diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,16492 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 27438,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.8716,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9994897215337516e-05,
+      "loss": 0.7215,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9987607522962533e-05,
+      "loss": 0.7342,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.998031783058755e-05,
+      "loss": 0.7118,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9973028138212568e-05,
+      "loss": 0.7299,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.996573844583759e-05,
+      "loss": 0.8007,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9958448753462603e-05,
+      "loss": 0.7032,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9951159061087624e-05,
+      "loss": 0.7953,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.994386936871264e-05,
+      "loss": 0.7844,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.993657967633766e-05,
+      "loss": 0.7249,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.992928998396268e-05,
+      "loss": 0.7273,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9922000291587697e-05,
+      "loss": 0.7399,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9914710599212715e-05,
+      "loss": 0.7727,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9907420906837732e-05,
+      "loss": 0.7471,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9900131214462753e-05,
+      "loss": 0.723,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.989284152208777e-05,
+      "loss": 0.6925,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9885551829712788e-05,
+      "loss": 0.7393,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.987826213733781e-05,
+      "loss": 0.7139,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9870972444962823e-05,
+      "loss": 0.7345,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9863682752587844e-05,
+      "loss": 0.7953,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.985639306021286e-05,
+      "loss": 0.7016,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.984910336783788e-05,
+      "loss": 0.7333,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9841813675462896e-05,
+      "loss": 0.7344,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9834523983087913e-05,
+      "loss": 0.7694,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9827234290712934e-05,
+      "loss": 0.7246,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9819944598337952e-05,
+      "loss": 0.7286,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.981265490596297e-05,
+      "loss": 0.7326,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9805365213587987e-05,
+      "loss": 0.7033,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9798075521213007e-05,
+      "loss": 0.6866,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9790785828838025e-05,
+      "loss": 0.7855,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9783496136463042e-05,
+      "loss": 0.6743,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9776206444088063e-05,
+      "loss": 0.7403,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9768916751713077e-05,
+      "loss": 0.7809,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9761627059338098e-05,
+      "loss": 0.7414,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9754337366963116e-05,
+      "loss": 0.7298,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9747047674588133e-05,
+      "loss": 0.8152,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9739757982213154e-05,
+      "loss": 0.7362,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.973246828983817e-05,
+      "loss": 0.7096,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.972517859746319e-05,
+      "loss": 0.745,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9717888905088206e-05,
+      "loss": 0.694,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9710599212713227e-05,
+      "loss": 0.7422,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9703309520338245e-05,
+      "loss": 0.7374,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9696019827963262e-05,
+      "loss": 0.7477,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.968873013558828e-05,
+      "loss": 0.7343,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9681440443213297e-05,
+      "loss": 0.6886,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9674150750838318e-05,
+      "loss": 0.7289,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9666861058463335e-05,
+      "loss": 0.8051,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9659571366088353e-05,
+      "loss": 0.6701,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.965228167371337e-05,
+      "loss": 0.7508,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9644991981338388e-05,
+      "loss": 0.7257,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.963770228896341e-05,
+      "loss": 0.7826,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9630412596588426e-05,
+      "loss": 0.7464,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9623122904213443e-05,
+      "loss": 0.6794,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.961583321183846e-05,
+      "loss": 0.6739,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.960854351946348e-05,
+      "loss": 0.7415,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.96012538270885e-05,
+      "loss": 0.7279,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9593964134713516e-05,
+      "loss": 0.7109,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9586674442338537e-05,
+      "loss": 0.771,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.957938474996355e-05,
+      "loss": 0.7249,
+      "step": 580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9572095057588572e-05,
+      "loss": 0.7507,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.956480536521359e-05,
+      "loss": 0.7018,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9557515672838607e-05,
+      "loss": 0.7137,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9550225980463628e-05,
+      "loss": 0.7634,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9542936288088645e-05,
+      "loss": 0.7361,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9535646595713663e-05,
+      "loss": 0.7304,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.952835690333868e-05,
+      "loss": 0.7729,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9521067210963698e-05,
+      "loss": 0.6814,
+      "step": 660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9513777518588715e-05,
+      "loss": 0.6833,
+      "step": 670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9506487826213736e-05,
+      "loss": 0.7199,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9499198133838754e-05,
+      "loss": 0.6873,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.949190844146377e-05,
+      "loss": 0.7428,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9484618749088792e-05,
+      "loss": 0.7146,
+      "step": 710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9477329056713806e-05,
+      "loss": 0.7472,
+      "step": 720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9470039364338827e-05,
+      "loss": 0.7382,
+      "step": 730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9462749671963844e-05,
+      "loss": 0.7704,
+      "step": 740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.945545997958886e-05,
+      "loss": 0.6516,
+      "step": 750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9448170287213882e-05,
+      "loss": 0.7077,
+      "step": 760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.94408805948389e-05,
+      "loss": 0.7184,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9433590902463917e-05,
+      "loss": 0.7306,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9426301210088935e-05,
+      "loss": 0.7219,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9419011517713956e-05,
+      "loss": 0.7617,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9411721825338973e-05,
+      "loss": 0.6958,
+      "step": 810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.940443213296399e-05,
+      "loss": 0.6949,
+      "step": 820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9397142440589008e-05,
+      "loss": 0.7268,
+      "step": 830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9389852748214025e-05,
+      "loss": 0.7314,
+      "step": 840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9382563055839046e-05,
+      "loss": 0.7353,
+      "step": 850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9375273363464064e-05,
+      "loss": 0.7027,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.936798367108908e-05,
+      "loss": 0.7673,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.93606939787141e-05,
+      "loss": 0.6789,
+      "step": 880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.935340428633912e-05,
+      "loss": 0.7044,
+      "step": 890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9346114593964137e-05,
+      "loss": 0.7702,
+      "step": 900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9338824901589154e-05,
+      "loss": 0.6909,
+      "step": 910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9331535209214172e-05,
+      "loss": 0.7016,
+      "step": 920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.932424551683919e-05,
+      "loss": 0.7188,
+      "step": 930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.931695582446421e-05,
+      "loss": 0.7167,
+      "step": 940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9309666132089228e-05,
+      "loss": 0.7507,
+      "step": 950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9302376439714245e-05,
+      "loss": 0.7241,
+      "step": 960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9295086747339266e-05,
+      "loss": 0.7299,
+      "step": 970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.928779705496428e-05,
+      "loss": 0.68,
+      "step": 980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.92805073625893e-05,
+      "loss": 0.7278,
+      "step": 990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9273217670214318e-05,
+      "loss": 0.7254,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9265927977839336e-05,
+      "loss": 0.7313,
+      "step": 1010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9258638285464357e-05,
+      "loss": 0.7285,
+      "step": 1020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9251348593089374e-05,
+      "loss": 0.7499,
+      "step": 1030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.924405890071439e-05,
+      "loss": 0.7053,
+      "step": 1040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.923676920833941e-05,
+      "loss": 0.6957,
+      "step": 1050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.922947951596443e-05,
+      "loss": 0.6935,
+      "step": 1060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9222189823589444e-05,
+      "loss": 0.6637,
+      "step": 1070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9214900131214465e-05,
+      "loss": 0.6745,
+      "step": 1080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9207610438839482e-05,
+      "loss": 0.6709,
+      "step": 1090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.92003207464645e-05,
+      "loss": 0.7171,
+      "step": 1100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.919303105408952e-05,
+      "loss": 0.7197,
+      "step": 1110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9185741361714538e-05,
+      "loss": 0.7135,
+      "step": 1120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9178451669339555e-05,
+      "loss": 0.6869,
+      "step": 1130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9171161976964573e-05,
+      "loss": 0.7442,
+      "step": 1140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.916387228458959e-05,
+      "loss": 0.7408,
+      "step": 1150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.915658259221461e-05,
+      "loss": 0.7034,
+      "step": 1160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.914929289983963e-05,
+      "loss": 0.7266,
+      "step": 1170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9142003207464646e-05,
+      "loss": 0.7224,
+      "step": 1180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9134713515089663e-05,
+      "loss": 0.6984,
+      "step": 1190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9127423822714684e-05,
+      "loss": 0.7134,
+      "step": 1200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9120134130339702e-05,
+      "loss": 0.7816,
+      "step": 1210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.911284443796472e-05,
+      "loss": 0.707,
+      "step": 1220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9105554745589737e-05,
+      "loss": 0.7517,
+      "step": 1230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9098265053214754e-05,
+      "loss": 0.7416,
+      "step": 1240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9090975360839775e-05,
+      "loss": 0.6792,
+      "step": 1250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9083685668464792e-05,
+      "loss": 0.7085,
+      "step": 1260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907639597608981e-05,
+      "loss": 0.7047,
+      "step": 1270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9069106283714827e-05,
+      "loss": 0.705,
+      "step": 1280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9061816591339848e-05,
+      "loss": 0.6827,
+      "step": 1290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9054526898964866e-05,
+      "loss": 0.6806,
+      "step": 1300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9047237206589883e-05,
+      "loss": 0.7041,
+      "step": 1310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9039947514214904e-05,
+      "loss": 0.7848,
+      "step": 1320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9032657821839918e-05,
+      "loss": 0.7032,
+      "step": 1330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.902536812946494e-05,
+      "loss": 0.7204,
+      "step": 1340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9018078437089956e-05,
+      "loss": 0.7604,
+      "step": 1350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9010788744714974e-05,
+      "loss": 0.7636,
+      "step": 1360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9003499052339995e-05,
+      "loss": 0.7457,
+      "step": 1370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899620935996501e-05,
+      "loss": 0.6483,
+      "step": 1380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.898891966759003e-05,
+      "loss": 0.688,
+      "step": 1390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8981629975215047e-05,
+      "loss": 0.7843,
+      "step": 1400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8974340282840064e-05,
+      "loss": 0.7497,
+      "step": 1410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8967050590465085e-05,
+      "loss": 0.6868,
+      "step": 1420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8959760898090103e-05,
+      "loss": 0.7694,
+      "step": 1430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.895247120571512e-05,
+      "loss": 0.7032,
+      "step": 1440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8945181513340138e-05,
+      "loss": 0.7426,
+      "step": 1450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.893789182096516e-05,
+      "loss": 0.7571,
+      "step": 1460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8930602128590176e-05,
+      "loss": 0.7011,
+      "step": 1470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8923312436215193e-05,
+      "loss": 0.7107,
+      "step": 1480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.891602274384021e-05,
+      "loss": 0.6953,
+      "step": 1490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8908733051465228e-05,
+      "loss": 0.6975,
+      "step": 1500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.890144335909025e-05,
+      "loss": 0.7042,
+      "step": 1510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8894153666715267e-05,
+      "loss": 0.7124,
+      "step": 1520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8886863974340284e-05,
+      "loss": 0.6806,
+      "step": 1530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.88795742819653e-05,
+      "loss": 0.7368,
+      "step": 1540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8872284589590322e-05,
+      "loss": 0.7062,
+      "step": 1550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.886499489721534e-05,
+      "loss": 0.6935,
+      "step": 1560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8857705204840357e-05,
+      "loss": 0.7529,
+      "step": 1570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8850415512465378e-05,
+      "loss": 0.71,
+      "step": 1580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8843125820090392e-05,
+      "loss": 0.7312,
+      "step": 1590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8835836127715413e-05,
+      "loss": 0.6777,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882854643534043e-05,
+      "loss": 0.7278,
+      "step": 1610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821256742965448e-05,
+      "loss": 0.6638,
+      "step": 1620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881396705059047e-05,
+      "loss": 0.7011,
+      "step": 1630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806677358215483e-05,
+      "loss": 0.7089,
+      "step": 1640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799387665840504e-05,
+      "loss": 0.7611,
+      "step": 1650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879209797346552e-05,
+      "loss": 0.6547,
+      "step": 1660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878480828109054e-05,
+      "loss": 0.6706,
+      "step": 1670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8777518588715556e-05,
+      "loss": 0.675,
+      "step": 1680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770228896340577e-05,
+      "loss": 0.6718,
+      "step": 1690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8762939203965594e-05,
+      "loss": 0.694,
+      "step": 1700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755649511590612e-05,
+      "loss": 0.7389,
+      "step": 1710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748359819215633e-05,
+      "loss": 0.6977,
+      "step": 1720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741070126840647e-05,
+      "loss": 0.7857,
+      "step": 1730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8733780434465667e-05,
+      "loss": 0.6589,
+      "step": 1740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8726490742090685e-05,
+      "loss": 0.717,
+      "step": 1750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719201049715702e-05,
+      "loss": 0.6797,
+      "step": 1760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711911357340723e-05,
+      "loss": 0.7416,
+      "step": 1770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870462166496574e-05,
+      "loss": 0.671,
+      "step": 1780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8697331972590758e-05,
+      "loss": 0.7334,
+      "step": 1790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8690042280215776e-05,
+      "loss": 0.6528,
+      "step": 1800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8682752587840796e-05,
+      "loss": 0.6659,
+      "step": 1810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8675462895465814e-05,
+      "loss": 0.7182,
+      "step": 1820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.866817320309083e-05,
+      "loss": 0.6772,
+      "step": 1830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.866088351071585e-05,
+      "loss": 0.6712,
+      "step": 1840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8653593818340866e-05,
+      "loss": 0.7245,
+      "step": 1850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8646304125965887e-05,
+      "loss": 0.7057,
+      "step": 1860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8639014433590904e-05,
+      "loss": 0.7301,
+      "step": 1870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8631724741215922e-05,
+      "loss": 0.7288,
+      "step": 1880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.862443504884094e-05,
+      "loss": 0.7052,
+      "step": 1890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8617145356465957e-05,
+      "loss": 0.7297,
+      "step": 1900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8609855664090978e-05,
+      "loss": 0.6716,
+      "step": 1910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8602565971715995e-05,
+      "loss": 0.7018,
+      "step": 1920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8595276279341013e-05,
+      "loss": 0.7234,
+      "step": 1930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.858798658696603e-05,
+      "loss": 0.6609,
+      "step": 1940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.858069689459105e-05,
+      "loss": 0.6921,
+      "step": 1950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.857340720221607e-05,
+      "loss": 0.7272,
+      "step": 1960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8566117509841086e-05,
+      "loss": 0.6964,
+      "step": 1970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8558827817466107e-05,
+      "loss": 0.6845,
+      "step": 1980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.855153812509112e-05,
+      "loss": 0.7416,
+      "step": 1990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.854424843271614e-05,
+      "loss": 0.6447,
+      "step": 2000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.853695874034116e-05,
+      "loss": 0.6836,
+      "step": 2010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8529669047966176e-05,
+      "loss": 0.6697,
+      "step": 2020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8522379355591197e-05,
+      "loss": 0.7192,
+      "step": 2030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8515089663216215e-05,
+      "loss": 0.6567,
+      "step": 2040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8507799970841232e-05,
+      "loss": 0.7209,
+      "step": 2050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.850051027846625e-05,
+      "loss": 0.7322,
+      "step": 2060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.849322058609127e-05,
+      "loss": 0.7108,
+      "step": 2070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8485930893716285e-05,
+      "loss": 0.6771,
+      "step": 2080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8478641201341305e-05,
+      "loss": 0.7319,
+      "step": 2090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8471351508966323e-05,
+      "loss": 0.6555,
+      "step": 2100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.846406181659134e-05,
+      "loss": 0.7454,
+      "step": 2110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.845677212421636e-05,
+      "loss": 0.6448,
+      "step": 2120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8449482431841375e-05,
+      "loss": 0.7145,
+      "step": 2130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8442192739466396e-05,
+      "loss": 0.7235,
+      "step": 2140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8434903047091414e-05,
+      "loss": 0.7994,
+      "step": 2150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.842761335471643e-05,
+      "loss": 0.6907,
+      "step": 2160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8420323662341452e-05,
+      "loss": 0.719,
+      "step": 2170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.841303396996647e-05,
+      "loss": 0.6913,
+      "step": 2180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8405744277591487e-05,
+      "loss": 0.7236,
+      "step": 2190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8398454585216504e-05,
+      "loss": 0.6936,
+      "step": 2200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8391164892841525e-05,
+      "loss": 0.6668,
+      "step": 2210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8383875200466542e-05,
+      "loss": 0.6916,
+      "step": 2220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.837658550809156e-05,
+      "loss": 0.7131,
+      "step": 2230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8369295815716577e-05,
+      "loss": 0.7072,
+      "step": 2240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8362006123341595e-05,
+      "loss": 0.7254,
+      "step": 2250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8354716430966616e-05,
+      "loss": 0.6884,
+      "step": 2260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8347426738591633e-05,
+      "loss": 0.6955,
+      "step": 2270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.834013704621665e-05,
+      "loss": 0.6711,
+      "step": 2280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8332847353841668e-05,
+      "loss": 0.6748,
+      "step": 2290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.832555766146669e-05,
+      "loss": 0.7006,
+      "step": 2300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8318267969091706e-05,
+      "loss": 0.6596,
+      "step": 2310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8310978276716724e-05,
+      "loss": 0.7564,
+      "step": 2320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8303688584341745e-05,
+      "loss": 0.6619,
+      "step": 2330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.829639889196676e-05,
+      "loss": 0.7095,
+      "step": 2340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.828910919959178e-05,
+      "loss": 0.722,
+      "step": 2350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8281819507216797e-05,
+      "loss": 0.701,
+      "step": 2360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8274529814841814e-05,
+      "loss": 0.6885,
+      "step": 2370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8267240122466835e-05,
+      "loss": 0.655,
+      "step": 2380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.825995043009185e-05,
+      "loss": 0.686,
+      "step": 2390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.825266073771687e-05,
+      "loss": 0.6759,
+      "step": 2400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8245371045341888e-05,
+      "loss": 0.7027,
+      "step": 2410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8238081352966905e-05,
+      "loss": 0.6804,
+      "step": 2420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8230791660591926e-05,
+      "loss": 0.6638,
+      "step": 2430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8223501968216943e-05,
+      "loss": 0.647,
+      "step": 2440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.821621227584196e-05,
+      "loss": 0.6629,
+      "step": 2450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8208922583466978e-05,
+      "loss": 0.7035,
+      "step": 2460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8201632891092e-05,
+      "loss": 0.6519,
+      "step": 2470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8194343198717017e-05,
+      "loss": 0.7487,
+      "step": 2480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8187053506342034e-05,
+      "loss": 0.6836,
+      "step": 2490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.817976381396705e-05,
+      "loss": 0.7127,
+      "step": 2500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.817247412159207e-05,
+      "loss": 0.7423,
+      "step": 2510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.816518442921709e-05,
+      "loss": 0.6603,
+      "step": 2520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8157894736842107e-05,
+      "loss": 0.6075,
+      "step": 2530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8150605044467125e-05,
+      "loss": 0.6539,
+      "step": 2540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8143315352092142e-05,
+      "loss": 0.6858,
+      "step": 2550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8136025659717163e-05,
+      "loss": 0.6496,
+      "step": 2560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.812873596734218e-05,
+      "loss": 0.7,
+      "step": 2570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8121446274967198e-05,
+      "loss": 0.6128,
+      "step": 2580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8114156582592215e-05,
+      "loss": 0.7398,
+      "step": 2590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8106866890217233e-05,
+      "loss": 0.6215,
+      "step": 2600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8099577197842254e-05,
+      "loss": 0.7296,
+      "step": 2610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.809228750546727e-05,
+      "loss": 0.6951,
+      "step": 2620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.808499781309229e-05,
+      "loss": 0.7587,
+      "step": 2630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.807770812071731e-05,
+      "loss": 0.7172,
+      "step": 2640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8070418428342323e-05,
+      "loss": 0.6965,
+      "step": 2650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8063128735967344e-05,
+      "loss": 0.6934,
+      "step": 2660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8055839043592362e-05,
+      "loss": 0.6648,
+      "step": 2670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.804854935121738e-05,
+      "loss": 0.6774,
+      "step": 2680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8041259658842397e-05,
+      "loss": 0.7116,
+      "step": 2690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8033969966467417e-05,
+      "loss": 0.7074,
+      "step": 2700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8026680274092435e-05,
+      "loss": 0.6915,
+      "step": 2710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8019390581717452e-05,
+      "loss": 0.7178,
+      "step": 2720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8012100889342473e-05,
+      "loss": 0.687,
+      "step": 2730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8004811196967487e-05,
+      "loss": 0.6583,
+      "step": 2740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7997521504592508e-05,
+      "loss": 0.6635,
+      "step": 2750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7990231812217526e-05,
+      "loss": 0.6745,
+      "step": 2760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7982942119842543e-05,
+      "loss": 0.6934,
+      "step": 2770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7975652427467564e-05,
+      "loss": 0.6414,
+      "step": 2780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.796836273509258e-05,
+      "loss": 0.7035,
+      "step": 2790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.79610730427176e-05,
+      "loss": 0.7028,
+      "step": 2800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7953783350342616e-05,
+      "loss": 0.6968,
+      "step": 2810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7946493657967637e-05,
+      "loss": 0.6716,
+      "step": 2820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7939203965592655e-05,
+      "loss": 0.685,
+      "step": 2830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7931914273217672e-05,
+      "loss": 0.6171,
+      "step": 2840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.792462458084269e-05,
+      "loss": 0.7359,
+      "step": 2850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7917334888467707e-05,
+      "loss": 0.6663,
+      "step": 2860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7910045196092728e-05,
+      "loss": 0.7192,
+      "step": 2870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7902755503717745e-05,
+      "loss": 0.7587,
+      "step": 2880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7895465811342763e-05,
+      "loss": 0.6273,
+      "step": 2890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.788817611896778e-05,
+      "loss": 0.6548,
+      "step": 2900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7880886426592798e-05,
+      "loss": 0.6791,
+      "step": 2910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.787359673421782e-05,
+      "loss": 0.7084,
+      "step": 2920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7866307041842836e-05,
+      "loss": 0.7389,
+      "step": 2930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7859017349467853e-05,
+      "loss": 0.6766,
+      "step": 2940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.785172765709287e-05,
+      "loss": 0.6952,
+      "step": 2950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.784443796471789e-05,
+      "loss": 0.6294,
+      "step": 2960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.783714827234291e-05,
+      "loss": 0.6808,
+      "step": 2970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7829858579967927e-05,
+      "loss": 0.7313,
+      "step": 2980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7822568887592947e-05,
+      "loss": 0.7059,
+      "step": 2990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.781527919521796e-05,
+      "loss": 0.6692,
+      "step": 3000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7807989502842982e-05,
+      "loss": 0.6765,
+      "step": 3010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7800699810468e-05,
+      "loss": 0.636,
+      "step": 3020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7793410118093017e-05,
+      "loss": 0.6953,
+      "step": 3030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7786120425718038e-05,
+      "loss": 0.6875,
+      "step": 3040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7778830733343055e-05,
+      "loss": 0.7463,
+      "step": 3050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7771541040968073e-05,
+      "loss": 0.6927,
+      "step": 3060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.776425134859309e-05,
+      "loss": 0.748,
+      "step": 3070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7756961656218108e-05,
+      "loss": 0.6515,
+      "step": 3080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7749671963843125e-05,
+      "loss": 0.7043,
+      "step": 3090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7742382271468146e-05,
+      "loss": 0.7037,
+      "step": 3100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7735092579093164e-05,
+      "loss": 0.7294,
+      "step": 3110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.772780288671818e-05,
+      "loss": 0.7295,
+      "step": 3120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7720513194343202e-05,
+      "loss": 0.6691,
+      "step": 3130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7713223501968216e-05,
+      "loss": 0.6742,
+      "step": 3140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7705933809593237e-05,
+      "loss": 0.6547,
+      "step": 3150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7698644117218254e-05,
+      "loss": 0.6443,
+      "step": 3160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.769135442484327e-05,
+      "loss": 0.6845,
+      "step": 3170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7684064732468293e-05,
+      "loss": 0.6852,
+      "step": 3180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.767677504009331e-05,
+      "loss": 0.6504,
+      "step": 3190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7669485347718327e-05,
+      "loss": 0.6367,
+      "step": 3200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7662195655343345e-05,
+      "loss": 0.6499,
+      "step": 3210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7654905962968366e-05,
+      "loss": 0.6954,
+      "step": 3220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7647616270593383e-05,
+      "loss": 0.6141,
+      "step": 3230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.76403265782184e-05,
+      "loss": 0.675,
+      "step": 3240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.7633036885843418e-05,
+      "loss": 0.657,
+      "step": 3250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.7625747193468436e-05,
+      "loss": 0.6555,
+      "step": 3260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.7618457501093456e-05,
+      "loss": 0.731,
+      "step": 3270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.7611167808718474e-05,
+      "loss": 0.6978,
+      "step": 3280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.760387811634349e-05,
+      "loss": 0.6897,
+      "step": 3290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.759658842396851e-05,
+      "loss": 0.7093,
+      "step": 3300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.758929873159353e-05,
+      "loss": 0.633,
+      "step": 3310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.7582009039218547e-05,
+      "loss": 0.6484,
+      "step": 3320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.7574719346843564e-05,
+      "loss": 0.6693,
+      "step": 3330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7567429654468582e-05,
+      "loss": 0.6715,
+      "step": 3340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.75601399620936e-05,
+      "loss": 0.6704,
+      "step": 3350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.755285026971862e-05,
+      "loss": 0.6946,
+      "step": 3360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7545560577343638e-05,
+      "loss": 0.6845,
+      "step": 3370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7538270884968655e-05,
+      "loss": 0.6447,
+      "step": 3380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.7530981192593676e-05,
+      "loss": 0.6419,
+      "step": 3390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.752369150021869e-05,
+      "loss": 0.646,
+      "step": 3400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.751640180784371e-05,
+      "loss": 0.6246,
+      "step": 3410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.750911211546873e-05,
+      "loss": 0.6903,
+      "step": 3420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7501822423093746e-05,
+      "loss": 0.665,
+      "step": 3430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7494532730718767e-05,
+      "loss": 0.6625,
+      "step": 3440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7487243038343784e-05,
+      "loss": 0.7383,
+      "step": 3450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.74799533459688e-05,
+      "loss": 0.6895,
+      "step": 3460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.747266365359382e-05,
+      "loss": 0.646,
+      "step": 3470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.746537396121884e-05,
+      "loss": 0.64,
+      "step": 3480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7458084268843857e-05,
+      "loss": 0.6192,
+      "step": 3490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7450794576468875e-05,
+      "loss": 0.6433,
+      "step": 3500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7443504884093892e-05,
+      "loss": 0.654,
+      "step": 3510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.743621519171891e-05,
+      "loss": 0.731,
+      "step": 3520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.742892549934393e-05,
+      "loss": 0.6701,
+      "step": 3530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7421635806968948e-05,
+      "loss": 0.6049,
+      "step": 3540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7414346114593965e-05,
+      "loss": 0.6281,
+      "step": 3550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7407056422218983e-05,
+      "loss": 0.608,
+      "step": 3560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7399766729844e-05,
+      "loss": 0.6601,
