diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,22738 @@
+{
+  "best_metric": 0.8097558194444616,
+  "best_model_checkpoint": "beit-sketch-classifier-3/checkpoint-37818",
+  "epoch": 2.9999405080611576,
+  "global_step": 37818,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3220518244315177e-07,
+      "loss": 5.9891,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6441036488630353e-07,
+      "loss": 5.9912,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.966155473294554e-07,
+      "loss": 5.9855,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.288207297726071e-07,
+      "loss": 5.9777,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.61025912215759e-07,
+      "loss": 5.9682,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.932310946589108e-07,
+      "loss": 5.9583,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.254362771020624e-07,
+      "loss": 5.9642,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0576414595452141e-06,
+      "loss": 5.9244,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.189846641988366e-06,
+      "loss": 5.9327,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.322051824431518e-06,
+      "loss": 5.9028,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4542570068746696e-06,
+      "loss": 5.8884,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5864621893178215e-06,
+      "loss": 5.8929,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7186673717609732e-06,
+      "loss": 5.8864,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8508725542041249e-06,
+      "loss": 5.8835,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9830777366472766e-06,
+      "loss": 5.8472,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1152829190904283e-06,
+      "loss": 5.8354,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2474881015335804e-06,
+      "loss": 5.8199,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.379693283976732e-06,
+      "loss": 5.7909,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5118984664198837e-06,
+      "loss": 5.8001,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.644103648863036e-06,
+      "loss": 5.8095,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7763088313061875e-06,
+      "loss": 5.7756,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9085140137493392e-06,
+      "loss": 5.7659,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.040719196192491e-06,
+      "loss": 5.7628,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.172924378635643e-06,
+      "loss": 5.7628,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3051295610787943e-06,
+      "loss": 5.749,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4373347435219464e-06,
+      "loss": 5.7458,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5695399259650977e-06,
+      "loss": 5.6993,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7017451084082498e-06,
+      "loss": 5.7048,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.833950290851401e-06,
+      "loss": 5.6754,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.966155473294553e-06,
+      "loss": 5.674,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.098360655737704e-06,
+      "loss": 5.6554,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2305658381808565e-06,
+      "loss": 5.6303,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.362771020624009e-06,
+      "loss": 5.6263,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.494976203067161e-06,
+      "loss": 5.6009,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.627181385510313e-06,
+      "loss": 5.5668,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.759386567953464e-06,
+      "loss": 5.5595,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.891591750396616e-06,
+      "loss": 5.5141,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.0237969328397675e-06,
+      "loss": 5.5082,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.15600211528292e-06,
+      "loss": 5.4657,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.288207297726072e-06,
+      "loss": 5.4296,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.420412480169223e-06,
+      "loss": 5.4103,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.552617662612375e-06,
+      "loss": 5.3909,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.684822845055526e-06,
+      "loss": 5.3804,
+      "step": 430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8170280274986785e-06,
+      "loss": 5.3371,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.94923320994183e-06,
+      "loss": 5.2939,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.081438392384982e-06,
+      "loss": 5.2944,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.213643574828133e-06,
+      "loss": 5.2787,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.345848757271286e-06,
+      "loss": 5.2274,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.478053939714437e-06,
+      "loss": 5.1705,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.6102591221575886e-06,
+      "loss": 5.1352,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.74246430460074e-06,
+      "loss": 5.1156,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.874669487043893e-06,
+      "loss": 5.0545,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.006874669487044e-06,
+      "loss": 5.0058,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.139079851930195e-06,
+      "loss": 4.991,
+      "step": 540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.2712850343733474e-06,
+      "loss": 4.9273,
+      "step": 550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.4034902168164995e-06,
+      "loss": 4.9101,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.535695399259652e-06,
+      "loss": 4.8585,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.667900581702802e-06,
+      "loss": 4.8144,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.800105764145956e-06,
+      "loss": 4.7873,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.932310946589106e-06,
+      "loss": 4.7252,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 4.6762,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.196721311475409e-06,
+      "loss": 4.6263,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.328926493918563e-06,
+      "loss": 4.5897,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.461131676361713e-06,
+      "loss": 4.5432,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.593336858804865e-06,
+      "loss": 4.4465,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.725542041248017e-06,
+      "loss": 4.4202,
+      "step": 660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.85774722369117e-06,
+      "loss": 4.3874,
+      "step": 670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.989952406134321e-06,
+      "loss": 4.3087,
+      "step": 680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.122157588577472e-06,
+      "loss": 4.284,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.254362771020626e-06,
+      "loss": 4.2607,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.386567953463776e-06,
+      "loss": 4.1384,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.518773135906928e-06,
+      "loss": 4.1016,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.65097831835008e-06,
+      "loss": 4.0673,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.783183500793232e-06,
+      "loss": 4.0227,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915388683236383e-06,
+      "loss": 3.9688,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0047593865679535e-05,
+      "loss": 3.9706,
+      "step": 760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0179799048122687e-05,
+      "loss": 3.8952,
+      "step": 770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.031200423056584e-05,
+      "loss": 3.832,
+      "step": 780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.044420941300899e-05,
+      "loss": 3.7726,
+      "step": 790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0576414595452143e-05,
+      "loss": 3.7676,
+      "step": 800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0708619777895294e-05,
+      "loss": 3.6796,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0840824960338446e-05,
+      "loss": 3.6294,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0973030142781596e-05,
+      "loss": 3.5967,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.110523532522475e-05,
+      "loss": 3.5669,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.12374405076679e-05,
+      "loss": 3.4788,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1369645690111053e-05,
+      "loss": 3.4677,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1501850872554205e-05,
+      "loss": 3.3864,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1634056054997357e-05,
+      "loss": 3.3375,
+      "step": 880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1766261237440509e-05,
+      "loss": 3.3317,
+      "step": 890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.189846641988366e-05,
+      "loss": 3.2866,
+      "step": 900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2030671602326812e-05,
+      "loss": 3.2172,
+      "step": 910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2162876784769964e-05,
+      "loss": 3.1842,
+      "step": 920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2295081967213116e-05,
+      "loss": 3.1644,
+      "step": 930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2427287149656266e-05,
+      "loss": 3.0838,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2559492332099418e-05,
+      "loss": 3.1015,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2691697514542572e-05,
+      "loss": 2.9638,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2823902696985724e-05,
+      "loss": 2.952,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2956107879428875e-05,
+      "loss": 2.973,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3088313061872027e-05,
+      "loss": 2.8985,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3220518244315177e-05,
+      "loss": 2.9158,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.335272342675833e-05,
+      "loss": 2.8414,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.348492860920148e-05,
+      "loss": 2.8071,
+      "step": 1020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3617133791644632e-05,
+      "loss": 2.7327,
+      "step": 1030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3749338974087786e-05,
+      "loss": 2.7512,
+      "step": 1040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3881544156530938e-05,
+      "loss": 2.7351,
+      "step": 1050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4013749338974088e-05,
+      "loss": 2.6775,
+      "step": 1060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.414595452141724e-05,
+      "loss": 2.6238,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.427815970386039e-05,
+      "loss": 2.5239,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4410364886303543e-05,
+      "loss": 2.5905,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4542570068746695e-05,
+      "loss": 2.5621,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4674775251189849e-05,
+      "loss": 2.5286,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4806980433632999e-05,
+      "loss": 2.4871,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4939185616076151e-05,
+      "loss": 2.4122,
+      "step": 1130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5071390798519303e-05,
+      "loss": 2.4369,
+      "step": 1140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5203595980962454e-05,
+      "loss": 2.4167,
+      "step": 1150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5335801163405604e-05,
+      "loss": 2.4037,
+      "step": 1160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5468006345848758e-05,
+      "loss": 2.3923,
+      "step": 1170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5600211528291912e-05,
+      "loss": 2.3209,
+      "step": 1180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5732416710735062e-05,
+      "loss": 2.3418,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.5864621893178213e-05,
+      "loss": 2.2481,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.5996827075621366e-05,
+      "loss": 2.2666,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 2.2052,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6261237440507667e-05,
+      "loss": 2.2008,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6393442622950818e-05,
+      "loss": 2.2345,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6525647805393975e-05,
+      "loss": 2.1584,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6657852987837125e-05,
+      "loss": 2.0938,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6790058170280276e-05,
+      "loss": 2.123,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6922263352723426e-05,
+      "loss": 2.0644,
+      "step": 1280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.705446853516658e-05,
+      "loss": 2.085,
+      "step": 1290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.718667371760973e-05,
+      "loss": 2.1379,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.731887890005288e-05,
+      "loss": 2.0422,
+      "step": 1310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7451084082496035e-05,
+      "loss": 2.0226,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.758328926493919e-05,
+      "loss": 2.013,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.771549444738234e-05,
+      "loss": 1.9788,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.784769962982549e-05,
+      "loss": 2.0106,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7979904812268643e-05,
+      "loss": 2.001,
+      "step": 1360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8112109994711793e-05,
+      "loss": 1.9276,
+      "step": 1370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8244315177154944e-05,
+      "loss": 1.9313,
+      "step": 1380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8376520359598098e-05,
+      "loss": 1.9863,
+      "step": 1390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.850872554204125e-05,
+      "loss": 1.9149,
+      "step": 1400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8640930724484402e-05,
+      "loss": 1.8749,
+      "step": 1410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8773135906927552e-05,
+      "loss": 1.8602,
+      "step": 1420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8905341089370703e-05,
+      "loss": 1.9082,
+      "step": 1430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9037546271813856e-05,
+      "loss": 1.8898,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9169751454257007e-05,
+      "loss": 1.824,
+      "step": 1450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.930195663670016e-05,
+      "loss": 1.848,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.943416181914331e-05,
+      "loss": 1.8598,
+      "step": 1470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566367001586465e-05,
+      "loss": 1.8003,
+      "step": 1480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9698572184029615e-05,
+      "loss": 1.7974,
+      "step": 1490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9830777366472766e-05,
+      "loss": 1.8118,
+      "step": 1500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9962982548915916e-05,
+      "loss": 1.7324,
+      "step": 1510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.009518773135907e-05,
+      "loss": 1.7799,
+      "step": 1520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.0227392913802224e-05,
+      "loss": 1.7088,
+      "step": 1530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.0359598096245374e-05,
+      "loss": 1.745,
+      "step": 1540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.0491803278688525e-05,
+      "loss": 1.7512,
+      "step": 1550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.062400846113168e-05,
+      "loss": 1.7203,
+      "step": 1560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.075621364357483e-05,
+      "loss": 1.7191,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.088841882601798e-05,
+      "loss": 1.7021,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1020624008461133e-05,
+      "loss": 1.6556,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1152829190904287e-05,
+      "loss": 1.6529,
+      "step": 1600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1285034373347437e-05,
+      "loss": 1.687,
+      "step": 1610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1417239555790588e-05,
+      "loss": 1.675,
+      "step": 1620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.154944473823374e-05,
+      "loss": 1.6889,
+      "step": 1630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1681649920676892e-05,
+      "loss": 1.6073,
+      "step": 1640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1813855103120042e-05,
+      "loss": 1.5961,
+      "step": 1650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1946060285563193e-05,
+      "loss": 1.672,
+      "step": 1660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.2078265468006347e-05,
+      "loss": 1.6347,
+      "step": 1670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.22104706504495e-05,
+      "loss": 1.6274,
+      "step": 1680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.234267583289265e-05,
+      "loss": 1.6514,
+      "step": 1690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.24748810153358e-05,
+      "loss": 1.5666,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2607086197778955e-05,
+      "loss": 1.6136,
+      "step": 1710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2739291380222105e-05,
+      "loss": 1.5622,
+      "step": 1720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2871496562665256e-05,
+      "loss": 1.5735,
+      "step": 1730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.300370174510841e-05,
+      "loss": 1.5654,
+      "step": 1740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3135906927551563e-05,
+      "loss": 1.5277,
+      "step": 1750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3268112109994714e-05,
+      "loss": 1.664,
+      "step": 1760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3400317292437864e-05,
+      "loss": 1.5339,
+      "step": 1770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3532522474881018e-05,
+      "loss": 1.6034,
+      "step": 1780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.366472765732417e-05,
+      "loss": 1.6094,
+      "step": 1790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.379693283976732e-05,
+      "loss": 1.5654,
+      "step": 1800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.392913802221047e-05,
+      "loss": 1.5229,
+      "step": 1810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.4061343204653623e-05,
+      "loss": 1.5363,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 1.5467,
+      "step": 1830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.4325753569539927e-05,
+      "loss": 1.5177,
+      "step": 1840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.4457958751983078e-05,
+      "loss": 1.5886,
+      "step": 1850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.459016393442623e-05,
+      "loss": 1.4747,
+      "step": 1860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.4722369116869382e-05,
+      "loss": 1.548,
+      "step": 1870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.4854574299312532e-05,
+      "loss": 1.4489,
+      "step": 1880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.4986779481755686e-05,
+      "loss": 1.5143,
+      "step": 1890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5118984664198837e-05,
+      "loss": 1.4454,
+      "step": 1900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5251189846641987e-05,
+      "loss": 1.4729,
+      "step": 1910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5383395029085144e-05,
+      "loss": 1.464,
+      "step": 1920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5515600211528295e-05,
+      "loss": 1.4249,
+      "step": 1930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.564780539397145e-05,
+      "loss": 1.4309,
+      "step": 1940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.57800105764146e-05,
+      "loss": 1.4612,
+      "step": 1950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.591221575885775e-05,
+      "loss": 1.4222,
+      "step": 1960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.60444209413009e-05,
+      "loss": 1.4511,
+      "step": 1970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.6176626123744053e-05,
+      "loss": 1.4665,
+      "step": 1980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.6308831306187204e-05,
+      "loss": 1.4681,
+      "step": 1990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.6441036488630354e-05,
+      "loss": 1.4197,
+      "step": 2000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.6573241671073508e-05,
+      "loss": 1.4299,
+      "step": 2010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.670544685351666e-05,
+      "loss": 1.4147,
+      "step": 2020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.683765203595981e-05,
+      "loss": 1.4352,
+      "step": 2030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.696985721840296e-05,
+      "loss": 1.4725,
+      "step": 2040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7102062400846113e-05,
+      "loss": 1.4084,
+      "step": 2050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7234267583289264e-05,
+      "loss": 1.4442,
+      "step": 2060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.736647276573242e-05,
+      "loss": 1.397,
+      "step": 2070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.749867794817557e-05,
+      "loss": 1.3977,
+      "step": 2080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.763088313061872e-05,
+      "loss": 1.3826,
+      "step": 2090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7763088313061875e-05,
+      "loss": 1.3964,
+      "step": 2100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7895293495505026e-05,
+      "loss": 1.3914,
+      "step": 2110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8027498677948176e-05,
+      "loss": 1.3374,
+      "step": 2120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.815970386039133e-05,
+      "loss": 1.3985,
+      "step": 2130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.829190904283448e-05,
+      "loss": 1.3384,
+      "step": 2140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.842411422527763e-05,
+      "loss": 1.3818,
+      "step": 2150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.855631940772078e-05,
+      "loss": 1.3785,
+      "step": 2160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8688524590163935e-05,
+      "loss": 1.2767,
+      "step": 2170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8820729772607085e-05,
+      "loss": 1.3576,
+      "step": 2180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8952934955050236e-05,
+      "loss": 1.2954,
+      "step": 2190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.908514013749339e-05,
+      "loss": 1.3496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9217345319936547e-05,
+      "loss": 1.3519,
+      "step": 2210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9349550502379697e-05,
+      "loss": 1.3026,
+      "step": 2220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9481755684822848e-05,
+      "loss": 1.3694,
+      "step": 2230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9613960867265998e-05,
+      "loss": 1.3022,
+      "step": 2240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9746166049709152e-05,
+      "loss": 1.3564,
+      "step": 2250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9878371232152302e-05,
+      "loss": 1.3699,
+      "step": 2260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.0010576414595453e-05,
+      "loss": 1.3269,
+      "step": 2270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.0142781597038607e-05,
+      "loss": 1.4007,
+      "step": 2280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.0274986779481757e-05,
+      "loss": 1.2685,
+      "step": 2290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.0407191961924907e-05,
+      "loss": 1.328,
+      "step": 2300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.053939714436806e-05,
+      "loss": 1.2992,
+      "step": 2310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.067160232681121e-05,
+      "loss": 1.3495,
+      "step": 2320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.0803807509254365e-05,
+      "loss": 1.3019,
+      "step": 2330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.0936012691697516e-05,
+      "loss": 1.2776,
+      "step": 2340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.106821787414067e-05,
+      "loss": 1.3255,
+      "step": 2350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.1200423056583823e-05,
+      "loss": 1.3343,
+      "step": 2360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.1332628239026974e-05,
+      "loss": 1.3054,
+      "step": 2370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.1464833421470124e-05,
+      "loss": 1.3025,
+      "step": 2380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.1597038603913275e-05,
+      "loss": 1.2936,
+      "step": 2390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.1729243786356425e-05,
+      "loss": 1.2325,
+      "step": 2400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.1861448968799576e-05,
+      "loss": 1.2984,
+      "step": 2410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.199365415124273e-05,
+      "loss": 1.3088,
+      "step": 2420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.212585933368588e-05,
+      "loss": 1.2412,
+      "step": 2430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 1.2971,
+      "step": 2440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.2390269698572184e-05,
+      "loss": 1.2618,
+      "step": 2450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.2522474881015334e-05,
+      "loss": 1.2842,
+      "step": 2460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.2654680063458485e-05,
+      "loss": 1.2856,
+      "step": 2470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.2786885245901635e-05,
+      "loss": 1.2906,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.291909042834479e-05,
+      "loss": 1.306,
+      "step": 2490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.305129561078795e-05,
+      "loss": 1.2524,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.31835007932311e-05,
+      "loss": 1.236,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.331570597567425e-05,
+      "loss": 1.2734,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.34479111581174e-05,
+      "loss": 1.2939,
+      "step": 2530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.358011634056055e-05,
+      "loss": 1.2459,
+      "step": 2540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.37123215230037e-05,
+      "loss": 1.2604,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.384452670544685e-05,
+      "loss": 1.3138,
+      "step": 2560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.397673188789001e-05,
+      "loss": 1.2595,
+      "step": 2570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.410893707033316e-05,
+      "loss": 1.2752,
+      "step": 2580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.424114225277631e-05,
+      "loss": 1.1998,
+      "step": 2590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.437334743521946e-05,
+      "loss": 1.2713,
+      "step": 2600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.450555261766261e-05,
+      "loss": 1.2832,
+      "step": 2610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.463775780010576e-05,
+      "loss": 1.2062,
+      "step": 2620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.476996298254891e-05,
+      "loss": 1.2631,
+      "step": 2630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.490216816499207e-05,
+      "loss": 1.2551,
+      "step": 2640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.5034373347435226e-05,
+      "loss": 1.2405,
+      "step": 2650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.516657852987838e-05,
+      "loss": 1.1769,
+      "step": 2660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.529878371232153e-05,
+      "loss": 1.2036,
+      "step": 2670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.543098889476468e-05,
+      "loss": 1.2572,
+      "step": 2680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.556319407720783e-05,
+      "loss": 1.2428,
+      "step": 2690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.569539925965098e-05,
+      "loss": 1.2261,
+      "step": 2700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.582760444209413e-05,
+      "loss": 1.2696,
+      "step": 2710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.5959809624537286e-05,
+      "loss": 1.211,
+      "step": 2720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6092014806980436e-05,
+      "loss": 1.2475,
+      "step": 2730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.622421998942359e-05,
+      "loss": 1.2716,
+      "step": 2740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.635642517186674e-05,
+      "loss": 1.2766,
+      "step": 2750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.648863035430989e-05,
+      "loss": 1.2155,
+      "step": 2760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.662083553675304e-05,
+      "loss": 1.2578,
+      "step": 2770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6753040719196195e-05,
+      "loss": 1.1874,
+      "step": 2780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.6885245901639346e-05,
+      "loss": 1.1888,
+      "step": 2790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.70174510840825e-05,
+      "loss": 1.2269,
+      "step": 2800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.714965626652565e-05,
+      "loss": 1.1628,
+      "step": 2810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.7281861448968804e-05,
+      "loss": 1.1903,
+      "step": 2820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.7414066631411954e-05,
+      "loss": 1.1944,
+      "step": 2830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.7546271813855104e-05,
+      "loss": 1.2125,
+      "step": 2840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.7678476996298255e-05,
+      "loss": 1.1776,
+      "step": 2850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.7810682178741405e-05,
+      "loss": 1.2057,
+      "step": 2860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.794288736118456e-05,
+      "loss": 1.2036,
+      "step": 2870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.807509254362771e-05,
+      "loss": 1.2108,
+      "step": 2880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.820729772607086e-05,
+      "loss": 1.2276,
+      "step": 2890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8339502908514014e-05,
+      "loss": 1.1962,
+      "step": 2900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8471708090957164e-05,
+      "loss": 1.2506,
+      "step": 2910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.860391327340032e-05,
+      "loss": 1.2194,
+      "step": 2920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.873611845584347e-05,
+      "loss": 1.2596,
+      "step": 2930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.886832363828662e-05,
+      "loss": 1.1833,
+      "step": 2940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.900052882072977e-05,
+      "loss": 1.1901,
+      "step": 2950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.913273400317293e-05,
+      "loss": 1.2302,
+      "step": 2960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.926493918561608e-05,
+      "loss": 1.2149,
+      "step": 2970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.939714436805923e-05,
+      "loss": 1.2127,
+      "step": 2980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.952934955050238e-05,
+      "loss": 1.1905,
+      "step": 2990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.966155473294553e-05,
+      "loss": 1.1154,
+      "step": 3000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.979375991538868e-05,
+      "loss": 1.158,
+      "step": 3010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.992596509783183e-05,
+      "loss": 1.1861,
+      "step": 3020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.005817028027499e-05,
+      "loss": 1.2127,
+      "step": 3030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.019037546271814e-05,
+      "loss": 1.1856,
+      "step": 3040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.032258064516129e-05,
+      "loss": 1.2378,
+      "step": 3050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.045478582760445e-05,
+      "loss": 1.1807,
+      "step": 3060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.05869910100476e-05,
+      "loss": 1.1552,
+      "step": 3070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.071919619249075e-05,
+      "loss": 1.1293,
+      "step": 3080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.08514013749339e-05,
+      "loss": 1.1768,
+      "step": 3090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.098360655737705e-05,
+      "loss": 1.1249,
+      "step": 3100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.1115811739820206e-05,
+      "loss": 1.1174,
+      "step": 3110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.124801692226336e-05,
+      "loss": 1.1545,
+      "step": 3120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.138022210470651e-05,
+      "loss": 1.1336,
+      "step": 3130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.151242728714966e-05,
+      "loss": 1.1194,
+      "step": 3140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.164463246959281e-05,
+      "loss": 1.1712,
+      "step": 3150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.177683765203596e-05,
+      "loss": 1.1527,
+      "step": 3160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.190904283447911e-05,
+      "loss": 1.1586,
+      "step": 3170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2041248016922266e-05,
+      "loss": 1.1713,
+      "step": 3180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2173453199365416e-05,
+      "loss": 1.1287,
+      "step": 3190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2305658381808574e-05,
+      "loss": 1.1501,
+      "step": 3200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2437863564251724e-05,
+      "loss": 1.1235,
+      "step": 3210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2570068746694874e-05,
+      "loss": 1.1414,
+      "step": 3220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2702273929138025e-05,
+      "loss": 1.1427,
+      "step": 3230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2834479111581175e-05,
+      "loss": 1.0933,
+      "step": 3240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2966684294024326e-05,
+      "loss": 1.2132,
+      "step": 3250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.309888947646748e-05,
+      "loss": 1.0981,
+      "step": 3260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.323109465891063e-05,
+      "loss": 1.1652,
+      "step": 3270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3363299841353784e-05,
+      "loss": 1.0994,
+      "step": 3280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3495505023796934e-05,
+      "loss": 1.1346,
+      "step": 3290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3627710206240085e-05,
+      "loss": 1.1514,
+      "step": 3300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3759915388683235e-05,
+      "loss": 1.0706,
+      "step": 3310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3892120571126385e-05,
+      "loss": 1.1173,
+      "step": 3320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.402432575356954e-05,
+      "loss": 1.1702,
+      "step": 3330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.415653093601269e-05,
+      "loss": 1.1624,
+      "step": 3340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.428873611845585e-05,
+      "loss": 1.1331,
+      "step": 3350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.4420941300899e-05,
+      "loss": 1.1668,
+      "step": 3360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.455314648334215e-05,
+      "loss": 1.1069,
+      "step": 3370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.46853516657853e-05,
+      "loss": 1.1323,
+      "step": 3380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.481755684822845e-05,
+      "loss": 1.1333,
+      "step": 3390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.49497620306716e-05,
+      "loss": 1.0968,
+      "step": 3400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.508196721311476e-05,
+      "loss": 1.1326,
+      "step": 3410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.521417239555791e-05,
+      "loss": 1.1067,
+      "step": 3420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.534637757800106e-05,
+      "loss": 1.117,
+      "step": 3430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.547858276044421e-05,
+      "loss": 1.1481,
+      "step": 3440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.561078794288736e-05,
+      "loss": 1.1456,
+      "step": 3450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.574299312533051e-05,
+      "loss": 1.1022,
+      "step": 3460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.587519830777366e-05,
+      "loss": 1.1551,
+      "step": 3470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.600740349021682e-05,
+      "loss": 1.1018,
+      "step": 3480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.6139608672659976e-05,
+      "loss": 1.1383,
+      "step": 3490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.627181385510313e-05,
+      "loss": 1.1411,
+      "step": 3500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.640401903754628e-05,
+      "loss": 1.0941,
+      "step": 3510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.653622421998943e-05,
+      "loss": 1.0946,
+      "step": 3520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.666842940243258e-05,
+      "loss": 1.1184,
+      "step": 3530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.680063458487573e-05,
+      "loss": 1.1084,
+      "step": 3540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.693283976731888e-05,
+      "loss": 1.1608,
+      "step": 3550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7065044949762036e-05,
+      "loss": 1.0914,
+      "step": 3560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7197250132205186e-05,
+      "loss": 1.1114,
+      "step": 3570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.732945531464834e-05,
+      "loss": 1.1296,
+      "step": 3580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.746166049709149e-05,
+      "loss": 1.1169,
+      "step": 3590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.759386567953464e-05,
+      "loss": 1.1281,
+      "step": 3600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.772607086197779e-05,
+      "loss": 1.1119,
+      "step": 3610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.785827604442094e-05,
+      "loss": 1.1357,
+      "step": 3620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7990481226864096e-05,
+      "loss": 1.1205,
+      "step": 3630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8122686409307246e-05,
+      "loss": 1.1445,
+      "step": 3640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.82548915917504e-05,
+      "loss": 1.1026,
+      "step": 3650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8387096774193554e-05,
+      "loss": 1.1373,
+      "step": 3660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8519301956636704e-05,
+      "loss": 1.1154,
+      "step": 3670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8651507139079855e-05,
+      "loss": 1.1175,
+      "step": 3680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8783712321523005e-05,
+      "loss": 1.0834,
+      "step": 3690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8915917503966155e-05,
+      "loss": 1.0919,
+      "step": 3700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9048122686409306e-05,
+      "loss": 1.0724,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.918032786885246e-05,
+      "loss": 1.1008,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9312533051295613e-05,
+      "loss": 1.1422,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9444738233738764e-05,
+      "loss": 1.1226,
+      "step": 3740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9576943416181914e-05,
+      "loss": 1.0252,
+      "step": 3750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9709148598625065e-05,
+      "loss": 1.1189,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.984135378106822e-05,
+      "loss": 1.1417,
+      "step": 3770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.997355896351137e-05,
+      "loss": 1.0808,
+      "step": 3780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.998824773768951e-05,
+      "loss": 1.0934,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.997355740980139e-05,
+      "loss": 1.0674,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9958867081913267e-05,
+      "loss": 1.1266,
+      "step": 3810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9944176754025155e-05,
+      "loss": 1.1221,
+      "step": 3820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9929486426137036e-05,
+      "loss": 1.1156,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.991479609824892e-05,
+      "loss": 1.1261,
+      "step": 3840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.990010577036079e-05,
+      "loss": 1.1171,
+      "step": 3850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9885415442472674e-05,
+      "loss": 1.1303,
+      "step": 3860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987072511458456e-05,
+      "loss": 1.1185,
+      "step": 3870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9856034786696444e-05,
+      "loss": 1.1262,
+      "step": 3880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9841344458808325e-05,
+      "loss": 1.1094,
+      "step": 3890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.98266541309202e-05,
+      "loss": 1.1499,
+      "step": 3900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.981196380303209e-05,
+      "loss": 1.0938,
+      "step": 3910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.979727347514397e-05,
+      "loss": 1.1202,
+      "step": 3920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.978258314725585e-05,
+      "loss": 1.1034,
+      "step": 3930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9767892819367726e-05,
+      "loss": 1.0568,
+      "step": 3940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9753202491479614e-05,
+      "loss": 1.1165,
+      "step": 3950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9738512163591496e-05,
+      "loss": 1.0565,
+      "step": 3960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.972382183570338e-05,
+      "loss": 1.1031,
+      "step": 3970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.970913150781525e-05,
+      "loss": 1.0915,
+      "step": 3980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9694441179927134e-05,
+      "loss": 1.0822,
+      "step": 3990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.967975085203902e-05,
+      "loss": 1.033,
+      "step": 4000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9665060524150903e-05,
+      "loss": 1.0827,
+      "step": 4010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9650370196262785e-05,
+      "loss": 1.1098,
+      "step": 4020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.963567986837466e-05,
+      "loss": 1.1144,
+      "step": 4030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.962098954048655e-05,
+      "loss": 1.1156,
+      "step": 4040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.960629921259843e-05,
+      "loss": 1.0344,
+      "step": 4050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.959160888471031e-05,
+      "loss": 1.0137,
+      "step": 4060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9576918556822186e-05,
+      "loss": 1.064,
+      "step": 4070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9562228228934074e-05,
+      "loss": 1.0881,
+      "step": 4080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9547537901045955e-05,
+      "loss": 1.0326,
+      "step": 4090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.953284757315784e-05,
+      "loss": 1.0487,
+      "step": 4100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.951815724526972e-05,
+      "loss": 1.0543,
+      "step": 4110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.95034669173816e-05,
+      "loss": 1.033,
+      "step": 4120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.948877658949348e-05,
+      "loss": 1.0469,
+      "step": 4130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.947408626160536e-05,
+      "loss": 1.054,
+      "step": 4140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9459395933717244e-05,
+      "loss": 1.0159,
+      "step": 4150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.944470560582912e-05,
+      "loss": 1.1134,
+      "step": 4160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.943001527794101e-05,
+      "loss": 1.013,
+      "step": 4170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.941532495005289e-05,
+      "loss": 1.0067,
+      "step": 4180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.940063462216477e-05,
+      "loss": 1.069,
+      "step": 4190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9385944294276645e-05,
+      "loss": 1.0707,
+      "step": 4200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9371253966388534e-05,
+      "loss": 1.04,
+      "step": 4210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9356563638500415e-05,
+      "loss": 1.0992,
+      "step": 4220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9341873310612297e-05,
+      "loss": 1.0531,
+      "step": 4230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.932718298272418e-05,
+      "loss": 1.0138,
+      "step": 4240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.931249265483606e-05,
+      "loss": 1.0672,
+      "step": 4250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.929780232694794e-05,
+      "loss": 1.0637,
+      "step": 4260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.928311199905982e-05,
+      "loss": 1.064,
+      "step": 4270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9268421671171704e-05,
+      "loss": 1.0332,
+      "step": 4280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9253731343283586e-05,
+      "loss": 1.0582,
+      "step": 4290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.923904101539547e-05,
+      "loss": 1.0621,
+      "step": 4300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.922435068750735e-05,
+      "loss": 1.05,
+      "step": 4310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.920966035961923e-05,
+      "loss": 1.0859,
+      "step": 4320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9194970031731105e-05,
+      "loss": 1.0393,
+      "step": 4330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.918027970384299e-05,
+      "loss": 1.024,
+      "step": 4340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9165589375954875e-05,
+      "loss": 1.0263,
+      "step": 4350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9150899048066756e-05,
+      "loss": 1.0618,
+      "step": 4360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.913620872017864e-05,
+      "loss": 1.0673,
+      "step": 4370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.912151839229052e-05,
+      "loss": 0.9843,
+      "step": 4380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.91068280644024e-05,
+      "loss": 1.043,
+      "step": 4390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.909213773651428e-05,
+      "loss": 1.0507,
+      "step": 4400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9077447408626164e-05,
+      "loss": 1.1045,
+      "step": 4410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9062757080738045e-05,
+      "loss": 1.0863,
+      "step": 4420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.904806675284993e-05,
+      "loss": 1.0353,
+      "step": 4430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.903337642496181e-05,
+      "loss": 1.0651,
+      "step": 4440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.901868609707369e-05,
+      "loss": 1.0464,
+      "step": 4450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.900399576918557e-05,
+      "loss": 1.114,
+      "step": 4460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.898930544129745e-05,
+      "loss": 1.0555,
+      "step": 4470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8974615113409334e-05,
+      "loss": 1.0387,
+      "step": 4480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8959924785521216e-05,
+      "loss": 1.0201,
+      "step": 4490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.89452344576331e-05,
+      "loss": 1.0673,
+      "step": 4500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.893054412974498e-05,
+      "loss": 1.0352,
+      "step": 4510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.891585380185686e-05,
+      "loss": 0.9945,
+      "step": 4520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.890116347396874e-05,
+      "loss": 0.9549,
+      "step": 4530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.888647314608062e-05,
+      "loss": 1.0581,
+      "step": 4540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8871782818192505e-05,
+      "loss": 1.04,
+      "step": 4550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8857092490304386e-05,
+      "loss": 1.073,
+      "step": 4560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.884240216241627e-05,
+      "loss": 1.0464,
+      "step": 4570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.882771183452815e-05,
+      "loss": 1.0713,
+      "step": 4580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.881302150664003e-05,
+      "loss": 0.9706,
+      "step": 4590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.879833117875191e-05,
+      "loss": 0.984,
+      "step": 4600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8783640850863794e-05,
+      "loss": 1.0161,
+      "step": 4610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8768950522975675e-05,
+      "loss": 1.0286,
+      "step": 4620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.875426019508756e-05,
+      "loss": 1.0362,
+      "step": 4630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.873956986719944e-05,
+      "loss": 1.0545,
+      "step": 4640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.872487953931132e-05,
+      "loss": 1.0084,
+      "step": 4650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.87101892114232e-05,
+      "loss": 1.0738,
+      "step": 4660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.869549888353508e-05,
+      "loss": 1.038,
+      "step": 4670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8680808555646964e-05,
+      "loss": 1.043,
+      "step": 4680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8666118227758846e-05,
+      "loss": 0.9706,
+      "step": 4690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.865142789987073e-05,
+      "loss": 1.0215,
+      "step": 4700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.863673757198261e-05,
+      "loss": 0.9936,
+      "step": 4710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.862204724409449e-05,
+      "loss": 0.9612,
+      "step": 4720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.860735691620637e-05,
+      "loss": 1.0079,
+      "step": 4730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.859266658831825e-05,
+      "loss": 0.9488,
+      "step": 4740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8577976260430135e-05,
+      "loss": 1.0819,
+      "step": 4750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8563285932542016e-05,
+      "loss": 1.0074,
+      "step": 4760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.85485956046539e-05,
+      "loss": 1.0684,
+      "step": 4770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.853390527676578e-05,
+      "loss": 0.9414,
+      "step": 4780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.851921494887766e-05,
+      "loss": 1.0441,
+      "step": 4790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.850452462098954e-05,
+      "loss": 1.0685,
+      "step": 4800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8489834293101424e-05,
+      "loss": 0.9934,
+      "step": 4810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8475143965213305e-05,
+      "loss": 0.9733,
+      "step": 4820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.846045363732519e-05,
+      "loss": 0.9903,
+      "step": 4830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.844576330943707e-05,
+      "loss": 1.0722,
+      "step": 4840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.843107298154895e-05,
+      "loss": 1.0111,
+      "step": 4850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.841638265366083e-05,
+      "loss": 1.0381,
+      "step": 4860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.840169232577271e-05,
+      "loss": 0.9838,
+      "step": 4870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8387001997884594e-05,
+      "loss": 1.0219,
+      "step": 4880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8372311669996476e-05,
+      "loss": 0.9365,
