
+      "step": 474500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.444221647148574e-06,
+      "loss": 0.8349,
+      "step": 475000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.400478722566624e-06,
+      "loss": 0.8293,
+      "step": 475500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.356735797984676e-06,
+      "loss": 0.836,
+      "step": 476000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.312992873402727e-06,
+      "loss": 0.8404,
+      "step": 476500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.269249948820779e-06,
+      "loss": 0.8408,
+      "step": 477000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.225507024238829e-06,
+      "loss": 0.8314,
+      "step": 477500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.181764099656882e-06,
+      "loss": 0.8255,
+      "step": 478000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.138021175074932e-06,
+      "loss": 0.831,
+      "step": 478500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.094278250492984e-06,
+      "loss": 0.8367,
+      "step": 479000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.050535325911033e-06,
+      "loss": 0.8299,
+      "step": 479500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.006792401329085e-06,
+      "loss": 0.8307,
+      "step": 480000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.963049476747137e-06,
+      "loss": 0.8369,
+      "step": 480500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.919306552165186e-06,
+      "loss": 0.8365,
+      "step": 481000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.87556362758324e-06,
+      "loss": 0.8342,
+      "step": 481500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.83182070300129e-06,
+      "loss": 0.8304,
+      "step": 482000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.788077778419341e-06,
+      "loss": 0.8213,
+      "step": 482500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.744334853837393e-06,
+      "loss": 0.8243,
+      "step": 483000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.700591929255444e-06,
+      "loss": 0.8282,
+      "step": 483500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.656849004673494e-06,
+      "loss": 0.8259,
+      "step": 484000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.613106080091545e-06,
+      "loss": 0.8319,
+      "step": 484500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.569363155509597e-06,
+      "loss": 0.8231,
+      "step": 485000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.525620230927648e-06,
+      "loss": 0.8326,
+      "step": 485500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.481877306345699e-06,
+      "loss": 0.8369,
+      "step": 486000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.43813438176375e-06,
+      "loss": 0.8308,
+      "step": 486500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.394391457181801e-06,
+      "loss": 0.8235,
+      "step": 487000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.350648532599852e-06,
+      "loss": 0.8321,
+      "step": 487500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.306905608017904e-06,
+      "loss": 0.8365,
+      "step": 488000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.263162683435955e-06,
+      "loss": 0.8414,
+      "step": 488500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.2194197588540055e-06,
+      "loss": 0.8346,
+      "step": 489000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.175676834272056e-06,
+      "loss": 0.827,
+      "step": 489500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.131933909690109e-06,
+      "loss": 0.8266,
+      "step": 490000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.088190985108159e-06,
+      "loss": 0.8295,
+      "step": 490500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.04444806052621e-06,
+      "loss": 0.8398,
+      "step": 491000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.0007051359442616e-06,
+      "loss": 0.8333,
+      "step": 491500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.956962211362312e-06,
+      "loss": 0.8306,
+      "step": 492000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.913219286780363e-06,
+      "loss": 0.8242,
+      "step": 492500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.869476362198415e-06,
+      "loss": 0.8344,
+      "step": 493000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.825733437616466e-06,
+      "loss": 0.828,
+      "step": 493500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.781990513034517e-06,
+      "loss": 0.8288,
+      "step": 494000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.7382475884525675e-06,
+      "loss": 0.8403,
+      "step": 494500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.69450466387062e-06,
+      "loss": 0.8371,
+      "step": 495000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.650761739288671e-06,
+      "loss": 0.8215,
+      "step": 495500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.607018814706721e-06,
+      "loss": 0.8324,
+      "step": 496000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.563275890124773e-06,
+      "loss": 0.8349,
+      "step": 496500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.519532965542824e-06,
+      "loss": 0.8337,
+      "step": 497000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.475790040960874e-06,
+      "loss": 0.8319,
+      "step": 497500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.432047116378927e-06,
+      "loss": 0.8257,
+      "step": 498000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.3883041917969775e-06,
+      "loss": 0.8314,
+      "step": 498500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.344561267215028e-06,
+      "loss": 0.831,