+      "step": 3570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.739247703746902e-05,
+      "loss": 0.5658,
+      "step": 3580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.738518734509404e-05,
+      "loss": 0.6395,
+      "step": 3590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7377897652719056e-05,
+      "loss": 0.706,
+      "step": 3600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7370607960344074e-05,
+      "loss": 0.6762,
+      "step": 3610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7363318267969094e-05,
+      "loss": 0.7146,
+      "step": 3620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7356028575594112e-05,
+      "loss": 0.7056,
+      "step": 3630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.734873888321913e-05,
+      "loss": 0.6818,
+      "step": 3640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.734144919084415e-05,
+      "loss": 0.6431,
+      "step": 3650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7334159498469164e-05,
+      "loss": 0.6807,
+      "step": 3660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7326869806094185e-05,
+      "loss": 0.7085,
+      "step": 3670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7319580113719202e-05,
+      "loss": 0.6723,
+      "step": 3680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.731229042134422e-05,
+      "loss": 0.7141,
+      "step": 3690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7305000728969237e-05,
+      "loss": 0.6622,
+      "step": 3700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7297711036594258e-05,
+      "loss": 0.7122,
+      "step": 3710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7290421344219276e-05,
+      "loss": 0.6915,
+      "step": 3720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7283131651844293e-05,
+      "loss": 0.7064,
+      "step": 3730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7275841959469314e-05,
+      "loss": 0.701,
+      "step": 3740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7268552267094328e-05,
+      "loss": 0.7001,
+      "step": 3750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.726126257471935e-05,
+      "loss": 0.6418,
+      "step": 3760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7253972882344366e-05,
+      "loss": 0.6327,
+      "step": 3770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7246683189969384e-05,
+      "loss": 0.7287,
+      "step": 3780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7239393497594405e-05,
+      "loss": 0.656,
+      "step": 3790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.723210380521942e-05,
+      "loss": 0.6429,
+      "step": 3800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.722481411284444e-05,
+      "loss": 0.6866,
+      "step": 3810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7217524420469457e-05,
+      "loss": 0.6755,
+      "step": 3820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7210234728094474e-05,
+      "loss": 0.6685,
+      "step": 3830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7202945035719495e-05,
+      "loss": 0.6692,
+      "step": 3840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7195655343344513e-05,
+      "loss": 0.6329,
+      "step": 3850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.718836565096953e-05,
+      "loss": 0.5945,
+      "step": 3860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7181075958594548e-05,
+      "loss": 0.6705,
+      "step": 3870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.717378626621957e-05,
+      "loss": 0.7204,
+      "step": 3880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7166496573844586e-05,
+      "loss": 0.7004,
+      "step": 3890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7159206881469603e-05,
+      "loss": 0.7343,
+      "step": 3900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.715191718909462e-05,
+      "loss": 0.6666,
+      "step": 3910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7144627496719638e-05,
+      "loss": 0.6353,
+      "step": 3920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.713733780434466e-05,
+      "loss": 0.6257,
+      "step": 3930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7130048111969677e-05,
+      "loss": 0.6461,
+      "step": 3940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7122758419594694e-05,
+      "loss": 0.6812,
+      "step": 3950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.711546872721971e-05,
+      "loss": 0.6279,
+      "step": 3960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7108179034844732e-05,
+      "loss": 0.6609,
+      "step": 3970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.710088934246975e-05,
+      "loss": 0.6643,
+      "step": 3980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7093599650094767e-05,
+      "loss": 0.6613,
+      "step": 3990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7086309957719788e-05,
+      "loss": 0.691,
+      "step": 4000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7079020265344802e-05,
+      "loss": 0.6475,
+      "step": 4010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7071730572969823e-05,
+      "loss": 0.661,
+      "step": 4020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.706444088059484e-05,
+      "loss": 0.7105,
+      "step": 4030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7057151188219858e-05,
+      "loss": 0.6432,
+      "step": 4040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.704986149584488e-05,
+      "loss": 0.7312,
+      "step": 4050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7042571803469893e-05,
+      "loss": 0.6524,
+      "step": 4060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7035282111094914e-05,
+      "loss": 0.6126,
+      "step": 4070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.702799241871993e-05,
+      "loss": 0.7031,
+      "step": 4080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.702070272634495e-05,
+      "loss": 0.7028,
+      "step": 4090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.701341303396997e-05,
+      "loss": 0.6662,
+      "step": 4100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7006123341594987e-05,
+      "loss": 0.6316,
+      "step": 4110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6998833649220004e-05,
+      "loss": 0.6327,
+      "step": 4120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6991543956845022e-05,
+      "loss": 0.6444,
+      "step": 4130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6984254264470043e-05,
+      "loss": 0.7096,
+      "step": 4140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6976964572095057e-05,
+      "loss": 0.6522,
+      "step": 4150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6969674879720077e-05,
+      "loss": 0.6689,
+      "step": 4160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6962385187345095e-05,
+      "loss": 0.6719,
+      "step": 4170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6955095494970112e-05,
+      "loss": 0.6818,
+      "step": 4180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6947805802595133e-05,
+      "loss": 0.6758,
+      "step": 4190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.694051611022015e-05,
+      "loss": 0.6189,
+      "step": 4200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6933226417845168e-05,
+      "loss": 0.668,
+      "step": 4210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6925936725470186e-05,
+      "loss": 0.5942,
+      "step": 4220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6918647033095206e-05,
+      "loss": 0.692,
+      "step": 4230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6911357340720224e-05,
+      "loss": 0.6892,
+      "step": 4240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.690406764834524e-05,
+      "loss": 0.647,
+      "step": 4250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6896777955970262e-05,
+      "loss": 0.645,
+      "step": 4260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6889488263595276e-05,
+      "loss": 0.6645,
+      "step": 4270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6882198571220297e-05,
+      "loss": 0.7113,
+      "step": 4280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6874908878845315e-05,
+      "loss": 0.6835,
+      "step": 4290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6867619186470332e-05,
+      "loss": 0.6203,
+      "step": 4300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.686032949409535e-05,
+      "loss": 0.6357,
+      "step": 4310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6853039801720367e-05,
+      "loss": 0.6494,
+      "step": 4320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6845750109345388e-05,
+      "loss": 0.6776,
+      "step": 4330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6838460416970405e-05,
+      "loss": 0.6456,
+      "step": 4340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6831170724595423e-05,
+      "loss": 0.6431,
+      "step": 4350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.682388103222044e-05,
+      "loss": 0.5721,
+      "step": 4360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.681659133984546e-05,
+      "loss": 0.6882,
+      "step": 4370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.680930164747048e-05,
+      "loss": 0.6043,
+      "step": 4380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6802011955095496e-05,
+      "loss": 0.5861,
+      "step": 4390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6794722262720517e-05,
+      "loss": 0.6537,
+      "step": 4400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.678743257034553e-05,
+      "loss": 0.694,
+      "step": 4410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.678014287797055e-05,
+      "loss": 0.6452,
+      "step": 4420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.677285318559557e-05,
+      "loss": 0.6567,
+      "step": 4430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6765563493220587e-05,
+      "loss": 0.6501,
+      "step": 4440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6758273800845607e-05,
+      "loss": 0.7245,
+      "step": 4450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6750984108470625e-05,
+      "loss": 0.645,
+      "step": 4460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6743694416095642e-05,
+      "loss": 0.7063,
+      "step": 4470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.673640472372066e-05,
+      "loss": 0.6678,
+      "step": 4480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.672911503134568e-05,
+      "loss": 0.6856,
+      "step": 4490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6721825338970698e-05,
+      "loss": 0.6292,
+      "step": 4500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6714535646595715e-05,
+      "loss": 0.6366,
+      "step": 4510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6707245954220733e-05,
+      "loss": 0.6213,
+      "step": 4520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.669995626184575e-05,
+      "loss": 0.6009,
+      "step": 4530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.669266656947077e-05,
+      "loss": 0.6367,
+      "step": 4540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6685376877095785e-05,
+      "loss": 0.599,
+      "step": 4550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6678087184720806e-05,
+      "loss": 0.5963,
+      "step": 4560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6670797492345824e-05,
+      "loss": 0.6839,
+      "step": 4570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.666350779997084e-05,
+      "loss": 0.5932,
+      "step": 4580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6656218107595862e-05,
+      "loss": 0.6111,
+      "step": 4590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.664892841522088e-05,
+      "loss": 0.6865,
+      "step": 4600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6641638722845897e-05,
+      "loss": 0.6671,
+      "step": 4610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6634349030470914e-05,
+      "loss": 0.6828,
+      "step": 4620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6627059338095935e-05,
+      "loss": 0.6601,
+      "step": 4630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6619769645720953e-05,
+      "loss": 0.6368,
+      "step": 4640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.661247995334597e-05,
+      "loss": 0.6212,
+      "step": 4650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.660519026097099e-05,
+      "loss": 0.6669,
+      "step": 4660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6597900568596005e-05,
+      "loss": 0.6806,
+      "step": 4670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6590610876221026e-05,
+      "loss": 0.613,
+      "step": 4680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6583321183846043e-05,
+      "loss": 0.66,
+      "step": 4690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.657603149147106e-05,
+      "loss": 0.6771,
+      "step": 4700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.6568741799096078e-05,
+      "loss": 0.5866,
+      "step": 4710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.65614521067211e-05,
+      "loss": 0.6881,
+      "step": 4720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6554162414346116e-05,
+      "loss": 0.6532,
+      "step": 4730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6546872721971134e-05,
+      "loss": 0.6529,
+      "step": 4740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6539583029596155e-05,
+      "loss": 0.6661,
+      "step": 4750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.653229333722117e-05,
+      "loss": 0.6791,
+      "step": 4760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.652500364484619e-05,
+      "loss": 0.6528,
+      "step": 4770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6517713952471207e-05,
+      "loss": 0.7449,
+      "step": 4780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6510424260096224e-05,
+      "loss": 0.6692,
+      "step": 4790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.6503134567721245e-05,
+      "loss": 0.6404,
+      "step": 4800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.649584487534626e-05,
+      "loss": 0.6349,
+      "step": 4810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.648855518297128e-05,
+      "loss": 0.6142,
+      "step": 4820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6481265490596298e-05,
+      "loss": 0.74,
+      "step": 4830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6473975798221315e-05,
+      "loss": 0.6356,
+      "step": 4840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6466686105846336e-05,
+      "loss": 0.6107,
+      "step": 4850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6459396413471353e-05,
+      "loss": 0.6595,
+      "step": 4860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.645210672109637e-05,
+      "loss": 0.6624,
+      "step": 4870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.644481702872139e-05,
+      "loss": 0.6411,
+      "step": 4880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.643752733634641e-05,
+      "loss": 0.5952,
+      "step": 4890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6430237643971427e-05,
+      "loss": 0.6477,
+      "step": 4900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6422947951596444e-05,
+      "loss": 0.6325,
+      "step": 4910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.641565825922146e-05,
+      "loss": 0.5601,
+      "step": 4920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.640836856684648e-05,
+      "loss": 0.6118,
+      "step": 4930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.64010788744715e-05,
+      "loss": 0.6132,
+      "step": 4940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6393789182096517e-05,
+      "loss": 0.7,
+      "step": 4950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6386499489721535e-05,
+      "loss": 0.6667,
+      "step": 4960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6379209797346552e-05,
+      "loss": 0.658,
+      "step": 4970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6371920104971573e-05,
+      "loss": 0.6485,
+      "step": 4980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.636463041259659e-05,
+      "loss": 0.6403,
+      "step": 4990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6357340720221608e-05,
+      "loss": 0.6611,
+      "step": 5000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6350051027846625e-05,
+      "loss": 0.7022,
+      "step": 5010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6342761335471643e-05,
+      "loss": 0.6339,
+      "step": 5020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6335471643096664e-05,
+      "loss": 0.6851,
+      "step": 5030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.632818195072168e-05,
+      "loss": 0.6963,
+      "step": 5040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.63208922583467e-05,
+      "loss": 0.6362,
+      "step": 5050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.631360256597172e-05,
+      "loss": 0.6828,
+      "step": 5060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6306312873596733e-05,
+      "loss": 0.7035,
+      "step": 5070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6299023181221754e-05,
+      "loss": 0.6544,
+      "step": 5080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6291733488846772e-05,
+      "loss": 0.6282,
+      "step": 5090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.628444379647179e-05,
+      "loss": 0.6957,
+      "step": 5100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.627715410409681e-05,
+      "loss": 0.6422,
+      "step": 5110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6269864411721828e-05,
+      "loss": 0.6363,
+      "step": 5120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6262574719346845e-05,
+      "loss": 0.6119,
+      "step": 5130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6255285026971862e-05,
+      "loss": 0.626,
+      "step": 5140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6247995334596883e-05,
+      "loss": 0.6403,
+      "step": 5150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6240705642221897e-05,
+      "loss": 0.636,
+      "step": 5160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6233415949846918e-05,
+      "loss": 0.6627,
+      "step": 5170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6226126257471936e-05,
+      "loss": 0.6631,
+      "step": 5180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6218836565096953e-05,
+      "loss": 0.6428,
+      "step": 5190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6211546872721974e-05,
+      "loss": 0.5688,
+      "step": 5200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.620425718034699e-05,
+      "loss": 0.6143,
+      "step": 5210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.619696748797201e-05,
+      "loss": 0.6219,
+      "step": 5220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6189677795597026e-05,
+      "loss": 0.6165,
+      "step": 5230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6182388103222047e-05,
+      "loss": 0.6477,
+      "step": 5240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6175098410847065e-05,
+      "loss": 0.6315,
+      "step": 5250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6167808718472082e-05,
+      "loss": 0.606,
+      "step": 5260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.61605190260971e-05,
+      "loss": 0.626,
+      "step": 5270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6153229333722117e-05,
+      "loss": 0.6309,
+      "step": 5280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6145939641347138e-05,
+      "loss": 0.6075,
+      "step": 5290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6138649948972155e-05,
+      "loss": 0.6056,
+      "step": 5300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6131360256597173e-05,
+      "loss": 0.6796,
+      "step": 5310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.612407056422219e-05,
+      "loss": 0.6645,
+      "step": 5320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6116780871847208e-05,
+      "loss": 0.6389,
+      "step": 5330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.610949117947223e-05,
+      "loss": 0.6865,
+      "step": 5340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6102201487097246e-05,
+      "loss": 0.6343,
+      "step": 5350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6094911794722263e-05,
+      "loss": 0.5814,
+      "step": 5360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.608762210234728e-05,
+      "loss": 0.6229,
+      "step": 5370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.60803324099723e-05,
+      "loss": 0.5994,
+      "step": 5380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.607304271759732e-05,
+      "loss": 0.5955,
+      "step": 5390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6065753025222337e-05,
+      "loss": 0.6011,
+      "step": 5400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6058463332847357e-05,
+      "loss": 0.6533,
+      "step": 5410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.605117364047237e-05,
+      "loss": 0.5982,
+      "step": 5420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6043883948097392e-05,
+      "loss": 0.6814,
+      "step": 5430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.603659425572241e-05,
+      "loss": 0.5518,
+      "step": 5440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6029304563347427e-05,
+      "loss": 0.6575,
+      "step": 5450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6022014870972448e-05,
+      "loss": 0.6056,
+      "step": 5460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6014725178597466e-05,
+      "loss": 0.6549,
+      "step": 5470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6007435486222483e-05,
+      "loss": 0.6202,
+      "step": 5480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.60001457938475e-05,
+      "loss": 0.6138,
+      "step": 5490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5992856101472518e-05,
+      "loss": 0.5677,
+      "step": 5500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.598556640909754e-05,
+      "loss": 0.6915,
+      "step": 5510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5978276716722556e-05,
+      "loss": 0.646,
+      "step": 5520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.5970987024347574e-05,
+      "loss": 0.6593,
+      "step": 5530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.596369733197259e-05,
+      "loss": 0.5973,
+      "step": 5540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5956407639597612e-05,
+      "loss": 0.618,
+      "step": 5550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5949117947222626e-05,
+      "loss": 0.6547,
+      "step": 5560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5941828254847647e-05,
+      "loss": 0.5603,
+      "step": 5570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5934538562472664e-05,
+      "loss": 0.6347,
+      "step": 5580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5927248870097682e-05,
+      "loss": 0.604,
+      "step": 5590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5919959177722703e-05,
+      "loss": 0.6162,
+      "step": 5600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.591266948534772e-05,
+      "loss": 0.6452,
+      "step": 5610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.5905379792972737e-05,
+      "loss": 0.647,
+      "step": 5620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5898090100597755e-05,
+      "loss": 0.6016,
+      "step": 5630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5890800408222776e-05,
+      "loss": 0.6027,
+      "step": 5640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5883510715847793e-05,
+      "loss": 0.6347,
+      "step": 5650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.587622102347281e-05,
+      "loss": 0.6555,
+      "step": 5660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.586893133109783e-05,
+      "loss": 0.6661,
+      "step": 5670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5861641638722846e-05,
+      "loss": 0.6621,
+      "step": 5680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5854351946347866e-05,
+      "loss": 0.6247,
+      "step": 5690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5847062253972884e-05,
+      "loss": 0.5752,
+      "step": 5700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.58397725615979e-05,
+      "loss": 0.6995,
+      "step": 5710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.583248286922292e-05,
+      "loss": 0.59,
+      "step": 5720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5825193176847936e-05,
+      "loss": 0.5735,
+      "step": 5730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5817903484472957e-05,
+      "loss": 0.623,
+      "step": 5740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5810613792097975e-05,
+      "loss": 0.6714,
+      "step": 5750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5803324099722992e-05,
+      "loss": 0.6011,
+      "step": 5760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.579603440734801e-05,
+      "loss": 0.6381,
+      "step": 5770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.578874471497303e-05,
+      "loss": 0.6189,
+      "step": 5780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5781455022598048e-05,
+      "loss": 0.668,
+      "step": 5790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5774165330223065e-05,
+      "loss": 0.6612,
+      "step": 5800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5766875637848086e-05,
+      "loss": 0.6208,
+      "step": 5810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.57595859454731e-05,
+      "loss": 0.6142,
+      "step": 5820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.575229625309812e-05,
+      "loss": 0.6559,
+      "step": 5830
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.574500656072314e-05,
+      "loss": 0.7055,
+      "step": 5840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5737716868348156e-05,
+      "loss": 0.6275,
+      "step": 5850
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5730427175973177e-05,
+      "loss": 0.6232,
+      "step": 5860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5723137483598194e-05,
+      "loss": 0.6881,
+      "step": 5870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.571584779122321e-05,
+      "loss": 0.6319,
+      "step": 5880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.570855809884823e-05,
+      "loss": 0.6214,
+      "step": 5890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.570126840647325e-05,
+      "loss": 0.6712,
+      "step": 5900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5693978714098267e-05,
+      "loss": 0.6433,
+      "step": 5910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5686689021723285e-05,
+      "loss": 0.6549,
+      "step": 5920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5679399329348302e-05,
+      "loss": 0.6669,
+      "step": 5930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.567210963697332e-05,
+      "loss": 0.6086,
+      "step": 5940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.566481994459834e-05,
+      "loss": 0.8295,
+      "step": 5950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5657530252223358e-05,
+      "loss": 0.5987,
+      "step": 5960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5650240559848375e-05,
+      "loss": 0.6571,
+      "step": 5970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5642950867473393e-05,
+      "loss": 0.6699,
+      "step": 5980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.563566117509841e-05,
+      "loss": 0.6635,
+      "step": 5990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.562837148272343e-05,
+      "loss": 0.6244,
+      "step": 6000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.562108179034845e-05,
+      "loss": 0.6579,
+      "step": 6010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5613792097973466e-05,
+      "loss": 0.589,
+      "step": 6020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5606502405598484e-05,
+      "loss": 0.6992,
+      "step": 6030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5599212713223504e-05,
+      "loss": 0.6574,
+      "step": 6040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5591923020848522e-05,
+      "loss": 0.5911,
+      "step": 6050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.558463332847354e-05,
+      "loss": 0.6146,
+      "step": 6060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.557734363609856e-05,
+      "loss": 0.6502,
+      "step": 6070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5570053943723574e-05,
+      "loss": 0.6783,
+      "step": 6080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5562764251348595e-05,
+      "loss": 0.6016,
+      "step": 6090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5555474558973613e-05,
+      "loss": 0.6672,
+      "step": 6100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.554818486659863e-05,
+      "loss": 0.6317,
+      "step": 6110
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.554089517422365e-05,
+      "loss": 0.5728,
+      "step": 6120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5533605481848668e-05,
+      "loss": 0.5969,
+      "step": 6130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5526315789473686e-05,
+      "loss": 0.6371,
+      "step": 6140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5519026097098703e-05,
+      "loss": 0.6856,
+      "step": 6150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5511736404723724e-05,
+      "loss": 0.6939,
+      "step": 6160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5504446712348738e-05,
+      "loss": 0.5996,
+      "step": 6170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.549715701997376e-05,
+      "loss": 0.5708,
+      "step": 6180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5489867327598776e-05,
+      "loss": 0.6608,
+      "step": 6190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5482577635223794e-05,
+      "loss": 0.644,
+      "step": 6200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5475287942848815e-05,
+      "loss": 0.6212,
+      "step": 6210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.546799825047383e-05,
+      "loss": 0.5643,
+      "step": 6220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.546070855809885e-05,
+      "loss": 0.6582,
+      "step": 6230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5453418865723867e-05,
+      "loss": 0.6833,
+      "step": 6240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5446129173348884e-05,
+      "loss": 0.6635,
+      "step": 6250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5438839480973905e-05,
+      "loss": 0.5706,
+      "step": 6260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5431549788598923e-05,
+      "loss": 0.6156,
+      "step": 6270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.542426009622394e-05,
+      "loss": 0.5967,
+      "step": 6280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5416970403848958e-05,
+      "loss": 0.5605,
+      "step": 6290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.540968071147398e-05,
+      "loss": 0.6557,
+      "step": 6300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5402391019098996e-05,
+      "loss": 0.6128,
+      "step": 6310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5395101326724013e-05,
+      "loss": 0.6335,
+      "step": 6320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.538781163434903e-05,
+      "loss": 0.5728,
+      "step": 6330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.538052194197405e-05,
+      "loss": 0.6091,
+      "step": 6340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.537323224959907e-05,
+      "loss": 0.6215,
+      "step": 6350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5365942557224087e-05,
+      "loss": 0.6196,
+      "step": 6360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5358652864849104e-05,
+      "loss": 0.6161,
+      "step": 6370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.535136317247412e-05,
+      "loss": 0.663,
+      "step": 6380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5344073480099142e-05,
+      "loss": 0.6515,
+      "step": 6390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.533678378772416e-05,
+      "loss": 0.6819,
+      "step": 6400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5329494095349177e-05,
+      "loss": 0.6584,
+      "step": 6410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5322204402974198e-05,
+      "loss": 0.7116,
+      "step": 6420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5314914710599212e-05,
+      "loss": 0.6237,
+      "step": 6430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5307625018224233e-05,
+      "loss": 0.6485,
+      "step": 6440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.530033532584925e-05,
+      "loss": 0.5877,
+      "step": 6450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5293045633474268e-05,
+      "loss": 0.6145,
+      "step": 6460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.528575594109929e-05,
+      "loss": 0.6208,
+      "step": 6470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5278466248724303e-05,
+      "loss": 0.6702,
+      "step": 6480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5271176556349324e-05,
+      "loss": 0.7025,
+      "step": 6490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.526388686397434e-05,
+      "loss": 0.669,
+      "step": 6500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5256597171599359e-05,
+      "loss": 0.65,
+      "step": 6510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5249307479224378e-05,
+      "loss": 0.5884,
+      "step": 6520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5242017786849397e-05,
+      "loss": 0.6086,
+      "step": 6530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5234728094474414e-05,
+      "loss": 0.6676,
+      "step": 6540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5227438402099432e-05,
+      "loss": 0.6153,
+      "step": 6550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5220148709724451e-05,
+      "loss": 0.6363,
+      "step": 6560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5212859017349468e-05,
+      "loss": 0.6243,
+      "step": 6570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5205569324974488e-05,
+      "loss": 0.6297,
+      "step": 6580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5198279632599507e-05,
+      "loss": 0.5903,
+      "step": 6590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5190989940224522e-05,
+      "loss": 0.5576,
+      "step": 6600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5183700247849542e-05,
+      "loss": 0.589,
+      "step": 6610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.517641055547456e-05,
+      "loss": 0.6402,
+      "step": 6620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5169120863099578e-05,
+      "loss": 0.6858,
+      "step": 6630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5161831170724597e-05,
+      "loss": 0.6056,
+      "step": 6640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5154541478349616e-05,
+      "loss": 0.6075,
+      "step": 6650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5147251785974632e-05,
+      "loss": 0.599,
+      "step": 6660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5139962093599651e-05,
+      "loss": 0.5842,
+      "step": 6670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.513267240122467e-05,
+      "loss": 0.6433,
+      "step": 6680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5125382708849688e-05,
+      "loss": 0.649,
+      "step": 6690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5118093016474707e-05,
+      "loss": 0.6478,
+      "step": 6700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5110803324099723e-05,
+      "loss": 0.5762,
+      "step": 6710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5103513631724742e-05,
+      "loss": 0.6911,
+      "step": 6720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5096223939349761e-05,
+      "loss": 0.6099,
+      "step": 6730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5088934246974779e-05,
+      "loss": 0.6318,
+      "step": 6740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5081644554599796e-05,
+      "loss": 0.655,
+      "step": 6750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5074354862224815e-05,
+      "loss": 0.6004,
+      "step": 6760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5067065169849833e-05,
+      "loss": 0.6186,
+      "step": 6770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5059775477474852e-05,
+      "loss": 0.5856,
+      "step": 6780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5052485785099871e-05,
+      "loss": 0.6387,
+      "step": 6790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5045196092724887e-05,
+      "loss": 0.611,
+      "step": 6800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5037906400349906e-05,
+      "loss": 0.6599,
+      "step": 6810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5030616707974925e-05,
+      "loss": 0.6011,
+      "step": 6820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5023327015599942e-05,
+      "loss": 0.5893,
+      "step": 6830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5016037323224962e-05,
+      "loss": 0.6336,
+      "step": 6840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.500874763084998e-05,
+      "loss": 0.6398,
+      "step": 6850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5001457938474997e-05,
+      "loss": 0.6141,
+      "step": 6860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4994168246100016e-05,
+      "loss": 0.6317,
+      "step": 6870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4986878553725035e-05,
+      "loss": 0.6522,
+      "step": 6880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4979588861350052e-05,
+      "loss": 0.5752,
+      "step": 6890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4972299168975071e-05,
+      "loss": 0.5841,
+      "step": 6900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4965009476600089e-05,
+      "loss": 0.6287,
+      "step": 6910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4957719784225106e-05,
+      "loss": 0.6426,
+      "step": 6920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4950430091850125e-05,
+      "loss": 0.662,
+      "step": 6930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4943140399475143e-05,
+      "loss": 0.6806,
+      "step": 6940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4935850707100162e-05,
+      "loss": 0.6352,
+      "step": 6950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.492856101472518e-05,
+      "loss": 0.5837,
+      "step": 6960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4921271322350197e-05,
+      "loss": 0.6173,
+      "step": 6970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4913981629975216e-05,
+      "loss": 0.6143,
+      "step": 6980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4906691937600235e-05,
+      "loss": 0.6144,
+      "step": 6990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4899402245225251e-05,
+      "loss": 0.6397,
+      "step": 7000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.489211255285027e-05,
+      "loss": 0.5703,
+      "step": 7010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.488482286047529e-05,
+      "loss": 0.6862,
+      "step": 7020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4877533168100307e-05,
+      "loss": 0.5957,