+      "step": 4890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.835762134210836e-05,
+      "loss": 1.0131,
+      "step": 4900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.834293101422024e-05,
+      "loss": 0.9864,
+      "step": 4910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.832824068633212e-05,
+      "loss": 0.9772,
+      "step": 4920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8313550358444e-05,
+      "loss": 1.0298,
+      "step": 4930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.829886003055589e-05,
+      "loss": 1.0244,
+      "step": 4940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8284169702667765e-05,
+      "loss": 0.9372,
+      "step": 4950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8269479374779646e-05,
+      "loss": 1.0092,
+      "step": 4960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.825478904689153e-05,
+      "loss": 1.0105,
+      "step": 4970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.824009871900341e-05,
+      "loss": 1.0607,
+      "step": 4980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.822540839111529e-05,
+      "loss": 1.0366,
+      "step": 4990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.821071806322717e-05,
+      "loss": 1.0177,
+      "step": 5000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8196027735339054e-05,
+      "loss": 1.0202,
+      "step": 5010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8181337407450935e-05,
+      "loss": 1.0351,
+      "step": 5020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.816664707956282e-05,
+      "loss": 0.9626,
+      "step": 5030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.81519567516747e-05,
+      "loss": 0.9936,
+      "step": 5040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.813726642378658e-05,
+      "loss": 0.9953,
+      "step": 5050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.812257609589846e-05,
+      "loss": 1.0137,
+      "step": 5060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.810788576801035e-05,
+      "loss": 0.9925,
+      "step": 5070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8093195440122224e-05,
+      "loss": 1.012,
+      "step": 5080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8078505112234106e-05,
+      "loss": 1.0516,
+      "step": 5090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.806381478434599e-05,
+      "loss": 0.9799,
+      "step": 5100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8049124456457876e-05,
+      "loss": 1.0051,
+      "step": 5110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.803443412856975e-05,
+      "loss": 1.0288,
+      "step": 5120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.801974380068163e-05,
+      "loss": 1.0292,
+      "step": 5130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.8005053472793513e-05,
+      "loss": 1.0574,
+      "step": 5140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7990363144905395e-05,
+      "loss": 1.0203,
+      "step": 5150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.797567281701728e-05,
+      "loss": 0.9617,
+      "step": 5160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.796098248912916e-05,
+      "loss": 0.9966,
+      "step": 5170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.794629216124104e-05,
+      "loss": 1.004,
+      "step": 5180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.793160183335292e-05,
+      "loss": 1.0078,
+      "step": 5190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.791691150546481e-05,
+      "loss": 0.9553,
+      "step": 5200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7902221177576684e-05,
+      "loss": 0.9471,
+      "step": 5210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7887530849688566e-05,
+      "loss": 1.014,
+      "step": 5220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.787284052180045e-05,
+      "loss": 1.024,
+      "step": 5230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7858150193912335e-05,
+      "loss": 1.054,
+      "step": 5240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.784345986602421e-05,
+      "loss": 1.0298,
+      "step": 5250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.782876953813609e-05,
+      "loss": 1.0335,
+      "step": 5260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.781407921024797e-05,
+      "loss": 0.9946,
+      "step": 5270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7799388882359855e-05,
+      "loss": 1.0251,
+      "step": 5280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.778469855447174e-05,
+      "loss": 0.9556,
+      "step": 5290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.777000822658362e-05,
+      "loss": 1.0366,
+      "step": 5300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.77553178986955e-05,
+      "loss": 0.9919,
+      "step": 5310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.774062757080738e-05,
+      "loss": 0.9954,
+      "step": 5320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.772593724291927e-05,
+      "loss": 0.9361,
+      "step": 5330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7711246915031144e-05,
+      "loss": 1.0363,
+      "step": 5340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7696556587143025e-05,
+      "loss": 1.033,
+      "step": 5350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7681866259254907e-05,
+      "loss": 0.9587,
+      "step": 5360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7667175931366795e-05,
+      "loss": 0.9678,
+      "step": 5370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.765248560347867e-05,
+      "loss": 1.0181,
+      "step": 5380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.763779527559055e-05,
+      "loss": 1.0325,
+      "step": 5390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.762310494770243e-05,
+      "loss": 0.9837,
+      "step": 5400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.760841461981432e-05,
+      "loss": 0.9769,
+      "step": 5410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.75937242919262e-05,
+      "loss": 1.0457,
+      "step": 5420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.757903396403808e-05,
+      "loss": 0.9652,
+      "step": 5430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.756434363614996e-05,
+      "loss": 1.0128,
+      "step": 5440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.754965330826184e-05,
+      "loss": 0.9725,
+      "step": 5450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.753496298037373e-05,
+      "loss": 0.9434,
+      "step": 5460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.75202726524856e-05,
+      "loss": 0.9897,
+      "step": 5470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7505582324597485e-05,
+      "loss": 0.9589,
+      "step": 5480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7490891996709366e-05,
+      "loss": 0.9509,
+      "step": 5490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7476201668821254e-05,
+      "loss": 0.933,
+      "step": 5500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7461511340933136e-05,
+      "loss": 0.9842,
+      "step": 5510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.744682101304501e-05,
+      "loss": 1.0169,
+      "step": 5520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.743213068515689e-05,
+      "loss": 0.9944,
+      "step": 5530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.741744035726878e-05,
+      "loss": 0.969,
+      "step": 5540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.740275002938066e-05,
+      "loss": 0.9813,
+      "step": 5550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.738805970149254e-05,
+      "loss": 1.0462,
+      "step": 5560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.737336937360442e-05,
+      "loss": 1.0136,
+      "step": 5570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7358679045716306e-05,
+      "loss": 0.8983,
+      "step": 5580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.734398871782819e-05,
+      "loss": 0.9633,
+      "step": 5590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.732929838994006e-05,
+      "loss": 0.93,
+      "step": 5600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7314608062051944e-05,
+      "loss": 0.9222,
+      "step": 5610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7299917734163826e-05,
+      "loss": 0.937,
+      "step": 5620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7285227406275714e-05,
+      "loss": 0.9741,
+      "step": 5630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7270537078387596e-05,
+      "loss": 0.9735,
+      "step": 5640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.725584675049947e-05,
+      "loss": 1.0259,
+      "step": 5650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.724115642261135e-05,
+      "loss": 0.9429,
+      "step": 5660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.722646609472324e-05,
+      "loss": 1.0342,
+      "step": 5670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.721177576683512e-05,
+      "loss": 0.9521,
+      "step": 5680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7197085438946996e-05,
+      "loss": 0.9738,
+      "step": 5690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.718239511105888e-05,
+      "loss": 0.9651,
+      "step": 5700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7167704783170766e-05,
+      "loss": 0.9507,
+      "step": 5710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.715301445528265e-05,
+      "loss": 0.9656,
+      "step": 5720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.713832412739452e-05,
+      "loss": 0.9874,
+      "step": 5730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7123633799506404e-05,
+      "loss": 0.9169,
+      "step": 5740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.710894347161829e-05,
+      "loss": 0.9815,
+      "step": 5750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7094253143730174e-05,
+      "loss": 0.9882,
+      "step": 5760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7079562815842055e-05,
+      "loss": 0.9512,
+      "step": 5770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.706487248795393e-05,
+      "loss": 0.9917,
+      "step": 5780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.705018216006581e-05,
+      "loss": 0.9798,
+      "step": 5790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.70354918321777e-05,
+      "loss": 0.9033,
+      "step": 5800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.702080150428958e-05,
+      "loss": 0.9585,
+      "step": 5810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7006111176401456e-05,
+      "loss": 0.967,
+      "step": 5820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.699142084851334e-05,
+      "loss": 0.9508,
+      "step": 5830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.6976730520625226e-05,
+      "loss": 0.9627,
+      "step": 5840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.696204019273711e-05,
+      "loss": 0.9609,
+      "step": 5850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.694734986484899e-05,
+      "loss": 0.9506,
+      "step": 5860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.693265953696086e-05,
+      "loss": 1.0114,
+      "step": 5870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.691796920907275e-05,
+      "loss": 0.9508,
+      "step": 5880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.690327888118463e-05,
+      "loss": 0.9998,
+      "step": 5890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.6888588553296515e-05,
+      "loss": 1.0045,
+      "step": 5900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.687389822540839e-05,
+      "loss": 0.9326,
+      "step": 5910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.685920789752027e-05,
+      "loss": 0.9839,
+      "step": 5920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.684451756963216e-05,
+      "loss": 0.9873,
+      "step": 5930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.682982724174404e-05,
+      "loss": 0.929,
+      "step": 5940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.6815136913855915e-05,
+      "loss": 0.9378,
+      "step": 5950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.68004465859678e-05,
+      "loss": 1.065,
+      "step": 5960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.6785756258079685e-05,
+      "loss": 0.9591,
+      "step": 5970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.677106593019157e-05,
+      "loss": 0.9753,
+      "step": 5980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.675637560230345e-05,
+      "loss": 0.9507,
+      "step": 5990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.674168527441532e-05,
+      "loss": 0.9719,
+      "step": 6000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.672699494652721e-05,
+      "loss": 0.9679,
+      "step": 6010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.671230461863909e-05,
+      "loss": 0.905,
+      "step": 6020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6697614290750974e-05,
+      "loss": 0.9225,
+      "step": 6030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.668292396286285e-05,
+      "loss": 0.9592,
+      "step": 6040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.666823363497474e-05,
+      "loss": 0.9796,
+      "step": 6050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.665354330708662e-05,
+      "loss": 0.9152,
+      "step": 6060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.66388529791985e-05,
+      "loss": 0.9971,
+      "step": 6070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6624162651310375e-05,
+      "loss": 0.9679,
+      "step": 6080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6609472323422256e-05,
+      "loss": 0.949,
+      "step": 6090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6594781995534145e-05,
+      "loss": 0.9938,
+      "step": 6100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6580091667646026e-05,
+      "loss": 0.9326,
+      "step": 6110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.656540133975791e-05,
+      "loss": 0.9419,
+      "step": 6120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.655071101186978e-05,
+      "loss": 0.9924,
+      "step": 6130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.653602068398167e-05,
+      "loss": 0.9591,
+      "step": 6140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.652133035609355e-05,
+      "loss": 0.9745,
+      "step": 6150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6506640028205434e-05,
+      "loss": 0.9435,
+      "step": 6160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.649194970031731e-05,
+      "loss": 0.9713,
+      "step": 6170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.64772593724292e-05,
+      "loss": 0.9649,
+      "step": 6180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.646256904454108e-05,
+      "loss": 0.9409,
+      "step": 6190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.644787871665296e-05,
+      "loss": 0.9773,
+      "step": 6200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.643318838876484e-05,
+      "loss": 0.9754,
+      "step": 6210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.641849806087672e-05,
+      "loss": 0.9371,
+      "step": 6220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6403807732988604e-05,
+      "loss": 0.9408,
+      "step": 6230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6389117405100486e-05,
+      "loss": 0.9785,
+      "step": 6240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.637442707721237e-05,
+      "loss": 0.9332,
+      "step": 6250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.635973674932424e-05,
+      "loss": 0.9876,
+      "step": 6260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.634504642143613e-05,
+      "loss": 0.8978,
+      "step": 6270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.633035609354801e-05,
+      "loss": 0.9341,
+      "step": 6280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.631566576565989e-05,
+      "loss": 0.9424,
+      "step": 6290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.630097543777177e-05,
+      "loss": 0.9913,
+      "step": 6300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6286285109883656e-05,
+      "loss": 0.9746,
+      "step": 6310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.627159478199554e-05,
+      "loss": 0.9322,
+      "step": 6320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.625690445410742e-05,
+      "loss": 0.9803,
+      "step": 6330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.62422141262193e-05,
+      "loss": 0.9984,
+      "step": 6340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.622752379833118e-05,
+      "loss": 0.9598,
+      "step": 6350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6212833470443064e-05,
+      "loss": 0.8992,
+      "step": 6360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6198143142554945e-05,
+      "loss": 1.0201,
+      "step": 6370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.618345281466683e-05,
+      "loss": 0.9779,
+      "step": 6380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.61687624867787e-05,
+      "loss": 0.8707,
+      "step": 6390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.615407215889059e-05,
+      "loss": 1.0432,
+      "step": 6400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.613938183100247e-05,
+      "loss": 0.9638,
+      "step": 6410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.612469150311435e-05,
+      "loss": 0.9591,
+      "step": 6420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.611000117522623e-05,
+      "loss": 0.9387,
+      "step": 6430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6095310847338116e-05,
+      "loss": 0.9881,
+      "step": 6440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.608062051945e-05,
+      "loss": 0.9809,
+      "step": 6450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.606593019156188e-05,
+      "loss": 0.9667,
+      "step": 6460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.605123986367376e-05,
+      "loss": 0.9324,
+      "step": 6470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.603654953578564e-05,
+      "loss": 0.9186,
+      "step": 6480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6021859207897523e-05,
+      "loss": 0.9335,
+      "step": 6490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6007168880009405e-05,
+      "loss": 0.971,
+      "step": 6500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.5992478552121286e-05,
+      "loss": 0.9463,
+      "step": 6510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.597778822423317e-05,
+      "loss": 0.9284,
+      "step": 6520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.596309789634505e-05,
+      "loss": 0.9396,
+      "step": 6530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.594840756845693e-05,
+      "loss": 0.997,
+      "step": 6540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.593371724056881e-05,
+      "loss": 0.9607,
+      "step": 6550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.5919026912680694e-05,
+      "loss": 0.9635,
+      "step": 6560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.5904336584792575e-05,
+      "loss": 0.95,
+      "step": 6570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.588964625690446e-05,
+      "loss": 0.9612,
+      "step": 6580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.587495592901634e-05,
+      "loss": 0.8854,
+      "step": 6590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.586026560112822e-05,
+      "loss": 0.9338,
+      "step": 6600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.58455752732401e-05,
+      "loss": 0.9546,
+      "step": 6610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.583088494535198e-05,
+      "loss": 1.0048,
+      "step": 6620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5816194617463865e-05,
+      "loss": 0.9257,
+      "step": 6630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5801504289575746e-05,
+      "loss": 0.9288,
+      "step": 6640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.578681396168763e-05,
+      "loss": 0.9091,
+      "step": 6650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.577212363379951e-05,
+      "loss": 0.9783,
+      "step": 6660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.575743330591139e-05,
+      "loss": 0.9806,
+      "step": 6670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.574274297802327e-05,
+      "loss": 1.0132,
+      "step": 6680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5728052650135154e-05,
+      "loss": 0.896,
+      "step": 6690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5713362322247035e-05,
+      "loss": 0.8943,
+      "step": 6700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5698671994358917e-05,
+      "loss": 0.9175,
+      "step": 6710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.56839816664708e-05,
+      "loss": 0.961,
+      "step": 6720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.566929133858268e-05,
+      "loss": 1.0114,
+      "step": 6730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.565460101069456e-05,
+      "loss": 0.9752,
+      "step": 6740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.563991068280644e-05,
+      "loss": 1.006,
+      "step": 6750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5625220354918324e-05,
+      "loss": 0.9812,
+      "step": 6760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5610530027030206e-05,
+      "loss": 0.9565,
+      "step": 6770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.559583969914209e-05,
+      "loss": 0.9558,
+      "step": 6780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.558114937125397e-05,
+      "loss": 0.9055,
+      "step": 6790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.556645904336585e-05,
+      "loss": 0.9499,
+      "step": 6800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.555176871547773e-05,
+      "loss": 0.9898,
+      "step": 6810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.553707838758961e-05,
+      "loss": 0.9157,
+      "step": 6820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5522388059701495e-05,
+      "loss": 0.9098,
+      "step": 6830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5507697731813376e-05,
+      "loss": 0.8776,
+      "step": 6840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.549300740392526e-05,
+      "loss": 0.9748,
+      "step": 6850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.547831707603714e-05,
+      "loss": 0.9337,
+      "step": 6860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.546362674814902e-05,
+      "loss": 0.9267,
+      "step": 6870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.54489364202609e-05,
+      "loss": 0.9574,
+      "step": 6880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5434246092372784e-05,
+      "loss": 0.8962,
+      "step": 6890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5419555764484665e-05,
+      "loss": 0.9278,
+      "step": 6900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.540486543659655e-05,
+      "loss": 0.9702,
+      "step": 6910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.539017510870843e-05,
+      "loss": 0.9738,
+      "step": 6920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.537548478082031e-05,
+      "loss": 0.911,
+      "step": 6930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.536079445293219e-05,
+      "loss": 0.9238,
+      "step": 6940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.534610412504407e-05,
+      "loss": 0.96,
+      "step": 6950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5331413797155954e-05,
+      "loss": 1.0077,
+      "step": 6960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5316723469267836e-05,
+      "loss": 0.9736,
+      "step": 6970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.530203314137972e-05,
+      "loss": 0.9449,
+      "step": 6980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.52873428134916e-05,
+      "loss": 0.902,
+      "step": 6990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.527265248560348e-05,
+      "loss": 0.9773,
+      "step": 7000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.525796215771536e-05,
+      "loss": 0.9306,
+      "step": 7010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.524327182982724e-05,
+      "loss": 0.9666,
+      "step": 7020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5228581501939125e-05,
+      "loss": 0.8784,
+      "step": 7030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5213891174051006e-05,
+      "loss": 0.9508,
+      "step": 7040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.519920084616289e-05,
+      "loss": 0.888,
+      "step": 7050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.518451051827477e-05,
+      "loss": 0.9673,
+      "step": 7060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.516982019038665e-05,
+      "loss": 0.9693,
+      "step": 7070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.515512986249853e-05,
+      "loss": 0.8666,
+      "step": 7080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5140439534610414e-05,
+      "loss": 0.9438,
+      "step": 7090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5125749206722295e-05,
+      "loss": 0.9002,
+      "step": 7100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.511105887883418e-05,
+      "loss": 0.9949,
+      "step": 7110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.509636855094606e-05,
+      "loss": 0.908,
+      "step": 7120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.508167822305794e-05,
+      "loss": 0.9315,
+      "step": 7130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.506698789516982e-05,
+      "loss": 0.9248,
+      "step": 7140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.50522975672817e-05,
+      "loss": 0.901,
+      "step": 7150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5037607239393584e-05,
+      "loss": 0.9349,
+      "step": 7160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.502291691150547e-05,
+      "loss": 0.8925,
+      "step": 7170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.500822658361735e-05,
+      "loss": 0.9785,
+      "step": 7180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.499353625572923e-05,
+      "loss": 0.9175,
+      "step": 7190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.497884592784111e-05,
+      "loss": 0.9563,
+      "step": 7200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.496415559995299e-05,
+      "loss": 0.9657,
+      "step": 7210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.494946527206487e-05,
+      "loss": 0.9751,
+      "step": 7220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4934774944176755e-05,
+      "loss": 0.9606,
+      "step": 7230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4920084616288636e-05,
+      "loss": 1.0009,
+      "step": 7240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.490539428840052e-05,
+      "loss": 0.919,
+      "step": 7250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.48907039605124e-05,
+      "loss": 0.9898,
+      "step": 7260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.487601363262428e-05,
+      "loss": 0.9732,
+      "step": 7270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.486132330473616e-05,
+      "loss": 0.9212,
+      "step": 7280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4846632976848044e-05,
+      "loss": 0.9719,
+      "step": 7290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.483194264895993e-05,
+      "loss": 0.9377,
+      "step": 7300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.481725232107181e-05,
+      "loss": 0.9362,
+      "step": 7310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.480256199318369e-05,
+      "loss": 0.9134,
+      "step": 7320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.478787166529557e-05,
+      "loss": 0.9442,
+      "step": 7330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.477318133740746e-05,
+      "loss": 0.9223,
+      "step": 7340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.475849100951933e-05,
+      "loss": 0.951,
+      "step": 7350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4743800681631214e-05,
+      "loss": 0.9853,
+      "step": 7360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4729110353743096e-05,
+      "loss": 0.8699,
+      "step": 7370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.471442002585498e-05,
+      "loss": 0.8852,
+      "step": 7380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4699729697966866e-05,
+      "loss": 0.9211,
+      "step": 7390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.468503937007874e-05,
+      "loss": 0.9008,
+      "step": 7400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.467034904219062e-05,
+      "loss": 0.9233,
+      "step": 7410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4655658714302503e-05,
+      "loss": 0.9673,
+      "step": 7420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.464096838641439e-05,
+      "loss": 0.8565,
+      "step": 7430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4626278058526266e-05,
+      "loss": 0.9539,
+      "step": 7440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.461158773063815e-05,
+      "loss": 0.8765,
+      "step": 7450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.459689740275003e-05,
+      "loss": 0.9546,
+      "step": 7460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.458220707486192e-05,
+      "loss": 0.9401,
+      "step": 7470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.456751674697379e-05,
+      "loss": 0.8975,
+      "step": 7480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4552826419085674e-05,
+      "loss": 0.9258,
+      "step": 7490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4538136091197555e-05,
+      "loss": 0.8865,
+      "step": 7500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4523445763309444e-05,
+      "loss": 0.989,
+      "step": 7510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4508755435421325e-05,
+      "loss": 0.8648,
+      "step": 7520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.44940651075332e-05,
+      "loss": 0.9117,
+      "step": 7530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.447937477964508e-05,
+      "loss": 0.9948,
+      "step": 7540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.446468445175696e-05,
+      "loss": 0.8837,
+      "step": 7550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.444999412386885e-05,
+      "loss": 0.8882,
+      "step": 7560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4435303795980726e-05,
+      "loss": 0.904,
+      "step": 7570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.442061346809261e-05,
+      "loss": 0.9619,
+      "step": 7580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.440592314020449e-05,
+      "loss": 0.9428,
+      "step": 7590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.439123281231638e-05,
+      "loss": 0.91,
+      "step": 7600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.437654248442825e-05,
+      "loss": 0.837,
+      "step": 7610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4361852156540134e-05,
+      "loss": 0.8955,
+      "step": 7620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4347161828652015e-05,
+      "loss": 0.9065,
+      "step": 7630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.43324715007639e-05,
+      "loss": 0.9223,
+      "step": 7640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4317781172875785e-05,
+      "loss": 0.9051,
+      "step": 7650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.430309084498766e-05,
+      "loss": 0.9357,
+      "step": 7660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.428840051709954e-05,
+      "loss": 0.9014,
+      "step": 7670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.427371018921142e-05,
+      "loss": 0.9404,
+      "step": 7680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.425901986132331e-05,
+      "loss": 0.8784,
+      "step": 7690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4244329533435186e-05,
+      "loss": 0.9412,
+      "step": 7700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.422963920554707e-05,
+      "loss": 0.9382,
+      "step": 7710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.421494887765895e-05,
+      "loss": 0.8882,
+      "step": 7720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.420025854977084e-05,
+      "loss": 0.9344,
+      "step": 7730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.418556822188272e-05,
+      "loss": 0.9074,
+      "step": 7740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.417087789399459e-05,
+      "loss": 0.9801,
+      "step": 7750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4156187566106475e-05,
+      "loss": 0.891,
+      "step": 7760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.414149723821836e-05,
+      "loss": 0.9394,
+      "step": 7770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4126806910330244e-05,
+      "loss": 0.8982,
+      "step": 7780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.411211658244212e-05,
+      "loss": 0.9174,
+      "step": 7790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4097426254554e-05,
+      "loss": 0.9312,
+      "step": 7800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.408273592666589e-05,
+      "loss": 0.9271,
+      "step": 7810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.406804559877777e-05,
+      "loss": 0.9608,
+      "step": 7820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4053355270889645e-05,
+      "loss": 0.8835,
+      "step": 7830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.403866494300153e-05,
+      "loss": 0.8768,
+      "step": 7840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.402397461511341e-05,
+      "loss": 0.8487,
+      "step": 7850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4009284287225296e-05,
+      "loss": 0.8823,
+      "step": 7860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.399459395933718e-05,
+      "loss": 0.9122,
+      "step": 7870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.397990363144905e-05,
+      "loss": 0.923,
+      "step": 7880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3965213303560934e-05,
+      "loss": 0.9462,
+      "step": 7890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.395052297567282e-05,
+      "loss": 0.8908,
+      "step": 7900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3935832647784704e-05,
+      "loss": 0.8534,
+      "step": 7910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.392114231989658e-05,
+      "loss": 0.9281,
+      "step": 7920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.390645199200846e-05,
+      "loss": 0.9104,
+      "step": 7930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.389176166412035e-05,
+      "loss": 0.9435,
+      "step": 7940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.387707133623223e-05,
+      "loss": 0.8823,
+      "step": 7950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3862381008344105e-05,
+      "loss": 0.9242,
+      "step": 7960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3847690680455986e-05,
+      "loss": 1.0079,
+      "step": 7970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3833000352567874e-05,
+      "loss": 0.9576,
+      "step": 7980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3818310024679756e-05,
+      "loss": 0.8616,
+      "step": 7990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.380361969679164e-05,
+      "loss": 0.9636,
+      "step": 8000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.378892936890351e-05,
+      "loss": 0.9245,
+      "step": 8010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3774239041015394e-05,
+      "loss": 0.8764,
+      "step": 8020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.375954871312728e-05,
+      "loss": 0.948,
+      "step": 8030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3744858385239164e-05,
+      "loss": 0.9433,
+      "step": 8040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.373016805735104e-05,
+      "loss": 0.9305,
+      "step": 8050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.371547772946292e-05,
+      "loss": 0.9117,
+      "step": 8060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.370078740157481e-05,
+      "loss": 0.9201,
+      "step": 8070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.368609707368669e-05,
+      "loss": 0.9129,
+      "step": 8080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.367140674579857e-05,
+      "loss": 0.9512,
+      "step": 8090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3656716417910446e-05,
+      "loss": 0.9316,
+      "step": 8100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3642026090022334e-05,
+      "loss": 0.9255,
+      "step": 8110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3627335762134216e-05,
+      "loss": 0.9549,
+      "step": 8120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.36126454342461e-05,
+      "loss": 0.8743,
+      "step": 8130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.359795510635797e-05,
+      "loss": 0.9069,
+      "step": 8140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.358326477846986e-05,
+      "loss": 0.9476,
+      "step": 8150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.356857445058174e-05,
+      "loss": 0.9085,
+      "step": 8160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.355388412269362e-05,
+      "loss": 0.913,
+      "step": 8170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.35391937948055e-05,
+      "loss": 0.9574,
+      "step": 8180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.352450346691738e-05,
+      "loss": 0.8911,
+      "step": 8190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.350981313902927e-05,
+      "loss": 0.9691,
+      "step": 8200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.349512281114115e-05,
+      "loss": 0.9385,
+      "step": 8210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.348043248325303e-05,
+      "loss": 0.9483,
+      "step": 8220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3465742155364905e-05,
+      "loss": 0.9085,
+      "step": 8230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3451051827476794e-05,
+      "loss": 0.8958,
+      "step": 8240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3436361499588675e-05,
+      "loss": 0.8855,
+      "step": 8250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.342167117170056e-05,
+      "loss": 0.8826,
+      "step": 8260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.340698084381243e-05,
+      "loss": 0.8989,
+      "step": 8270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.339229051592432e-05,
+      "loss": 0.8891,
+      "step": 8280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.33776001880362e-05,
+      "loss": 0.8953,
+      "step": 8290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.336290986014808e-05,
+      "loss": 0.8844,
+      "step": 8300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.334821953225996e-05,
+      "loss": 0.9399,
+      "step": 8310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.333352920437184e-05,
+      "loss": 0.8995,
+      "step": 8320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.331883887648373e-05,
+      "loss": 0.876,
+      "step": 8330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.330414854859561e-05,
+      "loss": 0.8953,
+      "step": 8340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.328945822070749e-05,
+      "loss": 0.9,
+      "step": 8350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3274767892819365e-05,
+      "loss": 0.891,
+      "step": 8360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.326007756493125e-05,
+      "loss": 0.9293,
+      "step": 8370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3245387237043135e-05,
+      "loss": 0.919,
+      "step": 8380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3230696909155016e-05,
+      "loss": 0.9141,
+      "step": 8390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.321600658126689e-05,
+      "loss": 0.8765,
+      "step": 8400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.320131625337878e-05,
+      "loss": 0.9044,
+      "step": 8410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.318662592549066e-05,
+      "loss": 0.9394,
+      "step": 8420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.317193559760254e-05,
+      "loss": 0.8588,
+      "step": 8430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3157245269714424e-05,
+      "loss": 0.8955,
+      "step": 8440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3142554941826305e-05,
+      "loss": 0.9535,
+      "step": 8450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.312786461393819e-05,
+      "loss": 0.9185,
+      "step": 8460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.311317428605007e-05,
+      "loss": 0.9551,
+      "step": 8470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.309848395816195e-05,
+      "loss": 0.8874,
+      "step": 8480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3083793630273824e-05,
+      "loss": 0.9225,
+      "step": 8490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.306910330238571e-05,
+      "loss": 0.9049,
+      "step": 8500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3054412974497594e-05,
+      "loss": 0.8725,
+      "step": 8510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3039722646609476e-05,
+      "loss": 0.9361,
+      "step": 8520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.302503231872135e-05,
+      "loss": 0.9158,
+      "step": 8530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.301034199083324e-05,
+      "loss": 0.8716,
+      "step": 8540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.299565166294512e-05,
+      "loss": 0.9241,
+      "step": 8550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.2980961335057e-05,
+      "loss": 0.8968,
+      "step": 8560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.296627100716888e-05,
+      "loss": 0.917,
+      "step": 8570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.2951580679280765e-05,
+      "loss": 0.9089,
+      "step": 8580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.2936890351392646e-05,
+      "loss": 0.9441,
+      "step": 8590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.292220002350453e-05,
+      "loss": 0.9028,
+      "step": 8600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.290750969561641e-05,
+      "loss": 0.9063,
+      "step": 8610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.289281936772829e-05,
+      "loss": 0.8952,
+      "step": 8620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.287812903984017e-05,
+      "loss": 0.8352,
+      "step": 8630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2863438711952054e-05,
+      "loss": 0.9463,
+      "step": 8640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2848748384063935e-05,
+      "loss": 0.9372,
+      "step": 8650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.283405805617581e-05,
+      "loss": 0.8842,
+      "step": 8660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.28193677282877e-05,
+      "loss": 0.9348,
+      "step": 8670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.280467740039958e-05,
+      "loss": 0.8542,
+      "step": 8680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.278998707251146e-05,
+      "loss": 0.8942,
+      "step": 8690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.277529674462334e-05,
+      "loss": 0.8821,
+      "step": 8700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2760606416735224e-05,
+      "loss": 0.936,
+      "step": 8710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2745916088847106e-05,
+      "loss": 0.9103,
+      "step": 8720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.273122576095899e-05,
+      "loss": 0.9163,
+      "step": 8730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.271653543307087e-05,
+      "loss": 0.8792,
+      "step": 8740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.270184510518275e-05,
+      "loss": 0.8487,
+      "step": 8750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.268715477729463e-05,
+      "loss": 0.9287,
+      "step": 8760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.267246444940651e-05,
+      "loss": 0.9108,
+      "step": 8770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2657774121518395e-05,
+      "loss": 0.8662,
+      "step": 8780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2643083793630276e-05,
+      "loss": 0.8881,
+      "step": 8790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262839346574216e-05,
+      "loss": 0.9276,
+      "step": 8800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.261370313785404e-05,
+      "loss": 0.9358,
+      "step": 8810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.259901280996592e-05,
+      "loss": 0.9402,
+      "step": 8820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.25843224820778e-05,
+      "loss": 0.8906,
+      "step": 8830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2569632154189684e-05,
+      "loss": 0.8678,
+      "step": 8840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2554941826301565e-05,
+      "loss": 0.853,
+      "step": 8850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.254025149841345e-05,
+      "loss": 0.8783,
+      "step": 8860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.252556117052533e-05,
+      "loss": 0.8481,
+      "step": 8870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.251087084263721e-05,
+      "loss": 0.8861,
+      "step": 8880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.249618051474909e-05,
+      "loss": 0.9329,
+      "step": 8890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.248149018686097e-05,
+      "loss": 0.8958,
+      "step": 8900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2466799858972854e-05,
+      "loss": 0.9297,
+      "step": 8910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2452109531084736e-05,
+      "loss": 0.8717,
+      "step": 8920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.243741920319662e-05,
+      "loss": 0.9116,
+      "step": 8930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.24227288753085e-05,
+      "loss": 0.8938,
+      "step": 8940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.240803854742038e-05,
+      "loss": 0.8845,
+      "step": 8950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.239334821953226e-05,
+      "loss": 0.9122,
+      "step": 8960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2378657891644143e-05,
+      "loss": 0.9119,
+      "step": 8970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2363967563756025e-05,
+      "loss": 0.8998,
+      "step": 8980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2349277235867906e-05,
+      "loss": 0.9286,
+      "step": 8990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.233458690797979e-05,
+      "loss": 0.871,
+      "step": 9000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.231989658009167e-05,
+      "loss": 0.905,
+      "step": 9010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.230520625220355e-05,
+      "loss": 0.9065,
+      "step": 9020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.229051592431543e-05,
+      "loss": 0.9361,
+      "step": 9030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2275825596427314e-05,
+      "loss": 0.882,
+      "step": 9040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2261135268539196e-05,
+      "loss": 0.917,
+      "step": 9050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.224644494065108e-05,
+      "loss": 0.9222,
+      "step": 9060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.223175461276296e-05,
+      "loss": 0.8754,
+      "step": 9070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.221706428487484e-05,
+      "loss": 0.9219,
+      "step": 9080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.220237395698672e-05,
+      "loss": 0.9044,
+      "step": 9090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.21876836290986e-05,
+      "loss": 0.9117,
+      "step": 9100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2172993301210485e-05,
+      "loss": 0.908,
+      "step": 9110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2158302973322366e-05,
+      "loss": 0.9373,
+      "step": 9120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.214361264543425e-05,
+      "loss": 0.8511,
+      "step": 9130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.212892231754613e-05,
+      "loss": 0.8257,
+      "step": 9140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.211423198965801e-05,
+      "loss": 0.8926,
+      "step": 9150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.209954166176989e-05,
+      "loss": 0.9297,
+      "step": 9160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2084851333881774e-05,
+      "loss": 0.8949,
+      "step": 9170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2070161005993655e-05,
+      "loss": 0.8762,
+      "step": 9180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2055470678105537e-05,
+      "loss": 0.9109,
+      "step": 9190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.204078035021742e-05,
+      "loss": 0.8877,
+      "step": 9200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.20260900223293e-05,
+      "loss": 0.8514,
+      "step": 9210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.201139969444118e-05,
+      "loss": 0.8681,
+      "step": 9220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.199670936655306e-05,
+      "loss": 0.9577,
+      "step": 9230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1982019038664944e-05,
+      "loss": 0.9046,
+      "step": 9240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1967328710776826e-05,
+      "loss": 0.9102,
+      "step": 9250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.195263838288871e-05,
+      "loss": 0.8569,
+      "step": 9260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1937948055000595e-05,
+      "loss": 0.95,
+      "step": 9270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.192325772711247e-05,
+      "loss": 0.8572,
+      "step": 9280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.190856739922435e-05,
+      "loss": 0.9061,
+      "step": 9290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.189387707133623e-05,
+      "loss": 0.9138,
+      "step": 9300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1879186743448115e-05,
+      "loss": 0.9223,
+      "step": 9310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1864496415559996e-05,
+      "loss": 0.9323,
+      "step": 9320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.184980608767188e-05,
+      "loss": 0.8403,
+      "step": 9330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.183511575978376e-05,
+      "loss": 0.8864,
+      "step": 9340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.182042543189564e-05,
+      "loss": 0.8819,
+      "step": 9350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.180573510400752e-05,
+      "loss": 0.9139,
+      "step": 9360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1791044776119404e-05,
+      "loss": 0.9142,
+      "step": 9370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.1776354448231285e-05,
+      "loss": 0.9197,
+      "step": 9380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.176166412034317e-05,
+      "loss": 0.8484,
+      "step": 9390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.1746973792455055e-05,
+      "loss": 0.874,
+      "step": 9400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.173228346456693e-05,
+      "loss": 0.94,
+      "step": 9410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.171759313667881e-05,
+      "loss": 0.8864,
+      "step": 9420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.170290280879069e-05,
+      "loss": 0.8698,
+      "step": 9430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.1688212480902574e-05,
+      "loss": 0.924,
+      "step": 9440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.1673522153014456e-05,
+      "loss": 0.8987,
+      "step": 9450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.165883182512634e-05,
+      "loss": 0.846,
+      "step": 9460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.164414149723822e-05,
+      "loss": 0.9335,
+      "step": 9470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.16294511693501e-05,
+      "loss": 0.903,
+      "step": 9480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.161476084146199e-05,
+      "loss": 0.8986,
+      "step": 9490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.160007051357386e-05,
+      "loss": 0.9032,
+      "step": 9500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.1585380185685745e-05,
+      "loss": 0.8794,
+      "step": 9510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1570689857797626e-05,
+      "loss": 0.9754,
+      "step": 9520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1555999529909515e-05,
+      "loss": 0.8878,
+      "step": 9530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.154130920202139e-05,
+      "loss": 0.8527,
+      "step": 9540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.152661887413327e-05,
+      "loss": 0.8302,
+      "step": 9550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.151192854624515e-05,
+      "loss": 0.8636,
+      "step": 9560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.149723821835704e-05,
+      "loss": 0.8775,
+      "step": 9570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1482547890468915e-05,
+      "loss": 0.936,