+      "step": 499000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.300818342633079e-06,
+      "loss": 0.8333,
+      "step": 499500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.257075418051131e-06,
+      "loss": 0.8313,
+      "step": 500000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.213332493469182e-06,
+      "loss": 0.833,
+      "step": 500500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.169589568887233e-06,
+      "loss": 0.8307,
+      "step": 501000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.1258466443052835e-06,
+      "loss": 0.827,
+      "step": 501500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.082103719723335e-06,
+      "loss": 0.8223,
+      "step": 502000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.038360795141386e-06,
+      "loss": 0.8262,
+      "step": 502500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.994617870559437e-06,
+      "loss": 0.8188,
+      "step": 503000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.950874945977489e-06,
+      "loss": 0.8281,
+      "step": 503500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.9071320213955396e-06,
+      "loss": 0.8239,
+      "step": 504000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.863389096813591e-06,
+      "loss": 0.8281,
+      "step": 504500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.819646172231642e-06,
+      "loss": 0.8267,
+      "step": 505000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.775903247649693e-06,
+      "loss": 0.8275,
+      "step": 505500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.732160323067744e-06,
+      "loss": 0.828,
+      "step": 506000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.688417398485795e-06,
+      "loss": 0.8293,
+      "step": 506500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.644674473903846e-06,
+      "loss": 0.8291,
+      "step": 507000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.600931549321897e-06,
+      "loss": 0.8267,
+      "step": 507500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.557188624739949e-06,
+      "loss": 0.8259,
+      "step": 508000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.513445700158e-06,
+      "loss": 0.8334,
+      "step": 508500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.469702775576051e-06,
+      "loss": 0.8209,
+      "step": 509000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.4259598509941025e-06,
+      "loss": 0.8227,
+      "step": 509500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.382216926412153e-06,
+      "loss": 0.825,
+      "step": 510000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.338474001830204e-06,
+      "loss": 0.8347,
+      "step": 510500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.294731077248255e-06,
+      "loss": 0.8264,
+      "step": 511000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.250988152666306e-06,
+      "loss": 0.8317,
+      "step": 511500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.207245228084358e-06,
+      "loss": 0.818,
+      "step": 512000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1635023035024085e-06,
+      "loss": 0.8199,
+      "step": 512500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.11975937892046e-06,
+      "loss": 0.8242,
+      "step": 513000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.076016454338511e-06,
+      "loss": 0.8269,
+      "step": 513500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.032273529756562e-06,
+      "loss": 0.8184,
+      "step": 514000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.988530605174614e-06,
+      "loss": 0.8223,
+      "step": 514500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9447876805926646e-06,
+      "loss": 0.8211,
+      "step": 515000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.901044756010715e-06,
+      "loss": 0.8279,
+      "step": 515500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.857301831428766e-06,
+      "loss": 0.8322,
+      "step": 516000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8135589068468176e-06,
+      "loss": 0.821,
+      "step": 516500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.769815982264869e-06,
+      "loss": 0.823,
+      "step": 517000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.72607305768292e-06,
+      "loss": 0.8268,
+      "step": 517500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.682330133100971e-06,
+      "loss": 0.8201,
+      "step": 518000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.638587208519022e-06,
+      "loss": 0.832,
+      "step": 518500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.594844283937074e-06,
+      "loss": 0.8217,
+      "step": 519000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.551101359355125e-06,
+      "loss": 0.828,
+      "step": 519500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.507358434773176e-06,
+      "loss": 0.8185,
+      "step": 520000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.463615510191227e-06,
+      "loss": 0.8196,
+      "step": 520500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.419872585609277e-06,
+      "loss": 0.8251,
+      "step": 521000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.376129661027329e-06,
+      "loss": 0.8153,
+      "step": 521500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3323867364453805e-06,
+      "loss": 0.819,
+      "step": 522000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.288643811863431e-06,
+      "loss": 0.8253,
+      "step": 522500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.244900887281483e-06,
+      "loss": 0.8197,
+      "step": 523000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2011579626995335e-06,
+      "loss": 0.8235,
+      "step": 523500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.157415038117585e-06,