+      "step": 7030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4870243475725326e-05,
+      "loss": 0.6381,
+      "step": 7040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4862953783350345e-05,
+      "loss": 0.6452,
+      "step": 7050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4855664090975361e-05,
+      "loss": 0.6473,
+      "step": 7060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.484837439860038e-05,
+      "loss": 0.6475,
+      "step": 7070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.48410847062254e-05,
+      "loss": 0.6394,
+      "step": 7080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4833795013850417e-05,
+      "loss": 0.5865,
+      "step": 7090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4826505321475436e-05,
+      "loss": 0.5962,
+      "step": 7100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4819215629100455e-05,
+      "loss": 0.596,
+      "step": 7110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.481192593672547e-05,
+      "loss": 0.5828,
+      "step": 7120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.480463624435049e-05,
+      "loss": 0.6042,
+      "step": 7130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4797346551975509e-05,
+      "loss": 0.6045,
+      "step": 7140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4790056859600526e-05,
+      "loss": 0.6051,
+      "step": 7150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4782767167225544e-05,
+      "loss": 0.5999,
+      "step": 7160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4775477474850563e-05,
+      "loss": 0.648,
+      "step": 7170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.476818778247558e-05,
+      "loss": 0.6523,
+      "step": 7180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.47608980901006e-05,
+      "loss": 0.5328,
+      "step": 7190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4753608397725615e-05,
+      "loss": 0.6622,
+      "step": 7200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4746318705350635e-05,
+      "loss": 0.6564,
+      "step": 7210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4739029012975654e-05,
+      "loss": 0.6295,
+      "step": 7220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4731739320600671e-05,
+      "loss": 0.6228,
+      "step": 7230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.472444962822569e-05,
+      "loss": 0.5671,
+      "step": 7240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.471715993585071e-05,
+      "loss": 0.6199,
+      "step": 7250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4709870243475725e-05,
+      "loss": 0.6526,
+      "step": 7260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.4702580551100744e-05,
+      "loss": 0.6812,
+      "step": 7270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4695290858725763e-05,
+      "loss": 0.6809,
+      "step": 7280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4688001166350781e-05,
+      "loss": 0.6295,
+      "step": 7290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.46807114739758e-05,
+      "loss": 0.6649,
+      "step": 7300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.467342178160082e-05,
+      "loss": 0.5537,
+      "step": 7310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4666132089225835e-05,
+      "loss": 0.6721,
+      "step": 7320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4658842396850854e-05,
+      "loss": 0.6558,
+      "step": 7330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4651552704475873e-05,
+      "loss": 0.5941,
+      "step": 7340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.464426301210089e-05,
+      "loss": 0.6562,
+      "step": 7350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.4636973319725908e-05,
+      "loss": 0.6117,
+      "step": 7360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4629683627350927e-05,
+      "loss": 0.5884,
+      "step": 7370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4622393934975945e-05,
+      "loss": 0.6081,
+      "step": 7380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4615104242600964e-05,
+      "loss": 0.6627,
+      "step": 7390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4607814550225983e-05,
+      "loss": 0.6224,
+      "step": 7400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4600524857850999e-05,
+      "loss": 0.6116,
+      "step": 7410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4593235165476018e-05,
+      "loss": 0.663,
+      "step": 7420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4585945473101035e-05,
+      "loss": 0.6339,
+      "step": 7430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4578655780726055e-05,
+      "loss": 0.5872,
+      "step": 7440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.4571366088351074e-05,
+      "loss": 0.6315,
+      "step": 7450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.456407639597609e-05,
+      "loss": 0.6097,
+      "step": 7460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4556786703601109e-05,
+      "loss": 0.6529,
+      "step": 7470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4549497011226128e-05,
+      "loss": 0.6694,
+      "step": 7480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4542207318851145e-05,
+      "loss": 0.6013,
+      "step": 7490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4534917626476164e-05,
+      "loss": 0.5854,
+      "step": 7500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4527627934101184e-05,
+      "loss": 0.6413,
+      "step": 7510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.45203382417262e-05,
+      "loss": 0.5764,
+      "step": 7520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4513048549351218e-05,
+      "loss": 0.5727,
+      "step": 7530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4505758856976238e-05,
+      "loss": 0.6175,
+      "step": 7540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4498469164601255e-05,
+      "loss": 0.6287,
+      "step": 7550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4491179472226272e-05,
+      "loss": 0.609,
+      "step": 7560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4483889779851292e-05,
+      "loss": 0.62,
+      "step": 7570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4476600087476309e-05,
+      "loss": 0.6036,
+      "step": 7580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4469310395101328e-05,
+      "loss": 0.6219,
+      "step": 7590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4462020702726347e-05,
+      "loss": 0.6115,
+      "step": 7600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4454731010351363e-05,
+      "loss": 0.6255,
+      "step": 7610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4447441317976382e-05,
+      "loss": 0.6535,
+      "step": 7620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4440151625601401e-05,
+      "loss": 0.6323,
+      "step": 7630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4432861933226419e-05,
+      "loss": 0.6411,
+      "step": 7640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4425572240851438e-05,
+      "loss": 0.6132,
+      "step": 7650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4418282548476454e-05,
+      "loss": 0.6106,
+      "step": 7660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4410992856101473e-05,
+      "loss": 0.6256,
+      "step": 7670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4403703163726492e-05,
+      "loss": 0.6327,
+      "step": 7680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.439641347135151e-05,
+      "loss": 0.6584,
+      "step": 7690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4389123778976529e-05,
+      "loss": 0.5828,
+      "step": 7700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4381834086601548e-05,
+      "loss": 0.6988,
+      "step": 7710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4374544394226564e-05,
+      "loss": 0.5815,
+      "step": 7720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4367254701851583e-05,
+      "loss": 0.5657,
+      "step": 7730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4359965009476602e-05,
+      "loss": 0.6006,
+      "step": 7740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.435267531710162e-05,
+      "loss": 0.5565,
+      "step": 7750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4345385624726637e-05,
+      "loss": 0.5214,
+      "step": 7760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4338095932351656e-05,
+      "loss": 0.6469,
+      "step": 7770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4330806239976673e-05,
+      "loss": 0.6331,
+      "step": 7780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4323516547601693e-05,
+      "loss": 0.6109,
+      "step": 7790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4316226855226712e-05,
+      "loss": 0.554,
+      "step": 7800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4308937162851727e-05,
+      "loss": 0.6066,
+      "step": 7810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4301647470476747e-05,
+      "loss": 0.6079,
+      "step": 7820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4294357778101766e-05,
+      "loss": 0.5946,
+      "step": 7830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4287068085726783e-05,
+      "loss": 0.5312,
+      "step": 7840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4279778393351802e-05,
+      "loss": 0.5878,
+      "step": 7850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4272488700976822e-05,
+      "loss": 0.577,
+      "step": 7860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4265199008601837e-05,
+      "loss": 0.575,
+      "step": 7870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4257909316226856e-05,
+      "loss": 0.671,
+      "step": 7880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4250619623851876e-05,
+      "loss": 0.6609,
+      "step": 7890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4243329931476893e-05,
+      "loss": 0.5604,
+      "step": 7900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4236040239101912e-05,
+      "loss": 0.5841,
+      "step": 7910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4228750546726928e-05,
+      "loss": 0.6105,
+      "step": 7920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4221460854351947e-05,
+      "loss": 0.5611,
+      "step": 7930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4214171161976966e-05,
+      "loss": 0.5848,
+      "step": 7940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4206881469601984e-05,
+      "loss": 0.545,
+      "step": 7950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4199591777227003e-05,
+      "loss": 0.6039,
+      "step": 7960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.419230208485202e-05,
+      "loss": 0.6141,
+      "step": 7970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4185012392477038e-05,
+      "loss": 0.5844,
+      "step": 7980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4177722700102057e-05,
+      "loss": 0.5301,
+      "step": 7990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4170433007727076e-05,
+      "loss": 0.5963,
+      "step": 8000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4163143315352092e-05,
+      "loss": 0.6129,
+      "step": 8010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4155853622977111e-05,
+      "loss": 0.6455,
+      "step": 8020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.414856393060213e-05,
+      "loss": 0.5293,
+      "step": 8030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4141274238227148e-05,
+      "loss": 0.5809,
+      "step": 8040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4133984545852167e-05,
+      "loss": 0.6143,
+      "step": 8050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4126694853477186e-05,
+      "loss": 0.5824,
+      "step": 8060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4119405161102202e-05,
+      "loss": 0.6727,
+      "step": 8070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.411211546872722e-05,
+      "loss": 0.635,
+      "step": 8080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.410482577635224e-05,
+      "loss": 0.5824,
+      "step": 8090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4097536083977257e-05,
+      "loss": 0.6074,
+      "step": 8100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4090246391602276e-05,
+      "loss": 0.6487,
+      "step": 8110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4082956699227296e-05,
+      "loss": 0.6239,
+      "step": 8120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4075667006852311e-05,
+      "loss": 0.6047,
+      "step": 8130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.406837731447733e-05,
+      "loss": 0.5848,
+      "step": 8140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4061087622102348e-05,
+      "loss": 0.6398,
+      "step": 8150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4053797929727367e-05,
+      "loss": 0.5414,
+      "step": 8160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4046508237352385e-05,
+      "loss": 0.6156,
+      "step": 8170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4039218544977402e-05,
+      "loss": 0.6241,
+      "step": 8180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4031928852602421e-05,
+      "loss": 0.5964,
+      "step": 8190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.402463916022744e-05,
+      "loss": 0.596,
+      "step": 8200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4017349467852456e-05,
+      "loss": 0.6107,
+      "step": 8210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4010059775477475e-05,
+      "loss": 0.5672,
+      "step": 8220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4002770083102494e-05,
+      "loss": 0.5837,
+      "step": 8230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3995480390727512e-05,
+      "loss": 0.5483,
+      "step": 8240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3988190698352531e-05,
+      "loss": 0.6155,
+      "step": 8250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.398090100597755e-05,
+      "loss": 0.6541,
+      "step": 8260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3973611313602566e-05,
+      "loss": 0.5876,
+      "step": 8270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3966321621227585e-05,
+      "loss": 0.6007,
+      "step": 8280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3959031928852604e-05,
+      "loss": 0.5771,
+      "step": 8290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3951742236477622e-05,
+      "loss": 0.5658,
+      "step": 8300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.394445254410264e-05,
+      "loss": 0.605,
+      "step": 8310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.393716285172766e-05,
+      "loss": 0.6405,
+      "step": 8320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3929873159352676e-05,
+      "loss": 0.5682,
+      "step": 8330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3922583466977695e-05,
+      "loss": 0.6121,
+      "step": 8340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3915293774602714e-05,
+      "loss": 0.5529,
+      "step": 8350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.3908004082227731e-05,
+      "loss": 0.5755,
+      "step": 8360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3900714389852749e-05,
+      "loss": 0.5811,
+      "step": 8370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3893424697477768e-05,
+      "loss": 0.6348,
+      "step": 8380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3886135005102785e-05,
+      "loss": 0.6187,
+      "step": 8390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3878845312727805e-05,
+      "loss": 0.6066,
+      "step": 8400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.387155562035282e-05,
+      "loss": 0.6404,
+      "step": 8410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.386426592797784e-05,
+      "loss": 0.5411,
+      "step": 8420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3856976235602859e-05,
+      "loss": 0.633,
+      "step": 8430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3849686543227876e-05,
+      "loss": 0.5889,
+      "step": 8440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3842396850852895e-05,
+      "loss": 0.6432,
+      "step": 8450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.3835107158477914e-05,
+      "loss": 0.5396,
+      "step": 8460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.382781746610293e-05,
+      "loss": 0.6607,
+      "step": 8470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.382052777372795e-05,
+      "loss": 0.5942,
+      "step": 8480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3813238081352968e-05,
+      "loss": 0.627,
+      "step": 8490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3805948388977986e-05,
+      "loss": 0.5483,
+      "step": 8500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3798658696603005e-05,
+      "loss": 0.6096,
+      "step": 8510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3791369004228024e-05,
+      "loss": 0.5637,
+      "step": 8520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.378407931185304e-05,
+      "loss": 0.6057,
+      "step": 8530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3776789619478059e-05,
+      "loss": 0.5378,
+      "step": 8540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3769499927103078e-05,
+      "loss": 0.6236,
+      "step": 8550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3762210234728096e-05,
+      "loss": 0.5967,
+      "step": 8560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3754920542353113e-05,
+      "loss": 0.6476,
+      "step": 8570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3747630849978132e-05,
+      "loss": 0.558,
+      "step": 8580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.374034115760315e-05,
+      "loss": 0.5448,
+      "step": 8590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3733051465228169e-05,
+      "loss": 0.5922,
+      "step": 8600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3725761772853188e-05,
+      "loss": 0.5697,
+      "step": 8610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3718472080478204e-05,
+      "loss": 0.5869,
+      "step": 8620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3711182388103223e-05,
+      "loss": 0.5668,
+      "step": 8630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.370389269572824e-05,
+      "loss": 0.5599,
+      "step": 8640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.369660300335326e-05,
+      "loss": 0.5346,
+      "step": 8650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3689313310978279e-05,
+      "loss": 0.6465,
+      "step": 8660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3682023618603295e-05,
+      "loss": 0.5857,
+      "step": 8670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3674733926228314e-05,
+      "loss": 0.6233,
+      "step": 8680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3667444233853333e-05,
+      "loss": 0.6056,
+      "step": 8690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.366015454147835e-05,
+      "loss": 0.5325,
+      "step": 8700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.365286484910337e-05,
+      "loss": 0.5882,
+      "step": 8710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3645575156728389e-05,
+      "loss": 0.6059,
+      "step": 8720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3638285464353404e-05,
+      "loss": 0.6712,
+      "step": 8730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3630995771978423e-05,
+      "loss": 0.6202,
+      "step": 8740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3623706079603443e-05,
+      "loss": 0.6197,
+      "step": 8750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.361641638722846e-05,
+      "loss": 0.5722,
+      "step": 8760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.360912669485348e-05,
+      "loss": 0.5902,
+      "step": 8770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3601837002478497e-05,
+      "loss": 0.597,
+      "step": 8780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3594547310103514e-05,
+      "loss": 0.5147,
+      "step": 8790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3587257617728533e-05,
+      "loss": 0.5723,
+      "step": 8800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3579967925353552e-05,
+      "loss": 0.6178,
+      "step": 8810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3572678232978568e-05,
+      "loss": 0.6315,
+      "step": 8820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3565388540603587e-05,
+      "loss": 0.6263,
+      "step": 8830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3558098848228606e-05,
+      "loss": 0.6178,
+      "step": 8840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3550809155853624e-05,
+      "loss": 0.5177,
+      "step": 8850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3543519463478643e-05,
+      "loss": 0.6145,
+      "step": 8860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3536229771103659e-05,
+      "loss": 0.6393,
+      "step": 8870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3528940078728678e-05,
+      "loss": 0.5732,
+      "step": 8880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3521650386353697e-05,
+      "loss": 0.6582,
+      "step": 8890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3514360693978715e-05,
+      "loss": 0.5912,
+      "step": 8900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3507071001603734e-05,
+      "loss": 0.5713,
+      "step": 8910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3499781309228753e-05,
+      "loss": 0.52,
+      "step": 8920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3492491616853769e-05,
+      "loss": 0.6827,
+      "step": 8930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3485201924478788e-05,
+      "loss": 0.6326,
+      "step": 8940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3477912232103807e-05,
+      "loss": 0.5945,
+      "step": 8950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3470622539728824e-05,
+      "loss": 0.5544,
+      "step": 8960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3463332847353844e-05,
+      "loss": 0.5613,
+      "step": 8970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3456043154978861e-05,
+      "loss": 0.5653,
+      "step": 8980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3448753462603878e-05,
+      "loss": 0.5792,
+      "step": 8990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3441463770228898e-05,
+      "loss": 0.5276,
+      "step": 9000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3434174077853917e-05,
+      "loss": 0.6064,
+      "step": 9010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3426884385478932e-05,
+      "loss": 0.5631,
+      "step": 9020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3419594693103952e-05,
+      "loss": 0.6084,
+      "step": 9030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.341230500072897e-05,
+      "loss": 0.5744,
+      "step": 9040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3405015308353988e-05,
+      "loss": 0.5633,
+      "step": 9050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3397725615979007e-05,
+      "loss": 0.5879,
+      "step": 9060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3390435923604027e-05,
+      "loss": 0.6053,
+      "step": 9070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3383146231229042e-05,
+      "loss": 0.6245,
+      "step": 9080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3375856538854061e-05,
+      "loss": 0.5594,
+      "step": 9090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.336856684647908e-05,
+      "loss": 0.5847,
+      "step": 9100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3361277154104098e-05,
+      "loss": 0.6013,
+      "step": 9110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3353987461729117e-05,
+      "loss": 0.6048,
+      "step": 9120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3346697769354133e-05,
+      "loss": 0.5651,
+      "step": 9130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3339408076979152e-05,
+      "loss": 0.5947,
+      "step": 9140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3332118384604171e-05,
+      "loss": 0.4878,
+      "step": 9150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3324828692229189e-05,
+      "loss": 0.3914,
+      "step": 9160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3317538999854208e-05,
+      "loss": 0.3927,
+      "step": 9170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3310249307479225e-05,
+      "loss": 0.3852,
+      "step": 9180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3302959615104243e-05,
+      "loss": 0.3599,
+      "step": 9190
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3295669922729262e-05,
+      "loss": 0.3485,
+      "step": 9200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3288380230354281e-05,
+      "loss": 0.3769,
+      "step": 9210
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3281090537979297e-05,
+      "loss": 0.3264,
+      "step": 9220
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3273800845604316e-05,
+      "loss": 0.3837,
+      "step": 9230
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3266511153229335e-05,
+      "loss": 0.3299,
+      "step": 9240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3259221460854353e-05,
+      "loss": 0.3482,
+      "step": 9250
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3251931768479372e-05,
+      "loss": 0.3514,
+      "step": 9260
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.324464207610439e-05,
+      "loss": 0.3453,
+      "step": 9270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3237352383729407e-05,
+      "loss": 0.305,
+      "step": 9280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3230062691354426e-05,
+      "loss": 0.3469,
+      "step": 9290
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3222772998979445e-05,
+      "loss": 0.3389,
+      "step": 9300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3215483306604462e-05,
+      "loss": 0.3853,
+      "step": 9310
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3208193614229481e-05,
+      "loss": 0.3878,
+      "step": 9320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.32009039218545e-05,
+      "loss": 0.3659,
+      "step": 9330
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3193614229479516e-05,
+      "loss": 0.3798,
+      "step": 9340
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3186324537104536e-05,
+      "loss": 0.3709,
+      "step": 9350
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3179034844729553e-05,
+      "loss": 0.3341,
+      "step": 9360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3171745152354572e-05,
+      "loss": 0.3408,
+      "step": 9370
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.316445545997959e-05,
+      "loss": 0.3526,
+      "step": 9380
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3157165767604607e-05,
+      "loss": 0.3495,
+      "step": 9390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3149876075229626e-05,
+      "loss": 0.2978,
+      "step": 9400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3142586382854645e-05,
+      "loss": 0.3801,
+      "step": 9410
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3135296690479661e-05,
+      "loss": 0.3485,
+      "step": 9420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.312800699810468e-05,
+      "loss": 0.2836,
+      "step": 9430
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.31207173057297e-05,
+      "loss": 0.3423,
+      "step": 9440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3113427613354717e-05,
+      "loss": 0.3481,
+      "step": 9450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3106137920979736e-05,
+      "loss": 0.3189,
+      "step": 9460
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3098848228604755e-05,
+      "loss": 0.3739,
+      "step": 9470
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3091558536229771e-05,
+      "loss": 0.373,
+      "step": 9480
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.308426884385479e-05,
+      "loss": 0.3411,
+      "step": 9490
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.307697915147981e-05,
+      "loss": 0.3348,
+      "step": 9500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3069689459104827e-05,
+      "loss": 0.3495,
+      "step": 9510
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3062399766729846e-05,
+      "loss": 0.3729,
+      "step": 9520
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3055110074354865e-05,
+      "loss": 0.3953,
+      "step": 9530
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.304782038197988e-05,
+      "loss": 0.3695,
+      "step": 9540
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.30405306896049e-05,
+      "loss": 0.4107,
+      "step": 9550
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3033240997229919e-05,
+      "loss": 0.3328,
+      "step": 9560
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3025951304854936e-05,
+      "loss": 0.3617,
+      "step": 9570
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3018661612479954e-05,
+      "loss": 0.3248,
+      "step": 9580
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3011371920104973e-05,
+      "loss": 0.3958,
+      "step": 9590
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.300408222772999e-05,
+      "loss": 0.3746,
+      "step": 9600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.299679253535501e-05,
+      "loss": 0.3755,
+      "step": 9610
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2989502842980027e-05,
+      "loss": 0.3695,
+      "step": 9620
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2982213150605045e-05,
+      "loss": 0.3738,
+      "step": 9630
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.2974923458230064e-05,
+      "loss": 0.3421,
+      "step": 9640
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2967633765855081e-05,
+      "loss": 0.3753,
+      "step": 9650
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.29603440734801e-05,
+      "loss": 0.3837,
+      "step": 9660
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.295305438110512e-05,
+      "loss": 0.3084,
+      "step": 9670
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2945764688730135e-05,
+      "loss": 0.3557,
+      "step": 9680
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2938474996355154e-05,
+      "loss": 0.3244,
+      "step": 9690
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2931185303980174e-05,
+      "loss": 0.3636,
+      "step": 9700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2923895611605191e-05,
+      "loss": 0.3559,
+      "step": 9710
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.291660591923021e-05,
+      "loss": 0.3333,
+      "step": 9720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.290931622685523e-05,
+      "loss": 0.3158,
+      "step": 9730
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.2902026534480245e-05,
+      "loss": 0.4051,
+      "step": 9740
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2894736842105264e-05,
+      "loss": 0.3376,
+      "step": 9750
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2887447149730283e-05,
+      "loss": 0.3505,
+      "step": 9760
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.28801574573553e-05,
+      "loss": 0.3467,
+      "step": 9770
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.287286776498032e-05,
+      "loss": 0.3431,
+      "step": 9780
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2865578072605337e-05,
+      "loss": 0.3519,
+      "step": 9790
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2858288380230355e-05,
+      "loss": 0.3406,
+      "step": 9800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2850998687855374e-05,
+      "loss": 0.3202,
+      "step": 9810
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2843708995480393e-05,
+      "loss": 0.3676,
+      "step": 9820
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.2836419303105409e-05,
+      "loss": 0.3376,
+      "step": 9830
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2829129610730428e-05,
+      "loss": 0.3456,
+      "step": 9840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2821839918355445e-05,
+      "loss": 0.3462,
+      "step": 9850
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2814550225980465e-05,
+      "loss": 0.3617,
+      "step": 9860
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2807260533605484e-05,
+      "loss": 0.355,
+      "step": 9870
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.27999708412305e-05,
+      "loss": 0.34,
+      "step": 9880
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2792681148855519e-05,
+      "loss": 0.3632,
+      "step": 9890
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2785391456480538e-05,
+      "loss": 0.3385,
+      "step": 9900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2778101764105555e-05,
+      "loss": 0.3417,
+      "step": 9910
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.2770812071730574e-05,
+      "loss": 0.3711,
+      "step": 9920
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2763522379355594e-05,
+      "loss": 0.3389,
+      "step": 9930
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.275623268698061e-05,
+      "loss": 0.3563,
+      "step": 9940
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2748942994605628e-05,
+      "loss": 0.3965,
+      "step": 9950
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2741653302230648e-05,
+      "loss": 0.3691,
+      "step": 9960
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2734363609855665e-05,
+      "loss": 0.3785,
+      "step": 9970
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2727073917480684e-05,
+      "loss": 0.3529,
+      "step": 9980
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2719784225105702e-05,
+      "loss": 0.3479,
+      "step": 9990
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2712494532730719e-05,
+      "loss": 0.3638,
+      "step": 10000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2705204840355738e-05,
+      "loss": 0.3765,
+      "step": 10010
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2697915147980757e-05,
+      "loss": 0.3625,
+      "step": 10020
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2690625455605773e-05,
+      "loss": 0.3869,
+      "step": 10030
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2683335763230792e-05,
+      "loss": 0.3299,
+      "step": 10040
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2676046070855811e-05,
+      "loss": 0.349,
+      "step": 10050
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2668756378480829e-05,
+      "loss": 0.3538,
+      "step": 10060
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2661466686105848e-05,
+      "loss": 0.3304,
+      "step": 10070
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2654176993730864e-05,
+      "loss": 0.3429,
+      "step": 10080
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2646887301355883e-05,
+      "loss": 0.3788,
+      "step": 10090
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2639597608980902e-05,
+      "loss": 0.3538,
+      "step": 10100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.263230791660592e-05,
+      "loss": 0.3444,
+      "step": 10110
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2625018224230939e-05,
+      "loss": 0.3463,
+      "step": 10120
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2617728531855958e-05,
+      "loss": 0.3327,
+      "step": 10130
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2610438839480974e-05,
+      "loss": 0.3072,
+      "step": 10140
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2603149147105993e-05,
+      "loss": 0.3691,
+      "step": 10150
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2595859454731012e-05,
+      "loss": 0.331,
+      "step": 10160
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.258856976235603e-05,
+      "loss": 0.3713,
+      "step": 10170
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2581280069981049e-05,
+      "loss": 0.3463,
+      "step": 10180
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2573990377606066e-05,
+      "loss": 0.3804,
+      "step": 10190
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2566700685231083e-05,
+      "loss": 0.3606,
+      "step": 10200
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2559410992856103e-05,
+      "loss": 0.3649,
+      "step": 10210
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2552121300481122e-05,
+      "loss": 0.3896,
+      "step": 10220
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2544831608106138e-05,
+      "loss": 0.3555,
+      "step": 10230
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2537541915731157e-05,
+      "loss": 0.3274,
+      "step": 10240
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2530252223356176e-05,
+      "loss": 0.3851,
+      "step": 10250
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2522962530981193e-05,
+      "loss": 0.3593,
+      "step": 10260
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2515672838606212e-05,
+      "loss": 0.3028,
+      "step": 10270
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2508383146231232e-05,
+      "loss": 0.3455,
+      "step": 10280
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2501093453856247e-05,
+      "loss": 0.3484,
+      "step": 10290
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2493803761481266e-05,
+      "loss": 0.3456,
+      "step": 10300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2486514069106286e-05,
+      "loss": 0.3645,
+      "step": 10310
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2479224376731303e-05,
+      "loss": 0.3532,
+      "step": 10320
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2471934684356322e-05,
+      "loss": 0.3803,
+      "step": 10330
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2464644991981338e-05,
+      "loss": 0.3922,
+      "step": 10340
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2457355299606357e-05,
+      "loss": 0.3798,
+      "step": 10350
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2450065607231376e-05,
+      "loss": 0.3818,
+      "step": 10360
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2442775914856394e-05,
+      "loss": 0.3334,
+      "step": 10370
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2435486222481413e-05,
+      "loss": 0.3314,
+      "step": 10380
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.242819653010643e-05,
+      "loss": 0.3786,
+      "step": 10390
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2420906837731448e-05,
+      "loss": 0.3557,
+      "step": 10400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2413617145356467e-05,
+      "loss": 0.3298,
+      "step": 10410
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2406327452981486e-05,
+      "loss": 0.3666,
+      "step": 10420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2399037760606502e-05,
+      "loss": 0.3688,
+      "step": 10430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2391748068231521e-05,
+      "loss": 0.378,
+      "step": 10440
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.238445837585654e-05,
+      "loss": 0.3517,
+      "step": 10450
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2377168683481558e-05,
+      "loss": 0.3533,
+      "step": 10460
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2369878991106577e-05,
+      "loss": 0.3457,