+      "step": 9580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.14678575625808e-05,
+      "loss": 0.8702,
+      "step": 9590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.145316723469268e-05,
+      "loss": 0.8496,
+      "step": 9600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.143847690680456e-05,
+      "loss": 0.9518,
+      "step": 9610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.142378657891645e-05,
+      "loss": 0.9027,
+      "step": 9620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.140909625102832e-05,
+      "loss": 0.8916,
+      "step": 9630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.1394405923140204e-05,
+      "loss": 0.8939,
+      "step": 9640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1379715595252086e-05,
+      "loss": 0.8696,
+      "step": 9650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1365025267363974e-05,
+      "loss": 0.9312,
+      "step": 9660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.135033493947585e-05,
+      "loss": 0.8905,
+      "step": 9670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.133564461158773e-05,
+      "loss": 0.8579,
+      "step": 9680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.132095428369961e-05,
+      "loss": 0.8597,
+      "step": 9690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.13062639558115e-05,
+      "loss": 0.8951,
+      "step": 9700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1291573627923375e-05,
+      "loss": 0.8688,
+      "step": 9710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1276883300035256e-05,
+      "loss": 0.8764,
+      "step": 9720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.126219297214714e-05,
+      "loss": 0.8369,
+      "step": 9730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1247502644259026e-05,
+      "loss": 0.9228,
+      "step": 9740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.123281231637091e-05,
+      "loss": 0.8963,
+      "step": 9750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.121812198848278e-05,
+      "loss": 0.9031,
+      "step": 9760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.1203431660594664e-05,
+      "loss": 0.8296,
+      "step": 9770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1188741332706545e-05,
+      "loss": 0.9386,
+      "step": 9780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1174051004818434e-05,
+      "loss": 0.8713,
+      "step": 9790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.115936067693031e-05,
+      "loss": 0.8901,
+      "step": 9800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.114467034904219e-05,
+      "loss": 0.9207,
+      "step": 9810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.112998002115407e-05,
+      "loss": 0.8851,
+      "step": 9820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.111528969326596e-05,
+      "loss": 0.9187,
+      "step": 9830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.110059936537784e-05,
+      "loss": 0.9037,
+      "step": 9840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1085909037489716e-05,
+      "loss": 0.9211,
+      "step": 9850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.10712187096016e-05,
+      "loss": 0.8531,
+      "step": 9860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.1056528381713486e-05,
+      "loss": 0.9026,
+      "step": 9870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.104183805382537e-05,
+      "loss": 0.8681,
+      "step": 9880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.102714772593724e-05,
+      "loss": 0.8986,
+      "step": 9890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1012457398049123e-05,
+      "loss": 0.8837,
+      "step": 9900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.099776707016101e-05,
+      "loss": 0.8816,
+      "step": 9910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.098307674227289e-05,
+      "loss": 0.9099,
+      "step": 9920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.096838641438477e-05,
+      "loss": 0.8693,
+      "step": 9930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.095369608649665e-05,
+      "loss": 0.9014,
+      "step": 9940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.093900575860853e-05,
+      "loss": 0.8561,
+      "step": 9950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.092431543072042e-05,
+      "loss": 0.8178,
+      "step": 9960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.09096251028323e-05,
+      "loss": 0.8695,
+      "step": 9970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.0894934774944175e-05,
+      "loss": 0.878,
+      "step": 9980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.088024444705606e-05,
+      "loss": 0.8924,
+      "step": 9990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.0865554119167945e-05,
+      "loss": 0.876,
+      "step": 10000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.085086379127983e-05,
+      "loss": 0.9255,
+      "step": 10010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.08361734633917e-05,
+      "loss": 0.8642,
+      "step": 10020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.082148313550358e-05,
+      "loss": 0.8676,
+      "step": 10030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.080679280761547e-05,
+      "loss": 0.8953,
+      "step": 10040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.079210247972735e-05,
+      "loss": 0.8712,
+      "step": 10050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.077741215183923e-05,
+      "loss": 0.9003,
+      "step": 10060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.076272182395111e-05,
+      "loss": 0.8641,
+      "step": 10070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.074803149606299e-05,
+      "loss": 0.8596,
+      "step": 10080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.073334116817488e-05,
+      "loss": 0.8826,
+      "step": 10090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.071865084028676e-05,
+      "loss": 0.9141,
+      "step": 10100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0703960512398635e-05,
+      "loss": 0.8354,
+      "step": 10110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0689270184510517e-05,
+      "loss": 0.8542,
+      "step": 10120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0674579856622405e-05,
+      "loss": 0.8413,
+      "step": 10130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0659889528734286e-05,
+      "loss": 0.923,
+      "step": 10140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.064519920084616e-05,
+      "loss": 0.9084,
+      "step": 10150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.063050887295804e-05,
+      "loss": 0.8891,
+      "step": 10160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.061581854506993e-05,
+      "loss": 0.9236,
+      "step": 10170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.060112821718181e-05,
+      "loss": 0.8644,
+      "step": 10180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.0586437889293694e-05,
+      "loss": 0.8662,
+      "step": 10190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.057174756140557e-05,
+      "loss": 0.8365,
+      "step": 10200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.055705723351746e-05,
+      "loss": 0.8493,
+      "step": 10210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.054236690562934e-05,
+      "loss": 0.8388,
+      "step": 10220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.052767657774122e-05,
+      "loss": 0.8607,
+      "step": 10230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.0512986249853095e-05,
+      "loss": 0.866,
+      "step": 10240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.0498295921964976e-05,
+      "loss": 0.8592,
+      "step": 10250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.0483605594076864e-05,
+      "loss": 0.8032,
+      "step": 10260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.0468915266188746e-05,
+      "loss": 0.8707,
+      "step": 10270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.045422493830062e-05,
+      "loss": 0.8648,
+      "step": 10280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.04395346104125e-05,
+      "loss": 0.8899,
+      "step": 10290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.042484428252439e-05,
+      "loss": 0.8637,
+      "step": 10300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.041015395463627e-05,
+      "loss": 0.9303,
+      "step": 10310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0395463626748153e-05,
+      "loss": 0.9301,
+      "step": 10320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.038077329886003e-05,
+      "loss": 0.8612,
+      "step": 10330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0366082970971916e-05,
+      "loss": 0.8664,
+      "step": 10340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.03513926430838e-05,
+      "loss": 0.8583,
+      "step": 10350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.033670231519568e-05,
+      "loss": 0.9188,
+      "step": 10360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0322011987307554e-05,
+      "loss": 0.9018,
+      "step": 10370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.030732165941944e-05,
+      "loss": 0.8611,
+      "step": 10380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0292631331531324e-05,
+      "loss": 0.8737,
+      "step": 10390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0277941003643205e-05,
+      "loss": 0.8986,
+      "step": 10400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.026325067575508e-05,
+      "loss": 0.8448,
+      "step": 10410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.024856034786696e-05,
+      "loss": 0.8693,
+      "step": 10420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.023387001997885e-05,
+      "loss": 0.8856,
+      "step": 10430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.021917969209073e-05,
+      "loss": 0.8367,
+      "step": 10440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.020448936420261e-05,
+      "loss": 0.9003,
+      "step": 10450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.018979903631449e-05,
+      "loss": 0.8471,
+      "step": 10460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0175108708426376e-05,
+      "loss": 0.8927,
+      "step": 10470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.016041838053826e-05,
+      "loss": 0.8289,
+      "step": 10480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.014572805265014e-05,
+      "loss": 0.8853,
+      "step": 10490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0131037724762014e-05,
+      "loss": 0.8393,
+      "step": 10500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.01163473968739e-05,
+      "loss": 0.8716,
+      "step": 10510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.0101657068985784e-05,
+      "loss": 0.8807,
+      "step": 10520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0086966741097665e-05,
+      "loss": 0.8733,
+      "step": 10530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0072276413209547e-05,
+      "loss": 0.8339,
+      "step": 10540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.005758608532142e-05,
+      "loss": 0.9074,
+      "step": 10550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.004289575743331e-05,
+      "loss": 0.9111,
+      "step": 10560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.002820542954519e-05,
+      "loss": 0.8481,
+      "step": 10570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.001351510165707e-05,
+      "loss": 0.8914,
+      "step": 10580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.999882477376895e-05,
+      "loss": 0.8561,
+      "step": 10590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9984134445880836e-05,
+      "loss": 0.9308,
+      "step": 10600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.996944411799272e-05,
+      "loss": 0.897,
+      "step": 10610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.99547537901046e-05,
+      "loss": 0.8875,
+      "step": 10620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.994006346221647e-05,
+      "loss": 0.896,
+      "step": 10630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.992537313432836e-05,
+      "loss": 0.8112,
+      "step": 10640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.991068280644024e-05,
+      "loss": 0.8924,
+      "step": 10650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9895992478552125e-05,
+      "loss": 0.8163,
+      "step": 10660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9881302150664006e-05,
+      "loss": 0.9103,
+      "step": 10670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.986661182277589e-05,
+      "loss": 0.8356,
+      "step": 10680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.985192149488777e-05,
+      "loss": 0.8728,
+      "step": 10690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.983723116699965e-05,
+      "loss": 0.8659,
+      "step": 10700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.982254083911153e-05,
+      "loss": 0.8867,
+      "step": 10710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.980785051122341e-05,
+      "loss": 0.8382,
+      "step": 10720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9793160183335295e-05,
+      "loss": 0.8952,
+      "step": 10730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.977846985544718e-05,
+      "loss": 0.8362,
+      "step": 10740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.976377952755906e-05,
+      "loss": 0.9009,
+      "step": 10750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.974908919967093e-05,
+      "loss": 0.9107,
+      "step": 10760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.973439887178282e-05,
+      "loss": 0.8573,
+      "step": 10770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.97197085438947e-05,
+      "loss": 0.8775,
+      "step": 10780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9705018216006584e-05,
+      "loss": 0.8896,
+      "step": 10790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9690327888118466e-05,
+      "loss": 0.8982,
+      "step": 10800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.967563756023035e-05,
+      "loss": 0.8407,
+      "step": 10810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.966094723234223e-05,
+      "loss": 0.834,
+      "step": 10820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.964625690445411e-05,
+      "loss": 0.8487,
+      "step": 10830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.963156657656599e-05,
+      "loss": 0.9157,
+      "step": 10840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.961687624867787e-05,
+      "loss": 0.885,
+      "step": 10850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9602185920789755e-05,
+      "loss": 0.8681,
+      "step": 10860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9587495592901636e-05,
+      "loss": 0.8972,
+      "step": 10870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.957280526501352e-05,
+      "loss": 0.8672,
+      "step": 10880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.95581149371254e-05,
+      "loss": 0.8967,
+      "step": 10890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.954342460923728e-05,
+      "loss": 0.8799,
+      "step": 10900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.952873428134916e-05,
+      "loss": 0.9148,
+      "step": 10910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9514043953461044e-05,
+      "loss": 0.8056,
+      "step": 10920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9499353625572925e-05,
+      "loss": 0.8541,
+      "step": 10930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.948466329768481e-05,
+      "loss": 0.8608,
+      "step": 10940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.946997296979669e-05,
+      "loss": 0.8124,
+      "step": 10950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.945528264190857e-05,
+      "loss": 0.8836,
+      "step": 10960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.944059231402045e-05,
+      "loss": 0.8788,
+      "step": 10970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.942590198613233e-05,
+      "loss": 0.8522,
+      "step": 10980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9411211658244214e-05,
+      "loss": 0.9229,
+      "step": 10990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9396521330356096e-05,
+      "loss": 0.8758,
+      "step": 11000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.938183100246798e-05,
+      "loss": 0.8955,
+      "step": 11010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.936714067457986e-05,
+      "loss": 0.8163,
+      "step": 11020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.935245034669174e-05,
+      "loss": 0.8753,
+      "step": 11030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.933776001880362e-05,
+      "loss": 0.8616,
+      "step": 11040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.93230696909155e-05,
+      "loss": 0.8597,
+      "step": 11050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9308379363027385e-05,
+      "loss": 0.8954,
+      "step": 11060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9293689035139266e-05,
+      "loss": 0.8548,
+      "step": 11070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.927899870725115e-05,
+      "loss": 0.8215,
+      "step": 11080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.926430837936303e-05,
+      "loss": 0.8658,
+      "step": 11090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.924961805147491e-05,
+      "loss": 0.8113,
+      "step": 11100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.923492772358679e-05,
+      "loss": 0.7995,
+      "step": 11110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9220237395698674e-05,
+      "loss": 0.8767,
+      "step": 11120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9205547067810555e-05,
+      "loss": 0.8863,
+      "step": 11130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.919085673992244e-05,
+      "loss": 0.8558,
+      "step": 11140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.917616641203432e-05,
+      "loss": 0.8121,
+      "step": 11150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.91614760841462e-05,
+      "loss": 0.8536,
+      "step": 11160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.914678575625808e-05,
+      "loss": 0.8644,
+      "step": 11170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.913209542836996e-05,
+      "loss": 0.8639,
+      "step": 11180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9117405100481844e-05,
+      "loss": 0.8745,
+      "step": 11190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9102714772593726e-05,
+      "loss": 0.8533,
+      "step": 11200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.908802444470561e-05,
+      "loss": 0.8474,
+      "step": 11210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.907333411681749e-05,
+      "loss": 0.8681,
+      "step": 11220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.905864378892937e-05,
+      "loss": 0.8564,
+      "step": 11230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.904395346104125e-05,
+      "loss": 0.8229,
+      "step": 11240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9029263133153133e-05,
+      "loss": 0.8659,
+      "step": 11250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9014572805265015e-05,
+      "loss": 0.8509,
+      "step": 11260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.8999882477376896e-05,
+      "loss": 0.898,
+      "step": 11270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.898519214948878e-05,
+      "loss": 0.8732,
+      "step": 11280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.897050182160066e-05,
+      "loss": 0.8895,
+      "step": 11290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.895581149371254e-05,
+      "loss": 0.8289,
+      "step": 11300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.894112116582442e-05,
+      "loss": 0.8409,
+      "step": 11310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.8926430837936304e-05,
+      "loss": 0.942,
+      "step": 11320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.8911740510048185e-05,
+      "loss": 0.9438,
+      "step": 11330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.889705018216007e-05,
+      "loss": 0.8123,
+      "step": 11340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.888235985427195e-05,
+      "loss": 0.8604,
+      "step": 11350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.886766952638383e-05,
+      "loss": 0.8256,
+      "step": 11360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.885297919849571e-05,
+      "loss": 0.8124,
+      "step": 11370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.883828887060759e-05,
+      "loss": 0.862,
+      "step": 11380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.8823598542719474e-05,
+      "loss": 0.8239,
+      "step": 11390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.8808908214831356e-05,
+      "loss": 0.8447,
+      "step": 11400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.879421788694324e-05,
+      "loss": 0.8425,
+      "step": 11410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.877952755905512e-05,
+      "loss": 0.8332,
+      "step": 11420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8764837231167e-05,
+      "loss": 0.8692,
+      "step": 11430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.875014690327888e-05,
+      "loss": 0.8102,
+      "step": 11440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8735456575390763e-05,
+      "loss": 0.867,
+      "step": 11450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8720766247502645e-05,
+      "loss": 0.8179,
+      "step": 11460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8706075919614527e-05,
+      "loss": 0.8777,
+      "step": 11470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.869138559172641e-05,
+      "loss": 0.8434,
+      "step": 11480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.867669526383829e-05,
+      "loss": 0.8357,
+      "step": 11490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.866200493595018e-05,
+      "loss": 0.8012,
+      "step": 11500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.864731460806205e-05,
+      "loss": 0.8696,
+      "step": 11510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8632624280173934e-05,
+      "loss": 0.8972,
+      "step": 11520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8617933952285816e-05,
+      "loss": 0.8675,
+      "step": 11530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.86032436243977e-05,
+      "loss": 0.8139,
+      "step": 11540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.858855329650958e-05,
+      "loss": 0.8702,
+      "step": 11550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.857386296862146e-05,
+      "loss": 0.898,
+      "step": 11560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.855917264073334e-05,
+      "loss": 0.8095,
+      "step": 11570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.854448231284522e-05,
+      "loss": 0.8472,
+      "step": 11580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.852979198495711e-05,
+      "loss": 0.8717,
+      "step": 11590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8515101657068986e-05,
+      "loss": 0.8423,
+      "step": 11600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.850041132918087e-05,
+      "loss": 0.8522,
+      "step": 11610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.848572100129275e-05,
+      "loss": 0.8577,
+      "step": 11620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.847103067340464e-05,
+      "loss": 0.8813,
+      "step": 11630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.845634034551651e-05,
+      "loss": 0.856,
+      "step": 11640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8441650017628394e-05,
+      "loss": 0.8749,
+      "step": 11650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8426959689740275e-05,
+      "loss": 0.8411,
+      "step": 11660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.841226936185216e-05,
+      "loss": 0.9075,
+      "step": 11670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.839757903396404e-05,
+      "loss": 0.8638,
+      "step": 11680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.838288870607592e-05,
+      "loss": 0.8181,
+      "step": 11690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.83681983781878e-05,
+      "loss": 0.848,
+      "step": 11700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.835350805029968e-05,
+      "loss": 0.841,
+      "step": 11710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.833881772241157e-05,
+      "loss": 0.8594,
+      "step": 11720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8324127394523446e-05,
+      "loss": 0.8535,
+      "step": 11730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.830943706663533e-05,
+      "loss": 0.9165,
+      "step": 11740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.829474673874721e-05,
+      "loss": 0.8424,
+      "step": 11750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.82800564108591e-05,
+      "loss": 0.8787,
+      "step": 11760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.826536608297097e-05,
+      "loss": 0.8674,
+      "step": 11770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.825067575508285e-05,
+      "loss": 0.836,
+      "step": 11780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8235985427194735e-05,
+      "loss": 0.7931,
+      "step": 11790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.822129509930662e-05,
+      "loss": 0.9458,
+      "step": 11800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.82066047714185e-05,
+      "loss": 0.8748,
+      "step": 11810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.819191444353038e-05,
+      "loss": 0.8691,
+      "step": 11820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.817722411564226e-05,
+      "loss": 0.8576,
+      "step": 11830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.816253378775414e-05,
+      "loss": 0.8425,
+      "step": 11840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.814784345986603e-05,
+      "loss": 0.9407,
+      "step": 11850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8133153131977905e-05,
+      "loss": 0.7633,
+      "step": 11860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.811846280408979e-05,
+      "loss": 0.8708,
+      "step": 11870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.810377247620167e-05,
+      "loss": 0.8314,
+      "step": 11880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8089082148313556e-05,
+      "loss": 0.9193,
+      "step": 11890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.807439182042543e-05,
+      "loss": 0.8265,
+      "step": 11900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.805970149253731e-05,
+      "loss": 0.8689,
+      "step": 11910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8045011164649194e-05,
+      "loss": 0.8512,
+      "step": 11920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.803032083676108e-05,
+      "loss": 0.8755,
+      "step": 11930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8015630508872964e-05,
+      "loss": 0.8362,
+      "step": 11940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.800094018098484e-05,
+      "loss": 0.8373,
+      "step": 11950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.798624985309672e-05,
+      "loss": 0.8696,
+      "step": 11960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.797155952520861e-05,
+      "loss": 0.7984,
+      "step": 11970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.795686919732049e-05,
+      "loss": 0.8312,
+      "step": 11980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7942178869432365e-05,
+      "loss": 0.8461,
+      "step": 11990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7927488541544246e-05,
+      "loss": 0.8632,
+      "step": 12000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.791279821365613e-05,
+      "loss": 0.8798,
+      "step": 12010
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7898107885768016e-05,
+      "loss": 0.8416,
+      "step": 12020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.788341755787989e-05,
+      "loss": 0.8531,
+      "step": 12030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.786872722999177e-05,
+      "loss": 0.8551,
+      "step": 12040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7854036902103654e-05,
+      "loss": 0.8286,
+      "step": 12050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.783934657421554e-05,
+      "loss": 0.833,
+      "step": 12060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7824656246327424e-05,
+      "loss": 0.8475,
+      "step": 12070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.78099659184393e-05,
+      "loss": 0.8436,
+      "step": 12080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.779527559055118e-05,
+      "loss": 0.8467,
+      "step": 12090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.778058526266307e-05,
+      "loss": 0.8707,
+      "step": 12100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.776589493477495e-05,
+      "loss": 0.8695,
+      "step": 12110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7751204606886824e-05,
+      "loss": 0.8497,
+      "step": 12120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7736514278998706e-05,
+      "loss": 0.8431,
+      "step": 12130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7721823951110594e-05,
+      "loss": 0.8808,
+      "step": 12140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7707133623222476e-05,
+      "loss": 0.8923,
+      "step": 12150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.769244329533435e-05,
+      "loss": 0.9064,
+      "step": 12160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.767775296744623e-05,
+      "loss": 0.8548,
+      "step": 12170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.766306263955811e-05,
+      "loss": 0.903,
+      "step": 12180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.764837231167e-05,
+      "loss": 0.8979,
+      "step": 12190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.763368198378188e-05,
+      "loss": 0.8716,
+      "step": 12200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.761899165589376e-05,
+      "loss": 0.8683,
+      "step": 12210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.760430132800564e-05,
+      "loss": 0.8444,
+      "step": 12220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.758961100011753e-05,
+      "loss": 0.8507,
+      "step": 12230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.757492067222941e-05,
+      "loss": 0.8798,
+      "step": 12240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7560230344341284e-05,
+      "loss": 0.8541,
+      "step": 12250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7545540016453165e-05,
+      "loss": 0.8167,
+      "step": 12260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7530849688565054e-05,
+      "loss": 0.9093,
+      "step": 12270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7516159360676935e-05,
+      "loss": 0.8632,
+      "step": 12280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.750146903278882e-05,
+      "loss": 0.8992,
+      "step": 12290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.748677870490069e-05,
+      "loss": 0.876,
+      "step": 12300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.747208837701258e-05,
+      "loss": 0.8237,
+      "step": 12310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.745739804912446e-05,
+      "loss": 0.8165,
+      "step": 12320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.744270772123634e-05,
+      "loss": 0.7832,
+      "step": 12330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.742801739334822e-05,
+      "loss": 0.8783,
+      "step": 12340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.74133270654601e-05,
+      "loss": 0.8415,
+      "step": 12350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.739863673757199e-05,
+      "loss": 0.8106,
+      "step": 12360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.738394640968387e-05,
+      "loss": 0.8615,
+      "step": 12370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7369256081795743e-05,
+      "loss": 0.8667,
+      "step": 12380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7354565753907625e-05,
+      "loss": 0.8609,
+      "step": 12390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.733987542601951e-05,
+      "loss": 0.8507,
+      "step": 12400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7325185098131395e-05,
+      "loss": 0.8864,
+      "step": 12410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7310494770243276e-05,
+      "loss": 0.8449,
+      "step": 12420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.729580444235515e-05,
+      "loss": 0.797,
+      "step": 12430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.728111411446704e-05,
+      "loss": 0.8783,
+      "step": 12440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.726642378657892e-05,
+      "loss": 0.9068,
+      "step": 12450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.72517334586908e-05,
+      "loss": 0.866,
+      "step": 12460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.723704313080268e-05,
+      "loss": 0.9002,
+      "step": 12470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.722235280291456e-05,
+      "loss": 0.7855,
+      "step": 12480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.720766247502645e-05,
+      "loss": 0.8548,
+      "step": 12490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.719297214713833e-05,
+      "loss": 0.8102,
+      "step": 12500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.71782818192502e-05,
+      "loss": 0.9192,
+      "step": 12510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7163591491362085e-05,
+      "loss": 0.8461,
+      "step": 12520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.714890116347397e-05,
+      "loss": 0.7977,
+      "step": 12530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7134210835585854e-05,
+      "loss": 0.8306,
+      "step": 12540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7119520507697736e-05,
+      "loss": 0.8085,
+      "step": 12550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.710483017980961e-05,
+      "loss": 0.8605,
+      "step": 12560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.70901398519215e-05,
+      "loss": 0.8643,
+      "step": 12570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.707544952403338e-05,
+      "loss": 0.7989,
+      "step": 12580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.706075919614526e-05,
+      "loss": 0.8722,
+      "step": 12590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7046068868257137e-05,
+      "loss": 0.939,
+      "step": 12600
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.785347679348951,
+      "eval_loss": 0.827536404132843,
+      "eval_runtime": 2008.6604,
+      "eval_samples_per_second": 401.671,
+      "eval_steps_per_second": 6.276,
+      "step": 12606
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7031378540369025e-05,
+      "loss": 0.7626,
+      "step": 12610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7016688212480906e-05,
+      "loss": 0.7947,
+      "step": 12620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.700199788459279e-05,
+      "loss": 0.8057,
+      "step": 12630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.698730755670467e-05,
+      "loss": 0.7983,
+      "step": 12640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6972617228816544e-05,
+      "loss": 0.7852,
+      "step": 12650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.695792690092843e-05,
+      "loss": 0.7845,
+      "step": 12660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.6943236573040314e-05,
+      "loss": 0.8596,
+      "step": 12670
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.6928546245152195e-05,
+      "loss": 0.8028,
+      "step": 12680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.691385591726407e-05,
+      "loss": 0.8007,
+      "step": 12690
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.689916558937596e-05,
+      "loss": 0.7881,
+      "step": 12700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.688447526148784e-05,
+      "loss": 0.7546,
+      "step": 12710
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.686978493359972e-05,
+      "loss": 0.7546,
+      "step": 12720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.6855094605711596e-05,
+      "loss": 0.83,
+      "step": 12730
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.6840404277823484e-05,
+      "loss": 0.8117,
+      "step": 12740
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.6825713949935366e-05,
+      "loss": 0.7148,
+      "step": 12750
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.681102362204725e-05,
+      "loss": 0.7356,
+      "step": 12760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.679633329415913e-05,
+      "loss": 0.7923,
+      "step": 12770
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.678164296627101e-05,
+      "loss": 0.8358,
+      "step": 12780
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.676695263838289e-05,
+      "loss": 0.8633,
+      "step": 12790
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6752262310494773e-05,
+      "loss": 0.7769,
+      "step": 12800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6737571982606655e-05,
+      "loss": 0.7376,
+      "step": 12810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.672288165471853e-05,
+      "loss": 0.7999,
+      "step": 12820
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.670819132683042e-05,
+      "loss": 0.7966,
+      "step": 12830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.66935009989423e-05,
+      "loss": 0.7778,
+      "step": 12840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.667881067105418e-05,
+      "loss": 0.7476,
+      "step": 12850
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6664120343166056e-05,
+      "loss": 0.7571,
+      "step": 12860
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6649430015277944e-05,
+      "loss": 0.8123,
+      "step": 12870
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.6634739687389825e-05,
+      "loss": 0.7667,
+      "step": 12880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.662004935950171e-05,
+      "loss": 0.7721,
+      "step": 12890
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.660535903161359e-05,
+      "loss": 0.819,
+      "step": 12900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.659066870372547e-05,
+      "loss": 0.7894,
+      "step": 12910
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.657597837583735e-05,
+      "loss": 0.7804,
+      "step": 12920
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.656128804794923e-05,
+      "loss": 0.7856,
+      "step": 12930
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6546597720061115e-05,
+      "loss": 0.7485,
+      "step": 12940
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.653190739217299e-05,
+      "loss": 0.7958,
+      "step": 12950
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.651721706428488e-05,
+      "loss": 0.8629,
+      "step": 12960
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.650252673639676e-05,
+      "loss": 0.8674,
+      "step": 12970
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.648783640850864e-05,
+      "loss": 0.8313,
+      "step": 12980
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.647314608062052e-05,
+      "loss": 0.7516,
+      "step": 12990
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6458455752732404e-05,
+      "loss": 0.7847,
+      "step": 13000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6443765424844285e-05,
+      "loss": 0.7597,
+      "step": 13010
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6429075096956167e-05,
+      "loss": 0.7538,
+      "step": 13020
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.641438476906805e-05,
+      "loss": 0.8422,
+      "step": 13030
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.639969444117993e-05,
+      "loss": 0.8162,
+      "step": 13040
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.638500411329181e-05,
+      "loss": 0.7517,
+      "step": 13050
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.637031378540369e-05,
+      "loss": 0.7413,
+      "step": 13060
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6355623457515574e-05,
+      "loss": 0.8089,
+      "step": 13070
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6340933129627456e-05,
+      "loss": 0.7642,
+      "step": 13080
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.632624280173934e-05,
+      "loss": 0.7959,
+      "step": 13090
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.631155247385122e-05,
+      "loss": 0.8499,
+      "step": 13100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.62968621459631e-05,
+      "loss": 0.7975,
+      "step": 13110
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.628217181807498e-05,
+      "loss": 0.782,
+      "step": 13120
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.626748149018686e-05,
+      "loss": 0.8235,
+      "step": 13130
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6252791162298745e-05,
+      "loss": 0.7995,
+      "step": 13140
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6238100834410626e-05,
+      "loss": 0.788,
+      "step": 13150
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.622341050652251e-05,
+      "loss": 0.7705,
+      "step": 13160
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.620872017863439e-05,
+      "loss": 0.7816,
+      "step": 13170
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.619402985074627e-05,
+      "loss": 0.8112,
+      "step": 13180
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.617933952285815e-05,
+      "loss": 0.776,
+      "step": 13190
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6164649194970034e-05,
+      "loss": 0.8111,
+      "step": 13200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6149958867081915e-05,
+      "loss": 0.7632,
+      "step": 13210
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.61352685391938e-05,
+      "loss": 0.8396,
+      "step": 13220
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.612057821130568e-05,
+      "loss": 0.7672,
+      "step": 13230
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.610588788341756e-05,
+      "loss": 0.7819,
+      "step": 13240
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.609119755552944e-05,
+      "loss": 0.7885,
+      "step": 13250
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.607650722764132e-05,
+      "loss": 0.799,
+      "step": 13260
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6061816899753204e-05,
+      "loss": 0.8394,
+      "step": 13270
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6047126571865086e-05,
+      "loss": 0.8041,
+      "step": 13280
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.603243624397697e-05,
+      "loss": 0.7693,
+      "step": 13290
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.601774591608885e-05,
+      "loss": 0.8347,
+      "step": 13300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.600305558820073e-05,
+      "loss": 0.6997,
+      "step": 13310
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.598836526031261e-05,
+      "loss": 0.7598,
+      "step": 13320
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.597367493242449e-05,
+      "loss": 0.8301,
+      "step": 13330
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.5958984604536375e-05,
+      "loss": 0.7611,
+      "step": 13340
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.5944294276648256e-05,
+      "loss": 0.773,
+      "step": 13350
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.592960394876014e-05,
+      "loss": 0.7519,
+      "step": 13360
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.591491362087202e-05,
+      "loss": 0.76,
+      "step": 13370
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.59002232929839e-05,
+      "loss": 0.7787,
+      "step": 13380
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.588553296509578e-05,
+      "loss": 0.7994,
+      "step": 13390
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.5870842637207664e-05,
+      "loss": 0.7624,
+      "step": 13400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.5856152309319545e-05,
+      "loss": 0.8144,
+      "step": 13410
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.584146198143143e-05,
+      "loss": 0.7929,
+      "step": 13420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.582677165354331e-05,
+      "loss": 0.75,
+      "step": 13430
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.581208132565519e-05,
+      "loss": 0.8071,
+      "step": 13440
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.579739099776707e-05,
+      "loss": 0.8175,
+      "step": 13450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.578270066987895e-05,
+      "loss": 0.7457,
+      "step": 13460
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.5768010341990834e-05,
+      "loss": 0.7808,
+      "step": 13470
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.5753320014102716e-05,
+      "loss": 0.7811,
+      "step": 13480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.57386296862146e-05,
+      "loss": 0.7732,
+      "step": 13490
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.572393935832648e-05,
+      "loss": 0.8023,
+      "step": 13500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.570924903043836e-05,
+      "loss": 0.8529,
+      "step": 13510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.569455870255024e-05,
+      "loss": 0.8067,
+      "step": 13520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.567986837466212e-05,
+      "loss": 0.819,
+      "step": 13530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.5665178046774005e-05,
+      "loss": 0.8085,
+      "step": 13540
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.5650487718885886e-05,
+      "loss": 0.762,
+      "step": 13550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.563579739099777e-05,
+      "loss": 0.8407,
+      "step": 13560
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.562110706310965e-05,
+      "loss": 0.8204,
+      "step": 13570
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.560641673522153e-05,
+      "loss": 0.7785,
+      "step": 13580
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.559172640733341e-05,
+      "loss": 0.7692,
+      "step": 13590
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.55770360794453e-05,
+      "loss": 0.8535,
+      "step": 13600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.5562345751557175e-05,
+      "loss": 0.7856,
+      "step": 13610
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.554765542366906e-05,
+      "loss": 0.7499,
+      "step": 13620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.553296509578094e-05,
+      "loss": 0.799,
+      "step": 13630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.551827476789282e-05,
+      "loss": 0.8215,
+      "step": 13640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.55035844400047e-05,
+      "loss": 0.7921,
+      "step": 13650
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.548889411211658e-05,
+      "loss": 0.787,
+      "step": 13660
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.5474203784228464e-05,
+      "loss": 0.7472,
+      "step": 13670
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.5459513456340346e-05,
+      "loss": 0.8201,
+      "step": 13680
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.5444823128452234e-05,
+      "loss": 0.7454,
+      "step": 13690
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.543013280056411e-05,
+      "loss": 0.7435,
+      "step": 13700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.541544247267599e-05,
+      "loss": 0.7876,
+      "step": 13710
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.540075214478787e-05,
+      "loss": 0.7836,
+      "step": 13720
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.538606181689976e-05,
+      "loss": 0.772,
+      "step": 13730
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.5371371489011635e-05,
+      "loss": 0.7928,
+      "step": 13740
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.5356681161123516e-05,
+      "loss": 0.8099,
+      "step": 13750
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.53419908332354e-05,
+      "loss": 0.8431,
+      "step": 13760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.532730050534728e-05,
+      "loss": 0.7482,
+      "step": 13770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.531261017745916e-05,
+      "loss": 0.8157,
+      "step": 13780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.529791984957104e-05,
+      "loss": 0.7578,
+      "step": 13790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.5283229521682924e-05,
+      "loss": 0.7637,
+      "step": 13800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.5268539193794805e-05,
+      "loss": 0.7676,
+      "step": 13810
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.5253848865906694e-05,
+      "loss": 0.8261,
+      "step": 13820
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.523915853801857e-05,
+      "loss": 0.8196,
+      "step": 13830
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.522446821013045e-05,
+      "loss": 0.7888,
+      "step": 13840
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.520977788224233e-05,
+      "loss": 0.8194,
+      "step": 13850
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.519508755435422e-05,
+      "loss": 0.8101,
+      "step": 13860
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.5180397226466094e-05,
+      "loss": 0.8086,
+      "step": 13870
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.5165706898577976e-05,
+      "loss": 0.7961,
+      "step": 13880
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.515101657068986e-05,
+      "loss": 0.8025,
+      "step": 13890
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.5136326242801746e-05,
+      "loss": 0.7914,
+      "step": 13900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.512163591491362e-05,
+      "loss": 0.8003,
+      "step": 13910
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.51069455870255e-05,
+      "loss": 0.759,
+      "step": 13920
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.5092255259137384e-05,
+      "loss": 0.8468,
+      "step": 13930
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.5077564931249265e-05,
+      "loss": 0.7909,
+      "step": 13940
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.506287460336115e-05,
+      "loss": 0.7592,
+      "step": 13950
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.504818427547303e-05,
+      "loss": 0.759,
+      "step": 13960
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.503349394758491e-05,
+      "loss": 0.7957,
+      "step": 13970
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.501880361969679e-05,
+      "loss": 0.7596,
+      "step": 13980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.500411329180868e-05,
+      "loss": 0.8093,
+      "step": 13990
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.4989422963920554e-05,
+      "loss": 0.7976,
+      "step": 14000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.4974732636032436e-05,
+      "loss": 0.7968,
+      "step": 14010
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.496004230814432e-05,
+      "loss": 0.792,
+      "step": 14020
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.4945351980256205e-05,
+      "loss": 0.7553,
+      "step": 14030
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.493066165236809e-05,
+      "loss": 0.8211,
+      "step": 14040
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.491597132447996e-05,
+      "loss": 0.8004,
+      "step": 14050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.490128099659184e-05,
+      "loss": 0.7964,
+      "step": 14060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.488659066870373e-05,
+      "loss": 0.8077,
+      "step": 14070
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.487190034081561e-05,
+      "loss": 0.7875,
+      "step": 14080
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.485721001292749e-05,
+      "loss": 0.7985,
+      "step": 14090
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.484251968503937e-05,
+      "loss": 0.801,
+      "step": 14100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.482782935715125e-05,
+      "loss": 0.751,
+      "step": 14110
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.481313902926314e-05,
+      "loss": 0.7704,
+      "step": 14120
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.4798448701375014e-05,
+      "loss": 0.8156,
+      "step": 14130
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.4783758373486895e-05,
+      "loss": 0.7657,
+      "step": 14140
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.476906804559878e-05,
+      "loss": 0.8065,
+      "step": 14150
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.4754377717710665e-05,
+      "loss": 0.7661,
+      "step": 14160
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.4739687389822546e-05,
+      "loss": 0.8392,
+      "step": 14170
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.472499706193442e-05,
+      "loss": 0.8343,
+      "step": 14180
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.47103067340463e-05,
+      "loss": 0.8342,
+      "step": 14190
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.469561640615819e-05,
+      "loss": 0.8214,