+      "loss": 0.8189,
+      "step": 524000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.113672113535636e-06,
+      "loss": 0.8251,
+      "step": 524500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.0699291889536865e-06,
+      "loss": 0.822,
+      "step": 525000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.026186264371738e-06,
+      "loss": 0.8125,
+      "step": 525500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.982443339789789e-06,
+      "loss": 0.8301,
+      "step": 526000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.93870041520784e-06,
+      "loss": 0.8184,
+      "step": 526500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.894957490625892e-06,
+      "loss": 0.8131,
+      "step": 527000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8512145660439426e-06,
+      "loss": 0.8171,
+      "step": 527500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8074716414619937e-06,
+      "loss": 0.8226,
+      "step": 528000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7637287168800444e-06,
+      "loss": 0.8221,
+      "step": 528500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.719985792298096e-06,
+      "loss": 0.8181,
+      "step": 529000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6762428677161475e-06,
+      "loss": 0.8301,
+      "step": 529500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6324999431341982e-06,
+      "loss": 0.8245,
+      "step": 530000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.5887570185522494e-06,
+      "loss": 0.8236,
+      "step": 530500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5450140939703e-06,
+      "loss": 0.831,
+      "step": 531000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5012711693883517e-06,
+      "loss": 0.819,
+      "step": 531500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4575282448064024e-06,
+      "loss": 0.8209,
+      "step": 532000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.413785320224454e-06,
+      "loss": 0.8279,
+      "step": 532500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.370042395642505e-06,
+      "loss": 0.8316,
+      "step": 533000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3262994710605558e-06,
+      "loss": 0.8281,
+      "step": 533500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2825565464786073e-06,
+      "loss": 0.8257,
+      "step": 534000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.238813621896658e-06,
+      "loss": 0.8316,
+      "step": 534500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1950706973147096e-06,
+      "loss": 0.8208,
+      "step": 535000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1513277727327607e-06,
+      "loss": 0.8255,
+      "step": 535500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.107584848150812e-06,
+      "loss": 0.8214,
+      "step": 536000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.063841923568863e-06,
+      "loss": 0.821,
+      "step": 536500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.020098998986914e-06,
+      "loss": 0.8209,
+      "step": 537000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9763560744049653e-06,
+      "loss": 0.8257,
+      "step": 537500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.932613149823016e-06,
+      "loss": 0.823,
+      "step": 538000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.888870225241067e-06,
+      "loss": 0.8192,
+      "step": 538500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8451273006591187e-06,
+      "loss": 0.8146,
+      "step": 539000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.80138437607717e-06,
+      "loss": 0.8219,
+      "step": 539500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.757641451495221e-06,
+      "loss": 0.8222,
+      "step": 540000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7138985269132717e-06,
+      "loss": 0.8227,
+      "step": 540500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.670155602331323e-06,
+      "loss": 0.8164,
+      "step": 541000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6264126777493744e-06,
+      "loss": 0.8226,
+      "step": 541500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5826697531674255e-06,
+      "loss": 0.818,
+      "step": 542000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5389268285854762e-06,
+      "loss": 0.8232,
+      "step": 542500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4951839040035274e-06,
+      "loss": 0.8193,
+      "step": 543000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4514409794215785e-06,
+      "loss": 0.8252,
+      "step": 543500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.40769805483963e-06,
+      "loss": 0.8244,
+      "step": 544000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.363955130257681e-06,
+      "loss": 0.8133,
+      "step": 544500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.320212205675732e-06,
+      "loss": 0.8172,
+      "step": 545000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.276469281093783e-06,
+      "loss": 0.8212,
+      "step": 545500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.232726356511834e-06,
+      "loss": 0.8183,
+      "step": 546000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1889834319298857e-06,
+      "loss": 0.8262,
+      "step": 546500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.145240507347937e-06,
+      "loss": 0.8225,
+      "step": 547000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1014975827659876e-06,
+      "loss": 0.8112,
+      "step": 547500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0577546581840387e-06,
+      "loss": 0.8277,
+      "step": 548000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.01401173360209e-06,
+      "loss": 0.818,