+      "step": 10470
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2362589298731596e-05,
+      "loss": 0.3633,
+      "step": 10480
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2355299606356612e-05,
+      "loss": 0.3825,
+      "step": 10490
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.234800991398163e-05,
+      "loss": 0.3417,
+      "step": 10500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.234072022160665e-05,
+      "loss": 0.3165,
+      "step": 10510
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2333430529231667e-05,
+      "loss": 0.3743,
+      "step": 10520
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2326140836856687e-05,
+      "loss": 0.3549,
+      "step": 10530
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2318851144481706e-05,
+      "loss": 0.3376,
+      "step": 10540
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2311561452106721e-05,
+      "loss": 0.3783,
+      "step": 10550
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.230427175973174e-05,
+      "loss": 0.3295,
+      "step": 10560
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2296982067356758e-05,
+      "loss": 0.3682,
+      "step": 10570
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2289692374981777e-05,
+      "loss": 0.3414,
+      "step": 10580
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2282402682606796e-05,
+      "loss": 0.3825,
+      "step": 10590
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2275112990231812e-05,
+      "loss": 0.342,
+      "step": 10600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2267823297856831e-05,
+      "loss": 0.3557,
+      "step": 10610
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.226053360548185e-05,
+      "loss": 0.3806,
+      "step": 10620
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2253243913106868e-05,
+      "loss": 0.3512,
+      "step": 10630
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2245954220731885e-05,
+      "loss": 0.3093,
+      "step": 10640
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2238664528356904e-05,
+      "loss": 0.3762,
+      "step": 10650
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2231374835981922e-05,
+      "loss": 0.3628,
+      "step": 10660
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2224085143606941e-05,
+      "loss": 0.3143,
+      "step": 10670
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.221679545123196e-05,
+      "loss": 0.417,
+      "step": 10680
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2209505758856976e-05,
+      "loss": 0.3108,
+      "step": 10690
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2202216066481995e-05,
+      "loss": 0.3475,
+      "step": 10700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2194926374107014e-05,
+      "loss": 0.3253,
+      "step": 10710
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2187636681732032e-05,
+      "loss": 0.3544,
+      "step": 10720
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.218034698935705e-05,
+      "loss": 0.3365,
+      "step": 10730
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.217305729698207e-05,
+      "loss": 0.3306,
+      "step": 10740
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2165767604607086e-05,
+      "loss": 0.3026,
+      "step": 10750
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2158477912232105e-05,
+      "loss": 0.3407,
+      "step": 10760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2151188219857124e-05,
+      "loss": 0.3655,
+      "step": 10770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2143898527482141e-05,
+      "loss": 0.3216,
+      "step": 10780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.213660883510716e-05,
+      "loss": 0.3376,
+      "step": 10790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2129319142732176e-05,
+      "loss": 0.3574,
+      "step": 10800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2122029450357196e-05,
+      "loss": 0.3383,
+      "step": 10810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2114739757982215e-05,
+      "loss": 0.3693,
+      "step": 10820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2107450065607232e-05,
+      "loss": 0.3243,
+      "step": 10830
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.210016037323225e-05,
+      "loss": 0.3378,
+      "step": 10840
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2092870680857269e-05,
+      "loss": 0.3575,
+      "step": 10850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2085580988482286e-05,
+      "loss": 0.364,
+      "step": 10860
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2078291296107305e-05,
+      "loss": 0.338,
+      "step": 10870
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2071001603732324e-05,
+      "loss": 0.3752,
+      "step": 10880
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.206371191135734e-05,
+      "loss": 0.3535,
+      "step": 10890
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.205642221898236e-05,
+      "loss": 0.3139,
+      "step": 10900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2049132526607379e-05,
+      "loss": 0.3467,
+      "step": 10910
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2041842834232396e-05,
+      "loss": 0.391,
+      "step": 10920
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2034553141857415e-05,
+      "loss": 0.3489,
+      "step": 10930
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2027263449482434e-05,
+      "loss": 0.3644,
+      "step": 10940
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.201997375710745e-05,
+      "loss": 0.3484,
+      "step": 10950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.201268406473247e-05,
+      "loss": 0.3523,
+      "step": 10960
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2005394372357488e-05,
+      "loss": 0.3282,
+      "step": 10970
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1998104679982506e-05,
+      "loss": 0.3444,
+      "step": 10980
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1990814987607525e-05,
+      "loss": 0.3004,
+      "step": 10990
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1983525295232542e-05,
+      "loss": 0.41,
+      "step": 11000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.197623560285756e-05,
+      "loss": 0.3301,
+      "step": 11010
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1968945910482579e-05,
+      "loss": 0.3422,
+      "step": 11020
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1961656218107598e-05,
+      "loss": 0.3455,
+      "step": 11030
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1954366525732614e-05,
+      "loss": 0.3572,
+      "step": 11040
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1947076833357633e-05,
+      "loss": 0.3546,
+      "step": 11050
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.193978714098265e-05,
+      "loss": 0.3588,
+      "step": 11060
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.193249744860767e-05,
+      "loss": 0.348,
+      "step": 11070
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1925207756232689e-05,
+      "loss": 0.3587,
+      "step": 11080
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1917918063857705e-05,
+      "loss": 0.3312,
+      "step": 11090
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1910628371482724e-05,
+      "loss": 0.3206,
+      "step": 11100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1903338679107743e-05,
+      "loss": 0.3249,
+      "step": 11110
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.189604898673276e-05,
+      "loss": 0.3811,
+      "step": 11120
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.188875929435778e-05,
+      "loss": 0.3283,
+      "step": 11130
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1881469601982799e-05,
+      "loss": 0.3437,
+      "step": 11140
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1874179909607814e-05,
+      "loss": 0.3489,
+      "step": 11150
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1866890217232834e-05,
+      "loss": 0.3664,
+      "step": 11160
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1859600524857853e-05,
+      "loss": 0.3348,
+      "step": 11170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.185231083248287e-05,
+      "loss": 0.2901,
+      "step": 11180
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.184502114010789e-05,
+      "loss": 0.3711,
+      "step": 11190
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1837731447732907e-05,
+      "loss": 0.3562,
+      "step": 11200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1830441755357924e-05,
+      "loss": 0.3619,
+      "step": 11210
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1823152062982943e-05,
+      "loss": 0.3821,
+      "step": 11220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1815862370607962e-05,
+      "loss": 0.3653,
+      "step": 11230
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1808572678232978e-05,
+      "loss": 0.3518,
+      "step": 11240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1801282985857997e-05,
+      "loss": 0.358,
+      "step": 11250
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1793993293483017e-05,
+      "loss": 0.3475,
+      "step": 11260
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1786703601108034e-05,
+      "loss": 0.3916,
+      "step": 11270
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1779413908733053e-05,
+      "loss": 0.3418,
+      "step": 11280
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1772124216358069e-05,
+      "loss": 0.3556,
+      "step": 11290
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1764834523983088e-05,
+      "loss": 0.3766,
+      "step": 11300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1757544831608107e-05,
+      "loss": 0.3053,
+      "step": 11310
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1750255139233125e-05,
+      "loss": 0.3641,
+      "step": 11320
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1742965446858144e-05,
+      "loss": 0.338,
+      "step": 11330
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1735675754483163e-05,
+      "loss": 0.3294,
+      "step": 11340
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1728386062108179e-05,
+      "loss": 0.3732,
+      "step": 11350
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1721096369733198e-05,
+      "loss": 0.3591,
+      "step": 11360
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1713806677358217e-05,
+      "loss": 0.392,
+      "step": 11370
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1706516984983234e-05,
+      "loss": 0.3433,
+      "step": 11380
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1699227292608254e-05,
+      "loss": 0.3689,
+      "step": 11390
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1691937600233271e-05,
+      "loss": 0.402,
+      "step": 11400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1684647907858288e-05,
+      "loss": 0.391,
+      "step": 11410
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1677358215483308e-05,
+      "loss": 0.3016,
+      "step": 11420
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1670068523108327e-05,
+      "loss": 0.3241,
+      "step": 11430
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1662778830733344e-05,
+      "loss": 0.3083,
+      "step": 11440
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1655489138358362e-05,
+      "loss": 0.3879,
+      "step": 11450
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.164819944598338e-05,
+      "loss": 0.3706,
+      "step": 11460
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1640909753608398e-05,
+      "loss": 0.3313,
+      "step": 11470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1633620061233417e-05,
+      "loss": 0.3406,
+      "step": 11480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1626330368858437e-05,
+      "loss": 0.4092,
+      "step": 11490
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1619040676483452e-05,
+      "loss": 0.3545,
+      "step": 11500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1611750984108471e-05,
+      "loss": 0.3054,
+      "step": 11510
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.160446129173349e-05,
+      "loss": 0.3835,
+      "step": 11520
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1597171599358508e-05,
+      "loss": 0.3121,
+      "step": 11530
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1589881906983527e-05,
+      "loss": 0.3536,
+      "step": 11540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1582592214608543e-05,
+      "loss": 0.3941,
+      "step": 11550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1575302522233562e-05,
+      "loss": 0.349,
+      "step": 11560
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1568012829858581e-05,
+      "loss": 0.4004,
+      "step": 11570
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1560723137483599e-05,
+      "loss": 0.3445,
+      "step": 11580
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1553433445108618e-05,
+      "loss": 0.3522,
+      "step": 11590
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1546143752733637e-05,
+      "loss": 0.3555,
+      "step": 11600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1538854060358653e-05,
+      "loss": 0.3434,
+      "step": 11610
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1531564367983672e-05,
+      "loss": 0.3711,
+      "step": 11620
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1524274675608691e-05,
+      "loss": 0.3268,
+      "step": 11630
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1516984983233709e-05,
+      "loss": 0.3515,
+      "step": 11640
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1509695290858726e-05,
+      "loss": 0.3605,
+      "step": 11650
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1502405598483745e-05,
+      "loss": 0.3388,
+      "step": 11660
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1495115906108763e-05,
+      "loss": 0.345,
+      "step": 11670
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1487826213733782e-05,
+      "loss": 0.3684,
+      "step": 11680
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1480536521358801e-05,
+      "loss": 0.3564,
+      "step": 11690
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1473246828983817e-05,
+      "loss": 0.3231,
+      "step": 11700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1465957136608836e-05,
+      "loss": 0.3071,
+      "step": 11710
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1458667444233855e-05,
+      "loss": 0.3154,
+      "step": 11720
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1451377751858872e-05,
+      "loss": 0.3306,
+      "step": 11730
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1444088059483892e-05,
+      "loss": 0.3403,
+      "step": 11740
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.143679836710891e-05,
+      "loss": 0.3838,
+      "step": 11750
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1429508674733926e-05,
+      "loss": 0.3494,
+      "step": 11760
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1422218982358946e-05,
+      "loss": 0.3248,
+      "step": 11770
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1414929289983963e-05,
+      "loss": 0.3063,
+      "step": 11780
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1407639597608982e-05,
+      "loss": 0.37,
+      "step": 11790
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1400349905234001e-05,
+      "loss": 0.3113,
+      "step": 11800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1393060212859017e-05,
+      "loss": 0.3453,
+      "step": 11810
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1385770520484036e-05,
+      "loss": 0.3018,
+      "step": 11820
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1378480828109055e-05,
+      "loss": 0.3576,
+      "step": 11830
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1371191135734073e-05,
+      "loss": 0.3839,
+      "step": 11840
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.136390144335909e-05,
+      "loss": 0.3658,
+      "step": 11850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.135661175098411e-05,
+      "loss": 0.3136,
+      "step": 11860
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1349322058609127e-05,
+      "loss": 0.3557,
+      "step": 11870
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1342032366234146e-05,
+      "loss": 0.3242,
+      "step": 11880
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1334742673859165e-05,
+      "loss": 0.3945,
+      "step": 11890
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1327452981484181e-05,
+      "loss": 0.3312,
+      "step": 11900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.13201632891092e-05,
+      "loss": 0.3501,
+      "step": 11910
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.131287359673422e-05,
+      "loss": 0.3499,
+      "step": 11920
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1305583904359237e-05,
+      "loss": 0.3681,
+      "step": 11930
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1298294211984256e-05,
+      "loss": 0.3513,
+      "step": 11940
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1291004519609275e-05,
+      "loss": 0.3407,
+      "step": 11950
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.128371482723429e-05,
+      "loss": 0.3423,
+      "step": 11960
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.127642513485931e-05,
+      "loss": 0.3413,
+      "step": 11970
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1269135442484329e-05,
+      "loss": 0.3972,
+      "step": 11980
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1261845750109346e-05,
+      "loss": 0.3455,
+      "step": 11990
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1254556057734366e-05,
+      "loss": 0.3321,
+      "step": 12000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1247266365359381e-05,
+      "loss": 0.3967,
+      "step": 12010
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.12399766729844e-05,
+      "loss": 0.3418,
+      "step": 12020
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.123268698060942e-05,
+      "loss": 0.3381,
+      "step": 12030
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1225397288234437e-05,
+      "loss": 0.344,
+      "step": 12040
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1218107595859455e-05,
+      "loss": 0.3303,
+      "step": 12050
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1210817903484474e-05,
+      "loss": 0.3589,
+      "step": 12060
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1203528211109491e-05,
+      "loss": 0.3554,
+      "step": 12070
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.119623851873451e-05,
+      "loss": 0.3373,
+      "step": 12080
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.118894882635953e-05,
+      "loss": 0.3324,
+      "step": 12090
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1181659133984545e-05,
+      "loss": 0.3567,
+      "step": 12100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1174369441609564e-05,
+      "loss": 0.3584,
+      "step": 12110
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1167079749234584e-05,
+      "loss": 0.3127,
+      "step": 12120
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1159790056859601e-05,
+      "loss": 0.3213,
+      "step": 12130
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.115250036448462e-05,
+      "loss": 0.3446,
+      "step": 12140
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.114521067210964e-05,
+      "loss": 0.3611,
+      "step": 12150
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1137920979734655e-05,
+      "loss": 0.3185,
+      "step": 12160
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1130631287359674e-05,
+      "loss": 0.3069,
+      "step": 12170
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1123341594984693e-05,
+      "loss": 0.3846,
+      "step": 12180
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.111605190260971e-05,
+      "loss": 0.3082,
+      "step": 12190
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.110876221023473e-05,
+      "loss": 0.3532,
+      "step": 12200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1101472517859747e-05,
+      "loss": 0.3106,
+      "step": 12210
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1094182825484765e-05,
+      "loss": 0.3399,
+      "step": 12220
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1086893133109784e-05,
+      "loss": 0.3054,
+      "step": 12230
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1079603440734803e-05,
+      "loss": 0.366,
+      "step": 12240
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1072313748359819e-05,
+      "loss": 0.3368,
+      "step": 12250
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1065024055984838e-05,
+      "loss": 0.306,
+      "step": 12260
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1057734363609856e-05,
+      "loss": 0.3162,
+      "step": 12270
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1050444671234875e-05,
+      "loss": 0.3797,
+      "step": 12280
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1043154978859894e-05,
+      "loss": 0.3414,
+      "step": 12290
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.103586528648491e-05,
+      "loss": 0.3159,
+      "step": 12300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1028575594109929e-05,
+      "loss": 0.3772,
+      "step": 12310
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1021285901734948e-05,
+      "loss": 0.32,
+      "step": 12320
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1013996209359965e-05,
+      "loss": 0.2964,
+      "step": 12330
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1006706516984984e-05,
+      "loss": 0.3562,
+      "step": 12340
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0999416824610004e-05,
+      "loss": 0.357,
+      "step": 12350
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.099212713223502e-05,
+      "loss": 0.3564,
+      "step": 12360
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0984837439860039e-05,
+      "loss": 0.3903,
+      "step": 12370
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0977547747485058e-05,
+      "loss": 0.3207,
+      "step": 12380
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0970258055110075e-05,
+      "loss": 0.3399,
+      "step": 12390
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0962968362735094e-05,
+      "loss": 0.3728,
+      "step": 12400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0955678670360112e-05,
+      "loss": 0.3791,
+      "step": 12410
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.094838897798513e-05,
+      "loss": 0.3732,
+      "step": 12420
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0941099285610148e-05,
+      "loss": 0.3436,
+      "step": 12430
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0933809593235167e-05,
+      "loss": 0.3158,
+      "step": 12440
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0926519900860185e-05,
+      "loss": 0.363,
+      "step": 12450
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0919230208485202e-05,
+      "loss": 0.294,
+      "step": 12460
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0911940516110222e-05,
+      "loss": 0.3528,
+      "step": 12470
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.0904650823735239e-05,
+      "loss": 0.3185,
+      "step": 12480
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0897361131360258e-05,
+      "loss": 0.3165,
+      "step": 12490
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0890071438985274e-05,
+      "loss": 0.3239,
+      "step": 12500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0882781746610293e-05,
+      "loss": 0.3702,
+      "step": 12510
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0875492054235312e-05,
+      "loss": 0.3403,
+      "step": 12520
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.086820236186033e-05,
+      "loss": 0.3263,
+      "step": 12530
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0860912669485349e-05,
+      "loss": 0.3428,
+      "step": 12540
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0853622977110368e-05,
+      "loss": 0.3255,
+      "step": 12550
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0846333284735384e-05,
+      "loss": 0.3553,
+      "step": 12560
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.0839043592360403e-05,
+      "loss": 0.3405,
+      "step": 12570
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0831753899985422e-05,
+      "loss": 0.3726,
+      "step": 12580
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.082446420761044e-05,
+      "loss": 0.3248,
+      "step": 12590
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0817174515235459e-05,
+      "loss": 0.33,
+      "step": 12600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0809884822860478e-05,
+      "loss": 0.3692,
+      "step": 12610
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0802595130485493e-05,
+      "loss": 0.3292,
+      "step": 12620
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0795305438110513e-05,
+      "loss": 0.3573,
+      "step": 12630
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0788015745735532e-05,
+      "loss": 0.3312,
+      "step": 12640
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.078072605336055e-05,
+      "loss": 0.2995,
+      "step": 12650
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.0773436360985567e-05,
+      "loss": 0.3958,
+      "step": 12660
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0766146668610586e-05,
+      "loss": 0.3531,
+      "step": 12670
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0758856976235603e-05,
+      "loss": 0.3426,
+      "step": 12680
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0751567283860622e-05,
+      "loss": 0.3215,
+      "step": 12690
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0744277591485642e-05,
+      "loss": 0.3316,
+      "step": 12700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0736987899110657e-05,
+      "loss": 0.3287,
+      "step": 12710
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0729698206735676e-05,
+      "loss": 0.3655,
+      "step": 12720
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0722408514360696e-05,
+      "loss": 0.3317,
+      "step": 12730
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0715118821985713e-05,
+      "loss": 0.3141,
+      "step": 12740
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.0707829129610732e-05,
+      "loss": 0.3494,
+      "step": 12750
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0700539437235748e-05,
+      "loss": 0.3272,
+      "step": 12760
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0693249744860767e-05,
+      "loss": 0.3542,
+      "step": 12770
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0685960052485786e-05,
+      "loss": 0.3113,
+      "step": 12780
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0678670360110804e-05,
+      "loss": 0.3443,
+      "step": 12790
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0671380667735823e-05,
+      "loss": 0.3481,
+      "step": 12800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0664090975360842e-05,
+      "loss": 0.3683,
+      "step": 12810
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0656801282985858e-05,
+      "loss": 0.3503,
+      "step": 12820
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0649511590610877e-05,
+      "loss": 0.3656,
+      "step": 12830
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0642221898235896e-05,
+      "loss": 0.3805,
+      "step": 12840
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0634932205860914e-05,
+      "loss": 0.3165,
+      "step": 12850
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0627642513485931e-05,
+      "loss": 0.3025,
+      "step": 12860
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.062035282111095e-05,
+      "loss": 0.2748,
+      "step": 12870
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0613063128735968e-05,
+      "loss": 0.335,
+      "step": 12880
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0605773436360987e-05,
+      "loss": 0.3235,
+      "step": 12890
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0598483743986006e-05,
+      "loss": 0.3788,
+      "step": 12900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0591194051611022e-05,
+      "loss": 0.3403,
+      "step": 12910
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.058390435923604e-05,
+      "loss": 0.3325,
+      "step": 12920
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.057661466686106e-05,
+      "loss": 0.353,
+      "step": 12930
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0569324974486077e-05,
+      "loss": 0.3543,
+      "step": 12940
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0562035282111097e-05,
+      "loss": 0.4068,
+      "step": 12950
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0554745589736116e-05,
+      "loss": 0.3335,
+      "step": 12960
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0547455897361131e-05,
+      "loss": 0.3844,
+      "step": 12970
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.054016620498615e-05,
+      "loss": 0.3397,
+      "step": 12980
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0532876512611168e-05,
+      "loss": 0.3549,
+      "step": 12990
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0525586820236187e-05,
+      "loss": 0.3386,
+      "step": 13000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0518297127861206e-05,
+      "loss": 0.3179,
+      "step": 13010
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0511007435486222e-05,
+      "loss": 0.3447,
+      "step": 13020
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0503717743111241e-05,
+      "loss": 0.3295,
+      "step": 13030
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.049642805073626e-05,
+      "loss": 0.3124,
+      "step": 13040
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0489138358361278e-05,
+      "loss": 0.3725,
+      "step": 13050
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0481848665986295e-05,
+      "loss": 0.3553,
+      "step": 13060
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0474558973611314e-05,
+      "loss": 0.3372,
+      "step": 13070
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0467269281236332e-05,
+      "loss": 0.3214,
+      "step": 13080
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0459979588861351e-05,
+      "loss": 0.3574,
+      "step": 13090
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.045268989648637e-05,
+      "loss": 0.3967,
+      "step": 13100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0445400204111386e-05,
+      "loss": 0.3306,
+      "step": 13110
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0438110511736405e-05,
+      "loss": 0.3286,
+      "step": 13120
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0430820819361424e-05,
+      "loss": 0.3437,
+      "step": 13130
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0423531126986442e-05,
+      "loss": 0.3465,
+      "step": 13140
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0416241434611461e-05,
+      "loss": 0.3331,
+      "step": 13150
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.040895174223648e-05,
+      "loss": 0.3135,
+      "step": 13160
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0401662049861496e-05,
+      "loss": 0.3029,
+      "step": 13170
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0394372357486515e-05,
+      "loss": 0.3671,
+      "step": 13180
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0387082665111534e-05,
+      "loss": 0.3239,
+      "step": 13190
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0379792972736552e-05,
+      "loss": 0.3657,
+      "step": 13200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.037250328036157e-05,
+      "loss": 0.3271,
+      "step": 13210
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0365213587986586e-05,
+      "loss": 0.3368,
+      "step": 13220
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0357923895611606e-05,
+      "loss": 0.3765,
+      "step": 13230
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0350634203236625e-05,
+      "loss": 0.3364,
+      "step": 13240
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0343344510861642e-05,
+      "loss": 0.3463,
+      "step": 13250
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0336054818486661e-05,
+      "loss": 0.353,
+      "step": 13260
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0328765126111679e-05,
+      "loss": 0.3458,
+      "step": 13270
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0321475433736696e-05,
+      "loss": 0.3451,
+      "step": 13280
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0314185741361715e-05,
+      "loss": 0.4082,
+      "step": 13290
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0306896048986735e-05,
+      "loss": 0.3257,
+      "step": 13300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.029960635661175e-05,
+      "loss": 0.2903,
+      "step": 13310
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.029231666423677e-05,
+      "loss": 0.3154,
+      "step": 13320
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0285026971861789e-05,
+      "loss": 0.3696,
+      "step": 13330
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0277737279486806e-05,
+      "loss": 0.303,
+      "step": 13340
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0270447587111825e-05,
+      "loss": 0.3655,
+      "step": 13350
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0263157894736844e-05,
+      "loss": 0.3634,
+      "step": 13360
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.025586820236186e-05,
+      "loss": 0.311,
+      "step": 13370
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.024857850998688e-05,
+      "loss": 0.3493,
+      "step": 13380
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0241288817611898e-05,
+      "loss": 0.2892,
+      "step": 13390
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0233999125236916e-05,
+      "loss": 0.383,
+      "step": 13400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0226709432861935e-05,
+      "loss": 0.3042,
+      "step": 13410
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0219419740486954e-05,
+      "loss": 0.364,
+      "step": 13420
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.021213004811197e-05,
+      "loss": 0.3427,
+      "step": 13430
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0204840355736989e-05,
+      "loss": 0.3686,
+      "step": 13440
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0197550663362008e-05,
+      "loss": 0.3916,
+      "step": 13450
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0190260970987026e-05,
+      "loss": 0.372,
+      "step": 13460
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0182971278612043e-05,
+      "loss": 0.3256,
+      "step": 13470
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.017568158623706e-05,
+      "loss": 0.3863,
+      "step": 13480
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.016839189386208e-05,
+      "loss": 0.3848,
+      "step": 13490
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0161102201487099e-05,
+      "loss": 0.3542,
+      "step": 13500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0153812509112115e-05,
+      "loss": 0.3501,
+      "step": 13510
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0146522816737134e-05,
+      "loss": 0.369,
+      "step": 13520
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0139233124362153e-05,
+      "loss": 0.3983,
+      "step": 13530
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.013194343198717e-05,
+      "loss": 0.2944,
+      "step": 13540
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.012465373961219e-05,
+      "loss": 0.2564,
+      "step": 13550
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0117364047237209e-05,
+      "loss": 0.3319,
+      "step": 13560
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0110074354862224e-05,
+      "loss": 0.3801,
+      "step": 13570
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0102784662487244e-05,
+      "loss": 0.2913,
+      "step": 13580
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0095494970112263e-05,
+      "loss": 0.3156,
+      "step": 13590
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.008820527773728e-05,
+      "loss": 0.3335,
+      "step": 13600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.00809155853623e-05,
+      "loss": 0.3485,
+      "step": 13610
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0073625892987318e-05,
+      "loss": 0.3431,
+      "step": 13620
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0066336200612334e-05,
+      "loss": 0.3268,
+      "step": 13630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0059046508237353e-05,
+      "loss": 0.3465,
+      "step": 13640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0051756815862372e-05,
+      "loss": 0.3392,
+      "step": 13650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.004446712348739e-05,
+      "loss": 0.3242,
+      "step": 13660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0037177431112407e-05,
+      "loss": 0.3208,
+      "step": 13670
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0029887738737427e-05,
+      "loss": 0.338,
+      "step": 13680
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0022598046362444e-05,
+      "loss": 0.3583,
+      "step": 13690
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0015308353987463e-05,
+      "loss": 0.3003,
+      "step": 13700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0008018661612479e-05,
+      "loss": 0.3645,
+      "step": 13710
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0000728969237498e-05,
+      "loss": 0.3498,
+      "step": 13720
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.993439276862517e-06,
+      "loss": 0.3268,
+      "step": 13730
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.986149584487536e-06,
+      "loss": 0.2929,
+      "step": 13740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.978859892112554e-06,
+      "loss": 0.3107,
+      "step": 13750
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.971570199737571e-06,
+      "loss": 0.3368,
+      "step": 13760
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.96428050736259e-06,
+      "loss": 0.3314,
+      "step": 13770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.956990814987608e-06,
+      "loss": 0.3263,
+      "step": 13780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.949701122612625e-06,
+      "loss": 0.3251,
+      "step": 13790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.942411430237644e-06,
+      "loss": 0.3213,
+      "step": 13800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.935121737862664e-06,
+      "loss": 0.2802,
+      "step": 13810
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.927832045487681e-06,
+      "loss": 0.3556,
+      "step": 13820
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.9205423531127e-06,
+      "loss": 0.3672,
+      "step": 13830
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.913252660737718e-06,
+      "loss": 0.3367,
+      "step": 13840
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.905962968362735e-06,
+      "loss": 0.3621,
+      "step": 13850
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.898673275987754e-06,
+      "loss": 0.3574,
+      "step": 13860
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.891383583612772e-06,
+      "loss": 0.331,
+      "step": 13870