+      "step": 14200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.468092607827007e-05,
+      "loss": 0.8165,
+      "step": 14210
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.466623575038195e-05,
+      "loss": 0.7446,
+      "step": 14220
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.465154542249383e-05,
+      "loss": 0.8131,
+      "step": 14230
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.463685509460571e-05,
+      "loss": 0.7777,
+      "step": 14240
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.46221647667176e-05,
+      "loss": 0.8249,
+      "step": 14250
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.460747443882947e-05,
+      "loss": 0.8022,
+      "step": 14260
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.4592784110941355e-05,
+      "loss": 0.7485,
+      "step": 14270
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.4578093783053236e-05,
+      "loss": 0.7813,
+      "step": 14280
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.4563403455165124e-05,
+      "loss": 0.7611,
+      "step": 14290
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.4548713127277006e-05,
+      "loss": 0.7662,
+      "step": 14300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.453402279938888e-05,
+      "loss": 0.7297,
+      "step": 14310
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.451933247150076e-05,
+      "loss": 0.7848,
+      "step": 14320
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.450464214361265e-05,
+      "loss": 0.7634,
+      "step": 14330
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.448995181572453e-05,
+      "loss": 0.8049,
+      "step": 14340
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.447526148783641e-05,
+      "loss": 0.8545,
+      "step": 14350
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.446057115994829e-05,
+      "loss": 0.7939,
+      "step": 14360
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.4445880832060177e-05,
+      "loss": 0.7736,
+      "step": 14370
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.443119050417206e-05,
+      "loss": 0.7202,
+      "step": 14380
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.441650017628394e-05,
+      "loss": 0.7817,
+      "step": 14390
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.4401809848395814e-05,
+      "loss": 0.7508,
+      "step": 14400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.4387119520507696e-05,
+      "loss": 0.7758,
+      "step": 14410
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.4372429192619584e-05,
+      "loss": 0.82,
+      "step": 14420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.4357738864731466e-05,
+      "loss": 0.7932,
+      "step": 14430
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.434304853684334e-05,
+      "loss": 0.791,
+      "step": 14440
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.432835820895522e-05,
+      "loss": 0.7964,
+      "step": 14450
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.431366788106711e-05,
+      "loss": 0.7915,
+      "step": 14460
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.429897755317899e-05,
+      "loss": 0.7984,
+      "step": 14470
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.4284287225290866e-05,
+      "loss": 0.7986,
+      "step": 14480
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.426959689740275e-05,
+      "loss": 0.7529,
+      "step": 14490
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.4254906569514636e-05,
+      "loss": 0.827,
+      "step": 14500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.424021624162652e-05,
+      "loss": 0.7992,
+      "step": 14510
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.42255259137384e-05,
+      "loss": 0.7921,
+      "step": 14520
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.4210835585850274e-05,
+      "loss": 0.7957,
+      "step": 14530
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.419614525796216e-05,
+      "loss": 0.7542,
+      "step": 14540
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.4181454930074044e-05,
+      "loss": 0.7681,
+      "step": 14550
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.4166764602185925e-05,
+      "loss": 0.7535,
+      "step": 14560
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.41520742742978e-05,
+      "loss": 0.8259,
+      "step": 14570
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.413738394640968e-05,
+      "loss": 0.7946,
+      "step": 14580
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.412269361852157e-05,
+      "loss": 0.7481,
+      "step": 14590
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.410800329063345e-05,
+      "loss": 0.7595,
+      "step": 14600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.4093312962745326e-05,
+      "loss": 0.7952,
+      "step": 14610
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.407862263485721e-05,
+      "loss": 0.8109,
+      "step": 14620
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.4063932306969096e-05,
+      "loss": 0.7911,
+      "step": 14630
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.404924197908098e-05,
+      "loss": 0.7977,
+      "step": 14640
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.403455165119286e-05,
+      "loss": 0.8181,
+      "step": 14650
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.4019861323304733e-05,
+      "loss": 0.7834,
+      "step": 14660
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.400517099541662e-05,
+      "loss": 0.8162,
+      "step": 14670
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.39904806675285e-05,
+      "loss": 0.8195,
+      "step": 14680
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.3975790339640385e-05,
+      "loss": 0.7814,
+      "step": 14690
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.396110001175226e-05,
+      "loss": 0.7669,
+      "step": 14700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.394640968386415e-05,
+      "loss": 0.7855,
+      "step": 14710
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.393171935597603e-05,
+      "loss": 0.8002,
+      "step": 14720
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.391702902808791e-05,
+      "loss": 0.7828,
+      "step": 14730
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.390233870019979e-05,
+      "loss": 0.7498,
+      "step": 14740
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.388764837231167e-05,
+      "loss": 0.8235,
+      "step": 14750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.3872958044423555e-05,
+      "loss": 0.7993,
+      "step": 14760
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.385826771653544e-05,
+      "loss": 0.7636,
+      "step": 14770
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.384357738864732e-05,
+      "loss": 0.7762,
+      "step": 14780
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.382888706075919e-05,
+      "loss": 0.8223,
+      "step": 14790
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.381419673287108e-05,
+      "loss": 0.8192,
+      "step": 14800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.379950640498296e-05,
+      "loss": 0.7805,
+      "step": 14810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.3784816077094844e-05,
+      "loss": 0.7895,
+      "step": 14820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.377012574920672e-05,
+      "loss": 0.7997,
+      "step": 14830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.375543542131861e-05,
+      "loss": 0.7398,
+      "step": 14840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.374074509343049e-05,
+      "loss": 0.7877,
+      "step": 14850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.372605476554237e-05,
+      "loss": 0.7355,
+      "step": 14860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.371136443765425e-05,
+      "loss": 0.7411,
+      "step": 14870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.3696674109766127e-05,
+      "loss": 0.7981,
+      "step": 14880
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.3681983781878015e-05,
+      "loss": 0.874,
+      "step": 14890
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.3667293453989896e-05,
+      "loss": 0.7455,
+      "step": 14900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.365260312610178e-05,
+      "loss": 0.7971,
+      "step": 14910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.363791279821365e-05,
+      "loss": 0.7776,
+      "step": 14920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.362322247032554e-05,
+      "loss": 0.7521,
+      "step": 14930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.360853214243742e-05,
+      "loss": 0.7874,
+      "step": 14940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.3593841814549304e-05,
+      "loss": 0.821,
+      "step": 14950
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.357915148666118e-05,
+      "loss": 0.7579,
+      "step": 14960
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.356446115877307e-05,
+      "loss": 0.8291,
+      "step": 14970
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.354977083088495e-05,
+      "loss": 0.7531,
+      "step": 14980
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.353508050299683e-05,
+      "loss": 0.7855,
+      "step": 14990
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.352039017510871e-05,
+      "loss": 0.805,
+      "step": 15000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.350569984722059e-05,
+      "loss": 0.7859,
+      "step": 15010
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.3491009519332474e-05,
+      "loss": 0.8134,
+      "step": 15020
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.3476319191444356e-05,
+      "loss": 0.7419,
+      "step": 15030
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.346162886355624e-05,
+      "loss": 0.7843,
+      "step": 15040
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.344693853566811e-05,
+      "loss": 0.7773,
+      "step": 15050
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.343224820778e-05,
+      "loss": 0.8025,
+      "step": 15060
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.341755787989188e-05,
+      "loss": 0.7496,
+      "step": 15070
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.340286755200376e-05,
+      "loss": 0.8022,
+      "step": 15080
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.3388177224115645e-05,
+      "loss": 0.8039,
+      "step": 15090
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.3373486896227526e-05,
+      "loss": 0.7441,
+      "step": 15100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.335879656833941e-05,
+      "loss": 0.7805,
+      "step": 15110
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.334410624045129e-05,
+      "loss": 0.7429,
+      "step": 15120
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.332941591256317e-05,
+      "loss": 0.8038,
+      "step": 15130
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.331472558467505e-05,
+      "loss": 0.7221,
+      "step": 15140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.3300035256786934e-05,
+      "loss": 0.7883,
+      "step": 15150
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.3285344928898815e-05,
+      "loss": 0.7668,
+      "step": 15160
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.32706546010107e-05,
+      "loss": 0.7807,
+      "step": 15170
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.325596427312258e-05,
+      "loss": 0.7461,
+      "step": 15180
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.324127394523446e-05,
+      "loss": 0.8001,
+      "step": 15190
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.322658361734634e-05,
+      "loss": 0.696,
+      "step": 15200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.321189328945822e-05,
+      "loss": 0.7795,
+      "step": 15210
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.3197202961570104e-05,
+      "loss": 0.819,
+      "step": 15220
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.3182512633681986e-05,
+      "loss": 0.7546,
+      "step": 15230
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.316782230579387e-05,
+      "loss": 0.7656,
+      "step": 15240
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.315313197790575e-05,
+      "loss": 0.7336,
+      "step": 15250
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.313844165001763e-05,
+      "loss": 0.7933,
+      "step": 15260
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.312375132212951e-05,
+      "loss": 0.8064,
+      "step": 15270
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.3109060994241393e-05,
+      "loss": 0.7418,
+      "step": 15280
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.3094370666353275e-05,
+      "loss": 0.7692,
+      "step": 15290
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.3079680338465156e-05,
+      "loss": 0.7916,
+      "step": 15300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.306499001057704e-05,
+      "loss": 0.7864,
+      "step": 15310
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.305029968268892e-05,
+      "loss": 0.8059,
+      "step": 15320
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.30356093548008e-05,
+      "loss": 0.783,
+      "step": 15330
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.302091902691268e-05,
+      "loss": 0.832,
+      "step": 15340
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.3006228699024564e-05,
+      "loss": 0.7731,
+      "step": 15350
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.2991538371136446e-05,
+      "loss": 0.8042,
+      "step": 15360
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.297684804324833e-05,
+      "loss": 0.7916,
+      "step": 15370
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.296215771536021e-05,
+      "loss": 0.7935,
+      "step": 15380
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.294746738747209e-05,
+      "loss": 0.8027,
+      "step": 15390
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.293277705958397e-05,
+      "loss": 0.7603,
+      "step": 15400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.291808673169585e-05,
+      "loss": 0.7586,
+      "step": 15410
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.2903396403807735e-05,
+      "loss": 0.8139,
+      "step": 15420
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.2888706075919616e-05,
+      "loss": 0.7953,
+      "step": 15430
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.28740157480315e-05,
+      "loss": 0.763,
+      "step": 15440
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.285932542014338e-05,
+      "loss": 0.7922,
+      "step": 15450
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.284463509225526e-05,
+      "loss": 0.7384,
+      "step": 15460
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.282994476436714e-05,
+      "loss": 0.7568,
+      "step": 15470
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.2815254436479024e-05,
+      "loss": 0.7261,
+      "step": 15480
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.2800564108590905e-05,
+      "loss": 0.7602,
+      "step": 15490
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.2785873780702787e-05,
+      "loss": 0.7976,
+      "step": 15500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.277118345281467e-05,
+      "loss": 0.8046,
+      "step": 15510
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.275649312492655e-05,
+      "loss": 0.8282,
+      "step": 15520
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.274180279703843e-05,
+      "loss": 0.7825,
+      "step": 15530
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.272711246915031e-05,
+      "loss": 0.7287,
+      "step": 15540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.2712422141262194e-05,
+      "loss": 0.7711,
+      "step": 15550
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.2697731813374076e-05,
+      "loss": 0.7563,
+      "step": 15560
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.268304148548596e-05,
+      "loss": 0.7618,
+      "step": 15570
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.266835115759784e-05,
+      "loss": 0.7971,
+      "step": 15580
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.265366082970972e-05,
+      "loss": 0.7441,
+      "step": 15590
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.26389705018216e-05,
+      "loss": 0.8184,
+      "step": 15600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.262428017393348e-05,
+      "loss": 0.763,
+      "step": 15610
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.2609589846045365e-05,
+      "loss": 0.8109,
+      "step": 15620
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.2594899518157246e-05,
+      "loss": 0.7928,
+      "step": 15630
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.258020919026913e-05,
+      "loss": 0.7618,
+      "step": 15640
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.256551886238101e-05,
+      "loss": 0.73,
+      "step": 15650
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.255082853449289e-05,
+      "loss": 0.7817,
+      "step": 15660
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.253613820660477e-05,
+      "loss": 0.8167,
+      "step": 15670
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.2521447878716654e-05,
+      "loss": 0.7449,
+      "step": 15680
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.2506757550828535e-05,
+      "loss": 0.8041,
+      "step": 15690
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.249206722294042e-05,
+      "loss": 0.7851,
+      "step": 15700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.24773768950523e-05,
+      "loss": 0.7868,
+      "step": 15710
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.246268656716418e-05,
+      "loss": 0.7751,
+      "step": 15720
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.244799623927606e-05,
+      "loss": 0.8006,
+      "step": 15730
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.243330591138794e-05,
+      "loss": 0.7635,
+      "step": 15740
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.2418615583499824e-05,
+      "loss": 0.7795,
+      "step": 15750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.2403925255611706e-05,
+      "loss": 0.6922,
+      "step": 15760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.238923492772359e-05,
+      "loss": 0.77,
+      "step": 15770
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.237454459983547e-05,
+      "loss": 0.7448,
+      "step": 15780
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.235985427194736e-05,
+      "loss": 0.722,
+      "step": 15790
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.234516394405923e-05,
+      "loss": 0.7027,
+      "step": 15800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.233047361617111e-05,
+      "loss": 0.7702,
+      "step": 15810
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.2315783288282995e-05,
+      "loss": 0.7975,
+      "step": 15820
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.230109296039488e-05,
+      "loss": 0.8216,
+      "step": 15830
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.228640263250676e-05,
+      "loss": 0.814,
+      "step": 15840
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.227171230461864e-05,
+      "loss": 0.7982,
+      "step": 15850
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.225702197673052e-05,
+      "loss": 0.7402,
+      "step": 15860
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.22423316488424e-05,
+      "loss": 0.7168,
+      "step": 15870
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.2227641320954284e-05,
+      "loss": 0.7907,
+      "step": 15880
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.2212950993066165e-05,
+      "loss": 0.8127,
+      "step": 15890
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.219826066517805e-05,
+      "loss": 0.7795,
+      "step": 15900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.218357033728993e-05,
+      "loss": 0.791,
+      "step": 15910
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.2168880009401817e-05,
+      "loss": 0.7916,
+      "step": 15920
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.215418968151369e-05,
+      "loss": 0.7312,
+      "step": 15930
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.213949935362557e-05,
+      "loss": 0.7762,
+      "step": 15940
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.2124809025737454e-05,
+      "loss": 0.7964,
+      "step": 15950
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.211011869784934e-05,
+      "loss": 0.7732,
+      "step": 15960
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.209542836996122e-05,
+      "loss": 0.7543,
+      "step": 15970
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.20807380420731e-05,
+      "loss": 0.8151,
+      "step": 15980
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.206604771418498e-05,
+      "loss": 0.7808,
+      "step": 15990
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.205135738629687e-05,
+      "loss": 0.7764,
+      "step": 16000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.203666705840874e-05,
+      "loss": 0.8093,
+      "step": 16010
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.2021976730520625e-05,
+      "loss": 0.7852,
+      "step": 16020
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.2007286402632506e-05,
+      "loss": 0.7664,
+      "step": 16030
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.199259607474439e-05,
+      "loss": 0.7489,
+      "step": 16040
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.1977905746856276e-05,
+      "loss": 0.7906,
+      "step": 16050
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.196321541896815e-05,
+      "loss": 0.8657,
+      "step": 16060
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.194852509108003e-05,
+      "loss": 0.7447,
+      "step": 16070
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.1933834763191914e-05,
+      "loss": 0.7491,
+      "step": 16080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.19191444353038e-05,
+      "loss": 0.753,
+      "step": 16090
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.190445410741568e-05,
+      "loss": 0.7976,
+      "step": 16100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.188976377952756e-05,
+      "loss": 0.7275,
+      "step": 16110
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.187507345163944e-05,
+      "loss": 0.7376,
+      "step": 16120
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.186038312375133e-05,
+      "loss": 0.7337,
+      "step": 16130
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.184569279586321e-05,
+      "loss": 0.7547,
+      "step": 16140
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.1831002467975084e-05,
+      "loss": 0.802,
+      "step": 16150
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.1816312140086966e-05,
+      "loss": 0.7964,
+      "step": 16160
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.180162181219885e-05,
+      "loss": 0.7803,
+      "step": 16170
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.1786931484310736e-05,
+      "loss": 0.7979,
+      "step": 16180
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.177224115642261e-05,
+      "loss": 0.7737,
+      "step": 16190
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.175755082853449e-05,
+      "loss": 0.7588,
+      "step": 16200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.1742860500646373e-05,
+      "loss": 0.7304,
+      "step": 16210
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.172817017275826e-05,
+      "loss": 0.7293,
+      "step": 16220
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.1713479844870136e-05,
+      "loss": 0.7998,
+      "step": 16230
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.169878951698202e-05,
+      "loss": 0.7428,
+      "step": 16240
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.16840991890939e-05,
+      "loss": 0.7852,
+      "step": 16250
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.166940886120579e-05,
+      "loss": 0.7732,
+      "step": 16260
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.165471853331767e-05,
+      "loss": 0.8086,
+      "step": 16270
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.1640028205429544e-05,
+      "loss": 0.7803,
+      "step": 16280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.1625337877541425e-05,
+      "loss": 0.7377,
+      "step": 16290
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.1610647549653314e-05,
+      "loss": 0.7679,
+      "step": 16300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.1595957221765195e-05,
+      "loss": 0.771,
+      "step": 16310
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.158126689387707e-05,
+      "loss": 0.7422,
+      "step": 16320
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.156657656598895e-05,
+      "loss": 0.8016,
+      "step": 16330
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.155188623810083e-05,
+      "loss": 0.7331,
+      "step": 16340
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.153719591021272e-05,
+      "loss": 0.7691,
+      "step": 16350
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.1522505582324596e-05,
+      "loss": 0.7675,
+      "step": 16360
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.150781525443648e-05,
+      "loss": 0.7798,
+      "step": 16370
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.149312492654836e-05,
+      "loss": 0.7505,
+      "step": 16380
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.147843459866025e-05,
+      "loss": 0.8041,
+      "step": 16390
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.146374427077213e-05,
+      "loss": 0.8581,
+      "step": 16400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.1449053942884004e-05,
+      "loss": 0.7074,
+      "step": 16410
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.1434363614995885e-05,
+      "loss": 0.8211,
+      "step": 16420
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.141967328710777e-05,
+      "loss": 0.7405,
+      "step": 16430
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.1404982959219655e-05,
+      "loss": 0.8445,
+      "step": 16440
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.139029263133153e-05,
+      "loss": 0.7511,
+      "step": 16450
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.137560230344341e-05,
+      "loss": 0.8151,
+      "step": 16460
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.13609119755553e-05,
+      "loss": 0.801,
+      "step": 16470
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.134622164766718e-05,
+      "loss": 0.8055,
+      "step": 16480
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.133153131977906e-05,
+      "loss": 0.7699,
+      "step": 16490
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.131684099189094e-05,
+      "loss": 0.7832,
+      "step": 16500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.130215066400282e-05,
+      "loss": 0.7135,
+      "step": 16510
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.128746033611471e-05,
+      "loss": 0.803,
+      "step": 16520
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.127277000822659e-05,
+      "loss": 0.7643,
+      "step": 16530
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.125807968033846e-05,
+      "loss": 0.7961,
+      "step": 16540
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.1243389352450345e-05,
+      "loss": 0.7559,
+      "step": 16550
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.122869902456223e-05,
+      "loss": 0.7063,
+      "step": 16560
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.1214008696674114e-05,
+      "loss": 0.8185,
+      "step": 16570
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.119931836878599e-05,
+      "loss": 0.7969,
+      "step": 16580
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.118462804089787e-05,
+      "loss": 0.7555,
+      "step": 16590
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.116993771300976e-05,
+      "loss": 0.7519,
+      "step": 16600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.115524738512164e-05,
+      "loss": 0.8568,
+      "step": 16610
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.114055705723352e-05,
+      "loss": 0.7535,
+      "step": 16620
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.11258667293454e-05,
+      "loss": 0.7772,
+      "step": 16630
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.111117640145728e-05,
+      "loss": 0.7592,
+      "step": 16640
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.1096486073569166e-05,
+      "loss": 0.7595,
+      "step": 16650
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.108179574568105e-05,
+      "loss": 0.7582,
+      "step": 16660
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.106710541779292e-05,
+      "loss": 0.8067,
+      "step": 16670
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.1052415089904804e-05,
+      "loss": 0.7188,
+      "step": 16680
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.103772476201669e-05,
+      "loss": 0.7388,
+      "step": 16690
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.1023034434128574e-05,
+      "loss": 0.7085,
+      "step": 16700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.100834410624045e-05,
+      "loss": 0.7294,
+      "step": 16710
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.099365377835233e-05,
+      "loss": 0.7785,
+      "step": 16720
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.097896345046422e-05,
+      "loss": 0.8212,
+      "step": 16730
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.09642731225761e-05,
+      "loss": 0.738,
+      "step": 16740
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.094958279468798e-05,
+      "loss": 0.7751,
+      "step": 16750
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.0934892466799856e-05,
+      "loss": 0.8189,
+      "step": 16760
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.0920202138911745e-05,
+      "loss": 0.7955,
+      "step": 16770
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.0905511811023626e-05,
+      "loss": 0.7584,
+      "step": 16780
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.089082148313551e-05,
+      "loss": 0.7406,
+      "step": 16790
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.087613115524738e-05,
+      "loss": 0.8233,
+      "step": 16800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.0861440827359264e-05,
+      "loss": 0.7893,
+      "step": 16810
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.084675049947115e-05,
+      "loss": 0.7896,
+      "step": 16820
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.0832060171583034e-05,
+      "loss": 0.8096,
+      "step": 16830
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.0817369843694915e-05,
+      "loss": 0.7719,
+      "step": 16840
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.080267951580679e-05,
+      "loss": 0.7614,
+      "step": 16850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.078798918791868e-05,
+      "loss": 0.8957,
+      "step": 16860
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.077329886003056e-05,
+      "loss": 0.854,
+      "step": 16870
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.075860853214244e-05,
+      "loss": 0.7604,
+      "step": 16880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.0743918204254316e-05,
+      "loss": 0.7925,
+      "step": 16890
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.0729227876366204e-05,
+      "loss": 0.7724,
+      "step": 16900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.0714537548478086e-05,
+      "loss": 0.7669,
+      "step": 16910
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.069984722058997e-05,
+      "loss": 0.7384,
+      "step": 16920
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.068515689270184e-05,
+      "loss": 0.7362,
+      "step": 16930
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.067046656481373e-05,
+      "loss": 0.8216,
+      "step": 16940
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.065577623692561e-05,
+      "loss": 0.6931,
+      "step": 16950
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.064108590903749e-05,
+      "loss": 0.7652,
+      "step": 16960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.0626395581149375e-05,
+      "loss": 0.7757,
+      "step": 16970
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.061170525326125e-05,
+      "loss": 0.7722,
+      "step": 16980
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.059701492537314e-05,
+      "loss": 0.7484,
+      "step": 16990
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.058232459748502e-05,
+      "loss": 0.7138,
+      "step": 17000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.05676342695969e-05,
+      "loss": 0.7354,
+      "step": 17010
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.0552943941708775e-05,
+      "loss": 0.7648,
+      "step": 17020
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.0538253613820664e-05,
+      "loss": 0.7771,
+      "step": 17030
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.0523563285932545e-05,
+      "loss": 0.7876,
+      "step": 17040
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.0508872958044427e-05,
+      "loss": 0.739,
+      "step": 17050
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.0494182630156305e-05,
+      "loss": 0.7413,
+      "step": 17060
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.0479492302268186e-05,
+      "loss": 0.752,
+      "step": 17070
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.0464801974380068e-05,
+      "loss": 0.7669,
+      "step": 17080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.0450111646491953e-05,
+      "loss": 0.8291,
+      "step": 17090
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.0435421318603834e-05,
+      "loss": 0.8195,
+      "step": 17100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.0420730990715712e-05,
+      "loss": 0.7926,
+      "step": 17110
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.0406040662827594e-05,
+      "loss": 0.7257,
+      "step": 17120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.039135033493948e-05,
+      "loss": 0.7665,
+      "step": 17130
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.037666000705136e-05,
+      "loss": 0.7807,
+      "step": 17140
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.036196967916324e-05,
+      "loss": 0.7648,
+      "step": 17150
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.034727935127512e-05,
+      "loss": 0.7391,
+      "step": 17160
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.0332589023387005e-05,
+      "loss": 0.8252,
+      "step": 17170
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.0317898695498886e-05,
+      "loss": 0.7116,
+      "step": 17180
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.0303208367610768e-05,
+      "loss": 0.7859,
+      "step": 17190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.0288518039722646e-05,
+      "loss": 0.7531,
+      "step": 17200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.027382771183453e-05,
+      "loss": 0.8152,
+      "step": 17210
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.0259137383946412e-05,
+      "loss": 0.7257,
+      "step": 17220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.0244447056058294e-05,
+      "loss": 0.7228,
+      "step": 17230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.0229756728170172e-05,
+      "loss": 0.7729,
+      "step": 17240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.0215066400282053e-05,
+      "loss": 0.7818,
+      "step": 17250
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.0200376072393938e-05,
+      "loss": 0.7644,
+      "step": 17260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.018568574450582e-05,
+      "loss": 0.7901,
+      "step": 17270
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.0170995416617698e-05,
+      "loss": 0.7414,
+      "step": 17280
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.015630508872958e-05,
+      "loss": 0.7494,
+      "step": 17290
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.0141614760841464e-05,
+      "loss": 0.7852,
+      "step": 17300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.0126924432953346e-05,
+      "loss": 0.7436,
+      "step": 17310
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.011223410506523e-05,
+      "loss": 0.7725,
+      "step": 17320
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.0097543777177105e-05,
+      "loss": 0.8177,
+      "step": 17330
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.008285344928899e-05,
+      "loss": 0.7464,
+      "step": 17340
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.0068163121400872e-05,
+      "loss": 0.7298,
+      "step": 17350
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.0053472793512753e-05,
+      "loss": 0.7295,
+      "step": 17360
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.003878246562463e-05,
+      "loss": 0.7741,
+      "step": 17370
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.0024092137736516e-05,
+      "loss": 0.7584,
+      "step": 17380
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.0009401809848398e-05,
+      "loss": 0.842,
+      "step": 17390
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.999471148196028e-05,
+      "loss": 0.7817,
+      "step": 17400
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.9980021154072157e-05,
+      "loss": 0.7779,
+      "step": 17410
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.996533082618404e-05,
+      "loss": 0.83,
+      "step": 17420
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.9950640498295924e-05,
+      "loss": 0.769,
+      "step": 17430
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.9935950170407805e-05,
+      "loss": 0.787,
+      "step": 17440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.992125984251969e-05,
+      "loss": 0.7655,
+      "step": 17450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.9906569514631565e-05,
+      "loss": 0.7266,
+      "step": 17460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.989187918674345e-05,
+      "loss": 0.7885,
+      "step": 17470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.987718885885533e-05,
+      "loss": 0.8088,
+      "step": 17480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.9862498530967213e-05,
+      "loss": 0.7561,
+      "step": 17490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.984780820307909e-05,
+      "loss": 0.7585,
+      "step": 17500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.9833117875190976e-05,
+      "loss": 0.7403,
+      "step": 17510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.9818427547302857e-05,
+      "loss": 0.8046,
+      "step": 17520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.980373721941474e-05,
+      "loss": 0.7932,
+      "step": 17530
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.9789046891526624e-05,
+      "loss": 0.7178,
+      "step": 17540
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.9774356563638502e-05,
+      "loss": 0.7259,
+      "step": 17550
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.9759666235750383e-05,
+      "loss": 0.7506,
+      "step": 17560
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.9744975907862265e-05,
+      "loss": 0.7754,
+      "step": 17570
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.973028557997415e-05,
+      "loss": 0.7556,
+      "step": 17580
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.9715595252086025e-05,
+      "loss": 0.7863,
+      "step": 17590
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.970090492419791e-05,
+      "loss": 0.7794,
+      "step": 17600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.968621459630979e-05,
+      "loss": 0.758,
+      "step": 17610
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.9671524268421676e-05,
+      "loss": 0.7956,
+      "step": 17620
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.965683394053355e-05,
+      "loss": 0.7906,
+      "step": 17630
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.9642143612645435e-05,
+      "loss": 0.756,
+      "step": 17640
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.9627453284757317e-05,
+      "loss": 0.7754,
+      "step": 17650
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.96127629568692e-05,
+      "loss": 0.8384,
+      "step": 17660
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.9598072628981083e-05,
+      "loss": 0.7799,
+      "step": 17670
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.958338230109296e-05,
+      "loss": 0.795,
+      "step": 17680
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.9568691973204843e-05,
+      "loss": 0.6872,
+      "step": 17690
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.9554001645316724e-05,
+      "loss": 0.8041,
+      "step": 17700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.953931131742861e-05,
+      "loss": 0.7729,
+      "step": 17710
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.9524620989540484e-05,
+      "loss": 0.7663,
+      "step": 17720
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.950993066165237e-05,
+      "loss": 0.7851,
+      "step": 17730
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.949524033376425e-05,
+      "loss": 0.8074,
+      "step": 17740
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.9480550005876135e-05,
+      "loss": 0.765,
+      "step": 17750
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.946585967798801e-05,
+      "loss": 0.742,
+      "step": 17760
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.9451169350099895e-05,
+      "loss": 0.7957,
+      "step": 17770
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.9436479022211777e-05,
+      "loss": 0.7754,
+      "step": 17780
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.942178869432366e-05,
+      "loss": 0.7761,
+      "step": 17790
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.9407098366435543e-05,
+      "loss": 0.7872,
+      "step": 17800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.939240803854742e-05,
+      "loss": 0.7198,
+      "step": 17810
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.9377717710659303e-05,
+      "loss": 0.7827,
+      "step": 17820
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.9363027382771184e-05,
+      "loss": 0.8286,
+      "step": 17830
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.934833705488307e-05,
+      "loss": 0.6937,
+      "step": 17840
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.9333646726994947e-05,
+      "loss": 0.8048,
+      "step": 17850
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.931895639910683e-05,
+      "loss": 0.7357,
+      "step": 17860
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.930426607121871e-05,
+      "loss": 0.8074,
+      "step": 17870
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.9289575743330595e-05,
+      "loss": 0.7959,
+      "step": 17880
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.9274885415442476e-05,
+      "loss": 0.764,
+      "step": 17890
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.9260195087554355e-05,
+      "loss": 0.7764,
+      "step": 17900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.9245504759666236e-05,
+      "loss": 0.7749,
+      "step": 17910
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.923081443177812e-05,
+      "loss": 0.7657,
+      "step": 17920
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.9216124103890002e-05,
+      "loss": 0.7492,
+      "step": 17930
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.920143377600188e-05,
+      "loss": 0.7597,
+      "step": 17940
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.9186743448113762e-05,
+      "loss": 0.7866,
+      "step": 17950
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.9172053120225644e-05,
+      "loss": 0.7872,
+      "step": 17960
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.915736279233753e-05,
+      "loss": 0.729,
+      "step": 17970
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.9142672464449407e-05,
+      "loss": 0.7709,
+      "step": 17980
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.9127982136561288e-05,
+      "loss": 0.8134,
+      "step": 17990
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.911329180867317e-05,
+      "loss": 0.7076,
+      "step": 18000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.9098601480785055e-05,
+      "loss": 0.7293,
+      "step": 18010
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.9083911152896936e-05,
+      "loss": 0.7655,
+      "step": 18020
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.9069220825008814e-05,
+      "loss": 0.795,
+      "step": 18030
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.9054530497120696e-05,
+      "loss": 0.7135,
+      "step": 18040
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.903984016923258e-05,
+      "loss": 0.7661,
+      "step": 18050
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.9025149841344462e-05,
+      "loss": 0.8056,
+      "step": 18060
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.901045951345634e-05,
+      "loss": 0.8362,
+      "step": 18070
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.899576918556822e-05,
+      "loss": 0.7502,
+      "step": 18080
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.8981078857680107e-05,
+      "loss": 0.7286,
+      "step": 18090
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.8966388529791988e-05,
+      "loss": 0.7444,
+      "step": 18100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.8951698201903866e-05,
+      "loss": 0.7578,
+      "step": 18110
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.8937007874015748e-05,
+      "loss": 0.7708,
+      "step": 18120
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.892231754612763e-05,
+      "loss": 0.7476,
+      "step": 18130
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.8907627218239514e-05,
+      "loss": 0.7752,
+      "step": 18140
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.8892936890351396e-05,
+      "loss": 0.8123,
+      "step": 18150
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.8878246562463274e-05,
+      "loss": 0.7413,
+      "step": 18160
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.8863556234575155e-05,
+      "loss": 0.7609,
+      "step": 18170
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.884886590668704e-05,
+      "loss": 0.7562,
+      "step": 18180
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.883417557879892e-05,
+      "loss": 0.7895,
+      "step": 18190
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.88194852509108e-05,
+      "loss": 0.8291,
+      "step": 18200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.880479492302268e-05,
+      "loss": 0.7626,
+      "step": 18210
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.8790104595134566e-05,
+      "loss": 0.8026,
+      "step": 18220
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.8775414267246448e-05,
+      "loss": 0.7011,
+      "step": 18230
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.876072393935833e-05,
+      "loss": 0.7679,
+      "step": 18240
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.8746033611470207e-05,
+      "loss": 0.8308,
+      "step": 18250
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.8731343283582092e-05,
+      "loss": 0.7722,
+      "step": 18260
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.8716652955693974e-05,
+      "loss": 0.7262,
+      "step": 18270
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.8701962627805855e-05,
+      "loss": 0.7708,
+      "step": 18280
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.8687272299917733e-05,
+      "loss": 0.737,
+      "step": 18290
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.8672581972029615e-05,
+      "loss": 0.7857,
+      "step": 18300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.86578916441415e-05,
+      "loss": 0.7985,
+      "step": 18310
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.864320131625338e-05,
+      "loss": 0.7109,
+      "step": 18320
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.862851098836526e-05,
+      "loss": 0.7399,
+      "step": 18330
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.861382066047714e-05,
+      "loss": 0.7626,
+      "step": 18340
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.8599130332589026e-05,
+      "loss": 0.7891,
+      "step": 18350
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.8584440004700907e-05,
+      "loss": 0.7767,
+      "step": 18360
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.856974967681279e-05,
+      "loss": 0.731,
+      "step": 18370
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.8555059348924667e-05,
+      "loss": 0.7879,
+      "step": 18380
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.8540369021036552e-05,
+      "loss": 0.789,
+      "step": 18390
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.8525678693148433e-05,
+      "loss": 0.7904,
+      "step": 18400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.8510988365260315e-05,
+      "loss": 0.7345,
+      "step": 18410
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.8496298037372193e-05,
+      "loss": 0.775,
+      "step": 18420
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.8481607709484078e-05,
+      "loss": 0.7591,
+      "step": 18430
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.846691738159596e-05,
+      "loss": 0.7842,
+      "step": 18440
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.845222705370784e-05,
+      "loss": 0.6846,
+      "step": 18450
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.843753672581972e-05,
+      "loss": 0.7997,
+      "step": 18460
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.84228463979316e-05,
+      "loss": 0.7549,
+      "step": 18470
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.8408156070043485e-05,
+      "loss": 0.7724,
+      "step": 18480
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.8393465742155367e-05,
+      "loss": 0.7542,
+      "step": 18490
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.837877541426725e-05,
+      "loss": 0.7749,
+      "step": 18500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.8364085086379126e-05,
+      "loss": 0.6993,
+      "step": 18510
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.834939475849101e-05,
+      "loss": 0.7434,
+      "step": 18520
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.8334704430602893e-05,
+      "loss": 0.7908,
+      "step": 18530
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.8320014102714774e-05,
+      "loss": 0.7554,
+      "step": 18540
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.8305323774826652e-05,
+      "loss": 0.7506,
+      "step": 18550
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.8290633446938537e-05,
+      "loss": 0.7941,
+      "step": 18560
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.827594311905042e-05,
+      "loss": 0.7235,
+      "step": 18570
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.82612527911623e-05,
+      "loss": 0.7477,
+      "step": 18580
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.8246562463274185e-05,
+      "loss": 0.7898,
+      "step": 18590
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.823187213538606e-05,
+      "loss": 0.7575,
+      "step": 18600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.8217181807497945e-05,
+      "loss": 0.7486,
+      "step": 18610
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.8202491479609826e-05,
+      "loss": 0.829,
+      "step": 18620
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.818780115172171e-05,
+      "loss": 0.7498,
+      "step": 18630
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.8173110823833586e-05,
+      "loss": 0.7496,
+      "step": 18640
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.815842049594547e-05,
+      "loss": 0.709,
+      "step": 18650
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.8143730168057352e-05,
+      "loss": 0.7064,
+      "step": 18660
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.8129039840169237e-05,
+      "loss": 0.7973,
+      "step": 18670
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.8114349512281112e-05,
+      "loss": 0.734,
+      "step": 18680
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.8099659184392997e-05,
+      "loss": 0.7901,
+      "step": 18690
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.808496885650488e-05,
+      "loss": 0.7062,
+      "step": 18700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.807027852861676e-05,
+      "loss": 0.7285,
+      "step": 18710
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.8055588200728645e-05,
+      "loss": 0.7864,
+      "step": 18720
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.8040897872840523e-05,
+      "loss": 0.7928,
+      "step": 18730
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.8026207544952404e-05,
+      "loss": 0.7933,
+      "step": 18740
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.8011517217064286e-05,
+      "loss": 0.8055,
+      "step": 18750
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.799682688917617e-05,
+      "loss": 0.7642,
+      "step": 18760
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.7982136561288046e-05,
+      "loss": 0.7322,
+      "step": 18770
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.796744623339993e-05,
+      "loss": 0.7325,
+      "step": 18780
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.7952755905511812e-05,
+      "loss": 0.8099,
+      "step": 18790
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.7938065577623697e-05,
+      "loss": 0.7439,
+      "step": 18800