+      "step": 548500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.970268809020141e-06,
+      "loss": 0.8196,
+      "step": 549000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.926525884438192e-06,
+      "loss": 0.8143,
+      "step": 549500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8827829598562435e-06,
+      "loss": 0.8179,
+      "step": 550000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8390400352742944e-06,
+      "loss": 0.8272,
+      "step": 550500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7952971106923456e-06,
+      "loss": 0.8193,
+      "step": 551000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7515541861103965e-06,
+      "loss": 0.8144,
+      "step": 551500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.707811261528448e-06,
+      "loss": 0.8234,
+      "step": 552000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.664068336946499e-06,
+      "loss": 0.8151,
+      "step": 552500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.62032541236455e-06,
+      "loss": 0.8185,
+      "step": 553000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5765824877826012e-06,
+      "loss": 0.8196,
+      "step": 553500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5328395632006524e-06,
+      "loss": 0.8214,
+      "step": 554000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4890966386187035e-06,
+      "loss": 0.8222,
+      "step": 554500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4453537140367546e-06,
+      "loss": 0.8186,
+      "step": 555000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4016107894548058e-06,
+      "loss": 0.8178,
+      "step": 555500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.357867864872857e-06,
+      "loss": 0.8196,
+      "step": 556000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.314124940290908e-06,
+      "loss": 0.8239,
+      "step": 556500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2703820157089592e-06,
+      "loss": 0.8283,
+      "step": 557000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2266390911270101e-06,
+      "loss": 0.8239,
+      "step": 557500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1828961665450615e-06,
+      "loss": 0.8149,
+      "step": 558000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1391532419631124e-06,
+      "loss": 0.8272,
+      "step": 558500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0954103173811637e-06,
+      "loss": 0.8121,
+      "step": 559000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0516673927992149e-06,
+      "loss": 0.8151,
+      "step": 559500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0079244682172658e-06,
+      "loss": 0.8234,
+      "step": 560000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.641815436353171e-07,
+      "loss": 0.8274,
+      "step": 560500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.204386190533682e-07,
+      "loss": 0.8173,
+      "step": 561000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.766956944714192e-07,
+      "loss": 0.8186,
+      "step": 561500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.329527698894705e-07,
+      "loss": 0.8211,
+      "step": 562000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.892098453075215e-07,
+      "loss": 0.8172,
+      "step": 562500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.454669207255726e-07,
+      "loss": 0.8203,
+      "step": 563000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.017239961436238e-07,
+      "loss": 0.8189,
+      "step": 563500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.57981071561675e-07,
+      "loss": 0.8158,
+      "step": 564000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.142381469797261e-07,
+      "loss": 0.8143,
+      "step": 564500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.704952223977772e-07,
+      "loss": 0.8102,
+      "step": 565000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.267522978158283e-07,
+      "loss": 0.8174,
+      "step": 565500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.830093732338794e-07,
+      "loss": 0.8156,
+      "step": 566000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3926644865193063e-07,
+      "loss": 0.8173,
+      "step": 566500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.9552352406998166e-07,
+      "loss": 0.8163,
+      "step": 567000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.517805994880328e-07,
+      "loss": 0.8122,
+      "step": 567500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.08037674906084e-07,
+      "loss": 0.8123,
+      "step": 568000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6429475032413507e-07,
+      "loss": 0.8161,
+      "step": 568500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2055182574218623e-07,
+      "loss": 0.8174,
+      "step": 569000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7680890116023734e-07,
+      "loss": 0.8173,
+      "step": 569500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3306597657828848e-07,
+      "loss": 0.8153,
+      "step": 570000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.93230519963396e-08,
+      "loss": 0.8166,
+      "step": 570500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.5580127414390725e-08,
+      "loss": 0.8105,
+      "step": 571000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8372028324418525e-09,
+      "loss": 0.8254,
+      "step": 571500
+    },
+    {
+      "epoch": 3.0,
+      "step": 571521,
+      "total_flos": 2.4130383698872074e+18,
+      "train_loss": 0.9593743994665668,
+      "train_runtime": 188052.0768,
+      "train_samples_per_second": 48.626,
+      "train_steps_per_second": 3.039
+    }
+  ],
+  "max_steps": 571521,
+  "num_train_epochs": 3,
+  "total_flos": 2.4130383698872074e+18,
+  "trial_name": null,
+  "trial_params": null
+}