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.884093891237791e-06,
+      "loss": 0.3467,
+      "step": 13880
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.876804198862808e-06,
+      "loss": 0.2879,
+      "step": 13890
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.869514506487827e-06,
+      "loss": 0.3329,
+      "step": 13900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.862224814112845e-06,
+      "loss": 0.3158,
+      "step": 13910
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.854935121737862e-06,
+      "loss": 0.3375,
+      "step": 13920
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.847645429362882e-06,
+      "loss": 0.3275,
+      "step": 13930
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.8403557369879e-06,
+      "loss": 0.3602,
+      "step": 13940
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.833066044612918e-06,
+      "loss": 0.311,
+      "step": 13950
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.825776352237937e-06,
+      "loss": 0.3258,
+      "step": 13960
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.818486659862955e-06,
+      "loss": 0.3551,
+      "step": 13970
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.811196967487972e-06,
+      "loss": 0.3543,
+      "step": 13980
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.803907275112991e-06,
+      "loss": 0.3193,
+      "step": 13990
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.796617582738009e-06,
+      "loss": 0.3516,
+      "step": 14000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.789327890363028e-06,
+      "loss": 0.368,
+      "step": 14010
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.782038197988045e-06,
+      "loss": 0.385,
+      "step": 14020
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.774748505613065e-06,
+      "loss": 0.3062,
+      "step": 14030
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.767458813238082e-06,
+      "loss": 0.3485,
+      "step": 14040
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.7601691208631e-06,
+      "loss": 0.3527,
+      "step": 14050
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.752879428488119e-06,
+      "loss": 0.332,
+      "step": 14060
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.745589736113136e-06,
+      "loss": 0.3341,
+      "step": 14070
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.738300043738155e-06,
+      "loss": 0.3727,
+      "step": 14080
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.731010351363174e-06,
+      "loss": 0.3437,
+      "step": 14090
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.723720658988192e-06,
+      "loss": 0.3638,
+      "step": 14100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.71643096661321e-06,
+      "loss": 0.3292,
+      "step": 14110
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.709141274238227e-06,
+      "loss": 0.3424,
+      "step": 14120
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.701851581863246e-06,
+      "loss": 0.3726,
+      "step": 14130
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.694561889488265e-06,
+      "loss": 0.369,
+      "step": 14140
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.687272197113282e-06,
+      "loss": 0.2965,
+      "step": 14150
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.679982504738302e-06,
+      "loss": 0.3255,
+      "step": 14160
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.672692812363319e-06,
+      "loss": 0.3337,
+      "step": 14170
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.665403119988336e-06,
+      "loss": 0.3549,
+      "step": 14180
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.658113427613356e-06,
+      "loss": 0.3613,
+      "step": 14190
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.650823735238373e-06,
+      "loss": 0.3311,
+      "step": 14200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.643534042863392e-06,
+      "loss": 0.3392,
+      "step": 14210
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.636244350488411e-06,
+      "loss": 0.3264,
+      "step": 14220
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.628954658113429e-06,
+      "loss": 0.3455,
+      "step": 14230
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.621664965738446e-06,
+      "loss": 0.3663,
+      "step": 14240
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.614375273363464e-06,
+      "loss": 0.3189,
+      "step": 14250
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.607085580988483e-06,
+      "loss": 0.2985,
+      "step": 14260
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.599795888613502e-06,
+      "loss": 0.3859,
+      "step": 14270
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.59250619623852e-06,
+      "loss": 0.3147,
+      "step": 14280
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.585216503863539e-06,
+      "loss": 0.3233,
+      "step": 14290
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.577926811488556e-06,
+      "loss": 0.3671,
+      "step": 14300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.570637119113574e-06,
+      "loss": 0.3348,
+      "step": 14310
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.563347426738593e-06,
+      "loss": 0.3476,
+      "step": 14320
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.55605773436361e-06,
+      "loss": 0.3483,
+      "step": 14330
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.54876804198863e-06,
+      "loss": 0.3244,
+      "step": 14340
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.541478349613648e-06,
+      "loss": 0.3229,
+      "step": 14350
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.534188657238666e-06,
+      "loss": 0.3743,
+      "step": 14360
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.526898964863683e-06,
+      "loss": 0.3673,
+      "step": 14370
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.5196092724887e-06,
+      "loss": 0.3331,
+      "step": 14380
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.51231958011372e-06,
+      "loss": 0.3168,
+      "step": 14390
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.505029887738737e-06,
+      "loss": 0.3474,
+      "step": 14400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.497740195363757e-06,
+      "loss": 0.35,
+      "step": 14410
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.490450502988776e-06,
+      "loss": 0.3203,
+      "step": 14420
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.483160810613793e-06,
+      "loss": 0.3358,
+      "step": 14430
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.47587111823881e-06,
+      "loss": 0.3677,
+      "step": 14440
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.46858142586383e-06,
+      "loss": 0.3171,
+      "step": 14450
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.461291733488847e-06,
+      "loss": 0.393,
+      "step": 14460
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.454002041113866e-06,
+      "loss": 0.2796,
+      "step": 14470
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.446712348738884e-06,
+      "loss": 0.3251,
+      "step": 14480
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.439422656363903e-06,
+      "loss": 0.3212,
+      "step": 14490
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.43213296398892e-06,
+      "loss": 0.3823,
+      "step": 14500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.424843271613938e-06,
+      "loss": 0.3276,
+      "step": 14510
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.417553579238957e-06,
+      "loss": 0.3582,
+      "step": 14520
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.410263886863974e-06,
+      "loss": 0.3109,
+      "step": 14530
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.402974194488994e-06,
+      "loss": 0.3227,
+      "step": 14540
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.395684502114013e-06,
+      "loss": 0.344,
+      "step": 14550
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.38839480973903e-06,
+      "loss": 0.3771,
+      "step": 14560
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.381105117364048e-06,
+      "loss": 0.3458,
+      "step": 14570
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.373815424989067e-06,
+      "loss": 0.3033,
+      "step": 14580
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.366525732614084e-06,
+      "loss": 0.3309,
+      "step": 14590
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.359236040239102e-06,
+      "loss": 0.3153,
+      "step": 14600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.351946347864121e-06,
+      "loss": 0.3104,
+      "step": 14610
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.34465665548914e-06,
+      "loss": 0.3224,
+      "step": 14620
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.337366963114157e-06,
+      "loss": 0.3479,
+      "step": 14630
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.330077270739175e-06,
+      "loss": 0.3509,
+      "step": 14640
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.322787578364194e-06,
+      "loss": 0.3484,
+      "step": 14650
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.315497885989212e-06,
+      "loss": 0.3426,
+      "step": 14660
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.30820819361423e-06,
+      "loss": 0.3299,
+      "step": 14670
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.300918501239248e-06,
+      "loss": 0.3561,
+      "step": 14680
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.293628808864267e-06,
+      "loss": 0.2985,
+      "step": 14690
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.286339116489285e-06,
+      "loss": 0.3413,
+      "step": 14700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.279049424114304e-06,
+      "loss": 0.301,
+      "step": 14710
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.271759731739321e-06,
+      "loss": 0.3407,
+      "step": 14720
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.264470039364339e-06,
+      "loss": 0.2984,
+      "step": 14730
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.257180346989358e-06,
+      "loss": 0.3811,
+      "step": 14740
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.249890654614377e-06,
+      "loss": 0.3129,
+      "step": 14750
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.242600962239395e-06,
+      "loss": 0.3044,
+      "step": 14760
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.235311269864412e-06,
+      "loss": 0.31,
+      "step": 14770
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.228021577489431e-06,
+      "loss": 0.3099,
+      "step": 14780
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.220731885114449e-06,
+      "loss": 0.3213,
+      "step": 14790
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.213442192739466e-06,
+      "loss": 0.318,
+      "step": 14800
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.206152500364485e-06,
+      "loss": 0.328,
+      "step": 14810
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.198862807989504e-06,
+      "loss": 0.3206,
+      "step": 14820
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.191573115614522e-06,
+      "loss": 0.3609,
+      "step": 14830
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.184283423239541e-06,
+      "loss": 0.4013,
+      "step": 14840
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.176993730864558e-06,
+      "loss": 0.3156,
+      "step": 14850
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.169704038489576e-06,
+      "loss": 0.2948,
+      "step": 14860
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.162414346114595e-06,
+      "loss": 0.2858,
+      "step": 14870
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.155124653739612e-06,
+      "loss": 0.3302,
+      "step": 14880
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.147834961364632e-06,
+      "loss": 0.3557,
+      "step": 14890
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.140545268989649e-06,
+      "loss": 0.3328,
+      "step": 14900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.133255576614668e-06,
+      "loss": 0.3054,
+      "step": 14910
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.125965884239686e-06,
+      "loss": 0.3224,
+      "step": 14920
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.118676191864703e-06,
+      "loss": 0.3359,
+      "step": 14930
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.111386499489722e-06,
+      "loss": 0.3263,
+      "step": 14940
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.104096807114741e-06,
+      "loss": 0.3856,
+      "step": 14950
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.096807114739759e-06,
+      "loss": 0.3347,
+      "step": 14960
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.089517422364776e-06,
+      "loss": 0.3331,
+      "step": 14970
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.082227729989795e-06,
+      "loss": 0.3209,
+      "step": 14980
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.074938037614813e-06,
+      "loss": 0.3194,
+      "step": 14990
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.06764834523983e-06,
+      "loss": 0.3377,
+      "step": 15000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.06035865286485e-06,
+      "loss": 0.3155,
+      "step": 15010
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.053068960489869e-06,
+      "loss": 0.365,
+      "step": 15020
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.045779268114886e-06,
+      "loss": 0.3391,
+      "step": 15030
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.038489575739905e-06,
+      "loss": 0.3316,
+      "step": 15040
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.031199883364923e-06,
+      "loss": 0.2979,
+      "step": 15050
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.02391019098994e-06,
+      "loss": 0.2898,
+      "step": 15060
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.01662049861496e-06,
+      "loss": 0.3304,
+      "step": 15070
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.009330806239977e-06,
+      "loss": 0.3505,
+      "step": 15080
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.002041113864996e-06,
+      "loss": 0.3357,
+      "step": 15090
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.994751421490013e-06,
+      "loss": 0.3817,
+      "step": 15100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.987461729115032e-06,
+      "loss": 0.3538,
+      "step": 15110
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.98017203674005e-06,
+      "loss": 0.3472,
+      "step": 15120
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.972882344365067e-06,
+      "loss": 0.3559,
+      "step": 15130
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.965592651990087e-06,
+      "loss": 0.3443,
+      "step": 15140
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.958302959615106e-06,
+      "loss": 0.31,
+      "step": 15150
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.951013267240123e-06,
+      "loss": 0.3084,
+      "step": 15160
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.943723574865142e-06,
+      "loss": 0.3259,
+      "step": 15170
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.93643388249016e-06,
+      "loss": 0.3464,
+      "step": 15180
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.929144190115177e-06,
+      "loss": 0.3128,
+      "step": 15190
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.921854497740196e-06,
+      "loss": 0.3604,
+      "step": 15200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.914564805365214e-06,
+      "loss": 0.3112,
+      "step": 15210
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.907275112990233e-06,
+      "loss": 0.3894,
+      "step": 15220
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.89998542061525e-06,
+      "loss": 0.3334,
+      "step": 15230
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.89269572824027e-06,
+      "loss": 0.3233,
+      "step": 15240
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.885406035865287e-06,
+      "loss": 0.2991,
+      "step": 15250
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.878116343490304e-06,
+      "loss": 0.3564,
+      "step": 15260
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.870826651115324e-06,
+      "loss": 0.3045,
+      "step": 15270
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.863536958740343e-06,
+      "loss": 0.3536,
+      "step": 15280
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.85624726636536e-06,
+      "loss": 0.3396,
+      "step": 15290
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.84895757399038e-06,
+      "loss": 0.3418,
+      "step": 15300
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.841667881615397e-06,
+      "loss": 0.3305,
+      "step": 15310
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.834378189240414e-06,
+      "loss": 0.3368,
+      "step": 15320
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.827088496865432e-06,
+      "loss": 0.3692,
+      "step": 15330
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.819798804490451e-06,
+      "loss": 0.3254,
+      "step": 15340
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.81250911211547e-06,
+      "loss": 0.3349,
+      "step": 15350
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.805219419740487e-06,
+      "loss": 0.2968,
+      "step": 15360
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.797929727365507e-06,
+      "loss": 0.3243,
+      "step": 15370
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.790640034990524e-06,
+      "loss": 0.3392,
+      "step": 15380
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.783350342615542e-06,
+      "loss": 0.3306,
+      "step": 15390
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.77606065024056e-06,
+      "loss": 0.34,
+      "step": 15400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.768770957865578e-06,
+      "loss": 0.3364,
+      "step": 15410
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.761481265490597e-06,
+      "loss": 0.339,
+      "step": 15420
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.754191573115616e-06,
+      "loss": 0.305,
+      "step": 15430
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.746901880740634e-06,
+      "loss": 0.3017,
+      "step": 15440
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.739612188365651e-06,
+      "loss": 0.3141,
+      "step": 15450
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.732322495990669e-06,
+      "loss": 0.2872,
+      "step": 15460
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.725032803615688e-06,
+      "loss": 0.3216,
+      "step": 15470
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.717743111240707e-06,
+      "loss": 0.3337,
+      "step": 15480
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.710453418865725e-06,
+      "loss": 0.3551,
+      "step": 15490
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.703163726490744e-06,
+      "loss": 0.3659,
+      "step": 15500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.695874034115761e-06,
+      "loss": 0.336,
+      "step": 15510
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.688584341740779e-06,
+      "loss": 0.3519,
+      "step": 15520
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.681294649365798e-06,
+      "loss": 0.3289,
+      "step": 15530
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.674004956990815e-06,
+      "loss": 0.3091,
+      "step": 15540
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.666715264615834e-06,
+      "loss": 0.3676,
+      "step": 15550
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.659425572240853e-06,
+      "loss": 0.3106,
+      "step": 15560
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.652135879865871e-06,
+      "loss": 0.2958,
+      "step": 15570
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.644846187490888e-06,
+      "loss": 0.297,
+      "step": 15580
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.637556495115906e-06,
+      "loss": 0.355,
+      "step": 15590
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.630266802740925e-06,
+      "loss": 0.2994,
+      "step": 15600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.622977110365942e-06,
+      "loss": 0.3371,
+      "step": 15610
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.615687417990962e-06,
+      "loss": 0.297,
+      "step": 15620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.60839772561598e-06,
+      "loss": 0.2966,
+      "step": 15630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.601108033240998e-06,
+      "loss": 0.3311,
+      "step": 15640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.593818340866016e-06,
+      "loss": 0.3683,
+      "step": 15650
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.586528648491035e-06,
+      "loss": 0.3299,
+      "step": 15660
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.579238956116052e-06,
+      "loss": 0.3286,
+      "step": 15670
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.571949263741071e-06,
+      "loss": 0.3527,
+      "step": 15680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.564659571366089e-06,
+      "loss": 0.3189,
+      "step": 15690
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.557369878991108e-06,
+      "loss": 0.3113,
+      "step": 15700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.550080186616125e-06,
+      "loss": 0.3605,
+      "step": 15710
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.542790494241143e-06,
+      "loss": 0.3182,
+      "step": 15720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.535500801866162e-06,
+      "loss": 0.3619,
+      "step": 15730
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.52821110949118e-06,
+      "loss": 0.3408,
+      "step": 15740
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.520921417116199e-06,
+      "loss": 0.3295,
+      "step": 15750
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.513631724741218e-06,
+      "loss": 0.3567,
+      "step": 15760
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.506342032366235e-06,
+      "loss": 0.336,
+      "step": 15770
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.499052339991253e-06,
+      "loss": 0.3732,
+      "step": 15780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.491762647616272e-06,
+      "loss": 0.3162,
+      "step": 15790
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.48447295524129e-06,
+      "loss": 0.3433,
+      "step": 15800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.477183262866307e-06,
+      "loss": 0.3709,
+      "step": 15810
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.469893570491326e-06,
+      "loss": 0.4008,
+      "step": 15820
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.462603878116345e-06,
+      "loss": 0.3382,
+      "step": 15830
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.455314185741362e-06,
+      "loss": 0.3291,
+      "step": 15840
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.44802449336638e-06,
+      "loss": 0.3086,
+      "step": 15850
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.440734800991399e-06,
+      "loss": 0.3798,
+      "step": 15860
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.433445108616417e-06,
+      "loss": 0.2931,
+      "step": 15870
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.426155416241436e-06,
+      "loss": 0.323,
+      "step": 15880
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.418865723866453e-06,
+      "loss": 0.3665,
+      "step": 15890
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.411576031491472e-06,
+      "loss": 0.2819,
+      "step": 15900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.40428633911649e-06,
+      "loss": 0.324,
+      "step": 15910
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.396996646741509e-06,
+      "loss": 0.2896,
+      "step": 15920
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.389706954366526e-06,
+      "loss": 0.3071,
+      "step": 15930
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.382417261991544e-06,
+      "loss": 0.3459,
+      "step": 15940
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.375127569616563e-06,
+      "loss": 0.3201,
+      "step": 15950
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.367837877241582e-06,
+      "loss": 0.3321,
+      "step": 15960
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.3605481848666e-06,
+      "loss": 0.3354,
+      "step": 15970
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.353258492491617e-06,
+      "loss": 0.3405,
+      "step": 15980
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.345968800116636e-06,
+      "loss": 0.3391,
+      "step": 15990
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.338679107741654e-06,
+      "loss": 0.3164,
+      "step": 16000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.331389415366673e-06,
+      "loss": 0.351,
+      "step": 16010
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.32409972299169e-06,
+      "loss": 0.3356,
+      "step": 16020
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.31681003061671e-06,
+      "loss": 0.2989,
+      "step": 16030
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.309520338241727e-06,
+      "loss": 0.3295,
+      "step": 16040
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.302230645866746e-06,
+      "loss": 0.3804,
+      "step": 16050
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.294940953491763e-06,
+      "loss": 0.3326,
+      "step": 16060
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.28765126111678e-06,
+      "loss": 0.3312,
+      "step": 16070
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.2803615687418e-06,
+      "loss": 0.3397,
+      "step": 16080
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.273071876366819e-06,
+      "loss": 0.3339,
+      "step": 16090
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.265782183991837e-06,
+      "loss": 0.3421,
+      "step": 16100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.258492491616854e-06,
+      "loss": 0.3165,
+      "step": 16110
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.251202799241873e-06,
+      "loss": 0.3453,
+      "step": 16120
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.24391310686689e-06,
+      "loss": 0.3286,
+      "step": 16130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.236623414491908e-06,
+      "loss": 0.3132,
+      "step": 16140
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.229333722116927e-06,
+      "loss": 0.3139,
+      "step": 16150
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.222044029741946e-06,
+      "loss": 0.3295,
+      "step": 16160
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.214754337366964e-06,
+      "loss": 0.3484,
+      "step": 16170
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.207464644991981e-06,
+      "loss": 0.3459,
+      "step": 16180
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.200174952617e-06,
+      "loss": 0.3809,
+      "step": 16190
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.192885260242018e-06,
+      "loss": 0.3262,
+      "step": 16200
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.185595567867037e-06,
+      "loss": 0.3202,
+      "step": 16210
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.178305875492055e-06,
+      "loss": 0.3217,
+      "step": 16220
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.171016183117074e-06,
+      "loss": 0.3281,
+      "step": 16230
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.163726490742091e-06,
+      "loss": 0.3349,
+      "step": 16240
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.15643679836711e-06,
+      "loss": 0.3231,
+      "step": 16250
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.149147105992128e-06,
+      "loss": 0.3224,
+      "step": 16260
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.141857413617145e-06,
+      "loss": 0.3356,
+      "step": 16270
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.134567721242164e-06,
+      "loss": 0.3057,
+      "step": 16280
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.127278028867183e-06,
+      "loss": 0.342,
+      "step": 16290
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.119988336492201e-06,
+      "loss": 0.3643,
+      "step": 16300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.112698644117218e-06,
+      "loss": 0.3352,
+      "step": 16310
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.105408951742238e-06,
+      "loss": 0.2953,
+      "step": 16320
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.098119259367255e-06,
+      "loss": 0.329,
+      "step": 16330
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.090829566992272e-06,
+      "loss": 0.3471,
+      "step": 16340
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.083539874617292e-06,
+      "loss": 0.3299,
+      "step": 16350
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.07625018224231e-06,
+      "loss": 0.3145,
+      "step": 16360
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.068960489867328e-06,
+      "loss": 0.3229,
+      "step": 16370
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.061670797492347e-06,
+      "loss": 0.4179,
+      "step": 16380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.054381105117365e-06,
+      "loss": 0.3292,
+      "step": 16390
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.047091412742382e-06,
+      "loss": 0.3373,
+      "step": 16400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.039801720367401e-06,
+      "loss": 0.2696,
+      "step": 16410
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.032512027992419e-06,
+      "loss": 0.3514,
+      "step": 16420
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.025222335617438e-06,
+      "loss": 0.2725,
+      "step": 16430
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.017932643242455e-06,
+      "loss": 0.3151,
+      "step": 16440
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.010642950867475e-06,
+      "loss": 0.3352,
+      "step": 16450
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.003353258492492e-06,
+      "loss": 0.3024,
+      "step": 16460
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.99606356611751e-06,
+      "loss": 0.3294,
+      "step": 16470
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.988773873742529e-06,
+      "loss": 0.3695,
+      "step": 16480
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.981484181367548e-06,
+      "loss": 0.3632,
+      "step": 16490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.974194488992565e-06,
+      "loss": 0.31,
+      "step": 16500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.966904796617584e-06,
+      "loss": 0.3294,
+      "step": 16510
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.959615104242602e-06,
+      "loss": 0.3158,
+      "step": 16520
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.95232541186762e-06,
+      "loss": 0.3365,
+      "step": 16530
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.945035719492637e-06,
+      "loss": 0.3529,
+      "step": 16540
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.937746027117656e-06,
+      "loss": 0.3223,
+      "step": 16550
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.930456334742675e-06,
+      "loss": 0.3341,
+      "step": 16560
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.923166642367692e-06,
+      "loss": 0.3373,
+      "step": 16570
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.915876949992712e-06,
+      "loss": 0.2813,
+      "step": 16580
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.908587257617729e-06,
+      "loss": 0.3108,
+      "step": 16590
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.901297565242747e-06,
+      "loss": 0.3044,
+      "step": 16600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.894007872867766e-06,
+      "loss": 0.3305,
+      "step": 16610
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.886718180492783e-06,
+      "loss": 0.3149,
+      "step": 16620
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.879428488117802e-06,
+      "loss": 0.3419,
+      "step": 16630
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.872138795742821e-06,
+      "loss": 0.3298,
+      "step": 16640
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.864849103367839e-06,
+      "loss": 0.3316,
+      "step": 16650
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.857559410992856e-06,
+      "loss": 0.2906,
+      "step": 16660
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.850269718617874e-06,
+      "loss": 0.3007,
+      "step": 16670
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.842980026242893e-06,
+      "loss": 0.3173,
+      "step": 16680
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.835690333867912e-06,
+      "loss": 0.2985,
+      "step": 16690
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.82840064149293e-06,
+      "loss": 0.3021,
+      "step": 16700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.821110949117949e-06,
+      "loss": 0.3238,
+      "step": 16710
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.813821256742966e-06,
+      "loss": 0.3508,
+      "step": 16720
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.806531564367984e-06,
+      "loss": 0.3335,
+      "step": 16730
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.799241871993003e-06,
+      "loss": 0.2531,
+      "step": 16740
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.79195217961802e-06,
+      "loss": 0.361,
+      "step": 16750
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.78466248724304e-06,
+      "loss": 0.342,
+      "step": 16760
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.777372794868058e-06,
+      "loss": 0.3044,
+      "step": 16770
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.770083102493076e-06,
+      "loss": 0.2813,
+      "step": 16780
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.762793410118093e-06,
+      "loss": 0.2928,
+      "step": 16790
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.75550371774311e-06,
+      "loss": 0.3103,
+      "step": 16800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.74821402536813e-06,
+      "loss": 0.2882,
+      "step": 16810
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.740924332993147e-06,
+      "loss": 0.3439,
+      "step": 16820
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.733634640618167e-06,
+      "loss": 0.3109,
+      "step": 16830
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.726344948243186e-06,
+      "loss": 0.3201,
+      "step": 16840
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.719055255868203e-06,
+      "loss": 0.3181,
+      "step": 16850
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.71176556349322e-06,
+      "loss": 0.3124,
+      "step": 16860
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.70447587111824e-06,
+      "loss": 0.3386,
+      "step": 16870
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.697186178743257e-06,
+      "loss": 0.2828,
+      "step": 16880
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.689896486368276e-06,
+      "loss": 0.3451,
+      "step": 16890
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.682606793993294e-06,
+      "loss": 0.3229,
+      "step": 16900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.675317101618313e-06,
+      "loss": 0.3496,
+      "step": 16910
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.66802740924333e-06,
+      "loss": 0.3259,
+      "step": 16920
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.660737716868348e-06,
+      "loss": 0.2879,
+      "step": 16930
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.653448024493367e-06,
+      "loss": 0.3225,
+      "step": 16940
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.646158332118384e-06,
+      "loss": 0.3302,
+      "step": 16950
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.638868639743404e-06,
+      "loss": 0.2826,
+      "step": 16960
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.631578947368423e-06,
+      "loss": 0.3275,
+      "step": 16970
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.62428925499344e-06,
+      "loss": 0.3918,
+      "step": 16980
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.616999562618458e-06,
+      "loss": 0.3702,
+      "step": 16990
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.609709870243477e-06,
+      "loss": 0.3636,
+      "step": 17000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.602420177868495e-06,
+      "loss": 0.3993,
+      "step": 17010
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.595130485493513e-06,
+      "loss": 0.3173,
+      "step": 17020
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.587840793118531e-06,
+      "loss": 0.2902,
+      "step": 17030
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.580551100743549e-06,
+      "loss": 0.3217,
+      "step": 17040
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.5732614083685675e-06,
+      "loss": 0.3052,
+      "step": 17050
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.565971715993585e-06,
+      "loss": 0.3175,
+      "step": 17060
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.558682023618604e-06,
+      "loss": 0.35,
+      "step": 17070
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.551392331243622e-06,
+      "loss": 0.3089,
+      "step": 17080
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.54410263886864e-06,
+      "loss": 0.3122,
+      "step": 17090
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.536812946493659e-06,
+      "loss": 0.3519,
+      "step": 17100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.529523254118677e-06,
+      "loss": 0.3457,
+      "step": 17110
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.522233561743695e-06,
+      "loss": 0.3108,
+      "step": 17120
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.514943869368714e-06,
+      "loss": 0.3196,
+      "step": 17130
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.507654176993731e-06,
+      "loss": 0.348,
+      "step": 17140
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.50036448461875e-06,
+      "loss": 0.3381,
+      "step": 17150
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.493074792243767e-06,
+      "loss": 0.3112,
+      "step": 17160
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.485785099868786e-06,
+      "loss": 0.3416,
+      "step": 17170
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.4784954074938046e-06,
+      "loss": 0.3518,
+      "step": 17180
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.471205715118822e-06,
+      "loss": 0.2901,
+      "step": 17190
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.463916022743841e-06,
+      "loss": 0.3279,
+      "step": 17200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.4566263303688595e-06,
+      "loss": 0.3212,
+      "step": 17210
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.449336637993877e-06,
+      "loss": 0.325,
+      "step": 17220
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.442046945618896e-06,
+      "loss": 0.3453,
+      "step": 17230
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.4347572532439135e-06,
+      "loss": 0.3018,
+      "step": 17240
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.427467560868932e-06,
+      "loss": 0.3229,
+      "step": 17250
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.420177868493949e-06,
+      "loss": 0.3125,
+      "step": 17260
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.412888176118968e-06,
+      "loss": 0.294,
+      "step": 17270
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.405598483743987e-06,