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.792337524973557e-05,
+      "loss": 0.7448,
+      "step": 18810
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.7908684921847456e-05,
+      "loss": 0.7609,
+      "step": 18820
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.7893994593959338e-05,
+      "loss": 0.7789,
+      "step": 18830
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.7879304266071223e-05,
+      "loss": 0.7669,
+      "step": 18840
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.7864613938183104e-05,
+      "loss": 0.7384,
+      "step": 18850
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.7849923610294982e-05,
+      "loss": 0.7368,
+      "step": 18860
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.7835233282406864e-05,
+      "loss": 0.7517,
+      "step": 18870
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.7820542954518745e-05,
+      "loss": 0.699,
+      "step": 18880
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.780585262663063e-05,
+      "loss": 0.8008,
+      "step": 18890
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.779116229874251e-05,
+      "loss": 0.7713,
+      "step": 18900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.777647197085439e-05,
+      "loss": 0.7775,
+      "step": 18910
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.776178164296627e-05,
+      "loss": 0.7295,
+      "step": 18920
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.7747091315078156e-05,
+      "loss": 0.745,
+      "step": 18930
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.7732400987190038e-05,
+      "loss": 0.7719,
+      "step": 18940
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.7717710659301916e-05,
+      "loss": 0.8328,
+      "step": 18950
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.7703020331413797e-05,
+      "loss": 0.7773,
+      "step": 18960
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.7688330003525682e-05,
+      "loss": 0.7321,
+      "step": 18970
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.7673639675637564e-05,
+      "loss": 0.7435,
+      "step": 18980
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.7658949347749442e-05,
+      "loss": 0.8293,
+      "step": 18990
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.7644259019861324e-05,
+      "loss": 0.7741,
+      "step": 19000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.7629568691973205e-05,
+      "loss": 0.7825,
+      "step": 19010
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.761487836408509e-05,
+      "loss": 0.7925,
+      "step": 19020
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.7600188036196968e-05,
+      "loss": 0.7334,
+      "step": 19030
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.758549770830885e-05,
+      "loss": 0.7622,
+      "step": 19040
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.757080738042073e-05,
+      "loss": 0.7564,
+      "step": 19050
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.7556117052532616e-05,
+      "loss": 0.7827,
+      "step": 19060
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.7541426724644497e-05,
+      "loss": 0.7625,
+      "step": 19070
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.7526736396756376e-05,
+      "loss": 0.7667,
+      "step": 19080
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.7512046068868257e-05,
+      "loss": 0.7737,
+      "step": 19090
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7497355740980142e-05,
+      "loss": 0.7493,
+      "step": 19100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7482665413092023e-05,
+      "loss": 0.7274,
+      "step": 19110
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.74679750852039e-05,
+      "loss": 0.8082,
+      "step": 19120
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7453284757315783e-05,
+      "loss": 0.7597,
+      "step": 19130
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7438594429427668e-05,
+      "loss": 0.7669,
+      "step": 19140
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.742390410153955e-05,
+      "loss": 0.7258,
+      "step": 19150
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7409213773651428e-05,
+      "loss": 0.8001,
+      "step": 19160
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.739452344576331e-05,
+      "loss": 0.7349,
+      "step": 19170
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.737983311787519e-05,
+      "loss": 0.7561,
+      "step": 19180
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7365142789987075e-05,
+      "loss": 0.7596,
+      "step": 19190
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7350452462098957e-05,
+      "loss": 0.8053,
+      "step": 19200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7335762134210835e-05,
+      "loss": 0.7497,
+      "step": 19210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.7321071806322717e-05,
+      "loss": 0.7399,
+      "step": 19220
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.73063814784346e-05,
+      "loss": 0.7678,
+      "step": 19230
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7291691150546483e-05,
+      "loss": 0.7709,
+      "step": 19240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.727700082265836e-05,
+      "loss": 0.7951,
+      "step": 19250
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7262310494770243e-05,
+      "loss": 0.7366,
+      "step": 19260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7247620166882128e-05,
+      "loss": 0.7363,
+      "step": 19270
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.723292983899401e-05,
+      "loss": 0.7751,
+      "step": 19280
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.721823951110589e-05,
+      "loss": 0.694,
+      "step": 19290
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.720354918321777e-05,
+      "loss": 0.7943,
+      "step": 19300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7188858855329654e-05,
+      "loss": 0.7311,
+      "step": 19310
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7174168527441535e-05,
+      "loss": 0.7951,
+      "step": 19320
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7159478199553417e-05,
+      "loss": 0.7337,
+      "step": 19330
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7144787871665295e-05,
+      "loss": 0.7096,
+      "step": 19340
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7130097543777176e-05,
+      "loss": 0.6996,
+      "step": 19350
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.711540721588906e-05,
+      "loss": 0.7646,
+      "step": 19360
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.7100716888000943e-05,
+      "loss": 0.7501,
+      "step": 19370
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.708602656011282e-05,
+      "loss": 0.7585,
+      "step": 19380
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.7071336232224702e-05,
+      "loss": 0.7552,
+      "step": 19390
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.7056645904336587e-05,
+      "loss": 0.7929,
+      "step": 19400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.704195557644847e-05,
+      "loss": 0.7785,
+      "step": 19410
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.702726524856035e-05,
+      "loss": 0.7557,
+      "step": 19420
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.7012574920672228e-05,
+      "loss": 0.7808,
+      "step": 19430
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6997884592784113e-05,
+      "loss": 0.7905,
+      "step": 19440
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6983194264895995e-05,
+      "loss": 0.7859,
+      "step": 19450
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6968503937007876e-05,
+      "loss": 0.7324,
+      "step": 19460
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6953813609119754e-05,
+      "loss": 0.7925,
+      "step": 19470
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6939123281231636e-05,
+      "loss": 0.7102,
+      "step": 19480
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.692443295334352e-05,
+      "loss": 0.7511,
+      "step": 19490
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6909742625455402e-05,
+      "loss": 0.7977,
+      "step": 19500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.689505229756728e-05,
+      "loss": 0.7988,
+      "step": 19510
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6880361969679162e-05,
+      "loss": 0.7423,
+      "step": 19520
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6865671641791047e-05,
+      "loss": 0.7891,
+      "step": 19530
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6850981313902928e-05,
+      "loss": 0.8041,
+      "step": 19540
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6836290986014813e-05,
+      "loss": 0.7648,
+      "step": 19550
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6821600658126688e-05,
+      "loss": 0.7879,
+      "step": 19560
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6806910330238573e-05,
+      "loss": 0.7817,
+      "step": 19570
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6792220002350454e-05,
+      "loss": 0.7726,
+      "step": 19580
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6777529674462336e-05,
+      "loss": 0.7431,
+      "step": 19590
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6762839346574214e-05,
+      "loss": 0.7806,
+      "step": 19600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.67481490186861e-05,
+      "loss": 0.7501,
+      "step": 19610
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.673345869079798e-05,
+      "loss": 0.7577,
+      "step": 19620
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.6718768362909862e-05,
+      "loss": 0.7902,
+      "step": 19630
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.6704078035021747e-05,
+      "loss": 0.7499,
+      "step": 19640
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.668938770713362e-05,
+      "loss": 0.7538,
+      "step": 19650
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.6674697379245506e-05,
+      "loss": 0.7874,
+      "step": 19660
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.6660007051357388e-05,
+      "loss": 0.7141,
+      "step": 19670
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.6645316723469273e-05,
+      "loss": 0.7615,
+      "step": 19680
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.6630626395581147e-05,
+      "loss": 0.7611,
+      "step": 19690
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.6615936067693032e-05,
+      "loss": 0.7344,
+      "step": 19700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.6601245739804914e-05,
+      "loss": 0.7498,
+      "step": 19710
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.65865554119168e-05,
+      "loss": 0.702,
+      "step": 19720
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6571865084028673e-05,
+      "loss": 0.7277,
+      "step": 19730
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6557174756140558e-05,
+      "loss": 0.777,
+      "step": 19740
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.654248442825244e-05,
+      "loss": 0.7607,
+      "step": 19750
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.652779410036432e-05,
+      "loss": 0.7541,
+      "step": 19760
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6513103772476206e-05,
+      "loss": 0.7859,
+      "step": 19770
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6498413444588084e-05,
+      "loss": 0.7603,
+      "step": 19780
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6483723116699966e-05,
+      "loss": 0.7785,
+      "step": 19790
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6469032788811847e-05,
+      "loss": 0.8143,
+      "step": 19800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6454342460923732e-05,
+      "loss": 0.7542,
+      "step": 19810
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6439652133035607e-05,
+      "loss": 0.7674,
+      "step": 19820
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6424961805147492e-05,
+      "loss": 0.7427,
+      "step": 19830
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6410271477259373e-05,
+      "loss": 0.7391,
+      "step": 19840
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6395581149371258e-05,
+      "loss": 0.7522,
+      "step": 19850
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6380890821483133e-05,
+      "loss": 0.7884,
+      "step": 19860
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6366200493595018e-05,
+      "loss": 0.743,
+      "step": 19870
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.63515101657069e-05,
+      "loss": 0.7769,
+      "step": 19880
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.633681983781878e-05,
+      "loss": 0.764,
+      "step": 19890
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6322129509930666e-05,
+      "loss": 0.7853,
+      "step": 19900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6307439182042544e-05,
+      "loss": 0.7442,
+      "step": 19910
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6292748854154425e-05,
+      "loss": 0.7389,
+      "step": 19920
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6278058526266307e-05,
+      "loss": 0.7431,
+      "step": 19930
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6263368198378192e-05,
+      "loss": 0.7494,
+      "step": 19940
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.624867787049007e-05,
+      "loss": 0.7756,
+      "step": 19950
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.623398754260195e-05,
+      "loss": 0.7971,
+      "step": 19960
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6219297214713833e-05,
+      "loss": 0.7963,
+      "step": 19970
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6204606886825718e-05,
+      "loss": 0.7874,
+      "step": 19980
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.61899165589376e-05,
+      "loss": 0.7517,
+      "step": 19990
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.6175226231049477e-05,
+      "loss": 0.7415,
+      "step": 20000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.616053590316136e-05,
+      "loss": 0.7447,
+      "step": 20010
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.6145845575273244e-05,
+      "loss": 0.7411,
+      "step": 20020
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.6131155247385125e-05,
+      "loss": 0.7268,
+      "step": 20030
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.6116464919497003e-05,
+      "loss": 0.7728,
+      "step": 20040
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.6101774591608885e-05,
+      "loss": 0.7276,
+      "step": 20050
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.6087084263720766e-05,
+      "loss": 0.7312,
+      "step": 20060
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.607239393583265e-05,
+      "loss": 0.7443,
+      "step": 20070
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.605770360794453e-05,
+      "loss": 0.7867,
+      "step": 20080
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.604301328005641e-05,
+      "loss": 0.7368,
+      "step": 20090
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.6028322952168292e-05,
+      "loss": 0.7978,
+      "step": 20100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.6013632624280177e-05,
+      "loss": 0.8223,
+      "step": 20110
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.599894229639206e-05,
+      "loss": 0.7241,
+      "step": 20120
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5984251968503937e-05,
+      "loss": 0.7338,
+      "step": 20130
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.596956164061582e-05,
+      "loss": 0.75,
+      "step": 20140
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5954871312727703e-05,
+      "loss": 0.7243,
+      "step": 20150
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5940180984839585e-05,
+      "loss": 0.7761,
+      "step": 20160
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5925490656951463e-05,
+      "loss": 0.7553,
+      "step": 20170
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5910800329063344e-05,
+      "loss": 0.737,
+      "step": 20180
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.589611000117523e-05,
+      "loss": 0.7723,
+      "step": 20190
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.588141967328711e-05,
+      "loss": 0.7655,
+      "step": 20200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.586672934539899e-05,
+      "loss": 0.7601,
+      "step": 20210
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.585203901751087e-05,
+      "loss": 0.7472,
+      "step": 20220
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5837348689622752e-05,
+      "loss": 0.7232,
+      "step": 20230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5822658361734637e-05,
+      "loss": 0.7941,
+      "step": 20240
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.580796803384652e-05,
+      "loss": 0.7655,
+      "step": 20250
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5793277705958397e-05,
+      "loss": 0.7502,
+      "step": 20260
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5778587378070278e-05,
+      "loss": 0.7322,
+      "step": 20270
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5763897050182163e-05,
+      "loss": 0.7366,
+      "step": 20280
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5749206722294044e-05,
+      "loss": 0.7535,
+      "step": 20290
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5734516394405923e-05,
+      "loss": 0.768,
+      "step": 20300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5719826066517804e-05,
+      "loss": 0.7663,
+      "step": 20310
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.570513573862969e-05,
+      "loss": 0.7251,
+      "step": 20320
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.569044541074157e-05,
+      "loss": 0.7807,
+      "step": 20330
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5675755082853452e-05,
+      "loss": 0.7574,
+      "step": 20340
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.566106475496533e-05,
+      "loss": 0.837,
+      "step": 20350
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.564637442707721e-05,
+      "loss": 0.7398,
+      "step": 20360
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5631684099189096e-05,
+      "loss": 0.7878,
+      "step": 20370
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5616993771300978e-05,
+      "loss": 0.8063,
+      "step": 20380
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5602303443412856e-05,
+      "loss": 0.7396,
+      "step": 20390
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5587613115524738e-05,
+      "loss": 0.7864,
+      "step": 20400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5572922787636623e-05,
+      "loss": 0.7561,
+      "step": 20410
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5558232459748504e-05,
+      "loss": 0.798,
+      "step": 20420
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5543542131860382e-05,
+      "loss": 0.7954,
+      "step": 20430
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5528851803972264e-05,
+      "loss": 0.7524,
+      "step": 20440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.551416147608415e-05,
+      "loss": 0.8193,
+      "step": 20450
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.549947114819603e-05,
+      "loss": 0.7255,
+      "step": 20460
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.548478082030791e-05,
+      "loss": 0.7976,
+      "step": 20470
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.547009049241979e-05,
+      "loss": 0.832,
+      "step": 20480
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5455400164531675e-05,
+      "loss": 0.7466,
+      "step": 20490
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5440709836643556e-05,
+      "loss": 0.7049,
+      "step": 20500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5426019508755438e-05,
+      "loss": 0.7194,
+      "step": 20510
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5411329180867316e-05,
+      "loss": 0.7478,
+      "step": 20520
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5396638852979197e-05,
+      "loss": 0.7275,
+      "step": 20530
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5381948525091082e-05,
+      "loss": 0.7629,
+      "step": 20540
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5367258197202964e-05,
+      "loss": 0.8226,
+      "step": 20550
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.535256786931484e-05,
+      "loss": 0.7572,
+      "step": 20560
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5337877541426723e-05,
+      "loss": 0.7659,
+      "step": 20570
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5323187213538608e-05,
+      "loss": 0.7169,
+      "step": 20580
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.530849688565049e-05,
+      "loss": 0.7577,
+      "step": 20590
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5293806557762374e-05,
+      "loss": 0.7345,
+      "step": 20600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.527911622987425e-05,
+      "loss": 0.7931,
+      "step": 20610
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5264425901986134e-05,
+      "loss": 0.7628,
+      "step": 20620
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5249735574098016e-05,
+      "loss": 0.7773,
+      "step": 20630
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5235045246209897e-05,
+      "loss": 0.7619,
+      "step": 20640
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5220354918321775e-05,
+      "loss": 0.7636,
+      "step": 20650
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.520566459043366e-05,
+      "loss": 0.7871,
+      "step": 20660
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.519097426254554e-05,
+      "loss": 0.7401,
+      "step": 20670
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5176283934657423e-05,
+      "loss": 0.7739,
+      "step": 20680
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5161593606769308e-05,
+      "loss": 0.7711,
+      "step": 20690
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5146903278881183e-05,
+      "loss": 0.759,
+      "step": 20700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5132212950993068e-05,
+      "loss": 0.7703,
+      "step": 20710
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.511752262310495e-05,
+      "loss": 0.7283,
+      "step": 20720
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5102832295216834e-05,
+      "loss": 0.7485,
+      "step": 20730
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.508814196732871e-05,
+      "loss": 0.7941,
+      "step": 20740
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5073451639440594e-05,
+      "loss": 0.7243,
+      "step": 20750
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5058761311552475e-05,
+      "loss": 0.7351,
+      "step": 20760
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5044070983664357e-05,
+      "loss": 0.7846,
+      "step": 20770
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5029380655776235e-05,
+      "loss": 0.6978,
+      "step": 20780
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.501469032788812e-05,
+      "loss": 0.7749,
+      "step": 20790
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5e-05,
+      "loss": 0.7521,
+      "step": 20800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4985309672111883e-05,
+      "loss": 0.7552,
+      "step": 20810
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4970619344223764e-05,
+      "loss": 0.7378,
+      "step": 20820
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4955929016335646e-05,
+      "loss": 0.7957,
+      "step": 20830
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.4941238688447527e-05,
+      "loss": 0.7934,
+      "step": 20840
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.492654836055941e-05,
+      "loss": 0.7451,
+      "step": 20850
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.491185803267129e-05,
+      "loss": 0.7281,
+      "step": 20860
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.4897167704783172e-05,
+      "loss": 0.7438,
+      "step": 20870
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.4882477376895053e-05,
+      "loss": 0.717,
+      "step": 20880
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.4867787049006935e-05,
+      "loss": 0.7535,
+      "step": 20890
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.4853096721118816e-05,
+      "loss": 0.7045,
+      "step": 20900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.4838406393230698e-05,
+      "loss": 0.7643,
+      "step": 20910
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.4823716065342583e-05,
+      "loss": 0.7302,
+      "step": 20920
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.480902573745446e-05,
+      "loss": 0.792,
+      "step": 20930
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.4794335409566342e-05,
+      "loss": 0.7822,
+      "step": 20940
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.4779645081678224e-05,
+      "loss": 0.7566,
+      "step": 20950
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.4764954753790105e-05,
+      "loss": 0.7825,
+      "step": 20960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.4750264425901987e-05,
+      "loss": 0.7717,
+      "step": 20970
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.4735574098013868e-05,
+      "loss": 0.7161,
+      "step": 20980
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.472088377012575e-05,
+      "loss": 0.7719,
+      "step": 20990
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.470619344223763e-05,
+      "loss": 0.8023,
+      "step": 21000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4691503114349513e-05,
+      "loss": 0.7677,
+      "step": 21010
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4676812786461394e-05,
+      "loss": 0.755,
+      "step": 21020
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4662122458573276e-05,
+      "loss": 0.7084,
+      "step": 21030
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4647432130685157e-05,
+      "loss": 0.7903,
+      "step": 21040
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4632741802797042e-05,
+      "loss": 0.7331,
+      "step": 21050
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.461805147490892e-05,
+      "loss": 0.7902,
+      "step": 21060
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4603361147020805e-05,
+      "loss": 0.7752,
+      "step": 21070
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4588670819132683e-05,
+      "loss": 0.7717,
+      "step": 21080
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4573980491244565e-05,
+      "loss": 0.7967,
+      "step": 21090
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4559290163356446e-05,
+      "loss": 0.7435,
+      "step": 21100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4544599835468328e-05,
+      "loss": 0.7967,
+      "step": 21110
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.452990950758021e-05,
+      "loss": 0.7653,
+      "step": 21120
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.451521917969209e-05,
+      "loss": 0.7723,
+      "step": 21130
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4500528851803972e-05,
+      "loss": 0.7502,
+      "step": 21140
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4485838523915854e-05,
+      "loss": 0.791,
+      "step": 21150
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.447114819602774e-05,
+      "loss": 0.7339,
+      "step": 21160
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4456457868139617e-05,
+      "loss": 0.7428,
+      "step": 21170
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4441767540251502e-05,
+      "loss": 0.7268,
+      "step": 21180
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.442707721236338e-05,
+      "loss": 0.7372,
+      "step": 21190
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4412386884475265e-05,
+      "loss": 0.7543,
+      "step": 21200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4397696556587143e-05,
+      "loss": 0.7305,
+      "step": 21210
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4383006228699028e-05,
+      "loss": 0.7233,
+      "step": 21220
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4368315900810906e-05,
+      "loss": 0.7654,
+      "step": 21230
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.435362557292279e-05,
+      "loss": 0.7509,
+      "step": 21240
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.433893524503467e-05,
+      "loss": 0.6923,
+      "step": 21250
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.432424491714655e-05,
+      "loss": 0.7136,
+      "step": 21260
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4309554589258435e-05,
+      "loss": 0.7432,
+      "step": 21270
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4294864261370313e-05,
+      "loss": 0.7059,
+      "step": 21280
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.42801739334822e-05,
+      "loss": 0.745,
+      "step": 21290
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4265483605594076e-05,
+      "loss": 0.7444,
+      "step": 21300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.425079327770596e-05,
+      "loss": 0.6992,
+      "step": 21310
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.423610294981784e-05,
+      "loss": 0.7564,
+      "step": 21320
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4221412621929724e-05,
+      "loss": 0.7685,
+      "step": 21330
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4206722294041602e-05,
+      "loss": 0.7601,
+      "step": 21340
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4192031966153487e-05,
+      "loss": 0.8118,
+      "step": 21350
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4177341638265365e-05,
+      "loss": 0.7625,
+      "step": 21360
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.416265131037725e-05,
+      "loss": 0.7753,
+      "step": 21370
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.414796098248913e-05,
+      "loss": 0.777,
+      "step": 21380
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4133270654601013e-05,
+      "loss": 0.7357,
+      "step": 21390
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4118580326712895e-05,
+      "loss": 0.7545,
+      "step": 21400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4103889998824773e-05,
+      "loss": 0.817,
+      "step": 21410
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4089199670936658e-05,
+      "loss": 0.7843,
+      "step": 21420
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4074509343048536e-05,
+      "loss": 0.7765,
+      "step": 21430
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.405981901516042e-05,
+      "loss": 0.745,
+      "step": 21440
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.40451286872723e-05,
+      "loss": 0.78,
+      "step": 21450
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4030438359384184e-05,
+      "loss": 0.7487,
+      "step": 21460
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4015748031496062e-05,
+      "loss": 0.7386,
+      "step": 21470
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4001057703607947e-05,
+      "loss": 0.7609,
+      "step": 21480
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.3986367375719825e-05,
+      "loss": 0.7655,
+      "step": 21490
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.397167704783171e-05,
+      "loss": 0.7871,
+      "step": 21500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.395698671994359e-05,
+      "loss": 0.7278,
+      "step": 21510
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3942296392055473e-05,
+      "loss": 0.7575,
+      "step": 21520
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3927606064167354e-05,
+      "loss": 0.7761,
+      "step": 21530
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3912915736279236e-05,
+      "loss": 0.7547,
+      "step": 21540
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3898225408391117e-05,
+      "loss": 0.7736,
+      "step": 21550
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3883535080502996e-05,
+      "loss": 0.7484,
+      "step": 21560
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.386884475261488e-05,
+      "loss": 0.7583,
+      "step": 21570
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.385415442472676e-05,
+      "loss": 0.7136,
+      "step": 21580
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3839464096838643e-05,
+      "loss": 0.7599,
+      "step": 21590
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.382477376895052e-05,
+      "loss": 0.7777,
+      "step": 21600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.3810083441062406e-05,
+      "loss": 0.7051,
+      "step": 21610
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3795393113174288e-05,
+      "loss": 0.701,
+      "step": 21620
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.378070278528617e-05,
+      "loss": 0.7327,
+      "step": 21630
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.376601245739805e-05,
+      "loss": 0.7947,
+      "step": 21640
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3751322129509933e-05,
+      "loss": 0.7019,
+      "step": 21650
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3736631801621814e-05,
+      "loss": 0.7618,
+      "step": 21660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3721941473733696e-05,
+      "loss": 0.7998,
+      "step": 21670
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3707251145845577e-05,
+      "loss": 0.7234,
+      "step": 21680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.369256081795746e-05,
+      "loss": 0.7796,
+      "step": 21690
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.367787049006934e-05,
+      "loss": 0.7569,
+      "step": 21700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.366318016218122e-05,
+      "loss": 0.7318,
+      "step": 21710
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3648489834293103e-05,
+      "loss": 0.7786,
+      "step": 21720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.363379950640498e-05,
+      "loss": 0.7494,
+      "step": 21730
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.3619109178516866e-05,
+      "loss": 0.7786,
+      "step": 21740
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3604418850628748e-05,
+      "loss": 0.803,
+      "step": 21750
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.358972852274063e-05,
+      "loss": 0.7237,
+      "step": 21760
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.357503819485251e-05,
+      "loss": 0.6989,
+      "step": 21770
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3560347866964392e-05,
+      "loss": 0.7763,
+      "step": 21780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3545657539076274e-05,
+      "loss": 0.7218,
+      "step": 21790
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3530967211188155e-05,
+      "loss": 0.7693,
+      "step": 21800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3516276883300037e-05,
+      "loss": 0.7117,
+      "step": 21810
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3501586555411918e-05,
+      "loss": 0.7138,
+      "step": 21820
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.34868962275238e-05,
+      "loss": 0.7924,
+      "step": 21830
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.347220589963568e-05,
+      "loss": 0.8146,
+      "step": 21840
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3457515571747563e-05,
+      "loss": 0.7534,
+      "step": 21850
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3442825243859444e-05,
+      "loss": 0.7547,
+      "step": 21860
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3428134915971326e-05,
+      "loss": 0.669,
+      "step": 21870
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3413444588083207e-05,
+      "loss": 0.7528,
+      "step": 21880
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.339875426019509e-05,
+      "loss": 0.747,
+      "step": 21890
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.338406393230697e-05,
+      "loss": 0.781,
+      "step": 21900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.336937360441885e-05,
+      "loss": 0.774,
+      "step": 21910
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3354683276530733e-05,
+      "loss": 0.709,
+      "step": 21920
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3339992948642615e-05,
+      "loss": 0.7768,
+      "step": 21930
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3325302620754496e-05,
+      "loss": 0.7167,
+      "step": 21940
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3310612292866378e-05,
+      "loss": 0.726,
+      "step": 21950
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.329592196497826e-05,
+      "loss": 0.7912,
+      "step": 21960
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.328123163709014e-05,
+      "loss": 0.7495,
+      "step": 21970
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3266541309202022e-05,
+      "loss": 0.7236,
+      "step": 21980
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.3251850981313904e-05,
+      "loss": 0.7508,
+      "step": 21990
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3237160653425785e-05,
+      "loss": 0.7474,
+      "step": 22000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3222470325537667e-05,
+      "loss": 0.7764,
+      "step": 22010
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3207779997649548e-05,
+      "loss": 0.7712,
+      "step": 22020
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.319308966976143e-05,
+      "loss": 0.7096,
+      "step": 22030
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.317839934187331e-05,
+      "loss": 0.7428,
+      "step": 22040
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3163709013985193e-05,
+      "loss": 0.7168,
+      "step": 22050
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3149018686097074e-05,
+      "loss": 0.764,
+      "step": 22060
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3134328358208956e-05,
+      "loss": 0.7789,
+      "step": 22070
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3119638030320837e-05,
+      "loss": 0.7127,
+      "step": 22080
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.310494770243272e-05,
+      "loss": 0.7171,
+      "step": 22090
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3090257374544604e-05,
+      "loss": 0.7676,
+      "step": 22100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3075567046656482e-05,
+      "loss": 0.8151,
+      "step": 22110
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.3060876718768367e-05,
+      "loss": 0.7795,
+      "step": 22120
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.3046186390880245e-05,
+      "loss": 0.8254,
+      "step": 22130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.3031496062992126e-05,
+      "loss": 0.7501,
+      "step": 22140
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.3016805735104008e-05,
+      "loss": 0.7965,
+      "step": 22150
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.300211540721589e-05,
+      "loss": 0.7934,
+      "step": 22160
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.298742507932777e-05,
+      "loss": 0.695,
+      "step": 22170
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2972734751439652e-05,
+      "loss": 0.7375,
+      "step": 22180
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2958044423551534e-05,
+      "loss": 0.7201,
+      "step": 22190
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2943354095663415e-05,
+      "loss": 0.7905,
+      "step": 22200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.29286637677753e-05,
+      "loss": 0.7746,
+      "step": 22210
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2913973439887178e-05,
+      "loss": 0.7793,
+      "step": 22220
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2899283111999063e-05,
+      "loss": 0.791,
+      "step": 22230
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.288459278411094e-05,
+      "loss": 0.7018,
+      "step": 22240
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.2869902456222826e-05,
+      "loss": 0.7488,
+      "step": 22250
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2855212128334704e-05,
+      "loss": 0.8065,
+      "step": 22260
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.284052180044659e-05,
+      "loss": 0.7764,
+      "step": 22270
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2825831472558467e-05,
+      "loss": 0.7566,
+      "step": 22280
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.281114114467035e-05,
+      "loss": 0.7466,
+      "step": 22290
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.279645081678223e-05,
+      "loss": 0.7766,
+      "step": 22300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2781760488894112e-05,
+      "loss": 0.7367,
+      "step": 22310
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2767070161005997e-05,
+      "loss": 0.7302,
+      "step": 22320
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2752379833117875e-05,
+      "loss": 0.7306,
+      "step": 22330
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.273768950522976e-05,
+      "loss": 0.7258,
+      "step": 22340
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2722999177341638e-05,
+      "loss": 0.7593,
+      "step": 22350
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2708308849453523e-05,
+      "loss": 0.7034,
+      "step": 22360
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.26936185215654e-05,
+      "loss": 0.7799,
+      "step": 22370
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2678928193677286e-05,
+      "loss": 0.7556,
+      "step": 22380
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2664237865789164e-05,
+      "loss": 0.7305,
+      "step": 22390
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.264954753790105e-05,
+      "loss": 0.7621,
+      "step": 22400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2634857210012927e-05,
+      "loss": 0.7069,
+      "step": 22410
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2620166882124812e-05,
+      "loss": 0.7472,
+      "step": 22420
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.260547655423669e-05,
+      "loss": 0.7963,
+      "step": 22430
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2590786226348575e-05,
+      "loss": 0.7483,
+      "step": 22440
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2576095898460456e-05,
+      "loss": 0.7288,
+      "step": 22450
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2561405570572334e-05,
+      "loss": 0.7732,
+      "step": 22460
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.254671524268422e-05,
+      "loss": 0.7369,
+      "step": 22470
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2532024914796097e-05,
+      "loss": 0.7411,
+      "step": 22480
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2517334586907982e-05,
+      "loss": 0.7241,
+      "step": 22490
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.250264425901986e-05,
+      "loss": 0.7702,
+      "step": 22500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2487953931131745e-05,
+      "loss": 0.761,
+      "step": 22510
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2473263603243623e-05,
+      "loss": 0.7797,
+      "step": 22520
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.245857327535551e-05,
+      "loss": 0.7344,
+      "step": 22530
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2443882947467386e-05,
+      "loss": 0.7423,
+      "step": 22540
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.242919261957927e-05,
+      "loss": 0.7547,
+      "step": 22550
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2414502291691153e-05,
+      "loss": 0.6623,
+      "step": 22560
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2399811963803034e-05,
+      "loss": 0.6756,
+      "step": 22570
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2385121635914916e-05,
+      "loss": 0.7307,
+      "step": 22580
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2370431308026797e-05,
+      "loss": 0.734,
+      "step": 22590
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.235574098013868e-05,
+      "loss": 0.7696,
+      "step": 22600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2341050652250557e-05,
+      "loss": 0.7712,
+      "step": 22610
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.2326360324362442e-05,
+      "loss": 0.6981,
+      "step": 22620
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.231166999647432e-05,
+      "loss": 0.7212,
+      "step": 22630
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.2296979668586205e-05,
+      "loss": 0.7154,
+      "step": 22640
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.2282289340698083e-05,
+      "loss": 0.736,
+      "step": 22650
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.2267599012809968e-05,
+      "loss": 0.7812,
+      "step": 22660
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.225290868492185e-05,
+      "loss": 0.7539,
+      "step": 22670
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.223821835703373e-05,
+      "loss": 0.7378,
+      "step": 22680
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.2223528029145612e-05,
+      "loss": 0.7328,
+      "step": 22690
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.2208837701257494e-05,
+      "loss": 0.7561,
+      "step": 22700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.2194147373369375e-05,
+      "loss": 0.8516,
+      "step": 22710
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.2179457045481257e-05,
+      "loss": 0.781,
+      "step": 22720
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.216476671759314e-05,
+      "loss": 0.742,
+      "step": 22730
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.215007638970502e-05,
+      "loss": 0.7365,
+      "step": 22740
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.21353860618169e-05,
+      "loss": 0.7708,
+      "step": 22750
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.212069573392878e-05,
+      "loss": 0.8052,
+      "step": 22760
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2106005406040664e-05,
+      "loss": 0.738,
+      "step": 22770
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2091315078152543e-05,
+      "loss": 0.7689,
+      "step": 22780
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2076624750264427e-05,
+      "loss": 0.7878,
+      "step": 22790
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.206193442237631e-05,
+      "loss": 0.7762,
+      "step": 22800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.204724409448819e-05,
+      "loss": 0.7701,
+      "step": 22810
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2032553766600072e-05,
+      "loss": 0.725,
+      "step": 22820
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2017863438711953e-05,
+      "loss": 0.7369,
+      "step": 22830
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.2003173110823835e-05,
+      "loss": 0.7514,
+      "step": 22840
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.1988482782935717e-05,
+      "loss": 0.7482,
+      "step": 22850
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.1973792455047598e-05,
+      "loss": 0.7404,
+      "step": 22860
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.195910212715948e-05,
+      "loss": 0.781,
+      "step": 22870
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.194441179927136e-05,
+      "loss": 0.7844,
+      "step": 22880
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1929721471383243e-05,
+      "loss": 0.7613,
+      "step": 22890
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1915031143495124e-05,
+      "loss": 0.716,
+      "step": 22900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1900340815607006e-05,
+      "loss": 0.7478,
+      "step": 22910
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1885650487718887e-05,
+      "loss": 0.7587,
+      "step": 22920
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.187096015983077e-05,
+      "loss": 0.7206,
+      "step": 22930
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.185626983194265e-05,
+      "loss": 0.7219,
+      "step": 22940
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.184157950405453e-05,
+      "loss": 0.7818,
+      "step": 22950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1826889176166413e-05,
+      "loss": 0.7159,
+      "step": 22960
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1812198848278295e-05,
+      "loss": 0.642,
+      "step": 22970
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1797508520390176e-05,
+      "loss": 0.7979,
+      "step": 22980
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.1782818192502058e-05,
+      "loss": 0.7587,
+      "step": 22990
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.176812786461394e-05,
+      "loss": 0.7284,
+      "step": 23000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.175343753672582e-05,
+      "loss": 0.7886,
+      "step": 23010
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.1738747208837702e-05,
+      "loss": 0.7265,
+      "step": 23020
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.1724056880949584e-05,
+      "loss": 0.7995,
+      "step": 23030
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.1709366553061465e-05,
+      "loss": 0.8178,
+      "step": 23040
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.1694676225173347e-05,
+      "loss": 0.7415,
+      "step": 23050
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.1679985897285228e-05,
+      "loss": 0.7467,
+      "step": 23060
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.166529556939711e-05,
+      "loss": 0.734,
+      "step": 23070
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.165060524150899e-05,
+      "loss": 0.7585,
+      "step": 23080
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.1635914913620873e-05,
+      "loss": 0.741,
+      "step": 23090
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.1621224585732754e-05,
+      "loss": 0.694,
+      "step": 23100
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.1606534257844636e-05,
+      "loss": 0.717,
+      "step": 23110
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.1591843929956517e-05,
+      "loss": 0.7422,
+      "step": 23120
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.15771536020684e-05,
+      "loss": 0.7404,
+      "step": 23130
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.156246327418028e-05,
+      "loss": 0.7338,
+      "step": 23140
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.1547772946292165e-05,
+      "loss": 0.7783,
+      "step": 23150
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.1533082618404043e-05,
+      "loss": 0.7808,
+      "step": 23160
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.1518392290515925e-05,
+      "loss": 0.7718,
+      "step": 23170
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.1503701962627806e-05,
+      "loss": 0.7592,
+      "step": 23180
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.1489011634739688e-05,
+      "loss": 0.7402,
+      "step": 23190
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.147432130685157e-05,
+      "loss": 0.7244,
+      "step": 23200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.145963097896345e-05,
+      "loss": 0.7353,
+      "step": 23210
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.1444940651075332e-05,
+      "loss": 0.7237,
+      "step": 23220
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.1430250323187214e-05,
+      "loss": 0.751,
+      "step": 23230
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.1415559995299095e-05,
+      "loss": 0.714,
+      "step": 23240
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.1400869667410977e-05,
+      "loss": 0.775,
+      "step": 23250
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.138617933952286e-05,
+      "loss": 0.7381,
+      "step": 23260
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.137148901163474e-05,
+      "loss": 0.7317,
+      "step": 23270
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.1356798683746625e-05,
+      "loss": 0.7253,
+      "step": 23280
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.1342108355858503e-05,
+      "loss": 0.754,
+      "step": 23290
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.1327418027970388e-05,
+      "loss": 0.7491,
+      "step": 23300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.1312727700082266e-05,
+      "loss": 0.7975,
+      "step": 23310
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.129803737219415e-05,
+      "loss": 0.7298,
+      "step": 23320
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.128334704430603e-05,
+      "loss": 0.7368,
+      "step": 23330
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.126865671641791e-05,
+      "loss": 0.7765,
+      "step": 23340
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.1253966388529792e-05,
+      "loss": 0.7702,
+      "step": 23350
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.1239276060641673e-05,
+      "loss": 0.6927,
+      "step": 23360
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.1224585732753558e-05,
+      "loss": 0.7587,
+      "step": 23370
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.1209895404865436e-05,
+      "loss": 0.7954,
+      "step": 23380
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.119520507697732e-05,
+      "loss": 0.7976,
+      "step": 23390
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.11805147490892e-05,
+      "loss": 0.8043,
+      "step": 23400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1165824421201084e-05,
+      "loss": 0.7391,
+      "step": 23410
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1151134093312962e-05,
+      "loss": 0.7123,
+      "step": 23420
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1136443765424847e-05,
+      "loss": 0.7247,
+      "step": 23430
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1121753437536725e-05,
+      "loss": 0.7241,
+      "step": 23440
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.110706310964861e-05,
+      "loss": 0.7536,
+      "step": 23450
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.109237278176049e-05,
+      "loss": 0.7464,
+      "step": 23460
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1077682453872373e-05,
+      "loss": 0.8023,
+      "step": 23470
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.106299212598425e-05,
+      "loss": 0.7241,
+      "step": 23480
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1048301798096133e-05,
+      "loss": 0.7239,
+      "step": 23490
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1033611470208018e-05,
+      "loss": 0.7555,
+      "step": 23500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1018921142319896e-05,
+      "loss": 0.7315,
+      "step": 23510
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.100423081443178e-05,
+      "loss": 0.7728,
+      "step": 23520
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.098954048654366e-05,
+      "loss": 0.6956,
+      "step": 23530
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0974850158655544e-05,
+      "loss": 0.7548,
+      "step": 23540
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0960159830767422e-05,
+      "loss": 0.7445,
+      "step": 23550
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0945469502879307e-05,
+      "loss": 0.7388,
+      "step": 23560
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0930779174991185e-05,
+      "loss": 0.7378,
+      "step": 23570
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.091608884710307e-05,
+      "loss": 0.7234,
+      "step": 23580
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0901398519214948e-05,
+      "loss": 0.755,
+      "step": 23590
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0886708191326833e-05,
+      "loss": 0.7901,
+      "step": 23600
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0872017863438714e-05,
+      "loss": 0.7384,
+      "step": 23610
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0857327535550596e-05,
+      "loss": 0.7426,
+      "step": 23620
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0842637207662477e-05,
+      "loss": 0.7622,
+      "step": 23630
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.082794687977436e-05,
+      "loss": 0.7197,
+      "step": 23640
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.081325655188624e-05,
+      "loss": 0.7517,
+      "step": 23650
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.079856622399812e-05,
+      "loss": 0.6972,
+      "step": 23660
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0783875896110003e-05,
+      "loss": 0.7724,
+      "step": 23670
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.076918556822188e-05,
+      "loss": 0.6833,
+      "step": 23680
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0754495240333766e-05,
+      "loss": 0.761,
+      "step": 23690
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0739804912445644e-05,
+      "loss": 0.7525,
+      "step": 23700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.072511458455753e-05,
+      "loss": 0.7531,
+      "step": 23710
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.071042425666941e-05,
+      "loss": 0.756,
+      "step": 23720
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0695733928781292e-05,
+      "loss": 0.7619,
+      "step": 23730
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0681043600893174e-05,
+      "loss": 0.7509,
+      "step": 23740
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0666353273005055e-05,
+      "loss": 0.7011,
+      "step": 23750
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0651662945116937e-05,
+      "loss": 0.7592,
+      "step": 23760
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.063697261722882e-05,
+      "loss": 0.7454,
+      "step": 23770
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.06222822893407e-05,
+      "loss": 0.7716,
+      "step": 23780
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.060759196145258e-05,
+      "loss": 0.7074,
+      "step": 23790
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.0592901633564463e-05,
+      "loss": 0.7232,
+      "step": 23800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.057821130567634e-05,
+      "loss": 0.7793,
+      "step": 23810
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.0563520977788226e-05,
+      "loss": 0.7668,
+      "step": 23820
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.0548830649900104e-05,
+      "loss": 0.7506,
+      "step": 23830
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.053414032201199e-05,
+      "loss": 0.7564,
+      "step": 23840
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.051944999412387e-05,
+      "loss": 0.7538,
+      "step": 23850
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.0504759666235752e-05,
+      "loss": 0.6667,
+      "step": 23860
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.0490069338347633e-05,
+      "loss": 0.7474,
+      "step": 23870
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.0475379010459515e-05,
+      "loss": 0.837,
+      "step": 23880
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0460688682571396e-05,
+      "loss": 0.7505,
+      "step": 23890
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0445998354683278e-05,
+      "loss": 0.7782,
+      "step": 23900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.043130802679516e-05,
+      "loss": 0.7627,
+      "step": 23910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.041661769890704e-05,
+      "loss": 0.808,
+      "step": 23920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0401927371018922e-05,
+      "loss": 0.7724,
+      "step": 23930
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0387237043130804e-05,
+      "loss": 0.7256,
+      "step": 23940
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0372546715242685e-05,
+      "loss": 0.7131,
+      "step": 23950
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0357856387354567e-05,
+      "loss": 0.761,
+      "step": 23960
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.034316605946645e-05,
+      "loss": 0.7492,
+      "step": 23970
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.032847573157833e-05,
+      "loss": 0.7954,
+      "step": 23980
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.031378540369021e-05,
+      "loss": 0.7113,
+      "step": 23990
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0299095075802093e-05,
+      "loss": 0.7715,
+      "step": 24000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0284404747913974e-05,
+      "loss": 0.7499,
+      "step": 24010
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.0269714420025856e-05,
+      "loss": 0.7356,
+      "step": 24020
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.0255024092137737e-05,
+      "loss": 0.6597,
+      "step": 24030
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.024033376424962e-05,
+      "loss": 0.6806,
+      "step": 24040
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.02256434363615e-05,
+      "loss": 0.7257,
+      "step": 24050
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.0210953108473382e-05,
+      "loss": 0.7742,
+      "step": 24060
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.0196262780585264e-05,
+      "loss": 0.7631,
+      "step": 24070
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.0181572452697145e-05,
+      "loss": 0.7722,
+      "step": 24080
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.0166882124809027e-05,
+      "loss": 0.7297,
+      "step": 24090
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.0152191796920908e-05,
+      "loss": 0.7432,
+      "step": 24100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.013750146903279e-05,
+      "loss": 0.7492,
+      "step": 24110
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.012281114114467e-05,
+      "loss": 0.7254,
+      "step": 24120
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.0108120813256553e-05,
+      "loss": 0.7922,
+      "step": 24130
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.0093430485368434e-05,
+      "loss": 0.7433,
+      "step": 24140
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.0078740157480316e-05,
+      "loss": 0.7108,
+      "step": 24150
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.0064049829592197e-05,
+      "loss": 0.7192,
+      "step": 24160
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.004935950170408e-05,
+      "loss": 0.7975,
+      "step": 24170
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.003466917381596e-05,
+      "loss": 0.7285,
+      "step": 24180
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.001997884592784e-05,
+      "loss": 0.7278,
+      "step": 24190
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.0005288518039726e-05,
+      "loss": 0.7113,
+      "step": 24200
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9990598190151605e-05,
+      "loss": 0.6924,
+      "step": 24210
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9975907862263486e-05,
+      "loss": 0.7477,
+      "step": 24220
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9961217534375368e-05,
+      "loss": 0.6984,
+      "step": 24230
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.994652720648725e-05,
+      "loss": 0.7312,
+      "step": 24240
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.993183687859913e-05,
+      "loss": 0.7211,
+      "step": 24250
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9917146550711012e-05,
+      "loss": 0.749,
+      "step": 24260
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9902456222822894e-05,
+      "loss": 0.7339,
+      "step": 24270
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9887765894934775e-05,
+      "loss": 0.7177,
+      "step": 24280
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9873075567046657e-05,
+      "loss": 0.7569,
+      "step": 24290
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9858385239158538e-05,
+      "loss": 0.7442,
+      "step": 24300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9843694911270423e-05,
+      "loss": 0.7189,
+      "step": 24310
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.98290045833823e-05,
+      "loss": 0.7984,
+      "step": 24320
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9814314255494186e-05,
+      "loss": 0.7517,
+      "step": 24330
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9799623927606064e-05,
+      "loss": 0.7581,
+      "step": 24340
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.978493359971795e-05,
+      "loss": 0.7527,
+      "step": 24350
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9770243271829827e-05,
+      "loss": 0.795,
+      "step": 24360
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.975555294394171e-05,
+      "loss": 0.6981,
+      "step": 24370
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.974086261605359e-05,
+      "loss": 0.7304,
+      "step": 24380
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.972617228816547e-05,
+      "loss": 0.7429,
+      "step": 24390
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9711481960277353e-05,
+      "loss": 0.7742,
+      "step": 24400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9696791632389235e-05,
+      "loss": 0.7648,
+      "step": 24410
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.968210130450112e-05,
+      "loss": 0.7597,
+      "step": 24420
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9667410976612998e-05,
+      "loss": 0.7414,
+      "step": 24430
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9652720648724883e-05,
+      "loss": 0.7717,
+      "step": 24440
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.963803032083676e-05,
+      "loss": 0.736,
+      "step": 24450
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9623339992948646e-05,
+      "loss": 0.7485,
+      "step": 24460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9608649665060524e-05,
+      "loss": 0.7086,
+      "step": 24470
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.959395933717241e-05,
+      "loss": 0.7554,
+      "step": 24480
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9579269009284287e-05,
+      "loss": 0.7985,
+      "step": 24490
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.956457868139617e-05,
+      "loss": 0.7521,
+      "step": 24500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.954988835350805e-05,
+      "loss": 0.7326,
+      "step": 24510
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9535198025619935e-05,
+      "loss": 0.7698,
+      "step": 24520
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9520507697731813e-05,
+      "loss": 0.7959,
+      "step": 24530
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9505817369843694e-05,
+      "loss": 0.6924,
+      "step": 24540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.949112704195558e-05,
+      "loss": 0.7068,
+      "step": 24550
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9476436714067457e-05,
+      "loss": 0.7808,
+      "step": 24560
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9461746386179342e-05,
+      "loss": 0.722,
+      "step": 24570
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.944705605829122e-05,
+      "loss": 0.7731,
+      "step": 24580
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9432365730403105e-05,
+      "loss": 0.7289,
+      "step": 24590
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9417675402514983e-05,
+      "loss": 0.6736,
+      "step": 24600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9402985074626868e-05,
+      "loss": 0.7395,
+      "step": 24610
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9388294746738746e-05,
+      "loss": 0.7461,
+      "step": 24620
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.937360441885063e-05,
+      "loss": 0.702,
+      "step": 24630
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.935891409096251e-05,
+      "loss": 0.7538,
+      "step": 24640
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9344223763074394e-05,
+      "loss": 0.7514,
+      "step": 24650
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9329533435186276e-05,
+      "loss": 0.7606,
+      "step": 24660
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9314843107298157e-05,
+      "loss": 0.7736,
+      "step": 24670
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.930015277941004e-05,
+      "loss": 0.7628,
+      "step": 24680
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9285462451521917e-05,
+      "loss": 0.7286,
+      "step": 24690
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9270772123633802e-05,
+      "loss": 0.7117,
+      "step": 24700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.925608179574568e-05,
+      "loss": 0.7567,
+      "step": 24710
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9241391467857565e-05,
+      "loss": 0.7558,
+      "step": 24720
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9226701139969443e-05,
+      "loss": 0.7552,
+      "step": 24730
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9212010812081328e-05,
+      "loss": 0.8204,
+      "step": 24740
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9197320484193206e-05,
+      "loss": 0.7388,
+      "step": 24750
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.918263015630509e-05,
+      "loss": 0.7339,
+      "step": 24760
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9167939828416972e-05,
+      "loss": 0.7448,
+      "step": 24770
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9153249500528854e-05,
+      "loss": 0.7077,
+      "step": 24780
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9138559172640735e-05,
+      "loss": 0.7525,
+      "step": 24790
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9123868844752617e-05,
+      "loss": 0.7457,
+      "step": 24800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9109178516864498e-05,
+      "loss": 0.7877,
+      "step": 24810
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.909448818897638e-05,
+      "loss": 0.7263,
+      "step": 24820
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.907979786108826e-05,
+      "loss": 0.772,
+      "step": 24830
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9065107533200143e-05,
+      "loss": 0.7468,
+      "step": 24840
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9050417205312024e-05,
+      "loss": 0.7771,
+      "step": 24850
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9035726877423902e-05,
+      "loss": 0.7711,
+      "step": 24860
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9021036549535787e-05,
+      "loss": 0.692,
+      "step": 24870
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9006346221647665e-05,
+      "loss": 0.7428,
+      "step": 24880
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.899165589375955e-05,
+      "loss": 0.7184,
+      "step": 24890
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8976965565871432e-05,
+      "loss": 0.6927,
+      "step": 24900
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8962275237983313e-05,
+      "loss": 0.7689,
+      "step": 24910
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8947584910095195e-05,
+      "loss": 0.6945,
+      "step": 24920
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8932894582207076e-05,
+      "loss": 0.7541,
+      "step": 24930
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8918204254318958e-05,
+      "loss": 0.768,
+      "step": 24940
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.890351392643084e-05,
+      "loss": 0.7086,
+      "step": 24950
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.888882359854272e-05,
+      "loss": 0.736,
+      "step": 24960
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8874133270654602e-05,
+      "loss": 0.7154,
+      "step": 24970
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8859442942766484e-05,
+      "loss": 0.7834,
+      "step": 24980
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8844752614878365e-05,
+      "loss": 0.742,
+      "step": 24990
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8830062286990247e-05,
+      "loss": 0.7802,
+      "step": 25000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.881537195910213e-05,
+      "loss": 0.8041,
+      "step": 25010
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.880068163121401e-05,
+      "loss": 0.7574,
+      "step": 25020
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.878599130332589e-05,
+      "loss": 0.7585,
+      "step": 25030
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8771300975437773e-05,
+      "loss": 0.7909,
+      "step": 25040
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8756610647549654e-05,
+      "loss": 0.7227,
+      "step": 25050
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8741920319661536e-05,
+      "loss": 0.7632,
+      "step": 25060
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8727229991773417e-05,
+      "loss": 0.7909,
+      "step": 25070
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.87125396638853e-05,
+      "loss": 0.7645,
+      "step": 25080
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.869784933599718e-05,
+      "loss": 0.7515,
+      "step": 25090
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8683159008109062e-05,
+      "loss": 0.6884,
+      "step": 25100
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8668468680220943e-05,
+      "loss": 0.7236,
+      "step": 25110
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8653778352332825e-05,
+      "loss": 0.7562,
+      "step": 25120
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8639088024444706e-05,
+      "loss": 0.7531,
+      "step": 25130
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8624397696556588e-05,
+      "loss": 0.7865,
+      "step": 25140
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.860970736866847e-05,
+      "loss": 0.7078,
+      "step": 25150
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.859501704078035e-05,
+      "loss": 0.7941,
+      "step": 25160
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8580326712892232e-05,
+      "loss": 0.7385,
+      "step": 25170
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8565636385004114e-05,
+      "loss": 0.7125,
+      "step": 25180
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8550946057115995e-05,
+      "loss": 0.7301,
+      "step": 25190
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8536255729227877e-05,
+      "loss": 0.7622,
+      "step": 25200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.852156540133976e-05,
+      "loss": 0.7312,
+      "step": 25210
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8027232806285409,
+      "eval_loss": 0.7587032318115234,
+      "eval_runtime": 2003.038,
+      "eval_samples_per_second": 402.799,
+      "eval_steps_per_second": 6.294,
+      "step": 25212
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.850687507345164e-05,
+      "loss": 0.7423,
+      "step": 25220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.849218474556352e-05,
+      "loss": 0.6582,
+      "step": 25230
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8477494417675403e-05,
+      "loss": 0.6851,
+      "step": 25240
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8462804089787284e-05,
+      "loss": 0.6685,
+      "step": 25250
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8448113761899166e-05,
+      "loss": 0.6851,
+      "step": 25260
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8433423434011048e-05,
+      "loss": 0.6544,
+      "step": 25270
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.841873310612293e-05,
+      "loss": 0.6725,
+      "step": 25280
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.840404277823481e-05,
+      "loss": 0.6471,
+      "step": 25290
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8389352450346692e-05,
+      "loss": 0.6442,
+      "step": 25300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8374662122458574e-05,
+      "loss": 0.6527,
+      "step": 25310
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8359971794570455e-05,
+      "loss": 0.7022,
+      "step": 25320
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8345281466682337e-05,
+      "loss": 0.6761,
+      "step": 25330
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8330591138794218e-05,
+      "loss": 0.708,
+      "step": 25340
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.83159008109061e-05,
+      "loss": 0.6753,
+      "step": 25350
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8301210483017984e-05,
+      "loss": 0.6397,
+      "step": 25360
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8286520155129863e-05,
+      "loss": 0.6234,
+      "step": 25370
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8271829827241747e-05,
+      "loss": 0.7162,
+      "step": 25380
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8257139499353626e-05,
+      "loss": 0.6155,
+      "step": 25390
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.824244917146551e-05,
+      "loss": 0.6602,
+      "step": 25400
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.822775884357739e-05,
+      "loss": 0.6559,
+      "step": 25410
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.821306851568927e-05,
+      "loss": 0.6651,
+      "step": 25420
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.819837818780115e-05,
+      "loss": 0.6255,
+      "step": 25430
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8183687859913033e-05,
+      "loss": 0.6328,
+      "step": 25440
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8168997532024915e-05,
+      "loss": 0.6655,
+      "step": 25450
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8154307204136796e-05,
+      "loss": 0.6264,
+      "step": 25460
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.813961687624868e-05,
+      "loss": 0.66,
+      "step": 25470
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.812492654836056e-05,
+      "loss": 0.6738,
+      "step": 25480
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8110236220472444e-05,
+      "loss": 0.6672,
+      "step": 25490
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8095545892584322e-05,
+      "loss": 0.6506,
+      "step": 25500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8080855564696207e-05,
+      "loss": 0.6093,
+      "step": 25510
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8066165236808085e-05,
+      "loss": 0.7344,
+      "step": 25520
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.805147490891997e-05,
+      "loss": 0.6605,
+      "step": 25530
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8036784581031848e-05,
+      "loss": 0.668,
+      "step": 25540
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8022094253143733e-05,
+      "loss": 0.685,
+      "step": 25550
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.800740392525561e-05,
+      "loss": 0.6414,
+      "step": 25560
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.7992713597367493e-05,
+      "loss": 0.6526,
+      "step": 25570
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.7978023269479374e-05,
+      "loss": 0.6356,
+      "step": 25580
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.7963332941591256e-05,
+      "loss": 0.7199,
+      "step": 25590
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.794864261370314e-05,
+      "loss": 0.6835,
+      "step": 25600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.793395228581502e-05,
+      "loss": 0.6391,
+      "step": 25610
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.7919261957926904e-05,
+      "loss": 0.6395,
+      "step": 25620
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.790457163003878e-05,
+      "loss": 0.6952,
+      "step": 25630
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.7889881302150667e-05,
+      "loss": 0.6792,
+      "step": 25640
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.7875190974262545e-05,
+      "loss": 0.6602,
+      "step": 25650
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.786050064637443e-05,
+      "loss": 0.6952,
+      "step": 25660
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.7845810318486308e-05,
+      "loss": 0.6722,
+      "step": 25670
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.7831119990598193e-05,
+      "loss": 0.614,
+      "step": 25680
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.781642966271007e-05,
+      "loss": 0.6653,
+      "step": 25690
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.7801739334821956e-05,
+      "loss": 0.6824,
+      "step": 25700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.7787049006933837e-05,
+      "loss": 0.6867,
+      "step": 25710
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.777235867904572e-05,
+      "loss": 0.6584,
+      "step": 25720
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.77576683511576e-05,
+      "loss": 0.6633,
+      "step": 25730
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.7742978023269478e-05,
+      "loss": 0.6876,
+      "step": 25740
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.7728287695381363e-05,
+      "loss": 0.6274,
+      "step": 25750
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.771359736749324e-05,
+      "loss": 0.6488,
+      "step": 25760
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.7698907039605126e-05,
+      "loss": 0.6465,
+      "step": 25770
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.7684216711717004e-05,
+      "loss": 0.608,
+      "step": 25780
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.766952638382889e-05,
+      "loss": 0.6468,
+      "step": 25790
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.7654836055940767e-05,
+      "loss": 0.6986,
+      "step": 25800
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.7640145728052652e-05,
+      "loss": 0.718,
+      "step": 25810
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.7625455400164534e-05,
+      "loss": 0.6791,
+      "step": 25820
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.7610765072276415e-05,
+      "loss": 0.687,
+      "step": 25830
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.7596074744388297e-05,
+      "loss": 0.6707,
+      "step": 25840
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.7581384416500178e-05,
+      "loss": 0.6696,
+      "step": 25850
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.756669408861206e-05,
+      "loss": 0.6318,
+      "step": 25860
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.755200376072394e-05,
+      "loss": 0.677,
+      "step": 25870
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.7537313432835823e-05,
+      "loss": 0.6457,
+      "step": 25880
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.75226231049477e-05,
+      "loss": 0.6057,
+      "step": 25890
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.7507932777059586e-05,
+      "loss": 0.6446,
+      "step": 25900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.7493242449171464e-05,
+      "loss": 0.694,
+      "step": 25910
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.747855212128335e-05,
+      "loss": 0.6619,
+      "step": 25920
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.7463861793395227e-05,
+      "loss": 0.7286,
+      "step": 25930
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.7449171465507112e-05,
+      "loss": 0.6581,
+      "step": 25940
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.7434481137618993e-05,
+      "loss": 0.6883,
+      "step": 25950
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.7419790809730875e-05,
+      "loss": 0.7099,
+      "step": 25960
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.7405100481842756e-05,
+      "loss": 0.6769,
+      "step": 25970
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.7390410153954638e-05,
+      "loss": 0.6297,
+      "step": 25980
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.737571982606652e-05,
+      "loss": 0.6473,
+      "step": 25990
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.73610294981784e-05,
+      "loss": 0.6294,
+      "step": 26000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.7346339170290282e-05,
+      "loss": 0.6792,
+      "step": 26010
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.7331648842402164e-05,
+      "loss": 0.6338,
+      "step": 26020
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.7316958514514045e-05,
+      "loss": 0.6748,
+      "step": 26030
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.7302268186625923e-05,
+      "loss": 0.6837,
+      "step": 26040
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.7287577858737808e-05,
+      "loss": 0.6479,
+      "step": 26050
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.727288753084969e-05,
+      "loss": 0.6943,
+      "step": 26060
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.725819720296157e-05,
+      "loss": 0.6463,
+      "step": 26070
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.7243506875073453e-05,
+      "loss": 0.704,
+      "step": 26080
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.7228816547185334e-05,
+      "loss": 0.6564,
+      "step": 26090
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.7214126219297216e-05,
+      "loss": 0.6603,
+      "step": 26100
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.7199435891409097e-05,
+      "loss": 0.6174,
+      "step": 26110
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.718474556352098e-05,
+      "loss": 0.7269,
+      "step": 26120
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.717005523563286e-05,
+      "loss": 0.6505,
+      "step": 26130
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.7155364907744742e-05,
+      "loss": 0.6807,
+      "step": 26140
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.7140674579856623e-05,
+      "loss": 0.6543,
+      "step": 26150
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.7125984251968505e-05,
+      "loss": 0.6774,
+      "step": 26160
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.7111293924080386e-05,
+      "loss": 0.6533,
+      "step": 26170
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.7096603596192268e-05,
+      "loss": 0.6936,
+      "step": 26180
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.708191326830415e-05,
+      "loss": 0.6574,
+      "step": 26190
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.706722294041603e-05,
+      "loss": 0.6052,
+      "step": 26200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.7052532612527912e-05,
+      "loss": 0.7222,
+      "step": 26210
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.7037842284639794e-05,
+      "loss": 0.6158,
+      "step": 26220
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.7023151956751675e-05,
+      "loss": 0.6654,
+      "step": 26230
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.7008461628863557e-05,
+      "loss": 0.6047,
+      "step": 26240
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.699377130097544e-05,
+      "loss": 0.665,
+      "step": 26250
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.697908097308732e-05,
+      "loss": 0.6478,
+      "step": 26260
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.69643906451992e-05,
+      "loss": 0.6811,
+      "step": 26270
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.6949700317311083e-05,
+      "loss": 0.6586,
+      "step": 26280
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.6935009989422964e-05,
+      "loss": 0.6852,
+      "step": 26290
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.6920319661534846e-05,
+      "loss": 0.7225,
+      "step": 26300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.6905629333646727e-05,
+      "loss": 0.6681,
+      "step": 26310
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.689093900575861e-05,
+      "loss": 0.6568,
+      "step": 26320
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.687624867787049e-05,
+      "loss": 0.6801,
+      "step": 26330
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.6861558349982372e-05,
+      "loss": 0.6328,
+      "step": 26340
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.6846868022094253e-05,
+      "loss": 0.6246,
+      "step": 26350
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.6832177694206135e-05,
+      "loss": 0.6319,
+      "step": 26360
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.6817487366318016e-05,
+      "loss": 0.6434,
+      "step": 26370
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.6802797038429898e-05,
+      "loss": 0.6729,
+      "step": 26380
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.678810671054178e-05,
+      "loss": 0.6728,
+      "step": 26390
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.677341638265366e-05,
+      "loss": 0.6364,
+      "step": 26400
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.6758726054765546e-05,
+      "loss": 0.6378,
+      "step": 26410
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.6744035726877424e-05,
+      "loss": 0.6289,
+      "step": 26420
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.672934539898931e-05,
+      "loss": 0.654,
+      "step": 26430
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.6714655071101187e-05,
+      "loss": 0.6176,
+      "step": 26440
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.669996474321307e-05,
+      "loss": 0.6688,
+      "step": 26450
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.668527441532495e-05,
+      "loss": 0.7008,
+      "step": 26460
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.667058408743683e-05,
+      "loss": 0.6483,
+      "step": 26470
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.6655893759548713e-05,
+      "loss": 0.6786,
+      "step": 26480
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.6641203431660595e-05,
+      "loss": 0.7234,
+      "step": 26490
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.6626513103772476e-05,
+      "loss": 0.6537,
+      "step": 26500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.6611822775884358e-05,
+      "loss": 0.6545,
+      "step": 26510
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.6597132447996242e-05,
+      "loss": 0.6753,
+      "step": 26520
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.658244212010812e-05,
+      "loss": 0.6893,
+      "step": 26530
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.6567751792220005e-05,
+      "loss": 0.695,
+      "step": 26540
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.6553061464331884e-05,
+      "loss": 0.6706,
+      "step": 26550
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.653837113644377e-05,
+      "loss": 0.6983,
+      "step": 26560
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.6523680808555647e-05,
+      "loss": 0.6031,
+      "step": 26570
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.650899048066753e-05,
+      "loss": 0.6736,
+      "step": 26580
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.649430015277941e-05,
+      "loss": 0.6815,
+      "step": 26590
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.6479609824891294e-05,
+      "loss": 0.6559,
+      "step": 26600
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.6464919497003173e-05,
+      "loss": 0.7229,
+      "step": 26610
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.6450229169115054e-05,
+      "loss": 0.6846,
+      "step": 26620
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.6435538841226936e-05,
+      "loss": 0.6587,
+      "step": 26630
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.6420848513338817e-05,
+      "loss": 0.6684,
+      "step": 26640
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.6406158185450702e-05,
+      "loss": 0.6783,
+      "step": 26650
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.639146785756258e-05,
+      "loss": 0.646,
+      "step": 26660
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6376777529674465e-05,
+      "loss": 0.6384,
+      "step": 26670
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6362087201786343e-05,
+      "loss": 0.6565,
+      "step": 26680
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6347396873898228e-05,
+      "loss": 0.646,
+      "step": 26690
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6332706546010106e-05,
+      "loss": 0.6593,
+      "step": 26700
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.631801621812199e-05,
+      "loss": 0.6578,
+      "step": 26710
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.630332589023387e-05,
+      "loss": 0.6321,
+      "step": 26720
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6288635562345754e-05,
+      "loss": 0.6506,
+      "step": 26730
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6273945234457632e-05,
+      "loss": 0.6628,
+      "step": 26740
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6259254906569517e-05,
+      "loss": 0.6515,
+      "step": 26750
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.62445645786814e-05,
+      "loss": 0.6081,
+      "step": 26760
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6229874250793277e-05,
+      "loss": 0.6762,
+      "step": 26770
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.621518392290516e-05,
+      "loss": 0.7064,
+      "step": 26780
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.620049359501704e-05,
+      "loss": 0.6375,
+      "step": 26790
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6185803267128925e-05,
+      "loss": 0.6338,
+      "step": 26800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6171112939240803e-05,
+      "loss": 0.6385,
+      "step": 26810
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6156422611352688e-05,
+      "loss": 0.6648,
+      "step": 26820
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6141732283464566e-05,
+      "loss": 0.6331,
+      "step": 26830
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.612704195557645e-05,
+      "loss": 0.6712,
+      "step": 26840
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.611235162768833e-05,
+      "loss": 0.6267,
+      "step": 26850
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6097661299800214e-05,
+      "loss": 0.6605,
+      "step": 26860
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6082970971912095e-05,
+      "loss": 0.684,
+      "step": 26870
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6068280644023977e-05,
+      "loss": 0.6607,
+      "step": 26880
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6053590316135858e-05,
+      "loss": 0.6775,
+      "step": 26890
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.603889998824774e-05,
+      "loss": 0.6303,
+      "step": 26900
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.602420966035962e-05,
+      "loss": 0.6895,
+      "step": 26910
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.6009519332471503e-05,
+      "loss": 0.6539,
+      "step": 26920
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5994829004583384e-05,
+      "loss": 0.6482,
+      "step": 26930
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5980138676695262e-05,
+      "loss": 0.6433,
+      "step": 26940
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5965448348807147e-05,
+      "loss": 0.6057,
+      "step": 26950
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5950758020919025e-05,
+      "loss": 0.6243,
+      "step": 26960
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.593606769303091e-05,
+      "loss": 0.6668,
+      "step": 26970
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5921377365142788e-05,
+      "loss": 0.6863,
+      "step": 26980
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5906687037254673e-05,
+      "loss": 0.6584,
+      "step": 26990
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5891996709366555e-05,
+      "loss": 0.7188,
+      "step": 27000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5877306381478436e-05,
+      "loss": 0.6761,
+      "step": 27010
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5862616053590318e-05,
+      "loss": 0.685,
+      "step": 27020
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.58479257257022e-05,
+      "loss": 0.6768,
+      "step": 27030
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.583323539781408e-05,
+      "loss": 0.703,
+      "step": 27040
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5818545069925962e-05,
+      "loss": 0.728,
+      "step": 27050
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5803854742037844e-05,
+      "loss": 0.6748,
+      "step": 27060
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5789164414149725e-05,
+      "loss": 0.6247,
+      "step": 27070
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5774474086261607e-05,
+      "loss": 0.6449,
+      "step": 27080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5759783758373485e-05,
+      "loss": 0.6497,
+      "step": 27090
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.574509343048537e-05,
+      "loss": 0.6809,
+      "step": 27100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.573040310259725e-05,
+      "loss": 0.609,
+      "step": 27110
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5715712774709133e-05,
+      "loss": 0.6471,
+      "step": 27120
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5701022446821014e-05,
+      "loss": 0.6276,
+      "step": 27130
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5686332118932896e-05,
+      "loss": 0.6678,
+      "step": 27140
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5671641791044777e-05,
+      "loss": 0.6434,
+      "step": 27150
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.565695146315666e-05,
+      "loss": 0.6006,
+      "step": 27160
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.564226113526854e-05,
+      "loss": 0.6347,
+      "step": 27170
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5627570807380422e-05,
+      "loss": 0.7059,
+      "step": 27180
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5612880479492303e-05,
+      "loss": 0.735,
+      "step": 27190
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5598190151604185e-05,
+      "loss": 0.6631,
+      "step": 27200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5583499823716066e-05,
+      "loss": 0.6457,
+      "step": 27210
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5568809495827948e-05,
+      "loss": 0.6558,
+      "step": 27220
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.555411916793983e-05,
+      "loss": 0.6924,
+      "step": 27230
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.553942884005171e-05,
+      "loss": 0.6828,
+      "step": 27240
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5524738512163592e-05,
+      "loss": 0.6473,
+      "step": 27250
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5510048184275474e-05,
+      "loss": 0.6351,
+      "step": 27260
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5495357856387355e-05,
+      "loss": 0.6713,
+      "step": 27270
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5480667528499237e-05,
+      "loss": 0.7327,
+      "step": 27280
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5465977200611118e-05,
+      "loss": 0.745,
+      "step": 27290
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5451286872723e-05,
+      "loss": 0.6712,
+      "step": 27300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.543659654483488e-05,
+      "loss": 0.6925,
+      "step": 27310
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5421906216946763e-05,
+      "loss": 0.6846,
+      "step": 27320
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5407215889058644e-05,
+      "loss": 0.6426,
+      "step": 27330
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5392525561170526e-05,
+      "loss": 0.6718,
+      "step": 27340
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5377835233282407e-05,
+      "loss": 0.7128,
+      "step": 27350
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.536314490539429e-05,
+      "loss": 0.6751,
+      "step": 27360
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.534845457750617e-05,
+      "loss": 0.6214,
+      "step": 27370
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5333764249618052e-05,
+      "loss": 0.6754,
+      "step": 27380
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5319073921729933e-05,
+      "loss": 0.6446,
+      "step": 27390
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5304383593841815e-05,
+      "loss": 0.6276,
+      "step": 27400
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5289693265953696e-05,
+      "loss": 0.6621,
+      "step": 27410
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5275002938065578e-05,
+      "loss": 0.6167,
+      "step": 27420
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.526031261017746e-05,
+      "loss": 0.6714,
+      "step": 27430
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5245622282289341e-05,
+      "loss": 0.7169,
+      "step": 27440
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5230931954401222e-05,
+      "loss": 0.656,
+      "step": 27450
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5216241626513106e-05,
+      "loss": 0.6949,
+      "step": 27460
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5201551298624985e-05,
+      "loss": 0.7136,
+      "step": 27470
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5186860970736869e-05,
+      "loss": 0.6506,
+      "step": 27480
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5172170642848748e-05,
+      "loss": 0.6522,
+      "step": 27490
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5157480314960632e-05,
+      "loss": 0.6566,
+      "step": 27500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5142789987072511e-05,
+      "loss": 0.68,
+      "step": 27510
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5128099659184395e-05,
+      "loss": 0.642,
+      "step": 27520
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5113409331296274e-05,
+      "loss": 0.6977,
+      "step": 27530
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5098719003408158e-05,
+      "loss": 0.6076,
+      "step": 27540
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5084028675520037e-05,
+      "loss": 0.68,
+      "step": 27550
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.506933834763192e-05,
+      "loss": 0.6558,
+      "step": 27560
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5054648019743802e-05,
+      "loss": 0.6724,
+      "step": 27570
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5039957691855684e-05,
+      "loss": 0.7124,
+      "step": 27580
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5025267363967565e-05,
+      "loss": 0.6696,
+      "step": 27590
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5010577036079445e-05,
+      "loss": 0.6667,
+      "step": 27600
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4995886708191328e-05,
+      "loss": 0.6193,
+      "step": 27610
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4981196380303208e-05,
+      "loss": 0.6492,
+      "step": 27620
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4966506052415091e-05,
+      "loss": 0.6932,
+      "step": 27630
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4951815724526971e-05,
+      "loss": 0.664,
+      "step": 27640
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4937125396638854e-05,
+      "loss": 0.7153,
+      "step": 27650
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4922435068750734e-05,
+      "loss": 0.6961,
+      "step": 27660
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4907744740862617e-05,
+      "loss": 0.6198,
+      "step": 27670
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4893054412974497e-05,
+      "loss": 0.6267,
+      "step": 27680
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.487836408508638e-05,
+      "loss": 0.6625,
+      "step": 27690
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4863673757198263e-05,
+      "loss": 0.6515,
+      "step": 27700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4848983429310143e-05,
+      "loss": 0.6974,
+      "step": 27710
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4834293101422025e-05,
+      "loss": 0.6384,
+      "step": 27720
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4819602773533906e-05,
+      "loss": 0.6393,
+      "step": 27730
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4804912445645788e-05,
+      "loss": 0.6462,
+      "step": 27740
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4790222117757668e-05,
+      "loss": 0.6743,
+      "step": 27750
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.477553178986955e-05,
+      "loss": 0.6229,
+      "step": 27760
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.476084146198143e-05,
+      "loss": 0.5978,
+      "step": 27770
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4746151134093314e-05,
+      "loss": 0.6609,
+      "step": 27780
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4731460806205194e-05,
+      "loss": 0.605,
+      "step": 27790
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4716770478317077e-05,
+      "loss": 0.6702,
+      "step": 27800
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.470208015042896e-05,
+      "loss": 0.7103,
+      "step": 27810
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.468738982254084e-05,
+      "loss": 0.6681,
+      "step": 27820
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4672699494652723e-05,
+      "loss": 0.6461,
+      "step": 27830
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4658009166764603e-05,
+      "loss": 0.6516,
+      "step": 27840
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4643318838876486e-05,
+      "loss": 0.6393,
+      "step": 27850
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4628628510988366e-05,
+      "loss": 0.6495,
+      "step": 27860
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4613938183100247e-05,
+      "loss": 0.7269,
+      "step": 27870
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4599247855212129e-05,
+      "loss": 0.6738,
+      "step": 27880
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.458455752732401e-05,
+      "loss": 0.6129,
+      "step": 27890
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4569867199435892e-05,
+      "loss": 0.6889,
+      "step": 27900
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4555176871547773e-05,
+      "loss": 0.6696,
+      "step": 27910
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.4540486543659653e-05,
+      "loss": 0.6512,
+      "step": 27920
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4525796215771536e-05,
+      "loss": 0.6505,
+      "step": 27930
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.451110588788342e-05,
+      "loss": 0.6414,
+      "step": 27940
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.44964155599953e-05,
+      "loss": 0.6526,
+      "step": 27950
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4481725232107183e-05,
+      "loss": 0.6472,
+      "step": 27960
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4467034904219062e-05,
+      "loss": 0.693,
+      "step": 27970
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4452344576330946e-05,
+      "loss": 0.6644,
+      "step": 27980
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4437654248442825e-05,
+      "loss": 0.682,
+      "step": 27990