+      "loss": 0.3352,
+      "step": 17280
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.398308791369004e-06,
+      "loss": 0.2799,
+      "step": 17290
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.391019098994023e-06,
+      "loss": 0.3076,
+      "step": 17300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.383729406619042e-06,
+      "loss": 0.401,
+      "step": 17310
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.376439714244059e-06,
+      "loss": 0.331,
+      "step": 17320
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.369150021869078e-06,
+      "loss": 0.329,
+      "step": 17330
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.361860329494096e-06,
+      "loss": 0.3311,
+      "step": 17340
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.354570637119114e-06,
+      "loss": 0.3524,
+      "step": 17350
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.347280944744133e-06,
+      "loss": 0.326,
+      "step": 17360
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.3399912523691506e-06,
+      "loss": 0.3263,
+      "step": 17370
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.332701559994169e-06,
+      "loss": 0.3604,
+      "step": 17380
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.325411867619186e-06,
+      "loss": 0.2812,
+      "step": 17390
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.3181221752442055e-06,
+      "loss": 0.3298,
+      "step": 17400
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.310832482869224e-06,
+      "loss": 0.3266,
+      "step": 17410
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.303542790494241e-06,
+      "loss": 0.3386,
+      "step": 17420
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.29625309811926e-06,
+      "loss": 0.3518,
+      "step": 17430
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.288963405744278e-06,
+      "loss": 0.3821,
+      "step": 17440
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.281673713369296e-06,
+      "loss": 0.3122,
+      "step": 17450
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.274384020994315e-06,
+      "loss": 0.3335,
+      "step": 17460
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.267094328619333e-06,
+      "loss": 0.3202,
+      "step": 17470
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.259804636244351e-06,
+      "loss": 0.3139,
+      "step": 17480
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.25251494386937e-06,
+      "loss": 0.3185,
+      "step": 17490
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.245225251494388e-06,
+      "loss": 0.2882,
+      "step": 17500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.237935559119406e-06,
+      "loss": 0.3439,
+      "step": 17510
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.230645866744423e-06,
+      "loss": 0.3069,
+      "step": 17520
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.2233561743694425e-06,
+      "loss": 0.2746,
+      "step": 17530
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.21606648199446e-06,
+      "loss": 0.2752,
+      "step": 17540
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.208776789619478e-06,
+      "loss": 0.3128,
+      "step": 17550
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.2014870972444974e-06,
+      "loss": 0.3646,
+      "step": 17560
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.194197404869515e-06,
+      "loss": 0.2768,
+      "step": 17570
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.186907712494533e-06,
+      "loss": 0.2908,
+      "step": 17580
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.179618020119552e-06,
+      "loss": 0.3049,
+      "step": 17590
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.17232832774457e-06,
+      "loss": 0.3654,
+      "step": 17600
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.165038635369588e-06,
+      "loss": 0.347,
+      "step": 17610
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.157748942994606e-06,
+      "loss": 0.2979,
+      "step": 17620
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.150459250619625e-06,
+      "loss": 0.3326,
+      "step": 17630
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.143169558244642e-06,
+      "loss": 0.2854,
+      "step": 17640
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.1358798658696604e-06,
+      "loss": 0.3295,
+      "step": 17650
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.12859017349468e-06,
+      "loss": 0.3213,
+      "step": 17660
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.121300481119697e-06,
+      "loss": 0.3182,
+      "step": 17670
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.114010788744715e-06,
+      "loss": 0.3195,
+      "step": 17680
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.1067210963697345e-06,
+      "loss": 0.2614,
+      "step": 17690
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.099431403994752e-06,
+      "loss": 0.3313,
+      "step": 17700
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.09214171161977e-06,
+      "loss": 0.3544,
+      "step": 17710
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.0848520192447885e-06,
+      "loss": 0.3316,
+      "step": 17720
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.077562326869807e-06,
+      "loss": 0.366,
+      "step": 17730
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.070272634494824e-06,
+      "loss": 0.3439,
+      "step": 17740
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.062982942119843e-06,
+      "loss": 0.294,
+      "step": 17750
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.055693249744862e-06,
+      "loss": 0.3214,
+      "step": 17760
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.048403557369879e-06,
+      "loss": 0.3165,
+      "step": 17770
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.0411138649948975e-06,
+      "loss": 0.2991,
+      "step": 17780
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.033824172619917e-06,
+      "loss": 0.2888,
+      "step": 17790
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.026534480244934e-06,
+      "loss": 0.303,
+      "step": 17800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.019244787869952e-06,
+      "loss": 0.3101,
+      "step": 17810
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.011955095494971e-06,
+      "loss": 0.2638,
+      "step": 17820
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.004665403119989e-06,
+      "loss": 0.3408,
+      "step": 17830
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.997375710745007e-06,
+      "loss": 0.3642,
+      "step": 17840
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.990086018370026e-06,
+      "loss": 0.2947,
+      "step": 17850
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.982796325995044e-06,
+      "loss": 0.3379,
+      "step": 17860
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.975506633620061e-06,
+      "loss": 0.2956,
+      "step": 17870
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.96821694124508e-06,
+      "loss": 0.3168,
+      "step": 17880
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.960927248870099e-06,
+      "loss": 0.3587,
+      "step": 17890
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.953637556495116e-06,
+      "loss": 0.3471,
+      "step": 17900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.9463478641201346e-06,
+      "loss": 0.3169,
+      "step": 17910
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.939058171745154e-06,
+      "loss": 0.3454,
+      "step": 17920
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.931768479370171e-06,
+      "loss": 0.3302,
+      "step": 17930
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.9244787869951895e-06,
+      "loss": 0.3384,
+      "step": 17940
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.917189094620208e-06,
+      "loss": 0.3219,
+      "step": 17950
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.909899402245226e-06,
+      "loss": 0.3279,
+      "step": 17960
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.9026097098702435e-06,
+      "loss": 0.3317,
+      "step": 17970
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.895320017495263e-06,
+      "loss": 0.3043,
+      "step": 17980
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.888030325120281e-06,
+      "loss": 0.2552,
+      "step": 17990
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.880740632745298e-06,
+      "loss": 0.3115,
+      "step": 18000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.873450940370317e-06,
+      "loss": 0.3167,
+      "step": 18010
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.866161247995336e-06,
+      "loss": 0.3005,
+      "step": 18020
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.858871555620353e-06,
+      "loss": 0.3233,
+      "step": 18030
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.851581863245372e-06,
+      "loss": 0.3318,
+      "step": 18040
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.84429217087039e-06,
+      "loss": 0.3064,
+      "step": 18050
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.837002478495408e-06,
+      "loss": 0.3073,
+      "step": 18060
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.829712786120426e-06,
+      "loss": 0.2675,
+      "step": 18070
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.822423093745445e-06,
+      "loss": 0.3362,
+      "step": 18080
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.815133401370463e-06,
+      "loss": 0.2884,
+      "step": 18090
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.8078437089954806e-06,
+      "loss": 0.3069,
+      "step": 18100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.800554016620499e-06,
+      "loss": 0.3155,
+      "step": 18110
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.793264324245518e-06,
+      "loss": 0.348,
+      "step": 18120
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.7859746318705355e-06,
+      "loss": 0.274,
+      "step": 18130
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.778684939495554e-06,
+      "loss": 0.3419,
+      "step": 18140
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.771395247120572e-06,
+      "loss": 0.3461,
+      "step": 18150
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.76410555474559e-06,
+      "loss": 0.3442,
+      "step": 18160
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.756815862370608e-06,
+      "loss": 0.3072,
+      "step": 18170
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.749526169995627e-06,
+      "loss": 0.3258,
+      "step": 18180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.742236477620645e-06,
+      "loss": 0.2636,
+      "step": 18190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.734946785245663e-06,
+      "loss": 0.3061,
+      "step": 18200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.727657092870682e-06,
+      "loss": 0.2698,
+      "step": 18210
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.7203674004957e-06,
+      "loss": 0.3076,
+      "step": 18220
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.713077708120718e-06,
+      "loss": 0.2849,
+      "step": 18230
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.705788015745736e-06,
+      "loss": 0.2886,
+      "step": 18240
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.698498323370754e-06,
+      "loss": 0.3156,
+      "step": 18250
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.6912086309957725e-06,
+      "loss": 0.3134,
+      "step": 18260
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.68391893862079e-06,
+      "loss": 0.3225,
+      "step": 18270
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.676629246245809e-06,
+      "loss": 0.3054,
+      "step": 18280
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.669339553870827e-06,
+      "loss": 0.3243,
+      "step": 18290
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.662049861495845e-06,
+      "loss": 0.2086,
+      "step": 18300
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.654760169120864e-06,
+      "loss": 0.1277,
+      "step": 18310
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.647470476745882e-06,
+      "loss": 0.1382,
+      "step": 18320
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.6401807843709e-06,
+      "loss": 0.1266,
+      "step": 18330
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.632891091995919e-06,
+      "loss": 0.1204,
+      "step": 18340
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.625601399620936e-06,
+      "loss": 0.1442,
+      "step": 18350
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.618311707245955e-06,
+      "loss": 0.1425,
+      "step": 18360
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.611022014870972e-06,
+      "loss": 0.1412,
+      "step": 18370
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.603732322495991e-06,
+      "loss": 0.1503,
+      "step": 18380
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.59644263012101e-06,
+      "loss": 0.1606,
+      "step": 18390
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.589152937746027e-06,
+      "loss": 0.1318,
+      "step": 18400
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.581863245371046e-06,
+      "loss": 0.1586,
+      "step": 18410
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.5745735529960645e-06,
+      "loss": 0.1255,
+      "step": 18420
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.567283860621082e-06,
+      "loss": 0.1573,
+      "step": 18430
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.559994168246101e-06,
+      "loss": 0.1512,
+      "step": 18440
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.5527044758711185e-06,
+      "loss": 0.1531,
+      "step": 18450
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.545414783496137e-06,
+      "loss": 0.1501,
+      "step": 18460
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.538125091121154e-06,
+      "loss": 0.1707,
+      "step": 18470
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.5308353987461734e-06,
+      "loss": 0.1327,
+      "step": 18480
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.523545706371192e-06,
+      "loss": 0.1731,
+      "step": 18490
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.516256013996209e-06,
+      "loss": 0.1285,
+      "step": 18500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.508966321621228e-06,
+      "loss": 0.1367,
+      "step": 18510
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.501676629246247e-06,
+      "loss": 0.1377,
+      "step": 18520
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.494386936871264e-06,
+      "loss": 0.1467,
+      "step": 18530
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.487097244496283e-06,
+      "loss": 0.1293,
+      "step": 18540
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.479807552121301e-06,
+      "loss": 0.1466,
+      "step": 18550
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.472517859746319e-06,
+      "loss": 0.1304,
+      "step": 18560
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.465228167371338e-06,
+      "loss": 0.1239,
+      "step": 18570
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.457938474996356e-06,
+      "loss": 0.1211,
+      "step": 18580
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.450648782621374e-06,
+      "loss": 0.1445,
+      "step": 18590
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.443359090246391e-06,
+      "loss": 0.1505,
+      "step": 18600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.4360693978714105e-06,
+      "loss": 0.1314,
+      "step": 18610
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.428779705496429e-06,
+      "loss": 0.1405,
+      "step": 18620
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.421490013121446e-06,
+      "loss": 0.1186,
+      "step": 18630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.414200320746465e-06,
+      "loss": 0.1772,
+      "step": 18640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.406910628371483e-06,
+      "loss": 0.1494,
+      "step": 18650
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.399620935996501e-06,
+      "loss": 0.1517,
+      "step": 18660
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.39233124362152e-06,
+      "loss": 0.161,
+      "step": 18670
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.385041551246538e-06,
+      "loss": 0.1495,
+      "step": 18680
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.377751858871556e-06,
+      "loss": 0.1224,
+      "step": 18690
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.370462166496575e-06,
+      "loss": 0.1586,
+      "step": 18700
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.363172474121593e-06,
+      "loss": 0.1293,
+      "step": 18710
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.355882781746611e-06,
+      "loss": 0.1463,
+      "step": 18720
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.348593089371628e-06,
+      "loss": 0.1466,
+      "step": 18730
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.3413033969966475e-06,
+      "loss": 0.1711,
+      "step": 18740
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.334013704621666e-06,
+      "loss": 0.1267,
+      "step": 18750
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.326724012246683e-06,
+      "loss": 0.1432,
+      "step": 18760
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.3194343198717025e-06,
+      "loss": 0.1584,
+      "step": 18770
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.31214462749672e-06,
+      "loss": 0.1526,
+      "step": 18780
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.304854935121738e-06,
+      "loss": 0.1245,
+      "step": 18790
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.297565242746757e-06,
+      "loss": 0.148,
+      "step": 18800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.290275550371775e-06,
+      "loss": 0.1284,
+      "step": 18810
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.282985857996793e-06,
+      "loss": 0.1401,
+      "step": 18820
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.2756961656218106e-06,
+      "loss": 0.1424,
+      "step": 18830
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.26840647324683e-06,
+      "loss": 0.1508,
+      "step": 18840
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.261116780871848e-06,
+      "loss": 0.1378,
+      "step": 18850
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.2538270884968655e-06,
+      "loss": 0.1563,
+      "step": 18860
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.246537396121885e-06,
+      "loss": 0.1324,
+      "step": 18870
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.239247703746902e-06,
+      "loss": 0.1502,
+      "step": 18880
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.23195801137192e-06,
+      "loss": 0.1611,
+      "step": 18890
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.2246683189969395e-06,
+      "loss": 0.1463,
+      "step": 18900
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.217378626621957e-06,
+      "loss": 0.1505,
+      "step": 18910
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.210088934246975e-06,
+      "loss": 0.1505,
+      "step": 18920
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.202799241871994e-06,
+      "loss": 0.1313,
+      "step": 18930
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.195509549497012e-06,
+      "loss": 0.1361,
+      "step": 18940
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.18821985712203e-06,
+      "loss": 0.1406,
+      "step": 18950
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.180930164747048e-06,
+      "loss": 0.1761,
+      "step": 18960
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.173640472372067e-06,
+      "loss": 0.1464,
+      "step": 18970
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.166350779997084e-06,
+      "loss": 0.1749,
+      "step": 18980
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.1590610876221025e-06,
+      "loss": 0.1463,
+      "step": 18990
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.151771395247122e-06,
+      "loss": 0.1787,
+      "step": 19000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.144481702872139e-06,
+      "loss": 0.1373,
+      "step": 19010
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.137192010497157e-06,
+      "loss": 0.1526,
+      "step": 19020
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.1299023181221766e-06,
+      "loss": 0.158,
+      "step": 19030
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.122612625747194e-06,
+      "loss": 0.1595,
+      "step": 19040
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.115322933372212e-06,
+      "loss": 0.1422,
+      "step": 19050
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.108033240997231e-06,
+      "loss": 0.1457,
+      "step": 19060
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.100743548622249e-06,
+      "loss": 0.1745,
+      "step": 19070
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.093453856247266e-06,
+      "loss": 0.1375,
+      "step": 19080
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.086164163872285e-06,
+      "loss": 0.1572,
+      "step": 19090
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.078874471497304e-06,
+      "loss": 0.1558,
+      "step": 19100
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.071584779122321e-06,
+      "loss": 0.1457,
+      "step": 19110
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0642950867473396e-06,
+      "loss": 0.1731,
+      "step": 19120
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.057005394372359e-06,
+      "loss": 0.1579,
+      "step": 19130
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.049715701997376e-06,
+      "loss": 0.1627,
+      "step": 19140
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.0424260096223945e-06,
+      "loss": 0.1399,
+      "step": 19150
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.035136317247413e-06,
+      "loss": 0.1343,
+      "step": 19160
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.027846624872431e-06,
+      "loss": 0.1282,
+      "step": 19170
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.0205569324974485e-06,
+      "loss": 0.1601,
+      "step": 19180
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.013267240122468e-06,
+      "loss": 0.1559,
+      "step": 19190
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.005977547747486e-06,
+      "loss": 0.1491,
+      "step": 19200
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.9986878553725034e-06,
+      "loss": 0.1618,
+      "step": 19210
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.991398162997522e-06,
+      "loss": 0.1454,
+      "step": 19220
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.984108470622541e-06,
+      "loss": 0.1224,
+      "step": 19230
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.976818778247558e-06,
+      "loss": 0.1473,
+      "step": 19240
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.969529085872577e-06,
+      "loss": 0.1432,
+      "step": 19250
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.962239393497595e-06,
+      "loss": 0.1433,
+      "step": 19260
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.954949701122613e-06,
+      "loss": 0.1145,
+      "step": 19270
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.947660008747631e-06,
+      "loss": 0.1348,
+      "step": 19280
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.94037031637265e-06,
+      "loss": 0.1735,
+      "step": 19290
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.933080623997668e-06,
+      "loss": 0.1343,
+      "step": 19300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.925790931622686e-06,
+      "loss": 0.1704,
+      "step": 19310
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.918501239247704e-06,
+      "loss": 0.1144,
+      "step": 19320
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.911211546872723e-06,
+      "loss": 0.1624,
+      "step": 19330
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.9039218544977405e-06,
+      "loss": 0.1594,
+      "step": 19340
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.896632162122759e-06,
+      "loss": 0.1046,
+      "step": 19350
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.889342469747777e-06,
+      "loss": 0.1474,
+      "step": 19360
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.882052777372795e-06,
+      "loss": 0.1386,
+      "step": 19370
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.874763084997813e-06,
+      "loss": 0.1249,
+      "step": 19380
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.867473392622832e-06,
+      "loss": 0.1527,
+      "step": 19390
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.86018370024785e-06,
+      "loss": 0.147,
+      "step": 19400
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.852894007872868e-06,
+      "loss": 0.1326,
+      "step": 19410
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.845604315497887e-06,
+      "loss": 0.1373,
+      "step": 19420
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.838314623122905e-06,
+      "loss": 0.1426,
+      "step": 19430
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.831024930747923e-06,
+      "loss": 0.1413,
+      "step": 19440
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.823735238372941e-06,
+      "loss": 0.1694,
+      "step": 19450
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.816445545997959e-06,
+      "loss": 0.1282,
+      "step": 19460
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.8091558536229775e-06,
+      "loss": 0.1446,
+      "step": 19470
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.801866161247995e-06,
+      "loss": 0.1588,
+      "step": 19480
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.794576468873014e-06,
+      "loss": 0.147,
+      "step": 19490
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7872867764980324e-06,
+      "loss": 0.1501,
+      "step": 19500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.77999708412305e-06,
+      "loss": 0.1779,
+      "step": 19510
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.772707391748069e-06,
+      "loss": 0.1411,
+      "step": 19520
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.765417699373087e-06,
+      "loss": 0.1537,
+      "step": 19530
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.758128006998105e-06,
+      "loss": 0.1511,
+      "step": 19540
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.750838314623124e-06,
+      "loss": 0.1528,
+      "step": 19550
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.743548622248141e-06,
+      "loss": 0.1479,
+      "step": 19560
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.73625892987316e-06,
+      "loss": 0.1308,
+      "step": 19570
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.728969237498177e-06,
+      "loss": 0.1459,
+      "step": 19580
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.721679545123196e-06,
+      "loss": 0.1453,
+      "step": 19590
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.714389852748215e-06,
+      "loss": 0.1331,
+      "step": 19600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.707100160373232e-06,
+      "loss": 0.1504,
+      "step": 19610
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.699810467998251e-06,
+      "loss": 0.1647,
+      "step": 19620
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6925207756232695e-06,
+      "loss": 0.141,
+      "step": 19630
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.685231083248287e-06,
+      "loss": 0.1491,
+      "step": 19640
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.677941390873306e-06,
+      "loss": 0.1323,
+      "step": 19650
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6706516984983236e-06,
+      "loss": 0.1234,
+      "step": 19660
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.663362006123342e-06,
+      "loss": 0.1401,
+      "step": 19670
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.65607231374836e-06,
+      "loss": 0.1629,
+      "step": 19680
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.6487826213733785e-06,
+      "loss": 0.1271,
+      "step": 19690
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.641492928998397e-06,
+      "loss": 0.1546,
+      "step": 19700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.634203236623414e-06,
+      "loss": 0.1526,
+      "step": 19710
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.626913544248433e-06,
+      "loss": 0.1533,
+      "step": 19720
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.619623851873452e-06,
+      "loss": 0.1202,
+      "step": 19730
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.612334159498469e-06,
+      "loss": 0.1525,
+      "step": 19740
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.605044467123488e-06,
+      "loss": 0.1368,
+      "step": 19750
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.5977547747485066e-06,
+      "loss": 0.1508,
+      "step": 19760
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.590465082373524e-06,
+      "loss": 0.1229,
+      "step": 19770
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.583175389998543e-06,
+      "loss": 0.133,
+      "step": 19780
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.575885697623561e-06,
+      "loss": 0.1655,
+      "step": 19790
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.568596005248579e-06,
+      "loss": 0.1407,
+      "step": 19800
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.561306312873596e-06,
+      "loss": 0.1474,
+      "step": 19810
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.5540166204986155e-06,
+      "loss": 0.1576,
+      "step": 19820
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.546726928123634e-06,
+      "loss": 0.1198,
+      "step": 19830
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.539437235748651e-06,
+      "loss": 0.1395,
+      "step": 19840
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.53214754337367e-06,
+      "loss": 0.1597,
+      "step": 19850
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.524857850998689e-06,
+      "loss": 0.1585,
+      "step": 19860
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.517568158623706e-06,
+      "loss": 0.121,
+      "step": 19870
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.510278466248725e-06,
+      "loss": 0.1481,
+      "step": 19880
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.502988773873743e-06,
+      "loss": 0.1507,
+      "step": 19890
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.495699081498761e-06,
+      "loss": 0.1541,
+      "step": 19900
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.48840938912378e-06,
+      "loss": 0.1817,
+      "step": 19910
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.481119696748798e-06,
+      "loss": 0.1671,
+      "step": 19920
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.473830004373816e-06,
+      "loss": 0.1396,
+      "step": 19930
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.466540311998833e-06,
+      "loss": 0.177,
+      "step": 19940
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4592506196238526e-06,
+      "loss": 0.1495,
+      "step": 19950
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.451960927248871e-06,
+      "loss": 0.1552,
+      "step": 19960
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.444671234873888e-06,
+      "loss": 0.132,
+      "step": 19970
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.4373815424989075e-06,
+      "loss": 0.1406,
+      "step": 19980
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.430091850123925e-06,
+      "loss": 0.1386,
+      "step": 19990
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.422802157748943e-06,
+      "loss": 0.1335,
+      "step": 20000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.415512465373962e-06,
+      "loss": 0.1321,
+      "step": 20010
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.40822277299898e-06,
+      "loss": 0.1446,
+      "step": 20020
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.400933080623998e-06,
+      "loss": 0.1508,
+      "step": 20030
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.393643388249016e-06,
+      "loss": 0.1645,
+      "step": 20040
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.386353695874035e-06,
+      "loss": 0.1231,
+      "step": 20050
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.379064003499053e-06,
+      "loss": 0.153,
+      "step": 20060
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.3717743111240705e-06,
+      "loss": 0.1767,
+      "step": 20070
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.36448461874909e-06,
+      "loss": 0.1537,
+      "step": 20080
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.357194926374107e-06,
+      "loss": 0.1754,
+      "step": 20090
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.349905233999125e-06,
+      "loss": 0.1347,
+      "step": 20100
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3426155416241445e-06,
+      "loss": 0.1411,
+      "step": 20110
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.335325849249162e-06,
+      "loss": 0.1325,
+      "step": 20120
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.32803615687418e-06,
+      "loss": 0.136,
+      "step": 20130
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.3207464644991994e-06,
+      "loss": 0.1483,
+      "step": 20140
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.313456772124217e-06,
+      "loss": 0.1493,
+      "step": 20150
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.306167079749235e-06,
+      "loss": 0.1349,
+      "step": 20160
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.298877387374253e-06,
+      "loss": 0.1431,
+      "step": 20170
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.291587694999272e-06,
+      "loss": 0.1383,
+      "step": 20180
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.284298002624289e-06,
+      "loss": 0.1609,
+      "step": 20190
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2770083102493075e-06,
+      "loss": 0.1427,
+      "step": 20200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.269718617874327e-06,
+      "loss": 0.1701,
+      "step": 20210
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.262428925499344e-06,
+      "loss": 0.1532,
+      "step": 20220
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.2551392331243624e-06,
+      "loss": 0.1296,
+      "step": 20230
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.247849540749382e-06,
+      "loss": 0.1469,
+      "step": 20240
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.240559848374399e-06,
+      "loss": 0.1381,
+      "step": 20250
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.233270155999417e-06,
+      "loss": 0.1374,
+      "step": 20260
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.225980463624436e-06,
+      "loss": 0.1323,
+      "step": 20270
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.218690771249454e-06,
+      "loss": 0.145,
+      "step": 20280
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.211401078874471e-06,
+      "loss": 0.1429,
+      "step": 20290
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.20411138649949e-06,
+      "loss": 0.1692,
+      "step": 20300
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.196821694124509e-06,
+      "loss": 0.1567,
+      "step": 20310
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.189532001749526e-06,
+      "loss": 0.1331,
+      "step": 20320
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.182242309374545e-06,
+      "loss": 0.1415,
+      "step": 20330
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.174952616999564e-06,
+      "loss": 0.1345,
+      "step": 20340
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.167662924624581e-06,
+      "loss": 0.1324,
+      "step": 20350
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1603732322495995e-06,
+      "loss": 0.1293,
+      "step": 20360
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.153083539874618e-06,
+      "loss": 0.1431,
+      "step": 20370
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.145793847499636e-06,
+      "loss": 0.1385,
+      "step": 20380
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1385041551246535e-06,
+      "loss": 0.157,
+      "step": 20390
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.131214462749672e-06,
+      "loss": 0.1435,
+      "step": 20400
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.123924770374691e-06,
+      "loss": 0.1281,
+      "step": 20410
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.1166350779997085e-06,
+      "loss": 0.1738,
+      "step": 20420
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.109345385624727e-06,
+      "loss": 0.1606,
+      "step": 20430
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.102055693249746e-06,
+      "loss": 0.1528,
+      "step": 20440
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.094766000874763e-06,
+      "loss": 0.12,
+      "step": 20450
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.087476308499782e-06,
+      "loss": 0.1528,
+      "step": 20460
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.0801866161248e-06,
+      "loss": 0.1263,
+      "step": 20470
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.072896923749818e-06,
+      "loss": 0.1368,
+      "step": 20480
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.065607231374836e-06,
+      "loss": 0.143,
+      "step": 20490
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.058317538999855e-06,
+      "loss": 0.1583,
+      "step": 20500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.051027846624873e-06,
+      "loss": 0.1081,
+      "step": 20510
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.043738154249891e-06,
+      "loss": 0.1532,
+      "step": 20520
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.036448461874909e-06,
+      "loss": 0.1693,
+      "step": 20530
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.029158769499928e-06,
+      "loss": 0.1444,
+      "step": 20540
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.0218690771249455e-06,
+      "loss": 0.1616,
+      "step": 20550
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.014579384749964e-06,
+      "loss": 0.1498,
+      "step": 20560
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.007289692374982e-06,
+      "loss": 0.1433,
+      "step": 20570
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5e-06,
+      "loss": 0.1353,
+      "step": 20580
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.992710307625019e-06,
+      "loss": 0.1389,
+      "step": 20590
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.985420615250037e-06,
+      "loss": 0.1493,
+      "step": 20600
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.978130922875055e-06,
+      "loss": 0.1561,
+      "step": 20610
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.970841230500073e-06,
+      "loss": 0.1429,
+      "step": 20620
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.963551538125092e-06,
+      "loss": 0.1676,
+      "step": 20630
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.95626184575011e-06,
+      "loss": 0.1371,
+      "step": 20640
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.948972153375128e-06,
+      "loss": 0.142,
+      "step": 20650
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.941682461000146e-06,
+      "loss": 0.1403,
+      "step": 20660
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.934392768625165e-06,
+      "loss": 0.1269,
+      "step": 20670
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9271030762501826e-06,
+      "loss": 0.1277,
+      "step": 20680
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.919813383875201e-06,
+      "loss": 0.1288,
+      "step": 20690
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.912523691500219e-06,
+      "loss": 0.1642,
+      "step": 20700
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.9052339991252375e-06,
+      "loss": 0.1641,
+      "step": 20710
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.897944306750256e-06,
+      "loss": 0.1263,
+      "step": 20720
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.890654614375274e-06,
+      "loss": 0.1638,
+      "step": 20730
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.883364922000292e-06,
+      "loss": 0.1034,
+      "step": 20740
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.87607522962531e-06,
+      "loss": 0.1419,
+      "step": 20750
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.868785537250328e-06,
+      "loss": 0.1557,
+      "step": 20760
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.861495844875347e-06,
+      "loss": 0.144,
+      "step": 20770
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.854206152500365e-06,
+      "loss": 0.152,
+      "step": 20780
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.846916460125383e-06,
+      "loss": 0.1477,
+      "step": 20790
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.839626767750401e-06,
+      "loss": 0.1316,
+      "step": 20800
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.83233707537542e-06,
+      "loss": 0.1598,
+      "step": 20810
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.825047383000438e-06,
+      "loss": 0.1454,
+      "step": 20820
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.817757690625456e-06,
+      "loss": 0.1244,
+      "step": 20830
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.8104679982504745e-06,
+      "loss": 0.1463,
+      "step": 20840
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.803178305875493e-06,
+      "loss": 0.1292,
+      "step": 20850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.79588861350051e-06,
+      "loss": 0.1752,
+      "step": 20860
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.7885989211255294e-06,
+      "loss": 0.1265,
+      "step": 20870
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.781309228750547e-06,
+      "loss": 0.1372,
+      "step": 20880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.774019536375565e-06,
+      "loss": 0.1692,
+      "step": 20890
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7667298440005835e-06,
+      "loss": 0.1503,
+      "step": 20900
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.759440151625602e-06,
+      "loss": 0.1319,
+      "step": 20910
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.75215045925062e-06,
+      "loss": 0.1795,
+      "step": 20920
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.744860766875638e-06,
+      "loss": 0.1604,
+      "step": 20930
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.737571074500657e-06,
+      "loss": 0.1209,
+      "step": 20940
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.730281382125675e-06,
+      "loss": 0.1341,
+      "step": 20950
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.7229916897506924e-06,
+      "loss": 0.1194,
+      "step": 20960
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.715701997375712e-06,
+      "loss": 0.1461,
+      "step": 20970
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.708412305000729e-06,
+      "loss": 0.1214,
+      "step": 20980
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.701122612625747e-06,
+      "loss": 0.1524,
+      "step": 20990
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.693832920250766e-06,
+      "loss": 0.1402,
+      "step": 21000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.686543227875784e-06,
+      "loss": 0.12,
+      "step": 21010
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.679253535500802e-06,
+      "loss": 0.1417,
+      "step": 21020
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.6719638431258205e-06,
+      "loss": 0.1303,
+      "step": 21030
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.664674150750839e-06,
+      "loss": 0.1229,
+      "step": 21040
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.657384458375857e-06,
+      "loss": 0.155,
+      "step": 21050
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.650094766000875e-06,
+      "loss": 0.1708,
+      "step": 21060
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.642805073625894e-06,
+      "loss": 0.1353,
+      "step": 21070
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.635515381250912e-06,
+      "loss": 0.1372,
+      "step": 21080
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.6282256888759295e-06,
+      "loss": 0.1146,
+      "step": 21090
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.620935996500948e-06,
+      "loss": 0.1259,
+      "step": 21100
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.613646304125966e-06,
+      "loss": 0.1116,
+      "step": 21110
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.606356611750984e-06,
+      "loss": 0.1398,
+      "step": 21120
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.599066919376003e-06,
+      "loss": 0.1427,
+      "step": 21130
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.591777227001021e-06,
+      "loss": 0.1494,
+      "step": 21140
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.584487534626039e-06,
+      "loss": 0.1388,
+      "step": 21150
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.577197842251057e-06,
+      "loss": 0.14,
+      "step": 21160
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.569908149876076e-06,
+      "loss": 0.1338,
+      "step": 21170
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.562618457501094e-06,
+      "loss": 0.1477,
+      "step": 21180
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.555328765126112e-06,
+      "loss": 0.1336,
+      "step": 21190
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.54803907275113e-06,
+      "loss": 0.1435,
+      "step": 21200
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.540749380376149e-06,
+      "loss": 0.1301,
+      "step": 21210
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5334596880011665e-06,
+      "loss": 0.1423,
+      "step": 21220
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.526169995626185e-06,
+      "loss": 0.1489,
+      "step": 21230
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.518880303251203e-06,
+      "loss": 0.1334,
+      "step": 21240
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.5115906108762214e-06,
+      "loss": 0.1367,
+      "step": 21250
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.50430091850124e-06,
+      "loss": 0.146,
+      "step": 21260
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.497011226126258e-06,
+      "loss": 0.1812,
+      "step": 21270
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.489721533751276e-06,
+      "loss": 0.1122,
+      "step": 21280
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.482431841376294e-06,
+      "loss": 0.146,
+      "step": 21290
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.475142149001312e-06,
+      "loss": 0.1301,
+      "step": 21300
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.467852456626331e-06,
+      "loss": 0.1486,
+      "step": 21310
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.460562764251349e-06,
+      "loss": 0.1344,
+      "step": 21320
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.453273071876367e-06,
+      "loss": 0.1426,
+      "step": 21330
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.445983379501385e-06,
+      "loss": 0.1582,
+      "step": 21340
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.438693687126404e-06,
+      "loss": 0.1312,
+      "step": 21350
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.431403994751422e-06,
+      "loss": 0.1278,
+      "step": 21360
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.42411430237644e-06,
+      "loss": 0.1689,
+      "step": 21370
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.4168246100014585e-06,
+      "loss": 0.1556,
+      "step": 21380
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.409534917626477e-06,
+      "loss": 0.1184,
+      "step": 21390
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.402245225251494e-06,
+      "loss": 0.134,
+      "step": 21400
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.394955532876513e-06,
+      "loss": 0.1366,
+      "step": 21410
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.387665840501531e-06,
+      "loss": 0.1569,
+      "step": 21420
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.380376148126549e-06,
+      "loss": 0.1367,
+      "step": 21430
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.3730864557515675e-06,
+      "loss": 0.1357,
+      "step": 21440
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.365796763376586e-06,
+      "loss": 0.1683,
+      "step": 21450
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.358507071001604e-06,
+      "loss": 0.149,
+      "step": 21460
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.351217378626622e-06,
+      "loss": 0.1359,
+      "step": 21470
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.343927686251641e-06,
+      "loss": 0.1301,
+      "step": 21480
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.336637993876659e-06,
+      "loss": 0.137,
+      "step": 21490
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.329348301501677e-06,
+      "loss": 0.1565,
+      "step": 21500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.3220586091266956e-06,
+      "loss": 0.136,
+      "step": 21510
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.314768916751713e-06,
+      "loss": 0.1142,
+      "step": 21520
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.307479224376731e-06,
+      "loss": 0.1001,
+      "step": 21530
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.3001895320017505e-06,
+      "loss": 0.1184,
+      "step": 21540
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.292899839626768e-06,
+      "loss": 0.1271,
+      "step": 21550
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.285610147251786e-06,
+      "loss": 0.1561,
+      "step": 21560
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2783204548768045e-06,
+      "loss": 0.1485,
+      "step": 21570
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.271030762501823e-06,
+      "loss": 0.1408,
+      "step": 21580
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.263741070126841e-06,
+      "loss": 0.1347,
+      "step": 21590
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.256451377751859e-06,
+      "loss": 0.1423,
+      "step": 21600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.249161685376878e-06,
+      "loss": 0.1513,
+      "step": 21610
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.241871993001896e-06,
+      "loss": 0.1586,
+      "step": 21620
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2345823006269135e-06,
+      "loss": 0.1155,
+      "step": 21630
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.227292608251933e-06,
+      "loss": 0.1314,
+      "step": 21640
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.22000291587695e-06,
+      "loss": 0.1447,
+      "step": 21650
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.212713223501968e-06,
+      "loss": 0.1192,
+      "step": 21660
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.205423531126987e-06,
+      "loss": 0.1382,
+      "step": 21670
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.198133838752005e-06,
+      "loss": 0.1435,
+      "step": 21680
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.190844146377023e-06,
+      "loss": 0.1383,
+      "step": 21690
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.1835544540020416e-06,
+      "loss": 0.1595,
+      "step": 21700
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.17626476162706e-06,
+      "loss": 0.1339,
+      "step": 21710
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.168975069252078e-06,
+      "loss": 0.1356,
+      "step": 21720
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.161685376877096e-06,
+      "loss": 0.1687,
+      "step": 21730
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.154395684502115e-06,
+      "loss": 0.1276,
+      "step": 21740
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.147105992127133e-06,
+      "loss": 0.1534,
+      "step": 21750
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1398162997521505e-06,
+      "loss": 0.149,
+      "step": 21760
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.132526607377169e-06,
+      "loss": 0.1662,
+      "step": 21770
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.125236915002187e-06,
+      "loss": 0.1227,
+      "step": 21780
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.1179472226272054e-06,
+      "loss": 0.1175,
+      "step": 21790
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.110657530252224e-06,
+      "loss": 0.1366,
+      "step": 21800
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.103367837877242e-06,
+      "loss": 0.159,
+      "step": 21810
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.09607814550226e-06,
+      "loss": 0.1389,
+      "step": 21820
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.088788453127278e-06,
+      "loss": 0.1758,
+      "step": 21830
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.081498760752297e-06,
+      "loss": 0.172,
+      "step": 21840
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.074209068377315e-06,
+      "loss": 0.1395,
+      "step": 21850
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.066919376002333e-06,
+      "loss": 0.1478,
+      "step": 21860
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.059629683627351e-06,
+      "loss": 0.1604,
+      "step": 21870
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.052339991252369e-06,
+      "loss": 0.1617,
+      "step": 21880
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.045050298877388e-06,
+      "loss": 0.1521,
+      "step": 21890
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.037760606502406e-06,
+      "loss": 0.1137,
+      "step": 21900
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.030470914127424e-06,
+      "loss": 0.1491,
+      "step": 21910
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.0231812217524425e-06,
+      "loss": 0.1363,
+      "step": 21920
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.015891529377461e-06,
+      "loss": 0.1426,
+      "step": 21930
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.008601837002479e-06,
+      "loss": 0.1386,
+      "step": 21940
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.001312144627497e-06,
+      "loss": 0.1319,
+      "step": 21950
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.994022452252515e-06,
+      "loss": 0.119,
+      "step": 21960
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.986732759877533e-06,
+      "loss": 0.135,
+      "step": 21970
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.979443067502552e-06,
+      "loss": 0.1548,
+      "step": 21980
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.97215337512757e-06,
+      "loss": 0.125,
+      "step": 21990
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.964863682752588e-06,
+      "loss": 0.1288,
+      "step": 22000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.957573990377606e-06,
+      "loss": 0.1386,
+      "step": 22010
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.950284298002625e-06,
+      "loss": 0.1303,
+      "step": 22020
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.942994605627643e-06,
+      "loss": 0.1593,
+      "step": 22030
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.935704913252661e-06,
+      "loss": 0.1488,
+      "step": 22040
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9284152208776795e-06,
+      "loss": 0.1516,
+      "step": 22050
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.921125528502697e-06,
+      "loss": 0.1546,
+      "step": 22060
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.913835836127715e-06,
+      "loss": 0.1307,
+      "step": 22070
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.9065461437527344e-06,
+      "loss": 0.1461,
+      "step": 22080
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.899256451377752e-06,
+      "loss": 0.1648,
+      "step": 22090
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.89196675900277e-06,
+      "loss": 0.1461,
+      "step": 22100
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8846770666277885e-06,
+      "loss": 0.1639,
+      "step": 22110
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.877387374252807e-06,
+      "loss": 0.1539,
+      "step": 22120
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.870097681877825e-06,
+      "loss": 0.1509,
+      "step": 22130
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.862807989502843e-06,
+      "loss": 0.1269,
+      "step": 22140
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.855518297127862e-06,
+      "loss": 0.1346,
+      "step": 22150
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.84822860475288e-06,
+      "loss": 0.1307,
+      "step": 22160
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8409389123778975e-06,
+      "loss": 0.1526,
+      "step": 22170
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.833649220002917e-06,
+      "loss": 0.1727,
+      "step": 22180
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.826359527627934e-06,
+      "loss": 0.1698,
+      "step": 22190
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.819069835252952e-06,
+      "loss": 0.1311,
+      "step": 22200
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.811780142877971e-06,
+      "loss": 0.1402,
+      "step": 22210
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.804490450502989e-06,
+      "loss": 0.1577,
+      "step": 22220
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7972007581280073e-06,
+      "loss": 0.1556,
+      "step": 22230
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.789911065753026e-06,
+      "loss": 0.1472,
+      "step": 22240
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.782621373378044e-06,
+      "loss": 0.1614,
+      "step": 22250
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.775331681003062e-06,
+      "loss": 0.1376,
+      "step": 22260
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.76804198862808e-06,
+      "loss": 0.1544,
+      "step": 22270
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7607522962530983e-06,
+      "loss": 0.1467,
+      "step": 22280
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.753462603878117e-06,
+      "loss": 0.1548,
+      "step": 22290
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.746172911503135e-06,
+      "loss": 0.1345,
+      "step": 22300
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7388832191281532e-06,
+      "loss": 0.1201,
+      "step": 22310
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.731593526753171e-06,
+      "loss": 0.1285,
+      "step": 22320
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7243038343781894e-06,
+      "loss": 0.1303,
+      "step": 22330
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.717014142003208e-06,
+      "loss": 0.1405,
+      "step": 22340
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.709724449628226e-06,
+      "loss": 0.149,
+      "step": 22350
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.7024347572532443e-06,
+      "loss": 0.1406,
+      "step": 22360
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.695145064878262e-06,
+      "loss": 0.1453,
+      "step": 22370
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6878553725032805e-06,
+      "loss": 0.1613,
+      "step": 22380
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6805656801282992e-06,
+      "loss": 0.1398,
+      "step": 22390
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.673275987753317e-06,
+      "loss": 0.1493,
+      "step": 22400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6659862953783354e-06,
+      "loss": 0.1373,
+      "step": 22410
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6586966030033537e-06,
+      "loss": 0.1129,
+      "step": 22420
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6514069106283716e-06,
+      "loss": 0.1424,
+      "step": 22430
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.6441172182533903e-06,
+      "loss": 0.1416,
+      "step": 22440
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.636827525878408e-06,
+      "loss": 0.1437,
+      "step": 22450
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6295378335034265e-06,
+      "loss": 0.1462,
+      "step": 22460
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6222481411284448e-06,
+      "loss": 0.1514,
+      "step": 22470
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6149584487534626e-06,
+      "loss": 0.1335,
+      "step": 22480
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6076687563784814e-06,
+      "loss": 0.1514,
+      "step": 22490
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6003790640034992e-06,
+      "loss": 0.1507,
+      "step": 22500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5930893716285175e-06,
+      "loss": 0.1369,
+      "step": 22510
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.585799679253536e-06,
+      "loss": 0.1421,
+      "step": 22520
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5785099868785537e-06,
+      "loss": 0.1231,
+      "step": 22530
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5712202945035724e-06,
+      "loss": 0.1331,
+      "step": 22540
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5639306021285903e-06,
+      "loss": 0.1217,
+      "step": 22550
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5566409097536086e-06,
+      "loss": 0.1341,
+      "step": 22560
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.549351217378627e-06,
+      "loss": 0.1305,
+      "step": 22570
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.542061525003645e-06,
+      "loss": 0.154,
+      "step": 22580
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5347718326286635e-06,
+      "loss": 0.137,
+      "step": 22590
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.527482140253682e-06,
+      "loss": 0.1453,
+      "step": 22600
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5201924478786997e-06,
+      "loss": 0.1214,
+      "step": 22610
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.512902755503718e-06,
+      "loss": 0.1286,
+      "step": 22620
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.505613063128736e-06,
+      "loss": 0.1213,
+      "step": 22630
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4983233707537546e-06,
+      "loss": 0.1312,
+      "step": 22640
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.491033678378773e-06,
+      "loss": 0.1189,
+      "step": 22650
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4837439860037908e-06,
+      "loss": 0.1546,
+      "step": 22660
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.476454293628809e-06,
+      "loss": 0.1179,
+      "step": 22670
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.469164601253827e-06,
+      "loss": 0.1448,
+      "step": 22680
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4618749088788457e-06,
+      "loss": 0.1285,
+      "step": 22690
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.454585216503864e-06,
+      "loss": 0.1251,
+      "step": 22700
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.447295524128882e-06,
+      "loss": 0.1107,
+      "step": 22710
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4400058317539e-06,
+      "loss": 0.165,
+      "step": 22720
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.432716139378918e-06,
+      "loss": 0.1581,
+      "step": 22730
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4254264470039368e-06,
+      "loss": 0.147,
+      "step": 22740
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.418136754628955e-06,
+      "loss": 0.1359,
+      "step": 22750
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.410847062253973e-06,
+      "loss": 0.1486,
+      "step": 22760
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4035573698789912e-06,
+      "loss": 0.1378,
+      "step": 22770
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.39626767750401e-06,
+      "loss": 0.1274,
+      "step": 22780
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.388977985129028e-06,
+      "loss": 0.1378,
+      "step": 22790
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.381688292754046e-06,
+      "loss": 0.1441,
+      "step": 22800
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.374398600379064e-06,
+      "loss": 0.1195,
+      "step": 22810
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3671089080040823e-06,
+      "loss": 0.1548,
+      "step": 22820
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.359819215629101e-06,
+      "loss": 0.1255,
+      "step": 22830
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.352529523254119e-06,
+      "loss": 0.1413,
+      "step": 22840
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3452398308791372e-06,
+      "loss": 0.1321,
+      "step": 22850
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.337950138504155e-06,
+      "loss": 0.1429,
+      "step": 22860
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.330660446129174e-06,
+      "loss": 0.1452,
+      "step": 22870
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.323370753754192e-06,
+      "loss": 0.1303,
+      "step": 22880
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.31608106137921e-06,
+      "loss": 0.1308,
+      "step": 22890
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3087913690042283e-06,
+      "loss": 0.1615,
+      "step": 22900
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.301501676629246e-06,
+      "loss": 0.1517,
+      "step": 22910
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.294211984254265e-06,
+      "loss": 0.1409,
+      "step": 22920
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.286922291879283e-06,
+      "loss": 0.1376,
+      "step": 22930
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.279632599504301e-06,
+      "loss": 0.136,
+      "step": 22940
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2723429071293194e-06,
+      "loss": 0.1355,
+      "step": 22950
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.265053214754338e-06,
+      "loss": 0.1396,
+      "step": 22960
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.257763522379356e-06,
+      "loss": 0.1532,
+      "step": 22970
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2504738300043743e-06,
+      "loss": 0.1361,
+      "step": 22980
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.243184137629392e-06,
+      "loss": 0.1525,
+      "step": 22990
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2358944452544105e-06,
+      "loss": 0.1348,
+      "step": 23000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.228604752879429e-06,
+      "loss": 0.1311,
+      "step": 23010
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.221315060504447e-06,
+      "loss": 0.1244,
+      "step": 23020
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2140253681294654e-06,
+      "loss": 0.1294,
+      "step": 23030
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2067356757544832e-06,
+      "loss": 0.1017,
+      "step": 23040
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1994459833795015e-06,
+      "loss": 0.1292,
+      "step": 23050
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1921562910045203e-06,
+      "loss": 0.1326,
+      "step": 23060
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.184866598629538e-06,
+      "loss": 0.1289,
+      "step": 23070
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1775769062545564e-06,
+      "loss": 0.1281,
+      "step": 23080
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1702872138795743e-06,
+      "loss": 0.1345,
+      "step": 23090
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1629975215045926e-06,
+      "loss": 0.1365,
+      "step": 23100
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1557078291296113e-06,
+      "loss": 0.1504,
+      "step": 23110
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.148418136754629e-06,
+      "loss": 0.1127,
+      "step": 23120
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1411284443796475e-06,
+      "loss": 0.139,
+      "step": 23130
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.133838752004666e-06,
+      "loss": 0.1359,
+      "step": 23140
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1265490596296837e-06,
+      "loss": 0.1157,
+      "step": 23150
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1192593672547024e-06,
+      "loss": 0.153,
+      "step": 23160
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1119696748797203e-06,
+      "loss": 0.1774,
+      "step": 23170
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1046799825047386e-06,
+      "loss": 0.1298,
+      "step": 23180
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.097390290129757e-06,
+      "loss": 0.1667,
+      "step": 23190
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0901005977547748e-06,
+      "loss": 0.1275,
+      "step": 23200
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0828109053797935e-06,
+      "loss": 0.1523,
+      "step": 23210
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0755212130048114e-06,
+      "loss": 0.1492,
+      "step": 23220
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0682315206298297e-06,
+      "loss": 0.131,
+      "step": 23230
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.060941828254848e-06,
+      "loss": 0.1283,
+      "step": 23240
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.053652135879866e-06,
+      "loss": 0.1237,
+      "step": 23250
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0463624435048846e-06,
+      "loss": 0.1406,
+      "step": 23260
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.0390727511299024e-06,
+      "loss": 0.1371,
+      "step": 23270
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0317830587549207e-06,
+      "loss": 0.1334,
+      "step": 23280
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.024493366379939e-06,
+      "loss": 0.1448,
+      "step": 23290
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.017203674004957e-06,
+      "loss": 0.1472,
+      "step": 23300
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0099139816299756e-06,
+      "loss": 0.1416,
+      "step": 23310
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.002624289254994e-06,
+      "loss": 0.1863,
+      "step": 23320
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.995334596880012e-06,
+      "loss": 0.1658,
+      "step": 23330
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.98804490450503e-06,
+      "loss": 0.1611,
+      "step": 23340
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.980755212130048e-06,
+      "loss": 0.1498,
+      "step": 23350
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.9734655197550667e-06,
+      "loss": 0.1396,
+      "step": 23360
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.966175827380085e-06,
+      "loss": 0.1527,
+      "step": 23370
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.958886135005103e-06,
+      "loss": 0.1229,
+      "step": 23380
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.951596442630121e-06,
+      "loss": 0.167,
+      "step": 23390
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.944306750255139e-06,
+      "loss": 0.151,
+      "step": 23400
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.937017057880158e-06,
+      "loss": 0.1312,
+      "step": 23410
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.929727365505176e-06,
+      "loss": 0.1424,
+      "step": 23420
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.922437673130194e-06,
+      "loss": 0.146,
+      "step": 23430
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9151479807552123e-06,
+      "loss": 0.1421,
+      "step": 23440
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.90785828838023e-06,
+      "loss": 0.1522,
+      "step": 23450
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.900568596005249e-06,
+      "loss": 0.1273,
+      "step": 23460
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.893278903630267e-06,
+      "loss": 0.109,
+      "step": 23470
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.885989211255285e-06,
+      "loss": 0.1365,
+      "step": 23480
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8786995188803034e-06,
+      "loss": 0.1503,
+      "step": 23490
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.871409826505322e-06,
+      "loss": 0.1061,
+      "step": 23500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.86412013413034e-06,
+      "loss": 0.1144,
+      "step": 23510
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8568304417553583e-06,
+      "loss": 0.1311,
+      "step": 23520
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.849540749380376e-06,
+      "loss": 0.1272,
+      "step": 23530
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8422510570053944e-06,
+      "loss": 0.1278,
+      "step": 23540
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.834961364630413e-06,
+      "loss": 0.1348,
+      "step": 23550
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.827671672255431e-06,
+      "loss": 0.1175,
+      "step": 23560
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8203819798804493e-06,
+      "loss": 0.1182,
+      "step": 23570
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8130922875054672e-06,
+      "loss": 0.132,
+      "step": 23580
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8058025951304855e-06,
+      "loss": 0.1399,
+      "step": 23590
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7985129027555042e-06,
+      "loss": 0.1313,
+      "step": 23600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.791223210380522e-06,
+      "loss": 0.1477,
+      "step": 23610
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7839335180055404e-06,
+      "loss": 0.1478,
+      "step": 23620
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7766438256305583e-06,
+      "loss": 0.1123,
+      "step": 23630
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.7693541332555766e-06,
+      "loss": 0.1528,
+      "step": 23640
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7620644408805953e-06,
+      "loss": 0.1311,
+      "step": 23650
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.754774748505613e-06,
+      "loss": 0.144,
+      "step": 23660
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7474850561306315e-06,
+      "loss": 0.1538,
+      "step": 23670
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.74019536375565e-06,
+      "loss": 0.1368,
+      "step": 23680
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7329056713806677e-06,
+      "loss": 0.1188,
+      "step": 23690
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7256159790056864e-06,
+      "loss": 0.1465,
+      "step": 23700
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7183262866307043e-06,
+      "loss": 0.1291,
+      "step": 23710
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7110365942557226e-06,
+      "loss": 0.1282,
+      "step": 23720
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.703746901880741e-06,
+      "loss": 0.1398,
+      "step": 23730
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6964572095057587e-06,
+      "loss": 0.1394,
+      "step": 23740
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6891675171307775e-06,
+      "loss": 0.1202,
+      "step": 23750
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6818778247557953e-06,
+      "loss": 0.1229,
+      "step": 23760
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6745881323808136e-06,
+      "loss": 0.1282,
+      "step": 23770
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.667298440005832e-06,
+      "loss": 0.1601,
+      "step": 23780
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6600087476308502e-06,
+      "loss": 0.1259,
+      "step": 23790
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6527190552558685e-06,
+      "loss": 0.1287,
+      "step": 23800
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.645429362880887e-06,
+      "loss": 0.1174,
+      "step": 23810
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.6381396705059047e-06,
+      "loss": 0.1375,
+      "step": 23820
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6308499781309235e-06,
+      "loss": 0.1467,
+      "step": 23830
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6235602857559413e-06,
+      "loss": 0.1091,
+      "step": 23840
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6162705933809596e-06,
+      "loss": 0.1493,
+      "step": 23850
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.608980901005978e-06,
+      "loss": 0.1468,
+      "step": 23860
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.601691208630996e-06,
+      "loss": 0.1517,
+      "step": 23870
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5944015162560145e-06,
+      "loss": 0.1611,
+      "step": 23880
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5871118238810324e-06,
+      "loss": 0.1335,
+      "step": 23890
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.5798221315060507e-06,
+      "loss": 0.1386,
+      "step": 23900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.572532439131069e-06,
+      "loss": 0.1308,
+      "step": 23910
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.565242746756087e-06,
+      "loss": 0.1475,
+      "step": 23920
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5579530543811056e-06,
+      "loss": 0.118,
+      "step": 23930
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5506633620061235e-06,
+      "loss": 0.126,
+      "step": 23940
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5433736696311418e-06,
+      "loss": 0.1339,
+      "step": 23950
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.53608397725616e-06,
+      "loss": 0.1639,
+      "step": 23960
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.528794284881178e-06,
+      "loss": 0.13,
+      "step": 23970
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5215045925061967e-06,
+      "loss": 0.1269,
+      "step": 23980
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.514214900131215e-06,
+      "loss": 0.1288,
+      "step": 23990
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.506925207756233e-06,
+      "loss": 0.1376,
+      "step": 24000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.499635515381251e-06,
+      "loss": 0.14,
+      "step": 24010
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4923458230062695e-06,
+      "loss": 0.1238,
+      "step": 24020
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4850561306312878e-06,
+      "loss": 0.1342,
+      "step": 24030
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4777664382563056e-06,
+      "loss": 0.1277,
+      "step": 24040
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.470476745881324e-06,
+      "loss": 0.1471,
+      "step": 24050
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4631870535063422e-06,
+      "loss": 0.1458,
+      "step": 24060
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4558973611313605e-06,
+      "loss": 0.1295,
+      "step": 24070