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4422963920554709e-05,
+      "loss": 0.6452,
+      "step": 28000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4408273592666588e-05,
+      "loss": 0.6493,
+      "step": 28010
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4393583264778472e-05,
+      "loss": 0.656,
+      "step": 28020
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4378892936890351e-05,
+      "loss": 0.6498,
+      "step": 28030
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.4364202609002233e-05,
+      "loss": 0.6366,
+      "step": 28040
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4349512281114116e-05,
+      "loss": 0.6769,
+      "step": 28050
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4334821953225996e-05,
+      "loss": 0.6504,
+      "step": 28060
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4320131625337879e-05,
+      "loss": 0.6939,
+      "step": 28070
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4305441297449759e-05,
+      "loss": 0.6226,
+      "step": 28080
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4290750969561642e-05,
+      "loss": 0.6322,
+      "step": 28090
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4276060641673522e-05,
+      "loss": 0.6573,
+      "step": 28100
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4261370313785405e-05,
+      "loss": 0.6792,
+      "step": 28110
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4246679985897285e-05,
+      "loss": 0.6528,
+      "step": 28120
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4231989658009168e-05,
+      "loss": 0.6852,
+      "step": 28130
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4217299330121048e-05,
+      "loss": 0.6342,
+      "step": 28140
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4202609002232931e-05,
+      "loss": 0.6342,
+      "step": 28150
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4187918674344813e-05,
+      "loss": 0.6816,
+      "step": 28160
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.4173228346456694e-05,
+      "loss": 0.6299,
+      "step": 28170
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4158538018568576e-05,
+      "loss": 0.6203,
+      "step": 28180
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4143847690680455e-05,
+      "loss": 0.666,
+      "step": 28190
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4129157362792339e-05,
+      "loss": 0.6934,
+      "step": 28200
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4114467034904218e-05,
+      "loss": 0.6592,
+      "step": 28210
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4099776707016102e-05,
+      "loss": 0.6541,
+      "step": 28220
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4085086379127981e-05,
+      "loss": 0.6699,
+      "step": 28230
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4070396051239865e-05,
+      "loss": 0.669,
+      "step": 28240
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4055705723351744e-05,
+      "loss": 0.6945,
+      "step": 28250
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4041015395463628e-05,
+      "loss": 0.6352,
+      "step": 28260
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.4026325067575507e-05,
+      "loss": 0.6508,
+      "step": 28270
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.401163473968739e-05,
+      "loss": 0.6823,
+      "step": 28280
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.3996944411799274e-05,
+      "loss": 0.6199,
+      "step": 28290
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.3982254083911154e-05,
+      "loss": 0.5945,
+      "step": 28300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.3967563756023035e-05,
+      "loss": 0.6531,
+      "step": 28310
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.3952873428134917e-05,
+      "loss": 0.6366,
+      "step": 28320
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.3938183100246798e-05,
+      "loss": 0.6737,
+      "step": 28330
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.392349277235868e-05,
+      "loss": 0.7108,
+      "step": 28340
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.3908802444470561e-05,
+      "loss": 0.6456,
+      "step": 28350
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.3894112116582441e-05,
+      "loss": 0.638,
+      "step": 28360
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.3879421788694324e-05,
+      "loss": 0.6233,
+      "step": 28370
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.3864731460806204e-05,
+      "loss": 0.7249,
+      "step": 28380
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.3850041132918087e-05,
+      "loss": 0.7226,
+      "step": 28390
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.383535080502997e-05,
+      "loss": 0.6815,
+      "step": 28400
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.382066047714185e-05,
+      "loss": 0.6308,
+      "step": 28410
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.3805970149253733e-05,
+      "loss": 0.6955,
+      "step": 28420
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.3791279821365613e-05,
+      "loss": 0.6822,
+      "step": 28430
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.3776589493477496e-05,
+      "loss": 0.64,
+      "step": 28440
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.3761899165589376e-05,
+      "loss": 0.6801,
+      "step": 28450
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.374720883770126e-05,
+      "loss": 0.6936,
+      "step": 28460
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.373251850981314e-05,
+      "loss": 0.6752,
+      "step": 28470
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.371782818192502e-05,
+      "loss": 0.674,
+      "step": 28480
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.3703137854036902e-05,
+      "loss": 0.6349,
+      "step": 28490
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.3688447526148784e-05,
+      "loss": 0.6346,
+      "step": 28500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.3673757198260667e-05,
+      "loss": 0.6589,
+      "step": 28510
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.3659066870372547e-05,
+      "loss": 0.664,
+      "step": 28520
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.364437654248443e-05,
+      "loss": 0.6241,
+      "step": 28530
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.362968621459631e-05,
+      "loss": 0.6342,
+      "step": 28540
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.3614995886708193e-05,
+      "loss": 0.6282,
+      "step": 28550
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.3600305558820073e-05,
+      "loss": 0.6251,
+      "step": 28560
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.3585615230931956e-05,
+      "loss": 0.6143,
+      "step": 28570
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.3570924903043836e-05,
+      "loss": 0.6443,
+      "step": 28580
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.3556234575155719e-05,
+      "loss": 0.6858,
+      "step": 28590
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.3541544247267599e-05,
+      "loss": 0.6945,
+      "step": 28600
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.3526853919379482e-05,
+      "loss": 0.6641,
+      "step": 28610
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.3512163591491362e-05,
+      "loss": 0.6481,
+      "step": 28620
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.3497473263603243e-05,
+      "loss": 0.6184,
+      "step": 28630
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.3482782935715127e-05,
+      "loss": 0.7019,
+      "step": 28640
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.3468092607827006e-05,
+      "loss": 0.6936,
+      "step": 28650
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.345340227993889e-05,
+      "loss": 0.6413,
+      "step": 28660
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.343871195205077e-05,
+      "loss": 0.6684,
+      "step": 28670
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.3424021624162653e-05,
+      "loss": 0.6662,
+      "step": 28680
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.3409331296274532e-05,
+      "loss": 0.6945,
+      "step": 28690
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.3394640968386416e-05,
+      "loss": 0.6559,
+      "step": 28700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.3379950640498295e-05,
+      "loss": 0.6802,
+      "step": 28710
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.3365260312610179e-05,
+      "loss": 0.6736,
+      "step": 28720
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.3350569984722058e-05,
+      "loss": 0.6636,
+      "step": 28730
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.3335879656833942e-05,
+      "loss": 0.691,
+      "step": 28740
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.3321189328945823e-05,
+      "loss": 0.6559,
+      "step": 28750
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.3306499001057705e-05,
+      "loss": 0.6169,
+      "step": 28760
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.3291808673169586e-05,
+      "loss": 0.7204,
+      "step": 28770
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.3277118345281468e-05,
+      "loss": 0.6821,
+      "step": 28780
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.3262428017393349e-05,
+      "loss": 0.6551,
+      "step": 28790
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.3247737689505229e-05,
+      "loss": 0.6559,
+      "step": 28800
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.3233047361617112e-05,
+      "loss": 0.6881,
+      "step": 28810
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.3218357033728992e-05,
+      "loss": 0.6381,
+      "step": 28820
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.3203666705840875e-05,
+      "loss": 0.6786,
+      "step": 28830
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.3188976377952755e-05,
+      "loss": 0.7047,
+      "step": 28840
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.3174286050064638e-05,
+      "loss": 0.6732,
+      "step": 28850
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.3159595722176521e-05,
+      "loss": 0.6597,
+      "step": 28860
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.3144905394288401e-05,
+      "loss": 0.6075,
+      "step": 28870
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.3130215066400284e-05,
+      "loss": 0.6802,
+      "step": 28880
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.3115524738512164e-05,
+      "loss": 0.7286,
+      "step": 28890
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.3100834410624047e-05,
+      "loss": 0.6503,
+      "step": 28900
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.3086144082735927e-05,
+      "loss": 0.6519,
+      "step": 28910
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.3071453754847809e-05,
+      "loss": 0.6415,
+      "step": 28920
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.305676342695969e-05,
+      "loss": 0.7125,
+      "step": 28930
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.3042073099071572e-05,
+      "loss": 0.6355,
+      "step": 28940
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.3027382771183452e-05,
+      "loss": 0.578,
+      "step": 28950
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.3012692443295335e-05,
+      "loss": 0.6594,
+      "step": 28960
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.2998002115407215e-05,
+      "loss": 0.6992,
+      "step": 28970
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.2983311787519098e-05,
+      "loss": 0.6425,
+      "step": 28980
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.2968621459630981e-05,
+      "loss": 0.6611,
+      "step": 28990
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.295393113174286e-05,
+      "loss": 0.6147,
+      "step": 29000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.2939240803854744e-05,
+      "loss": 0.6679,
+      "step": 29010
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.2924550475966624e-05,
+      "loss": 0.6438,
+      "step": 29020
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.2909860148078507e-05,
+      "loss": 0.6587,
+      "step": 29030
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.2895169820190387e-05,
+      "loss": 0.6442,
+      "step": 29040
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.288047949230227e-05,
+      "loss": 0.6097,
+      "step": 29050
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.286578916441415e-05,
+      "loss": 0.6801,
+      "step": 29060
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.2851098836526031e-05,
+      "loss": 0.6534,
+      "step": 29070
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.2836408508637913e-05,
+      "loss": 0.6618,
+      "step": 29080
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.2821718180749794e-05,
+      "loss": 0.6163,
+      "step": 29090
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.2807027852861677e-05,
+      "loss": 0.6453,
+      "step": 29100
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.2792337524973557e-05,
+      "loss": 0.683,
+      "step": 29110
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.277764719708544e-05,
+      "loss": 0.6032,
+      "step": 29120
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.276295686919732e-05,
+      "loss": 0.6659,
+      "step": 29130
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.2748266541309204e-05,
+      "loss": 0.5959,
+      "step": 29140
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.2733576213421083e-05,
+      "loss": 0.667,
+      "step": 29150
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.2718885885532967e-05,
+      "loss": 0.6153,
+      "step": 29160
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.2704195557644846e-05,
+      "loss": 0.6296,
+      "step": 29170
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.268950522975673e-05,
+      "loss": 0.6169,
+      "step": 29180
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.267481490186861e-05,
+      "loss": 0.6608,
+      "step": 29190
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.2660124573980493e-05,
+      "loss": 0.6282,
+      "step": 29200
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.2645434246092374e-05,
+      "loss": 0.662,
+      "step": 29210
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.2630743918204256e-05,
+      "loss": 0.6708,
+      "step": 29220
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.2616053590316137e-05,
+      "loss": 0.6582,
+      "step": 29230
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.2601363262428017e-05,
+      "loss": 0.6406,
+      "step": 29240
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.25866729345399e-05,
+      "loss": 0.6802,
+      "step": 29250
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.257198260665178e-05,
+      "loss": 0.6329,
+      "step": 29260
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.2557292278763663e-05,
+      "loss": 0.7033,
+      "step": 29270
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.2542601950875543e-05,
+      "loss": 0.6823,
+      "step": 29280
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.2527911622987426e-05,
+      "loss": 0.6656,
+      "step": 29290
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.2513221295099306e-05,
+      "loss": 0.6632,
+      "step": 29300
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.2498530967211189e-05,
+      "loss": 0.6532,
+      "step": 29310
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.248384063932307e-05,
+      "loss": 0.6588,
+      "step": 29320
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.2469150311434952e-05,
+      "loss": 0.5994,
+      "step": 29330
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.2454459983546834e-05,
+      "loss": 0.657,
+      "step": 29340
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.2439769655658715e-05,
+      "loss": 0.6523,
+      "step": 29350
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.2425079327770597e-05,
+      "loss": 0.6657,
+      "step": 29360
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.2410388999882478e-05,
+      "loss": 0.6939,
+      "step": 29370
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.239569867199436e-05,
+      "loss": 0.6881,
+      "step": 29380
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.2381008344106241e-05,
+      "loss": 0.5984,
+      "step": 29390
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.2366318016218123e-05,
+      "loss": 0.657,
+      "step": 29400
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.2351627688330004e-05,
+      "loss": 0.7093,
+      "step": 29410
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.2336937360441886e-05,
+      "loss": 0.6474,
+      "step": 29420
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.2322247032553767e-05,
+      "loss": 0.6518,
+      "step": 29430
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.2307556704665649e-05,
+      "loss": 0.6693,
+      "step": 29440
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.229286637677753e-05,
+      "loss": 0.6398,
+      "step": 29450
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.2278176048889412e-05,
+      "loss": 0.6197,
+      "step": 29460
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.2263485721001293e-05,
+      "loss": 0.6709,
+      "step": 29470
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.2248795393113175e-05,
+      "loss": 0.691,
+      "step": 29480
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.2234105065225056e-05,
+      "loss": 0.6536,
+      "step": 29490
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.221941473733694e-05,
+      "loss": 0.6021,
+      "step": 29500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.220472440944882e-05,
+      "loss": 0.6492,
+      "step": 29510
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.21900340815607e-05,
+      "loss": 0.6962,
+      "step": 29520
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.2175343753672582e-05,
+      "loss": 0.6676,
+      "step": 29530
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.2160653425784464e-05,
+      "loss": 0.6105,
+      "step": 29540
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.2145963097896345e-05,
+      "loss": 0.6454,
+      "step": 29550
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.2131272770008227e-05,
+      "loss": 0.6693,
+      "step": 29560
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.2116582442120108e-05,
+      "loss": 0.633,
+      "step": 29570
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.210189211423199e-05,
+      "loss": 0.6581,
+      "step": 29580
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.2087201786343871e-05,
+      "loss": 0.6716,
+      "step": 29590
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.2072511458455753e-05,
+      "loss": 0.5959,
+      "step": 29600
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.2057821130567634e-05,
+      "loss": 0.6714,
+      "step": 29610
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.2043130802679517e-05,
+      "loss": 0.6402,
+      "step": 29620
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.2028440474791399e-05,
+      "loss": 0.6777,
+      "step": 29630
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.201375014690328e-05,
+      "loss": 0.6814,
+      "step": 29640
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1999059819015162e-05,
+      "loss": 0.6544,
+      "step": 29650
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1984369491127043e-05,
+      "loss": 0.6789,
+      "step": 29660
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1969679163238923e-05,
+      "loss": 0.6815,
+      "step": 29670
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1954988835350805e-05,
+      "loss": 0.6724,
+      "step": 29680
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1940298507462686e-05,
+      "loss": 0.683,
+      "step": 29690
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1925608179574568e-05,
+      "loss": 0.6694,
+      "step": 29700
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.191091785168645e-05,
+      "loss": 0.6731,
+      "step": 29710
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.189622752379833e-05,
+      "loss": 0.6876,
+      "step": 29720
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1881537195910214e-05,
+      "loss": 0.629,
+      "step": 29730
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1866846868022095e-05,
+      "loss": 0.6492,
+      "step": 29740
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1852156540133977e-05,
+      "loss": 0.6929,
+      "step": 29750
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1837466212245859e-05,
+      "loss": 0.6129,
+      "step": 29760
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.182277588435774e-05,
+      "loss": 0.6236,
+      "step": 29770
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1808085556469622e-05,
+      "loss": 0.6923,
+      "step": 29780
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1793395228581503e-05,
+      "loss": 0.6796,
+      "step": 29790
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1778704900693385e-05,
+      "loss": 0.6141,
+      "step": 29800
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1764014572805266e-05,
+      "loss": 0.6605,
+      "step": 29810
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1749324244917148e-05,
+      "loss": 0.6265,
+      "step": 29820
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1734633917029027e-05,
+      "loss": 0.6509,
+      "step": 29830
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1719943589140909e-05,
+      "loss": 0.6429,
+      "step": 29840
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1705253261252792e-05,
+      "loss": 0.6752,
+      "step": 29850
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1690562933364674e-05,
+      "loss": 0.6438,
+      "step": 29860
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1675872605476555e-05,
+      "loss": 0.6189,
+      "step": 29870
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1661182277588437e-05,
+      "loss": 0.5944,
+      "step": 29880
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1646491949700318e-05,
+      "loss": 0.7132,
+      "step": 29890
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.16318016218122e-05,
+      "loss": 0.6461,
+      "step": 29900
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1617111293924081e-05,
+      "loss": 0.6713,
+      "step": 29910
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1602420966035963e-05,
+      "loss": 0.6595,
+      "step": 29920
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1587730638147844e-05,
+      "loss": 0.6494,
+      "step": 29930
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1573040310259726e-05,
+      "loss": 0.6624,
+      "step": 29940
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1558349982371607e-05,
+      "loss": 0.66,
+      "step": 29950
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1543659654483489e-05,
+      "loss": 0.6735,
+      "step": 29960
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.152896932659537e-05,
+      "loss": 0.6041,
+      "step": 29970
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1514278998707252e-05,
+      "loss": 0.6832,
+      "step": 29980
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1499588670819133e-05,
+      "loss": 0.642,
+      "step": 29990
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1484898342931015e-05,
+      "loss": 0.647,
+      "step": 30000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1470208015042896e-05,
+      "loss": 0.6049,
+      "step": 30010
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1455517687154778e-05,
+      "loss": 0.7538,
+      "step": 30020
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1440827359266659e-05,
+      "loss": 0.6492,
+      "step": 30030
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.142613703137854e-05,
+      "loss": 0.685,
+      "step": 30040
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1411446703490422e-05,
+      "loss": 0.6739,
+      "step": 30050
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1396756375602304e-05,
+      "loss": 0.6465,
+      "step": 30060
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1382066047714185e-05,
+      "loss": 0.7189,
+      "step": 30070
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1367375719826068e-05,
+      "loss": 0.664,
+      "step": 30080
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.135268539193795e-05,
+      "loss": 0.705,
+      "step": 30090
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1337995064049831e-05,
+      "loss": 0.6825,
+      "step": 30100
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1323304736161711e-05,
+      "loss": 0.6475,
+      "step": 30110
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1308614408273593e-05,
+      "loss": 0.6499,
+      "step": 30120
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1293924080385474e-05,
+      "loss": 0.6285,
+      "step": 30130
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1279233752497356e-05,
+      "loss": 0.6744,
+      "step": 30140
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1264543424609237e-05,
+      "loss": 0.6673,
+      "step": 30150
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1249853096721119e-05,
+      "loss": 0.6419,
+      "step": 30160
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1235162768833e-05,
+      "loss": 0.6669,
+      "step": 30170
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1220472440944882e-05,
+      "loss": 0.7251,
+      "step": 30180
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1205782113056763e-05,
+      "loss": 0.6644,
+      "step": 30190
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1191091785168646e-05,
+      "loss": 0.6596,
+      "step": 30200
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1176401457280528e-05,
+      "loss": 0.6352,
+      "step": 30210
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.116171112939241e-05,
+      "loss": 0.6423,
+      "step": 30220
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1147020801504291e-05,
+      "loss": 0.6094,
+      "step": 30230
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1132330473616172e-05,
+      "loss": 0.6733,
+      "step": 30240
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1117640145728054e-05,
+      "loss": 0.7133,
+      "step": 30250
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1102949817839935e-05,
+      "loss": 0.6639,
+      "step": 30260
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1088259489951815e-05,
+      "loss": 0.5985,
+      "step": 30270
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1073569162063697e-05,
+      "loss": 0.685,
+      "step": 30280
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1058878834175578e-05,
+      "loss": 0.6713,
+      "step": 30290
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.104418850628746e-05,
+      "loss": 0.6535,
+      "step": 30300
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1029498178399341e-05,
+      "loss": 0.6743,
+      "step": 30310
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1014807850511224e-05,
+      "loss": 0.6879,
+      "step": 30320
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1000117522623106e-05,
+      "loss": 0.7006,
+      "step": 30330
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0985427194734987e-05,
+      "loss": 0.6685,
+      "step": 30340
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0970736866846869e-05,
+      "loss": 0.6823,
+      "step": 30350
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.095604653895875e-05,
+      "loss": 0.6702,
+      "step": 30360
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0941356211070632e-05,
+      "loss": 0.6457,
+      "step": 30370
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0926665883182514e-05,
+      "loss": 0.6904,
+      "step": 30380
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0911975555294395e-05,
+      "loss": 0.6842,
+      "step": 30390
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0897285227406277e-05,
+      "loss": 0.7015,
+      "step": 30400
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0882594899518158e-05,
+      "loss": 0.7024,
+      "step": 30410
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.086790457163004e-05,
+      "loss": 0.6616,
+      "step": 30420
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0853214243741921e-05,
+      "loss": 0.6841,
+      "step": 30430
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0838523915853803e-05,
+      "loss": 0.6679,
+      "step": 30440
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0823833587965684e-05,
+      "loss": 0.6891,
+      "step": 30450
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0809143260077566e-05,
+      "loss": 0.6796,
+      "step": 30460
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0794452932189447e-05,
+      "loss": 0.6475,
+      "step": 30470
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0779762604301329e-05,
+      "loss": 0.6524,
+      "step": 30480
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.076507227641321e-05,
+      "loss": 0.6607,
+      "step": 30490
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0750381948525092e-05,
+      "loss": 0.631,
+      "step": 30500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0735691620636973e-05,
+      "loss": 0.6549,
+      "step": 30510
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0721001292748855e-05,
+      "loss": 0.5863,
+      "step": 30520
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0706310964860736e-05,
+      "loss": 0.7171,
+      "step": 30530
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0691620636972618e-05,
+      "loss": 0.6045,
+      "step": 30540
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0676930309084499e-05,
+      "loss": 0.6721,
+      "step": 30550
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.066223998119638e-05,
+      "loss": 0.6533,
+      "step": 30560
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0647549653308262e-05,
+      "loss": 0.6521,
+      "step": 30570
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0632859325420144e-05,
+      "loss": 0.6503,
+      "step": 30580
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0618168997532025e-05,
+      "loss": 0.6657,
+      "step": 30590
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0603478669643907e-05,
+      "loss": 0.6871,
+      "step": 30600
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0588788341755788e-05,
+      "loss": 0.7025,
+      "step": 30610
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.057409801386767e-05,
+      "loss": 0.6749,
+      "step": 30620
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0559407685979551e-05,
+      "loss": 0.6243,
+      "step": 30630
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0544717358091433e-05,
+      "loss": 0.6773,
+      "step": 30640
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0530027030203314e-05,
+      "loss": 0.6438,
+      "step": 30650
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0515336702315196e-05,
+      "loss": 0.6428,
+      "step": 30660
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0500646374427079e-05,
+      "loss": 0.6648,
+      "step": 30670
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.048595604653896e-05,
+      "loss": 0.6636,
+      "step": 30680
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0471265718650842e-05,
+      "loss": 0.662,
+      "step": 30690
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0456575390762723e-05,
+      "loss": 0.6112,
+      "step": 30700
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0441885062874603e-05,
+      "loss": 0.6179,
+      "step": 30710
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0427194734986485e-05,
+      "loss": 0.7023,
+      "step": 30720
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0412504407098366e-05,
+      "loss": 0.632,
+      "step": 30730
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0397814079210248e-05,
+      "loss": 0.626,
+      "step": 30740
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.038312375132213e-05,
+      "loss": 0.7254,
+      "step": 30750
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.036843342343401e-05,
+      "loss": 0.6596,
+      "step": 30760
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0353743095545892e-05,
+      "loss": 0.6119,
+      "step": 30770
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0339052767657775e-05,
+      "loss": 0.6714,
+      "step": 30780
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0324362439769657e-05,
+      "loss": 0.6428,
+      "step": 30790
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0309672111881538e-05,
+      "loss": 0.6609,
+      "step": 30800
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.029498178399342e-05,
+      "loss": 0.6574,
+      "step": 30810
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0280291456105301e-05,
+      "loss": 0.6984,
+      "step": 30820
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0265601128217183e-05,
+      "loss": 0.6726,
+      "step": 30830
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0250910800329064e-05,
+      "loss": 0.6579,
+      "step": 30840
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0236220472440946e-05,
+      "loss": 0.6518,
+      "step": 30850
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0221530144552827e-05,
+      "loss": 0.6791,
+      "step": 30860
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0206839816664707e-05,
+      "loss": 0.6667,
+      "step": 30870
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0192149488776589e-05,
+      "loss": 0.6838,
+      "step": 30880
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.017745916088847e-05,
+      "loss": 0.6175,
+      "step": 30890
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0162768833000353e-05,
+      "loss": 0.7053,
+      "step": 30900
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0148078505112235e-05,
+      "loss": 0.6573,
+      "step": 30910
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0133388177224116e-05,
+      "loss": 0.6179,
+      "step": 30920
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0118697849335998e-05,
+      "loss": 0.674,
+      "step": 30930
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.010400752144788e-05,
+      "loss": 0.5972,
+      "step": 30940
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0089317193559761e-05,
+      "loss": 0.651,
+      "step": 30950
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0074626865671643e-05,
+      "loss": 0.6713,
+      "step": 30960
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0059936537783524e-05,
+      "loss": 0.6727,
+      "step": 30970
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0045246209895406e-05,
+      "loss": 0.6686,
+      "step": 30980
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0030555882007287e-05,
+      "loss": 0.6396,
+      "step": 30990
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0015865554119169e-05,
+      "loss": 0.5668,
+      "step": 31000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.000117522623105e-05,
+      "loss": 0.6901,
+      "step": 31010
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.986484898342932e-06,
+      "loss": 0.6807,
+      "step": 31020
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.971794570454813e-06,
+      "loss": 0.6439,
+      "step": 31030
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.957104242566695e-06,
+      "loss": 0.6652,
+      "step": 31040
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.942413914678576e-06,
+      "loss": 0.6987,
+      "step": 31050
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.927723586790458e-06,
+      "loss": 0.659,
+      "step": 31060
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.913033258902339e-06,
+      "loss": 0.6139,
+      "step": 31070
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.89834293101422e-06,
+      "loss": 0.6613,
+      "step": 31080
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.883652603126102e-06,
+      "loss": 0.6447,
+      "step": 31090
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.868962275237984e-06,
+      "loss": 0.6867,
+      "step": 31100
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.854271947349865e-06,
+      "loss": 0.6136,
+      "step": 31110
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.839581619461747e-06,
+      "loss": 0.648,
+      "step": 31120
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.82489129157363e-06,
+      "loss": 0.66,
+      "step": 31130
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.810200963685511e-06,
+      "loss": 0.6669,
+      "step": 31140
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.795510635797391e-06,
+      "loss": 0.6402,
+      "step": 31150
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.780820307909273e-06,
+      "loss": 0.6205,
+      "step": 31160
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.766129980021154e-06,
+      "loss": 0.6355,
+      "step": 31170
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.751439652133036e-06,
+      "loss": 0.691,
+      "step": 31180
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.736749324244917e-06,
+      "loss": 0.6513,
+      "step": 31190
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.722058996356799e-06,
+      "loss": 0.5862,
+      "step": 31200
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.70736866846868e-06,
+      "loss": 0.6447,
+      "step": 31210
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.692678340580562e-06,
+      "loss": 0.6616,
+      "step": 31220
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.677988012692443e-06,
+      "loss": 0.6577,
+      "step": 31230
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.663297684804325e-06,
+      "loss": 0.6365,
+      "step": 31240
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.648607356916208e-06,
+      "loss": 0.6696,
+      "step": 31250
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.63391702902809e-06,
+      "loss": 0.6354,
+      "step": 31260
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.619226701139971e-06,
+      "loss": 0.6559,
+      "step": 31270
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.604536373251852e-06,
+      "loss": 0.6366,
+      "step": 31280
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.589846045363734e-06,
+      "loss": 0.6393,
+      "step": 31290
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.575155717475615e-06,
+      "loss": 0.6426,
+      "step": 31300
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.560465389587495e-06,
+      "loss": 0.6443,
+      "step": 31310
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.545775061699377e-06,
+      "loss": 0.6816,
+      "step": 31320
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.531084733811258e-06,
+      "loss": 0.664,
+      "step": 31330
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.51639440592314e-06,
+      "loss": 0.6354,
+      "step": 31340
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.501704078035021e-06,
+      "loss": 0.66,
+      "step": 31350
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.487013750146903e-06,
+      "loss": 0.6827,
+      "step": 31360
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.472323422258786e-06,
+      "loss": 0.6101,
+      "step": 31370
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.457633094370667e-06,
+      "loss": 0.6911,
+      "step": 31380
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.442942766482549e-06,
+      "loss": 0.6055,
+      "step": 31390
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.42825243859443e-06,
+      "loss": 0.6607,
+      "step": 31400
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.413562110706312e-06,
+      "loss": 0.6878,
+      "step": 31410
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.398871782818193e-06,
+      "loss": 0.6589,
+      "step": 31420
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.384181454930075e-06,
+      "loss": 0.6714,
+      "step": 31430
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.369491127041956e-06,
+      "loss": 0.6091,
+      "step": 31440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.354800799153838e-06,
+      "loss": 0.6799,
+      "step": 31450
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.34011047126572e-06,
+      "loss": 0.6397,
+      "step": 31460
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.3254201433776e-06,
+      "loss": 0.6601,
+      "step": 31470
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.31072981548948e-06,
+      "loss": 0.6404,
+      "step": 31480
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.296039487601364e-06,
+      "loss": 0.615,
+      "step": 31490
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.281349159713245e-06,
+      "loss": 0.627,
+      "step": 31500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.266658831825127e-06,
+      "loss": 0.6456,
+      "step": 31510
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.251968503937008e-06,
+      "loss": 0.6677,
+      "step": 31520
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.23727817604889e-06,
+      "loss": 0.6627,
+      "step": 31530
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.222587848160771e-06,
+      "loss": 0.6301,
+      "step": 31540
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.207897520272653e-06,
+      "loss": 0.6129,
+      "step": 31550
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.193207192384534e-06,
+      "loss": 0.6806,
+      "step": 31560
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.178516864496416e-06,
+      "loss": 0.6197,
+      "step": 31570
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.163826536608298e-06,
+      "loss": 0.6905,
+      "step": 31580
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.149136208720179e-06,
+      "loss": 0.6136,
+      "step": 31590
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.13444588083206e-06,
+      "loss": 0.6575,
+      "step": 31600
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.119755552943942e-06,
+      "loss": 0.6477,
+      "step": 31610
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.105065225055824e-06,
+      "loss": 0.6577,
+      "step": 31620
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.090374897167705e-06,
+      "loss": 0.6787,
+      "step": 31630
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.075684569279587e-06,
+      "loss": 0.6358,
+      "step": 31640
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.060994241391468e-06,
+      "loss": 0.655,
+      "step": 31650
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.04630391350335e-06,
+      "loss": 0.6198,
+      "step": 31660
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.031613585615231e-06,
+      "loss": 0.6218,
+      "step": 31670
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.016923257727113e-06,
+      "loss": 0.6207,
+      "step": 31680
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.002232929838994e-06,
+      "loss": 0.6097,
+      "step": 31690
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.987542601950876e-06,
+      "loss": 0.6928,
+      "step": 31700
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.972852274062757e-06,
+      "loss": 0.6271,
+      "step": 31710
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.95816194617464e-06,
+      "loss": 0.6969,
+      "step": 31720
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.943471618286522e-06,
+      "loss": 0.6407,
+      "step": 31730
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.928781290398403e-06,
+      "loss": 0.5821,
+      "step": 31740
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.914090962510283e-06,
+      "loss": 0.6975,
+      "step": 31750
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.899400634622165e-06,
+      "loss": 0.674,
+      "step": 31760
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.884710306734046e-06,
+      "loss": 0.6358,
+      "step": 31770
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.870019978845928e-06,
+      "loss": 0.6318,
+      "step": 31780
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.855329650957809e-06,
+      "loss": 0.6757,
+      "step": 31790
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.84063932306969e-06,
+      "loss": 0.6397,
+      "step": 31800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.825948995181572e-06,
+      "loss": 0.6266,
+      "step": 31810
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.811258667293454e-06,
+      "loss": 0.6255,
+      "step": 31820
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.796568339405335e-06,
+      "loss": 0.6691,
+      "step": 31830
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.781878011517218e-06,
+      "loss": 0.6608,
+      "step": 31840
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.7671876836291e-06,
+      "loss": 0.6849,
+      "step": 31850
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.752497355740981e-06,
+      "loss": 0.5634,
+      "step": 31860
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.737807027852863e-06,
+      "loss": 0.6347,
+      "step": 31870
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.723116699964744e-06,
+      "loss": 0.6003,
+      "step": 31880
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.708426372076626e-06,
+      "loss": 0.6265,
+      "step": 31890
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.693736044188507e-06,
+      "loss": 0.6333,
+      "step": 31900
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.679045716300387e-06,
+      "loss": 0.6226,
+      "step": 31910
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.664355388412269e-06,
+      "loss": 0.6888,
+      "step": 31920
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.64966506052415e-06,
+      "loss": 0.6613,
+      "step": 31930
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.634974732636032e-06,
+      "loss": 0.6327,
+      "step": 31940
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.620284404747915e-06,
+      "loss": 0.6228,
+      "step": 31950
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.605594076859796e-06,
+      "loss": 0.6549,
+      "step": 31960
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.590903748971678e-06,
+      "loss": 0.6702,
+      "step": 31970
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.57621342108356e-06,
+      "loss": 0.6302,
+      "step": 31980
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.561523093195441e-06,
+      "loss": 0.5505,
+      "step": 31990
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.546832765307322e-06,
+      "loss": 0.6177,
+      "step": 32000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.532142437419204e-06,
+      "loss": 0.6789,
+      "step": 32010
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.517452109531085e-06,
+      "loss": 0.6502,
+      "step": 32020
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.502761781642967e-06,
+      "loss": 0.6215,
+      "step": 32030
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.488071453754848e-06,
+      "loss": 0.6479,
+      "step": 32040
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.47338112586673e-06,
+      "loss": 0.6293,
+      "step": 32050
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.458690797978611e-06,
+      "loss": 0.6115,
+      "step": 32060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.444000470090493e-06,
+      "loss": 0.6054,
+      "step": 32070
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 8.429310142202374e-06,
+      "loss": 0.663,
+      "step": 32080
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.414619814314256e-06,
+      "loss": 0.6378,
+      "step": 32090
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.399929486426137e-06,
+      "loss": 0.6294,
+      "step": 32100
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.385239158538019e-06,
+      "loss": 0.6398,
+      "step": 32110
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.3705488306499e-06,
+      "loss": 0.6289,
+      "step": 32120
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.355858502761782e-06,
+      "loss": 0.6342,
+      "step": 32130
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.341168174873663e-06,
+      "loss": 0.6266,
+      "step": 32140
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.326477846985545e-06,
+      "loss": 0.6094,
+      "step": 32150
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.311787519097426e-06,
+      "loss": 0.657,
+      "step": 32160
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.297097191209308e-06,
+      "loss": 0.6813,
+      "step": 32170
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.28240686332119e-06,
+      "loss": 0.6658,
+      "step": 32180
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.267716535433071e-06,
+      "loss": 0.6601,
+      "step": 32190
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.253026207544953e-06,
+      "loss": 0.6557,
+      "step": 32200
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.238335879656834e-06,
+      "loss": 0.6525,
+      "step": 32210
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.223645551768716e-06,
+      "loss": 0.681,
+      "step": 32220
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.208955223880597e-06,
+      "loss": 0.63,
+      "step": 32230
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.194264895992479e-06,
+      "loss": 0.6473,
+      "step": 32240
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.17957456810436e-06,
+      "loss": 0.628,
+      "step": 32250
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.164884240216242e-06,
+      "loss": 0.6194,
+      "step": 32260
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.150193912328123e-06,
+      "loss": 0.6393,
+      "step": 32270
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.135503584440005e-06,
+      "loss": 0.6494,
+      "step": 32280
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.120813256551886e-06,
+      "loss": 0.6192,
+      "step": 32290
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.10612292866377e-06,
+      "loss": 0.6404,
+      "step": 32300
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.09143260077565e-06,
+      "loss": 0.6137,
+      "step": 32310
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.076742272887532e-06,
+      "loss": 0.6403,
+      "step": 32320
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.062051944999414e-06,
+      "loss": 0.5781,
+      "step": 32330
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.047361617111295e-06,
+      "loss": 0.6513,
+      "step": 32340
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.032671289223175e-06,
+      "loss": 0.6984,
+      "step": 32350
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.017980961335057e-06,
+      "loss": 0.6276,
+      "step": 32360
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.003290633446938e-06,
+      "loss": 0.6373,
+      "step": 32370
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.98860030555882e-06,
+      "loss": 0.6498,
+      "step": 32380
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.973909977670701e-06,
+      "loss": 0.6654,
+      "step": 32390
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.959219649782583e-06,
+      "loss": 0.6759,
+      "step": 32400
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.944529321894464e-06,
+      "loss": 0.612,
+      "step": 32410
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.929838994006347e-06,
+      "loss": 0.6448,
+      "step": 32420
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.915148666118229e-06,
+      "loss": 0.6608,
+      "step": 32430
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.90045833823011e-06,
+      "loss": 0.6459,
+      "step": 32440
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.885768010341992e-06,
+      "loss": 0.6734,
+      "step": 32450
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.871077682453873e-06,
+      "loss": 0.6405,
+      "step": 32460
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.856387354565755e-06,
+      "loss": 0.6332,
+      "step": 32470
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.841697026677636e-06,
+      "loss": 0.6368,
+      "step": 32480
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.827006698789518e-06,
+      "loss": 0.6857,
+      "step": 32490
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.8123163709014e-06,
+      "loss": 0.6723,
+      "step": 32500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.79762604301328e-06,
+      "loss": 0.701,
+      "step": 32510
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.78293571512516e-06,
+      "loss": 0.6663,
+      "step": 32520
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.768245387237042e-06,
+      "loss": 0.6519,
+      "step": 32530
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.753555059348925e-06,
+      "loss": 0.6774,
+      "step": 32540
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.738864731460807e-06,
+      "loss": 0.6497,
+      "step": 32550
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.724174403572688e-06,
+      "loss": 0.6883,
+      "step": 32560
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.70948407568457e-06,
+      "loss": 0.6348,
+      "step": 32570
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.694793747796451e-06,
+      "loss": 0.6601,
+      "step": 32580
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.680103419908333e-06,
+      "loss": 0.6652,
+      "step": 32590
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.665413092020214e-06,