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.448607668756379e-06,
+      "loss": 0.1493,
+      "step": 24080
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4413179763813967e-06,
+      "loss": 0.1411,
+      "step": 24090
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.434028284006415e-06,
+      "loss": 0.1366,
+      "step": 24100
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4267385916314333e-06,
+      "loss": 0.1662,
+      "step": 24110
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4194488992564516e-06,
+      "loss": 0.1629,
+      "step": 24120
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.41215920688147e-06,
+      "loss": 0.1224,
+      "step": 24130
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.404869514506488e-06,
+      "loss": 0.1139,
+      "step": 24140
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3975798221315065e-06,
+      "loss": 0.1317,
+      "step": 24150
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3902901297565244e-06,
+      "loss": 0.1291,
+      "step": 24160
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3830004373815427e-06,
+      "loss": 0.1484,
+      "step": 24170
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.375710745006561e-06,
+      "loss": 0.1151,
+      "step": 24180
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.368421052631579e-06,
+      "loss": 0.1546,
+      "step": 24190
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3611313602565976e-06,
+      "loss": 0.1649,
+      "step": 24200
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3538416678816155e-06,
+      "loss": 0.1534,
+      "step": 24210
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3465519755066338e-06,
+      "loss": 0.131,
+      "step": 24220
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.339262283131652e-06,
+      "loss": 0.1309,
+      "step": 24230
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3319725907566704e-06,
+      "loss": 0.1741,
+      "step": 24240
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3246828983816887e-06,
+      "loss": 0.1383,
+      "step": 24250
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3173932060067066e-06,
+      "loss": 0.1307,
+      "step": 24260
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.310103513631725e-06,
+      "loss": 0.133,
+      "step": 24270
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.302813821256743e-06,
+      "loss": 0.1663,
+      "step": 24280
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2955241288817615e-06,
+      "loss": 0.1133,
+      "step": 24290
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2882344365067798e-06,
+      "loss": 0.1355,
+      "step": 24300
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2809447441317976e-06,
+      "loss": 0.1366,
+      "step": 24310
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.273655051756816e-06,
+      "loss": 0.1191,
+      "step": 24320
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2663653593818342e-06,
+      "loss": 0.1535,
+      "step": 24330
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2590756670068525e-06,
+      "loss": 0.1326,
+      "step": 24340
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.251785974631871e-06,
+      "loss": 0.1325,
+      "step": 24350
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.2444962822568887e-06,
+      "loss": 0.1258,
+      "step": 24360
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.237206589881907e-06,
+      "loss": 0.15,
+      "step": 24370
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2299168975069253e-06,
+      "loss": 0.1114,
+      "step": 24380
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2226272051319436e-06,
+      "loss": 0.122,
+      "step": 24390
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.215337512756962e-06,
+      "loss": 0.1564,
+      "step": 24400
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2080478203819798e-06,
+      "loss": 0.1236,
+      "step": 24410
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2007581280069985e-06,
+      "loss": 0.1466,
+      "step": 24420
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1934684356320164e-06,
+      "loss": 0.1563,
+      "step": 24430
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.1861787432570347e-06,
+      "loss": 0.1123,
+      "step": 24440
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.178889050882053e-06,
+      "loss": 0.1425,
+      "step": 24450
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.171599358507071e-06,
+      "loss": 0.1629,
+      "step": 24460
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1643096661320896e-06,
+      "loss": 0.1332,
+      "step": 24470
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1570199737571075e-06,
+      "loss": 0.1554,
+      "step": 24480
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1497302813821258e-06,
+      "loss": 0.1261,
+      "step": 24490
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.142440589007144e-06,
+      "loss": 0.1153,
+      "step": 24500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1351508966321624e-06,
+      "loss": 0.132,
+      "step": 24510
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1278612042571807e-06,
+      "loss": 0.1198,
+      "step": 24520
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1205715118821985e-06,
+      "loss": 0.1109,
+      "step": 24530
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.113281819507217e-06,
+      "loss": 0.1526,
+      "step": 24540
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.105992127132235e-06,
+      "loss": 0.1404,
+      "step": 24550
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0987024347572534e-06,
+      "loss": 0.1269,
+      "step": 24560
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0914127423822717e-06,
+      "loss": 0.1304,
+      "step": 24570
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0841230500072896e-06,
+      "loss": 0.1348,
+      "step": 24580
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.076833357632308e-06,
+      "loss": 0.1421,
+      "step": 24590
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0695436652573262e-06,
+      "loss": 0.133,
+      "step": 24600
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0622539728823445e-06,
+      "loss": 0.132,
+      "step": 24610
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.054964280507363e-06,
+      "loss": 0.1308,
+      "step": 24620
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0476745881323807e-06,
+      "loss": 0.1606,
+      "step": 24630
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0403848957573994e-06,
+      "loss": 0.132,
+      "step": 24640
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0330952033824173e-06,
+      "loss": 0.1043,
+      "step": 24650
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0258055110074356e-06,
+      "loss": 0.1314,
+      "step": 24660
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.018515818632454e-06,
+      "loss": 0.1489,
+      "step": 24670
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0112261262574718e-06,
+      "loss": 0.1216,
+      "step": 24680
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0039364338824905e-06,
+      "loss": 0.1425,
+      "step": 24690
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9966467415075084e-06,
+      "loss": 0.1401,
+      "step": 24700
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9893570491325267e-06,
+      "loss": 0.1304,
+      "step": 24710
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.982067356757545e-06,
+      "loss": 0.1381,
+      "step": 24720
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9747776643825633e-06,
+      "loss": 0.1495,
+      "step": 24730
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9674879720075816e-06,
+      "loss": 0.1239,
+      "step": 24740
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9601982796326e-06,
+      "loss": 0.1321,
+      "step": 24750
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9529085872576178e-06,
+      "loss": 0.1013,
+      "step": 24760
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.945618894882636e-06,
+      "loss": 0.1303,
+      "step": 24770
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9383292025076544e-06,
+      "loss": 0.1049,
+      "step": 24780
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9310395101326727e-06,
+      "loss": 0.1305,
+      "step": 24790
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.923749817757691e-06,
+      "loss": 0.1317,
+      "step": 24800
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.916460125382709e-06,
+      "loss": 0.1476,
+      "step": 24810
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9091704330077276e-06,
+      "loss": 0.1056,
+      "step": 24820
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9018807406327454e-06,
+      "loss": 0.1408,
+      "step": 24830
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8945910482577637e-06,
+      "loss": 0.0962,
+      "step": 24840
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8873013558827818e-06,
+      "loss": 0.1136,
+      "step": 24850
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8800116635078e-06,
+      "loss": 0.1452,
+      "step": 24860
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8727219711328184e-06,
+      "loss": 0.1621,
+      "step": 24870
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8654322787578365e-06,
+      "loss": 0.1225,
+      "step": 24880
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8581425863828548e-06,
+      "loss": 0.1236,
+      "step": 24890
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.850852894007873e-06,
+      "loss": 0.127,
+      "step": 24900
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8435632016328914e-06,
+      "loss": 0.1298,
+      "step": 24910
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8362735092579095e-06,
+      "loss": 0.1391,
+      "step": 24920
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8289838168829278e-06,
+      "loss": 0.1318,
+      "step": 24930
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8216941245079459e-06,
+      "loss": 0.1412,
+      "step": 24940
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.814404432132964e-06,
+      "loss": 0.1163,
+      "step": 24950
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8071147397579825e-06,
+      "loss": 0.1312,
+      "step": 24960
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7998250473830006e-06,
+      "loss": 0.1429,
+      "step": 24970
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7925353550080189e-06,
+      "loss": 0.1323,
+      "step": 24980
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.785245662633037e-06,
+      "loss": 0.1292,
+      "step": 24990
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7779559702580555e-06,
+      "loss": 0.1533,
+      "step": 25000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7706662778830736e-06,
+      "loss": 0.1423,
+      "step": 25010
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7633765855080917e-06,
+      "loss": 0.1356,
+      "step": 25020
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.75608689313311e-06,
+      "loss": 0.141,
+      "step": 25030
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.748797200758128e-06,
+      "loss": 0.1542,
+      "step": 25040
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7415075083831466e-06,
+      "loss": 0.1104,
+      "step": 25050
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7342178160081646e-06,
+      "loss": 0.146,
+      "step": 25060
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7269281236331827e-06,
+      "loss": 0.1207,
+      "step": 25070
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.719638431258201e-06,
+      "loss": 0.1343,
+      "step": 25080
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7123487388832193e-06,
+      "loss": 0.1586,
+      "step": 25090
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7050590465082376e-06,
+      "loss": 0.1403,
+      "step": 25100
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6977693541332557e-06,
+      "loss": 0.1156,
+      "step": 25110
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6904796617582738e-06,
+      "loss": 0.1284,
+      "step": 25120
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6831899693832921e-06,
+      "loss": 0.1441,
+      "step": 25130
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6759002770083104e-06,
+      "loss": 0.1088,
+      "step": 25140
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6686105846333287e-06,
+      "loss": 0.144,
+      "step": 25150
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6613208922583468e-06,
+      "loss": 0.1492,
+      "step": 25160
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6540311998833649e-06,
+      "loss": 0.1129,
+      "step": 25170
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6467415075083834e-06,
+      "loss": 0.1178,
+      "step": 25180
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6394518151334015e-06,
+      "loss": 0.1253,
+      "step": 25190
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6321621227584198e-06,
+      "loss": 0.1295,
+      "step": 25200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6248724303834379e-06,
+      "loss": 0.1469,
+      "step": 25210
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.617582738008456e-06,
+      "loss": 0.1542,
+      "step": 25220
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6102930456334745e-06,
+      "loss": 0.1356,
+      "step": 25230
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6030033532584926e-06,
+      "loss": 0.0985,
+      "step": 25240
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5957136608835109e-06,
+      "loss": 0.124,
+      "step": 25250
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.588423968508529e-06,
+      "loss": 0.1369,
+      "step": 25260
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5811342761335475e-06,
+      "loss": 0.1478,
+      "step": 25270
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5738445837585656e-06,
+      "loss": 0.1303,
+      "step": 25280
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5665548913835837e-06,
+      "loss": 0.1662,
+      "step": 25290
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.559265199008602e-06,
+      "loss": 0.1486,
+      "step": 25300
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.55197550663362e-06,
+      "loss": 0.1537,
+      "step": 25310
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5446858142586386e-06,
+      "loss": 0.1436,
+      "step": 25320
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5373961218836566e-06,
+      "loss": 0.1639,
+      "step": 25330
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5301064295086747e-06,
+      "loss": 0.138,
+      "step": 25340
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.522816737133693e-06,
+      "loss": 0.1308,
+      "step": 25350
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5155270447587113e-06,
+      "loss": 0.1238,
+      "step": 25360
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5082373523837296e-06,
+      "loss": 0.1215,
+      "step": 25370
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5009476600087477e-06,
+      "loss": 0.1669,
+      "step": 25380
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.493657967633766e-06,
+      "loss": 0.1455,
+      "step": 25390
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4863682752587841e-06,
+      "loss": 0.1343,
+      "step": 25400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4790785828838026e-06,
+      "loss": 0.1377,
+      "step": 25410
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4717888905088207e-06,
+      "loss": 0.1193,
+      "step": 25420
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4644991981338388e-06,
+      "loss": 0.1316,
+      "step": 25430
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.457209505758857e-06,
+      "loss": 0.1452,
+      "step": 25440
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4499198133838754e-06,
+      "loss": 0.1204,
+      "step": 25450
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4426301210088937e-06,
+      "loss": 0.1264,
+      "step": 25460
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4353404286339118e-06,
+      "loss": 0.1424,
+      "step": 25470
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4280507362589299e-06,
+      "loss": 0.1451,
+      "step": 25480
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4207610438839482e-06,
+      "loss": 0.1259,
+      "step": 25490
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4134713515089665e-06,
+      "loss": 0.1322,
+      "step": 25500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4061816591339848e-06,
+      "loss": 0.1368,
+      "step": 25510
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3988919667590029e-06,
+      "loss": 0.1381,
+      "step": 25520
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.391602274384021e-06,
+      "loss": 0.1124,
+      "step": 25530
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3843125820090395e-06,
+      "loss": 0.1309,
+      "step": 25540
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3770228896340576e-06,
+      "loss": 0.1268,
+      "step": 25550
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.3697331972590759e-06,
+      "loss": 0.1228,
+      "step": 25560
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.362443504884094e-06,
+      "loss": 0.1597,
+      "step": 25570
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.355153812509112e-06,
+      "loss": 0.143,
+      "step": 25580
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3478641201341305e-06,
+      "loss": 0.1621,
+      "step": 25590
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3405744277591486e-06,
+      "loss": 0.1398,
+      "step": 25600
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.333284735384167e-06,
+      "loss": 0.1362,
+      "step": 25610
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.325995043009185e-06,
+      "loss": 0.1373,
+      "step": 25620
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3187053506342035e-06,
+      "loss": 0.145,
+      "step": 25630
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3114156582592216e-06,
+      "loss": 0.1465,
+      "step": 25640
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3041259658842397e-06,
+      "loss": 0.1618,
+      "step": 25650
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.296836273509258e-06,
+      "loss": 0.1328,
+      "step": 25660
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.289546581134276e-06,
+      "loss": 0.1317,
+      "step": 25670
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2822568887592946e-06,
+      "loss": 0.1346,
+      "step": 25680
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2749671963843127e-06,
+      "loss": 0.1473,
+      "step": 25690
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2676775040093308e-06,
+      "loss": 0.1303,
+      "step": 25700
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.260387811634349e-06,
+      "loss": 0.1202,
+      "step": 25710
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2530981192593674e-06,
+      "loss": 0.1362,
+      "step": 25720
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2458084268843857e-06,
+      "loss": 0.114,
+      "step": 25730
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2385187345094038e-06,
+      "loss": 0.1388,
+      "step": 25740
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2312290421344219e-06,
+      "loss": 0.144,
+      "step": 25750
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2239393497594402e-06,
+      "loss": 0.1235,
+      "step": 25760
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2166496573844585e-06,
+      "loss": 0.1344,
+      "step": 25770
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2093599650094768e-06,
+      "loss": 0.1437,
+      "step": 25780
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2020702726344949e-06,
+      "loss": 0.1312,
+      "step": 25790
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1947805802595132e-06,
+      "loss": 0.1335,
+      "step": 25800
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1874908878845312e-06,
+      "loss": 0.1252,
+      "step": 25810
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1802011955095495e-06,
+      "loss": 0.1417,
+      "step": 25820
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1729115031345678e-06,
+      "loss": 0.1414,
+      "step": 25830
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1656218107595861e-06,
+      "loss": 0.1205,
+      "step": 25840
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1583321183846042e-06,
+      "loss": 0.1355,
+      "step": 25850
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1510424260096225e-06,
+      "loss": 0.1273,
+      "step": 25860
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1437527336346408e-06,
+      "loss": 0.1451,
+      "step": 25870
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.136463041259659e-06,
+      "loss": 0.1499,
+      "step": 25880
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1291733488846772e-06,
+      "loss": 0.1265,
+      "step": 25890
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1218836565096953e-06,
+      "loss": 0.1131,
+      "step": 25900
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1145939641347136e-06,
+      "loss": 0.1315,
+      "step": 25910
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.107304271759732e-06,
+      "loss": 0.1497,
+      "step": 25920
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1000145793847502e-06,
+      "loss": 0.1111,
+      "step": 25930
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0927248870097683e-06,
+      "loss": 0.1375,
+      "step": 25940
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0854351946347864e-06,
+      "loss": 0.1308,
+      "step": 25950
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0781455022598047e-06,
+      "loss": 0.1802,
+      "step": 25960
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.070855809884823e-06,
+      "loss": 0.1168,
+      "step": 25970
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0635661175098413e-06,
+      "loss": 0.1257,
+      "step": 25980
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0562764251348594e-06,
+      "loss": 0.1326,
+      "step": 25990
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0489867327598777e-06,
+      "loss": 0.1324,
+      "step": 26000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0416970403848958e-06,
+      "loss": 0.1409,
+      "step": 26010
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.034407348009914e-06,
+      "loss": 0.1145,
+      "step": 26020
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0271176556349324e-06,
+      "loss": 0.1328,
+      "step": 26030
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0198279632599505e-06,
+      "loss": 0.1179,
+      "step": 26040
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0125382708849688e-06,
+      "loss": 0.1475,
+      "step": 26050
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0052485785099868e-06,
+      "loss": 0.1225,
+      "step": 26060
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.979588861350051e-07,
+      "loss": 0.1096,
+      "step": 26070
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.906691937600234e-07,
+      "loss": 0.1208,
+      "step": 26080
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.833795013850417e-07,
+      "loss": 0.1156,
+      "step": 26090
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.760898090100598e-07,
+      "loss": 0.1379,
+      "step": 26100
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.688001166350781e-07,
+      "loss": 0.1541,
+      "step": 26110
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.615104242600962e-07,
+      "loss": 0.1219,
+      "step": 26120
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.542207318851145e-07,
+      "loss": 0.1472,
+      "step": 26130
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.469310395101327e-07,
+      "loss": 0.1252,
+      "step": 26140
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.396413471351509e-07,
+      "loss": 0.1051,
+      "step": 26150
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.323516547601692e-07,
+      "loss": 0.1192,
+      "step": 26160
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.250619623851874e-07,
+      "loss": 0.1311,
+      "step": 26170
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.177722700102057e-07,
+      "loss": 0.1374,
+      "step": 26180
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.104825776352239e-07,
+      "loss": 0.1481,
+      "step": 26190
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.031928852602422e-07,
+      "loss": 0.1485,
+      "step": 26200
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.959031928852603e-07,
+      "loss": 0.1583,
+      "step": 26210
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.886135005102785e-07,
+      "loss": 0.1215,
+      "step": 26220
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.813238081352968e-07,
+      "loss": 0.1162,
+      "step": 26230
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.74034115760315e-07,
+      "loss": 0.1119,
+      "step": 26240
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.667444233853333e-07,
+      "loss": 0.1239,
+      "step": 26250
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.594547310103514e-07,
+      "loss": 0.116,
+      "step": 26260
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.521650386353697e-07,
+      "loss": 0.1348,
+      "step": 26270
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.448753462603879e-07,
+      "loss": 0.1535,
+      "step": 26280
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.375856538854062e-07,
+      "loss": 0.1256,
+      "step": 26290
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.302959615104244e-07,
+      "loss": 0.1154,
+      "step": 26300
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.230062691354424e-07,
+      "loss": 0.1525,
+      "step": 26310
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.157165767604607e-07,
+      "loss": 0.1325,
+      "step": 26320
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.084268843854789e-07,
+      "loss": 0.1629,
+      "step": 26330
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.011371920104972e-07,
+      "loss": 0.1563,
+      "step": 26340
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.938474996355154e-07,
+      "loss": 0.1355,
+      "step": 26350
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.865578072605337e-07,
+      "loss": 0.1043,
+      "step": 26360
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.792681148855518e-07,
+      "loss": 0.1183,
+      "step": 26370
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.719784225105701e-07,
+      "loss": 0.1432,
+      "step": 26380
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.646887301355883e-07,
+      "loss": 0.1356,
+      "step": 26390
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.573990377606065e-07,
+      "loss": 0.1343,
+      "step": 26400
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.501093453856248e-07,
+      "loss": 0.149,
+      "step": 26410
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.42819653010643e-07,
+      "loss": 0.1483,
+      "step": 26420
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.355299606356613e-07,
+      "loss": 0.1311,
+      "step": 26430
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.282402682606794e-07,
+      "loss": 0.1621,
+      "step": 26440
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.209505758856977e-07,
+      "loss": 0.1523,
+      "step": 26450
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.136608835107159e-07,
+      "loss": 0.1333,
+      "step": 26460
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.063711911357342e-07,
+      "loss": 0.1419,
+      "step": 26470
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.990814987607524e-07,
+      "loss": 0.1331,
+      "step": 26480
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.917918063857705e-07,
+      "loss": 0.1378,
+      "step": 26490
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.845021140107888e-07,
+      "loss": 0.1439,
+      "step": 26500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.77212421635807e-07,
+      "loss": 0.1111,
+      "step": 26510
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.699227292608253e-07,
+      "loss": 0.1179,
+      "step": 26520
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.626330368858435e-07,
+      "loss": 0.1478,
+      "step": 26530
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.553433445108618e-07,
+      "loss": 0.1562,
+      "step": 26540
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.480536521358799e-07,
+      "loss": 0.1281,
+      "step": 26550
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.407639597608982e-07,
+      "loss": 0.138,
+      "step": 26560
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.334742673859164e-07,
+      "loss": 0.1563,
+      "step": 26570
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.261845750109347e-07,
+      "loss": 0.1423,
+      "step": 26580
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.188948826359528e-07,
+      "loss": 0.1517,
+      "step": 26590
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.11605190260971e-07,
+      "loss": 0.1162,
+      "step": 26600
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.043154978859892e-07,
+      "loss": 0.1132,
+      "step": 26610
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.970258055110075e-07,
+      "loss": 0.1344,
+      "step": 26620
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.897361131360256e-07,
+      "loss": 0.1579,
+      "step": 26630
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.824464207610439e-07,
+      "loss": 0.1335,
+      "step": 26640
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.751567283860621e-07,
+      "loss": 0.1292,
+      "step": 26650
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.678670360110804e-07,
+      "loss": 0.1124,
+      "step": 26660
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.605773436360986e-07,
+      "loss": 0.1335,
+      "step": 26670
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.532876512611168e-07,
+      "loss": 0.1194,
+      "step": 26680
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.459979588861351e-07,
+      "loss": 0.1355,
+      "step": 26690
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.387082665111533e-07,
+      "loss": 0.1286,
+      "step": 26700
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.314185741361715e-07,
+      "loss": 0.1398,
+      "step": 26710
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.241288817611897e-07,
+      "loss": 0.144,
+      "step": 26720
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.168391893862079e-07,
+      "loss": 0.1408,
+      "step": 26730
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.095494970112262e-07,
+      "loss": 0.1357,
+      "step": 26740
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.022598046362444e-07,
+      "loss": 0.1392,
+      "step": 26750
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.949701122612626e-07,
+      "loss": 0.1485,
+      "step": 26760
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.876804198862809e-07,
+      "loss": 0.1176,
+      "step": 26770
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.803907275112991e-07,
+      "loss": 0.1445,
+      "step": 26780
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.731010351363173e-07,
+      "loss": 0.1443,
+      "step": 26790
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.658113427613355e-07,
+      "loss": 0.1151,
+      "step": 26800
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5852165038635376e-07,
+      "loss": 0.1046,
+      "step": 26810
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5123195801137195e-07,
+      "loss": 0.1396,
+      "step": 26820
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.4394226563639015e-07,
+      "loss": 0.1273,
+      "step": 26830
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.366525732614084e-07,
+      "loss": 0.1317,
+      "step": 26840
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.2936288088642664e-07,
+      "loss": 0.1484,
+      "step": 26850
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.2207318851144484e-07,
+      "loss": 0.1422,
+      "step": 26860
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.147834961364631e-07,
+      "loss": 0.1183,
+      "step": 26870
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0749380376148133e-07,
+      "loss": 0.1348,
+      "step": 26880
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.002041113864995e-07,
+      "loss": 0.1488,
+      "step": 26890
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.929144190115178e-07,
+      "loss": 0.1084,
+      "step": 26900
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.856247266365359e-07,
+      "loss": 0.1323,
+      "step": 26910
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.7833503426155416e-07,
+      "loss": 0.1244,
+      "step": 26920
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.710453418865724e-07,
+      "loss": 0.1251,
+      "step": 26930
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.6375564951159066e-07,
+      "loss": 0.138,
+      "step": 26940
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5646595713660885e-07,
+      "loss": 0.1433,
+      "step": 26950
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.491762647616271e-07,
+      "loss": 0.1162,
+      "step": 26960
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4188657238664535e-07,
+      "loss": 0.1339,
+      "step": 26970
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3459688001166354e-07,
+      "loss": 0.1403,
+      "step": 26980
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.273071876366818e-07,
+      "loss": 0.1294,
+      "step": 26990
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.2001749526169993e-07,
+      "loss": 0.1457,
+      "step": 27000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.127278028867182e-07,
+      "loss": 0.1487,
+      "step": 27010
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.054381105117364e-07,
+      "loss": 0.121,
+      "step": 27020
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.981484181367546e-07,
+      "loss": 0.1565,
+      "step": 27030
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9085872576177287e-07,
+      "loss": 0.1436,
+      "step": 27040
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.835690333867911e-07,
+      "loss": 0.1335,
+      "step": 27050
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.7627934101180936e-07,
+      "loss": 0.1182,
+      "step": 27060
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.6898964863682756e-07,
+      "loss": 0.142,
+      "step": 27070
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.6169995626184575e-07,
+      "loss": 0.1593,
+      "step": 27080
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.54410263886864e-07,
+      "loss": 0.11,
+      "step": 27090
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.4712057151188224e-07,
+      "loss": 0.1329,
+      "step": 27100
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.3983087913690044e-07,
+      "loss": 0.1463,
+      "step": 27110
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.3254118676191866e-07,
+      "loss": 0.1391,
+      "step": 27120
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.2525149438693688e-07,
+      "loss": 0.1316,
+      "step": 27130
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1796180201195513e-07,
+      "loss": 0.132,
+      "step": 27140
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1067210963697335e-07,
+      "loss": 0.1269,
+      "step": 27150
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0338241726199154e-07,
+      "loss": 0.14,
+      "step": 27160
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9609272488700976e-07,
+      "loss": 0.0933,
+      "step": 27170
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.88803032512028e-07,
+      "loss": 0.1228,
+      "step": 27180
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8151334013704623e-07,
+      "loss": 0.123,
+      "step": 27190
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.7422364776206448e-07,
+      "loss": 0.1357,
+      "step": 27200
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6693395538708267e-07,
+      "loss": 0.1358,
+      "step": 27210
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.596442630121009e-07,
+      "loss": 0.1383,
+      "step": 27220
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5235457063711912e-07,
+      "loss": 0.1505,
+      "step": 27230
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4506487826213734e-07,
+      "loss": 0.1243,
+      "step": 27240
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3777518588715559e-07,
+      "loss": 0.1203,
+      "step": 27250
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.304854935121738e-07,
+      "loss": 0.1241,
+      "step": 27260
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2319580113719203e-07,
+      "loss": 0.1525,
+      "step": 27270
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1590610876221025e-07,
+      "loss": 0.1435,
+      "step": 27280
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0861641638722847e-07,
+      "loss": 0.114,
+      "step": 27290
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0132672401224669e-07,
+      "loss": 0.1159,
+      "step": 27300
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.403703163726491e-08,
+      "loss": 0.1174,
+      "step": 27310
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.674733926228314e-08,
+      "loss": 0.1274,
+      "step": 27320
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.945764688730137e-08,
+      "loss": 0.1272,
+      "step": 27330
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.216795451231959e-08,
+      "loss": 0.1216,
+      "step": 27340
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.487826213733781e-08,
+      "loss": 0.1042,
+      "step": 27350
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.758856976235603e-08,
+      "loss": 0.132,
+      "step": 27360
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.029887738737426e-08,
+      "loss": 0.1059,
+      "step": 27370
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.3009185012392483e-08,
+      "loss": 0.1157,
+      "step": 27380
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5719492637410704e-08,
+      "loss": 0.1331,
+      "step": 27390
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.842980026242893e-08,
+      "loss": 0.1189,
+      "step": 27400
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.114010788744715e-08,
+      "loss": 0.1236,
+      "step": 27410
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3850415512465375e-08,
+      "loss": 0.1356,
+      "step": 27420
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.560723137483599e-09,
+      "loss": 0.1467,
+      "step": 27430
+    },
+    {
+      "epoch": 3.0,
+      "step": 27438,
+      "total_flos": 1.8689862064238756e+19,
+      "train_loss": 0.3784581153609504,
+      "train_runtime": 122473.9577,
+      "train_samples_per_second": 2.24,
+      "train_steps_per_second": 0.224
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 27438,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 1.8689862064238756e+19,
+  "trial_name": null,
+  "trial_params": null
+}