+      "loss": 0.6192,
+      "step": 32600
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.650722764132096e-06,
+      "loss": 0.6522,
+      "step": 32610
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.636032436243977e-06,
+      "loss": 0.5849,
+      "step": 32620
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.621342108355858e-06,
+      "loss": 0.62,
+      "step": 32630
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.60665178046774e-06,
+      "loss": 0.6422,
+      "step": 32640
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.591961452579623e-06,
+      "loss": 0.6296,
+      "step": 32650
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.577271124691504e-06,
+      "loss": 0.683,
+      "step": 32660
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.562580796803386e-06,
+      "loss": 0.6515,
+      "step": 32670
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.5478904689152664e-06,
+      "loss": 0.6324,
+      "step": 32680
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.533200141027148e-06,
+      "loss": 0.6835,
+      "step": 32690
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.5185098131390295e-06,
+      "loss": 0.6466,
+      "step": 32700
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.503819485250911e-06,
+      "loss": 0.6211,
+      "step": 32710
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.4891291573627925e-06,
+      "loss": 0.6261,
+      "step": 32720
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.474438829474674e-06,
+      "loss": 0.6308,
+      "step": 32730
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.4597485015865555e-06,
+      "loss": 0.5964,
+      "step": 32740
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.445058173698437e-06,
+      "loss": 0.6421,
+      "step": 32750
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.4303678458103185e-06,
+      "loss": 0.6053,
+      "step": 32760
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.415677517922201e-06,
+      "loss": 0.6717,
+      "step": 32770
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.400987190034082e-06,
+      "loss": 0.6042,
+      "step": 32780
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.386296862145964e-06,
+      "loss": 0.6661,
+      "step": 32790
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.371606534257845e-06,
+      "loss": 0.6532,
+      "step": 32800
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.356916206369727e-06,
+      "loss": 0.6788,
+      "step": 32810
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.342225878481608e-06,
+      "loss": 0.6429,
+      "step": 32820
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.32753555059349e-06,
+      "loss": 0.6813,
+      "step": 32830
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.3128452227053705e-06,
+      "loss": 0.6367,
+      "step": 32840
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.298154894817252e-06,
+      "loss": 0.6054,
+      "step": 32850
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.2834645669291335e-06,
+      "loss": 0.6685,
+      "step": 32860
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.268774239041015e-06,
+      "loss": 0.6514,
+      "step": 32870
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.2540839111528966e-06,
+      "loss": 0.6285,
+      "step": 32880
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.23939358326478e-06,
+      "loss": 0.6642,
+      "step": 32890
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.22470325537666e-06,
+      "loss": 0.6474,
+      "step": 32900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.210012927488542e-06,
+      "loss": 0.6694,
+      "step": 32910
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.195322599600423e-06,
+      "loss": 0.6487,
+      "step": 32920
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.180632271712305e-06,
+      "loss": 0.627,
+      "step": 32930
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.1659419438241864e-06,
+      "loss": 0.6357,
+      "step": 32940
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.151251615936068e-06,
+      "loss": 0.6254,
+      "step": 32950
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 7.1365612880479494e-06,
+      "loss": 0.5891,
+      "step": 32960
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.121870960159831e-06,
+      "loss": 0.579,
+      "step": 32970
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.1071806322717124e-06,
+      "loss": 0.6525,
+      "step": 32980
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.092490304383594e-06,
+      "loss": 0.6505,
+      "step": 32990
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.077799976495476e-06,
+      "loss": 0.6544,
+      "step": 33000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.063109648607358e-06,
+      "loss": 0.6661,
+      "step": 33010
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.048419320719239e-06,
+      "loss": 0.6401,
+      "step": 33020
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.033728992831121e-06,
+      "loss": 0.6923,
+      "step": 33030
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.019038664943002e-06,
+      "loss": 0.6406,
+      "step": 33040
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.004348337054884e-06,
+      "loss": 0.6287,
+      "step": 33050
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.9896580091667645e-06,
+      "loss": 0.6588,
+      "step": 33060
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.974967681278646e-06,
+      "loss": 0.6256,
+      "step": 33070
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.9602773533905275e-06,
+      "loss": 0.5949,
+      "step": 33080
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.945587025502409e-06,
+      "loss": 0.6203,
+      "step": 33090
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.9308966976142905e-06,
+      "loss": 0.6554,
+      "step": 33100
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.916206369726172e-06,
+      "loss": 0.6018,
+      "step": 33110
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.901516041838054e-06,
+      "loss": 0.6755,
+      "step": 33120
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.886825713949936e-06,
+      "loss": 0.6352,
+      "step": 33130
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.872135386061817e-06,
+      "loss": 0.621,
+      "step": 33140
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.857445058173699e-06,
+      "loss": 0.6517,
+      "step": 33150
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.84275473028558e-06,
+      "loss": 0.6224,
+      "step": 33160
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.828064402397462e-06,
+      "loss": 0.6688,
+      "step": 33170
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.813374074509343e-06,
+      "loss": 0.6357,
+      "step": 33180
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.798683746621225e-06,
+      "loss": 0.683,
+      "step": 33190
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.783993418733106e-06,
+      "loss": 0.6513,
+      "step": 33200
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.769303090844988e-06,
+      "loss": 0.6641,
+      "step": 33210
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.7546127629568686e-06,
+      "loss": 0.6113,
+      "step": 33220
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.73992243506875e-06,
+      "loss": 0.6306,
+      "step": 33230
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.725232107180633e-06,
+      "loss": 0.6644,
+      "step": 33240
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.710541779292515e-06,
+      "loss": 0.6792,
+      "step": 33250
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.695851451404396e-06,
+      "loss": 0.6538,
+      "step": 33260
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.681161123516278e-06,
+      "loss": 0.6761,
+      "step": 33270
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.6664707956281584e-06,
+      "loss": 0.6531,
+      "step": 33280
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.65178046774004e-06,
+      "loss": 0.6175,
+      "step": 33290
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.6370901398519215e-06,
+      "loss": 0.6708,
+      "step": 33300
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.622399811963803e-06,
+      "loss": 0.6247,
+      "step": 33310
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.6077094840756845e-06,
+      "loss": 0.6432,
+      "step": 33320
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.593019156187566e-06,
+      "loss": 0.6495,
+      "step": 33330
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.5783288282994475e-06,
+      "loss": 0.6116,
+      "step": 33340
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.56363850041133e-06,
+      "loss": 0.6699,
+      "step": 33350
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.548948172523211e-06,
+      "loss": 0.6855,
+      "step": 33360
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.534257844635093e-06,
+      "loss": 0.6272,
+      "step": 33370
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.519567516746974e-06,
+      "loss": 0.6722,
+      "step": 33380
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.504877188858856e-06,
+      "loss": 0.6183,
+      "step": 33390
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.490186860970737e-06,
+      "loss": 0.613,
+      "step": 33400
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.475496533082619e-06,
+      "loss": 0.6374,
+      "step": 33410
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.4608062051945e-06,
+      "loss": 0.6395,
+      "step": 33420
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.446115877306382e-06,
+      "loss": 0.6751,
+      "step": 33430
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.4314255494182625e-06,
+      "loss": 0.6862,
+      "step": 33440
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.416735221530144e-06,
+      "loss": 0.6488,
+      "step": 33450
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.4020448936420255e-06,
+      "loss": 0.5763,
+      "step": 33460
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.387354565753909e-06,
+      "loss": 0.6784,
+      "step": 33470
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.37266423786579e-06,
+      "loss": 0.6572,
+      "step": 33480
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.357973909977672e-06,
+      "loss": 0.7097,
+      "step": 33490
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.343283582089552e-06,
+      "loss": 0.6716,
+      "step": 33500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.328593254201434e-06,
+      "loss": 0.6038,
+      "step": 33510
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.313902926313315e-06,
+      "loss": 0.6563,
+      "step": 33520
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.299212598425197e-06,
+      "loss": 0.6809,
+      "step": 33530
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.2845222705370784e-06,
+      "loss": 0.6436,
+      "step": 33540
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.26983194264896e-06,
+      "loss": 0.6456,
+      "step": 33550
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.2551416147608414e-06,
+      "loss": 0.6371,
+      "step": 33560
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.240451286872724e-06,
+      "loss": 0.6183,
+      "step": 33570
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.2257609589846044e-06,
+      "loss": 0.632,
+      "step": 33580
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.211070631096486e-06,
+      "loss": 0.603,
+      "step": 33590
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.1963803032083675e-06,
+      "loss": 0.6362,
+      "step": 33600
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.181689975320249e-06,
+      "loss": 0.641,
+      "step": 33610
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.166999647432131e-06,
+      "loss": 0.7042,
+      "step": 33620
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.152309319544013e-06,
+      "loss": 0.6166,
+      "step": 33630
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.137618991655894e-06,
+      "loss": 0.6507,
+      "step": 33640
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.122928663767776e-06,
+      "loss": 0.6712,
+      "step": 33650
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.1082383358796565e-06,
+      "loss": 0.6839,
+      "step": 33660
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.093548007991539e-06,
+      "loss": 0.6208,
+      "step": 33670
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.07885768010342e-06,
+      "loss": 0.6154,
+      "step": 33680
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.064167352215302e-06,
+      "loss": 0.6534,
+      "step": 33690
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.049477024327183e-06,
+      "loss": 0.628,
+      "step": 33700
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.034786696439065e-06,
+      "loss": 0.6157,
+      "step": 33710
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.020096368550946e-06,
+      "loss": 0.6378,
+      "step": 33720
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.005406040662828e-06,
+      "loss": 0.6138,
+      "step": 33730
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.990715712774709e-06,
+      "loss": 0.6098,
+      "step": 33740
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.976025384886591e-06,
+      "loss": 0.6728,
+      "step": 33750
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.961335056998472e-06,
+      "loss": 0.6337,
+      "step": 33760
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.946644729110354e-06,
+      "loss": 0.6499,
+      "step": 33770
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.931954401222235e-06,
+      "loss": 0.614,
+      "step": 33780
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.917264073334118e-06,
+      "loss": 0.5943,
+      "step": 33790
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.902573745445998e-06,
+      "loss": 0.6328,
+      "step": 33800
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.88788341755788e-06,
+      "loss": 0.6209,
+      "step": 33810
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.873193089669761e-06,
+      "loss": 0.62,
+      "step": 33820
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.858502761781643e-06,
+      "loss": 0.6482,
+      "step": 33830
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.843812433893524e-06,
+      "loss": 0.6724,
+      "step": 33840
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.829122106005407e-06,
+      "loss": 0.653,
+      "step": 33850
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.814431778117288e-06,
+      "loss": 0.6598,
+      "step": 33860
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.79974145022917e-06,
+      "loss": 0.643,
+      "step": 33870
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.7850511223410504e-06,
+      "loss": 0.6568,
+      "step": 33880
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.770360794452932e-06,
+      "loss": 0.6433,
+      "step": 33890
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.7556704665648135e-06,
+      "loss": 0.6404,
+      "step": 33900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.740980138676696e-06,
+      "loss": 0.6288,
+      "step": 33910
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.726289810788577e-06,
+      "loss": 0.6496,
+      "step": 33920
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.711599482900459e-06,
+      "loss": 0.6142,
+      "step": 33930
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.69690915501234e-06,
+      "loss": 0.6866,
+      "step": 33940
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.682218827124222e-06,
+      "loss": 0.6472,
+      "step": 33950
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6675284992361025e-06,
+      "loss": 0.6233,
+      "step": 33960
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.652838171347985e-06,
+      "loss": 0.6808,
+      "step": 33970
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.638147843459866e-06,
+      "loss": 0.6711,
+      "step": 33980
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.623457515571748e-06,
+      "loss": 0.6351,
+      "step": 33990
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.608767187683629e-06,
+      "loss": 0.6587,
+      "step": 34000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.594076859795511e-06,
+      "loss": 0.6717,
+      "step": 34010
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.579386531907392e-06,
+      "loss": 0.6258,
+      "step": 34020
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.564696204019274e-06,
+      "loss": 0.6367,
+      "step": 34030
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.550005876131155e-06,
+      "loss": 0.5942,
+      "step": 34040
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.535315548243037e-06,
+      "loss": 0.5863,
+      "step": 34050
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.520625220354918e-06,
+      "loss": 0.6668,
+      "step": 34060
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.5059348924668e-06,
+      "loss": 0.6876,
+      "step": 34070
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.491244564578682e-06,
+      "loss": 0.5955,
+      "step": 34080
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.476554236690564e-06,
+      "loss": 0.6641,
+      "step": 34090
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.461863908802444e-06,
+      "loss": 0.6275,
+      "step": 34100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.447173580914326e-06,
+      "loss": 0.6625,
+      "step": 34110
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.432483253026207e-06,
+      "loss": 0.638,
+      "step": 34120
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.417792925138089e-06,
+      "loss": 0.6584,
+      "step": 34130
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.403102597249971e-06,
+      "loss": 0.6152,
+      "step": 34140
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.388412269361853e-06,
+      "loss": 0.6469,
+      "step": 34150
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.373721941473734e-06,
+      "loss": 0.6624,
+      "step": 34160
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.359031613585616e-06,
+      "loss": 0.6716,
+      "step": 34170
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.3443412856974964e-06,
+      "loss": 0.6613,
+      "step": 34180
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.329650957809378e-06,
+      "loss": 0.6516,
+      "step": 34190
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.31496062992126e-06,
+      "loss": 0.6597,
+      "step": 34200
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.300270302033142e-06,
+      "loss": 0.6466,
+      "step": 34210
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.285579974145023e-06,
+      "loss": 0.6522,
+      "step": 34220
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.270889646256905e-06,
+      "loss": 0.6188,
+      "step": 34230
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.256199318368786e-06,
+      "loss": 0.6877,
+      "step": 34240
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.241508990480668e-06,
+      "loss": 0.691,
+      "step": 34250
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.226818662592549e-06,
+      "loss": 0.6273,
+      "step": 34260
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.212128334704431e-06,
+      "loss": 0.6048,
+      "step": 34270
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.197438006816312e-06,
+      "loss": 0.6174,
+      "step": 34280
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.182747678928194e-06,
+      "loss": 0.6198,
+      "step": 34290
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.168057351040075e-06,
+      "loss": 0.6524,
+      "step": 34300
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.153367023151957e-06,
+      "loss": 0.6345,
+      "step": 34310
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.138676695263838e-06,
+      "loss": 0.6783,
+      "step": 34320
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.12398636737572e-06,
+      "loss": 0.6472,
+      "step": 34330
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.109296039487601e-06,
+      "loss": 0.6641,
+      "step": 34340
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.094605711599483e-06,
+      "loss": 0.6123,
+      "step": 34350
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.079915383711364e-06,
+      "loss": 0.6515,
+      "step": 34360
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.065225055823247e-06,
+      "loss": 0.6831,
+      "step": 34370
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.050534727935128e-06,
+      "loss": 0.6484,
+      "step": 34380
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.03584440004701e-06,
+      "loss": 0.6401,
+      "step": 34390
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.02115407215889e-06,
+      "loss": 0.6351,
+      "step": 34400
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.006463744270772e-06,
+      "loss": 0.6077,
+      "step": 34410
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.991773416382653e-06,
+      "loss": 0.6492,
+      "step": 34420
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.977083088494536e-06,
+      "loss": 0.6717,
+      "step": 34430
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.962392760606417e-06,
+      "loss": 0.6537,
+      "step": 34440
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.947702432718299e-06,
+      "loss": 0.6733,
+      "step": 34450
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.93301210483018e-06,
+      "loss": 0.6133,
+      "step": 34460
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.918321776942062e-06,
+      "loss": 0.688,
+      "step": 34470
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.9036314490539424e-06,
+      "loss": 0.6123,
+      "step": 34480
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.888941121165825e-06,
+      "loss": 0.6349,
+      "step": 34490
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.874250793277706e-06,
+      "loss": 0.635,
+      "step": 34500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.859560465389588e-06,
+      "loss": 0.675,
+      "step": 34510
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.844870137501469e-06,
+      "loss": 0.6254,
+      "step": 34520
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.830179809613351e-06,
+      "loss": 0.6477,
+      "step": 34530
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.815489481725232e-06,
+      "loss": 0.6538,
+      "step": 34540
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.800799153837114e-06,
+      "loss": 0.6533,
+      "step": 34550
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.786108825948995e-06,
+      "loss": 0.6886,
+      "step": 34560
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.771418498060877e-06,
+      "loss": 0.6431,
+      "step": 34570
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.756728170172758e-06,
+      "loss": 0.6429,
+      "step": 34580
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.74203784228464e-06,
+      "loss": 0.6752,
+      "step": 34590
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.727347514396521e-06,
+      "loss": 0.6421,
+      "step": 34600
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.712657186508404e-06,
+      "loss": 0.6562,
+      "step": 34610
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.697966858620284e-06,
+      "loss": 0.6439,
+      "step": 34620
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.683276530732166e-06,
+      "loss": 0.6369,
+      "step": 34630
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.668586202844047e-06,
+      "loss": 0.6614,
+      "step": 34640
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.653895874955929e-06,
+      "loss": 0.679,
+      "step": 34650
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.63920554706781e-06,
+      "loss": 0.6437,
+      "step": 34660
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.624515219179693e-06,
+      "loss": 0.6033,
+      "step": 34670
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.609824891291574e-06,
+      "loss": 0.591,
+      "step": 34680
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.595134563403456e-06,
+      "loss": 0.6358,
+      "step": 34690
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.580444235515336e-06,
+      "loss": 0.5958,
+      "step": 34700
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.565753907627218e-06,
+      "loss": 0.6211,
+      "step": 34710
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.5510635797391e-06,
+      "loss": 0.6733,
+      "step": 34720
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.536373251850982e-06,
+      "loss": 0.6417,
+      "step": 34730
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.521682923962863e-06,
+      "loss": 0.6295,
+      "step": 34740
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.506992596074745e-06,
+      "loss": 0.6377,
+      "step": 34750
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.492302268186626e-06,
+      "loss": 0.6146,
+      "step": 34760
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.477611940298508e-06,
+      "loss": 0.647,
+      "step": 34770
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.462921612410389e-06,
+      "loss": 0.6545,
+      "step": 34780
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.448231284522271e-06,
+      "loss": 0.6617,
+      "step": 34790
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.433540956634152e-06,
+      "loss": 0.642,
+      "step": 34800
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.418850628746034e-06,
+      "loss": 0.6553,
+      "step": 34810
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.404160300857915e-06,
+      "loss": 0.682,
+      "step": 34820
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.389469972969797e-06,
+      "loss": 0.6523,
+      "step": 34830
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.374779645081678e-06,
+      "loss": 0.6327,
+      "step": 34840
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.36008931719356e-06,
+      "loss": 0.6834,
+      "step": 34850
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.345398989305441e-06,
+      "loss": 0.6328,
+      "step": 34860
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.330708661417323e-06,
+      "loss": 0.6373,
+      "step": 34870
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.316018333529204e-06,
+      "loss": 0.6773,
+      "step": 34880
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.301328005641086e-06,
+      "loss": 0.6397,
+      "step": 34890
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.286637677752968e-06,
+      "loss": 0.6309,
+      "step": 34900
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.27194734986485e-06,
+      "loss": 0.6705,
+      "step": 34910
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.25725702197673e-06,
+      "loss": 0.6318,
+      "step": 34920
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.242566694088612e-06,
+      "loss": 0.6234,
+      "step": 34930
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.227876366200493e-06,
+      "loss": 0.6724,
+      "step": 34940
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.213186038312375e-06,
+      "loss": 0.7067,
+      "step": 34950
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.198495710424257e-06,
+      "loss": 0.6482,
+      "step": 34960
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.183805382536139e-06,
+      "loss": 0.6464,
+      "step": 34970
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.16911505464802e-06,
+      "loss": 0.6158,
+      "step": 34980
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.154424726759902e-06,
+      "loss": 0.6447,
+      "step": 34990
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.139734398871782e-06,
+      "loss": 0.6608,
+      "step": 35000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.125044070983664e-06,
+      "loss": 0.623,
+      "step": 35010
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.110353743095546e-06,
+      "loss": 0.6629,
+      "step": 35020
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.095663415207428e-06,
+      "loss": 0.6245,
+      "step": 35030
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.080973087319309e-06,
+      "loss": 0.6402,
+      "step": 35040
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.066282759431191e-06,
+      "loss": 0.6438,
+      "step": 35050
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.051592431543072e-06,
+      "loss": 0.6822,
+      "step": 35060
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.036902103654954e-06,
+      "loss": 0.6068,
+      "step": 35070
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.022211775766835e-06,
+      "loss": 0.6089,
+      "step": 35080
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.007521447878717e-06,
+      "loss": 0.63,
+      "step": 35090
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.992831119990598e-06,
+      "loss": 0.6715,
+      "step": 35100
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.97814079210248e-06,
+      "loss": 0.6241,
+      "step": 35110
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.963450464214361e-06,
+      "loss": 0.6245,
+      "step": 35120
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.948760136326244e-06,
+      "loss": 0.6552,
+      "step": 35130
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.934069808438124e-06,
+      "loss": 0.6641,
+      "step": 35140
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.919379480550006e-06,
+      "loss": 0.6741,
+      "step": 35150
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.904689152661887e-06,
+      "loss": 0.6266,
+      "step": 35160
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.889998824773769e-06,
+      "loss": 0.626,
+      "step": 35170
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.87530849688565e-06,
+      "loss": 0.6638,
+      "step": 35180
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.860618168997533e-06,
+      "loss": 0.6722,
+      "step": 35190
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.845927841109414e-06,
+      "loss": 0.662,
+      "step": 35200
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.831237513221296e-06,
+      "loss": 0.6674,
+      "step": 35210
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.816547185333176e-06,
+      "loss": 0.6556,
+      "step": 35220
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.8018568574450583e-06,
+      "loss": 0.6132,
+      "step": 35230
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.7871665295569398e-06,
+      "loss": 0.6628,
+      "step": 35240
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.7724762016688217e-06,
+      "loss": 0.6281,
+      "step": 35250
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.7577858737807032e-06,
+      "loss": 0.689,
+      "step": 35260
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.7430955458925847e-06,
+      "loss": 0.6326,
+      "step": 35270
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.728405218004466e-06,
+      "loss": 0.6588,
+      "step": 35280
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.7137148901163473e-06,
+      "loss": 0.6664,
+      "step": 35290
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.699024562228229e-06,
+      "loss": 0.6098,
+      "step": 35300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6843342343401107e-06,
+      "loss": 0.6194,
+      "step": 35310
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6696439064519922e-06,
+      "loss": 0.6569,
+      "step": 35320
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6549535785638738e-06,
+      "loss": 0.6424,
+      "step": 35330
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6402632506757553e-06,
+      "loss": 0.5929,
+      "step": 35340
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6255729227876368e-06,
+      "loss": 0.6747,
+      "step": 35350
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.610882594899518e-06,
+      "loss": 0.6549,
+      "step": 35360
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5961922670114e-06,
+      "loss": 0.6486,
+      "step": 35370
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5815019391232817e-06,
+      "loss": 0.6409,
+      "step": 35380
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5668116112351628e-06,
+      "loss": 0.6057,
+      "step": 35390
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5521212833470443e-06,
+      "loss": 0.5649,
+      "step": 35400
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.537430955458926e-06,
+      "loss": 0.6743,
+      "step": 35410
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5227406275708077e-06,
+      "loss": 0.6188,
+      "step": 35420
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5080502996826892e-06,
+      "loss": 0.6369,
+      "step": 35430
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.4933599717945707e-06,
+      "loss": 0.6582,
+      "step": 35440
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.4786696439064522e-06,
+      "loss": 0.5874,
+      "step": 35450
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.4639793160183337e-06,
+      "loss": 0.6978,
+      "step": 35460
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.449288988130215e-06,
+      "loss": 0.64,
+      "step": 35470
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.434598660242097e-06,
+      "loss": 0.6687,
+      "step": 35480
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.4199083323539787e-06,
+      "loss": 0.6156,
+      "step": 35490
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.4052180044658598e-06,
+      "loss": 0.668,
+      "step": 35500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3905276765777413e-06,
+      "loss": 0.6322,
+      "step": 35510
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3758373486896228e-06,
+      "loss": 0.6544,
+      "step": 35520
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3611470208015043e-06,
+      "loss": 0.6537,
+      "step": 35530
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.346456692913386e-06,
+      "loss": 0.6418,
+      "step": 35540
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3317663650252677e-06,
+      "loss": 0.6286,
+      "step": 35550
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3170760371371492e-06,
+      "loss": 0.6902,
+      "step": 35560
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3023857092490307e-06,
+      "loss": 0.6274,
+      "step": 35570
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.287695381360912e-06,
+      "loss": 0.6194,
+      "step": 35580
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.2730050534727933e-06,
+      "loss": 0.6289,
+      "step": 35590
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.2583147255846757e-06,
+      "loss": 0.632,
+      "step": 35600
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.2436243976965567e-06,
+      "loss": 0.6616,
+      "step": 35610
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.2289340698084382e-06,
+      "loss": 0.6317,
+      "step": 35620
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.2142437419203198e-06,
+      "loss": 0.6534,
+      "step": 35630
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.1995534140322013e-06,
+      "loss": 0.6746,
+      "step": 35640
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.1848630861440828e-06,
+      "loss": 0.5763,
+      "step": 35650
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.1701727582559647e-06,
+      "loss": 0.6306,
+      "step": 35660
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.155482430367846e-06,
+      "loss": 0.6433,
+      "step": 35670
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.1407921024797277e-06,
+      "loss": 0.6521,
+      "step": 35680
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.1261017745916088e-06,
+      "loss": 0.5778,
+      "step": 35690
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.1114114467034907e-06,
+      "loss": 0.6644,
+      "step": 35700
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.0967211188153722e-06,
+      "loss": 0.6322,
+      "step": 35710
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.0820307909272537e-06,
+      "loss": 0.6784,
+      "step": 35720
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.0673404630391352e-06,
+      "loss": 0.5989,
+      "step": 35730
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.0526501351510167e-06,
+      "loss": 0.6556,
+      "step": 35740
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.0379598072628982e-06,
+      "loss": 0.653,
+      "step": 35750
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.0232694793747797e-06,
+      "loss": 0.6589,
+      "step": 35760
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.0085791514866612e-06,
+      "loss": 0.625,
+      "step": 35770
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.9938888235985428e-06,
+      "loss": 0.5994,
+      "step": 35780
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.9791984957104247e-06,
+      "loss": 0.6294,
+      "step": 35790
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.9645081678223058e-06,
+      "loss": 0.6076,
+      "step": 35800
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.9498178399341873e-06,
+      "loss": 0.6497,
+      "step": 35810
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.935127512046069e-06,
+      "loss": 0.6397,
+      "step": 35820
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.9204371841579507e-06,
+      "loss": 0.6549,
+      "step": 35830
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.9057468562698318e-06,
+      "loss": 0.6228,
+      "step": 35840
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8910565283817137e-06,
+      "loss": 0.6329,
+      "step": 35850
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.8763662004935952e-06,
+      "loss": 0.6337,
+      "step": 35860
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.8616758726054767e-06,
+      "loss": 0.6505,
+      "step": 35870
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.8469855447173582e-06,
+      "loss": 0.6496,
+      "step": 35880
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.8322952168292397e-06,
+      "loss": 0.6378,
+      "step": 35890
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.8176048889411212e-06,
+      "loss": 0.7137,
+      "step": 35900
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.8029145610530027e-06,
+      "loss": 0.6609,
+      "step": 35910
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.7882242331648842e-06,
+      "loss": 0.647,
+      "step": 35920
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.7735339052767658e-06,
+      "loss": 0.676,
+      "step": 35930
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.7588435773886477e-06,
+      "loss": 0.6353,
+      "step": 35940
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.7441532495005288e-06,
+      "loss": 0.6195,
+      "step": 35950
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.7294629216124107e-06,
+      "loss": 0.632,
+      "step": 35960
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.714772593724292e-06,
+      "loss": 0.6287,
+      "step": 35970
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.7000822658361737e-06,
+      "loss": 0.6762,
+      "step": 35980
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.685391937948055e-06,
+      "loss": 0.6911,
+      "step": 35990
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.6707016100599367e-06,
+      "loss": 0.624,
+      "step": 36000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.6560112821718182e-06,
+      "loss": 0.65,
+      "step": 36010
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.6413209542836997e-06,
+      "loss": 0.6169,
+      "step": 36020
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.6266306263955812e-06,
+      "loss": 0.656,
+      "step": 36030
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.6119402985074627e-06,
+      "loss": 0.6295,
+      "step": 36040
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.5972499706193447e-06,
+      "loss": 0.6621,
+      "step": 36050
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.5825596427312257e-06,
+      "loss": 0.617,
+      "step": 36060
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.5678693148431072e-06,
+      "loss": 0.698,
+      "step": 36070
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.553178986954989e-06,
+      "loss": 0.6489,
+      "step": 36080
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.5384886590668707e-06,
+      "loss": 0.6289,
+      "step": 36090
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.5237983311787518e-06,
+      "loss": 0.6672,
+      "step": 36100
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.5091080032906337e-06,
+      "loss": 0.5994,
+      "step": 36110
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.494417675402515e-06,
+      "loss": 0.6722,
+      "step": 36120
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.4797273475143967e-06,
+      "loss": 0.6685,
+      "step": 36130
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.465037019626278e-06,
+      "loss": 0.6753,
+      "step": 36140
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.4503466917381597e-06,
+      "loss": 0.666,
+      "step": 36150
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.4356563638500412e-06,
+      "loss": 0.6345,
+      "step": 36160
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.4209660359619227e-06,
+      "loss": 0.6519,
+      "step": 36170
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.4062757080738042e-06,
+      "loss": 0.6676,
+      "step": 36180
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3915853801856857e-06,
+      "loss": 0.6388,
+      "step": 36190
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3768950522975677e-06,
+      "loss": 0.5982,
+      "step": 36200
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3622047244094487e-06,
+      "loss": 0.6267,
+      "step": 36210
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3475143965213302e-06,
+      "loss": 0.695,
+      "step": 36220
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.332824068633212e-06,
+      "loss": 0.6378,
+      "step": 36230
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3181337407450937e-06,
+      "loss": 0.5946,
+      "step": 36240
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.3034434128569748e-06,
+      "loss": 0.6207,
+      "step": 36250
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.2887530849688567e-06,
+      "loss": 0.6639,
+      "step": 36260
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.274062757080738e-06,
+      "loss": 0.6364,
+      "step": 36270
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.2593724291926197e-06,
+      "loss": 0.6527,
+      "step": 36280
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.244682101304501e-06,
+      "loss": 0.6137,
+      "step": 36290
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.2299917734163827e-06,
+      "loss": 0.7045,
+      "step": 36300
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.2153014455282646e-06,
+      "loss": 0.6114,
+      "step": 36310
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.2006111176401457e-06,
+      "loss": 0.6264,
+      "step": 36320
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1859207897520272e-06,
+      "loss": 0.6379,
+      "step": 36330
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.171230461863909e-06,
+      "loss": 0.6203,
+      "step": 36340
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1565401339757907e-06,
+      "loss": 0.6345,
+      "step": 36350
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1418498060876717e-06,
+      "loss": 0.6106,
+      "step": 36360
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1271594781995537e-06,
+      "loss": 0.6274,
+      "step": 36370
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.112469150311435e-06,
+      "loss": 0.6737,
+      "step": 36380
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.0977788224233167e-06,
+      "loss": 0.6663,
+      "step": 36390
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.083088494535198e-06,
+      "loss": 0.6485,
+      "step": 36400
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.0683981666470797e-06,
+      "loss": 0.5949,
+      "step": 36410
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.053707838758961e-06,
+      "loss": 0.6312,
+      "step": 36420
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.0390175108708427e-06,
+      "loss": 0.6427,
+      "step": 36430
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.024327182982724e-06,
+      "loss": 0.594,
+      "step": 36440
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.0096368550946057e-06,
+      "loss": 0.6333,
+      "step": 36450
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9949465272064876e-06,
+      "loss": 0.6306,
+      "step": 36460
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9802561993183687e-06,
+      "loss": 0.6671,
+      "step": 36470
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9655658714302502e-06,
+      "loss": 0.6135,
+      "step": 36480
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.950875543542132e-06,
+      "loss": 0.632,
+      "step": 36490
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9361852156540137e-06,
+      "loss": 0.6223,
+      "step": 36500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9214948877658947e-06,
+      "loss": 0.6343,
+      "step": 36510
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9068045598777767e-06,
+      "loss": 0.7089,
+      "step": 36520
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8921142319896582e-06,
+      "loss": 0.6505,
+      "step": 36530
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8774239041015395e-06,
+      "loss": 0.6863,
+      "step": 36540
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8627335762134212e-06,
+      "loss": 0.6331,
+      "step": 36550
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8480432483253027e-06,
+      "loss": 0.6516,
+      "step": 36560
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8333529204371842e-06,
+      "loss": 0.6377,
+      "step": 36570
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.818662592549066e-06,
+      "loss": 0.6182,
+      "step": 36580
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8039722646609472e-06,
+      "loss": 0.648,
+      "step": 36590
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7892819367728287e-06,
+      "loss": 0.5884,
+      "step": 36600
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7745916088847104e-06,
+      "loss": 0.6054,
+      "step": 36610
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.759901280996592e-06,
+      "loss": 0.7095,
+      "step": 36620
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.7452109531084732e-06,
+      "loss": 0.6666,
+      "step": 36630
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.7305206252203551e-06,
+      "loss": 0.6171,
+      "step": 36640
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.7158302973322364e-06,
+      "loss": 0.6662,
+      "step": 36650
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.7011399694441182e-06,
+      "loss": 0.6646,
+      "step": 36660
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6864496415559997e-06,
+      "loss": 0.6447,
+      "step": 36670
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6717593136678812e-06,
+      "loss": 0.6541,
+      "step": 36680
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6570689857797629e-06,
+      "loss": 0.6096,
+      "step": 36690
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6423786578916442e-06,
+      "loss": 0.6875,
+      "step": 36700
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6276883300035257e-06,
+      "loss": 0.6588,
+      "step": 36710
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6129980021154074e-06,
+      "loss": 0.6692,
+      "step": 36720
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.598307674227289e-06,
+      "loss": 0.6738,
+      "step": 36730
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5836173463391702e-06,
+      "loss": 0.6533,
+      "step": 36740
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5689270184510521e-06,
+      "loss": 0.6265,
+      "step": 36750
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5542366905629334e-06,
+      "loss": 0.6731,
+      "step": 36760
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5395463626748151e-06,
+      "loss": 0.6181,
+      "step": 36770
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5248560347866964e-06,
+      "loss": 0.6302,
+      "step": 36780
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5101657068985781e-06,
+      "loss": 0.638,
+      "step": 36790
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4954753790104597e-06,
+      "loss": 0.622,
+      "step": 36800
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4807850511223412e-06,
+      "loss": 0.6894,
+      "step": 36810
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4660947232342227e-06,
+      "loss": 0.6106,
+      "step": 36820
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4514043953461042e-06,
+      "loss": 0.6447,
+      "step": 36830
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4367140674579857e-06,
+      "loss": 0.5941,
+      "step": 36840
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4220237395698672e-06,
+      "loss": 0.6446,
+      "step": 36850
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.407333411681749e-06,
+      "loss": 0.6791,
+      "step": 36860
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3926430837936302e-06,
+      "loss": 0.6278,
+      "step": 36870
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.377952755905512e-06,
+      "loss": 0.6626,
+      "step": 36880
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.3632624280173934e-06,
+      "loss": 0.6599,
+      "step": 36890
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.348572100129275e-06,
+      "loss": 0.6196,
+      "step": 36900
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.3338817722411564e-06,
+      "loss": 0.5903,
+      "step": 36910
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.3191914443530381e-06,
+      "loss": 0.7073,
+      "step": 36920
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.3045011164649196e-06,
+      "loss": 0.6723,
+      "step": 36930
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2898107885768011e-06,
+      "loss": 0.6363,
+      "step": 36940
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2751204606886827e-06,
+      "loss": 0.6356,
+      "step": 36950
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2604301328005642e-06,
+      "loss": 0.621,
+      "step": 36960
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2457398049124457e-06,
+      "loss": 0.6361,
+      "step": 36970
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2310494770243272e-06,
+      "loss": 0.6162,
+      "step": 36980
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2163591491362089e-06,
+      "loss": 0.6613,
+      "step": 36990
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.2016688212480902e-06,
+      "loss": 0.5994,
+      "step": 37000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.186978493359972e-06,
+      "loss": 0.5589,
+      "step": 37010
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.1722881654718534e-06,
+      "loss": 0.6388,
+      "step": 37020
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.157597837583735e-06,
+      "loss": 0.6607,
+      "step": 37030
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.1429075096956164e-06,
+      "loss": 0.5519,
+      "step": 37040
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.1282171818074981e-06,
+      "loss": 0.6297,
+      "step": 37050
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.1135268539193794e-06,
+      "loss": 0.6206,
+      "step": 37060
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0988365260312611e-06,
+      "loss": 0.6372,
+      "step": 37070
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0841461981431426e-06,
+      "loss": 0.5882,
+      "step": 37080
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0694558702550241e-06,
+      "loss": 0.7245,
+      "step": 37090
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0547655423669057e-06,
+      "loss": 0.6804,
+      "step": 37100
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0400752144787872e-06,
+      "loss": 0.6254,
+      "step": 37110
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0253848865906689e-06,
+      "loss": 0.6484,
+      "step": 37120
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0106945587025502e-06,
+      "loss": 0.6792,
+      "step": 37130
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.960042308144319e-07,
+      "loss": 0.5987,
+      "step": 37140
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.813139029263134e-07,
+      "loss": 0.607,
+      "step": 37150
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.666235750381949e-07,
+      "loss": 0.6681,
+      "step": 37160
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.519332471500764e-07,
+      "loss": 0.5945,
+      "step": 37170
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.37242919261958e-07,
+      "loss": 0.6401,
+      "step": 37180
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.225525913738394e-07,
+      "loss": 0.6585,
+      "step": 37190
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.07862263485721e-07,
+      "loss": 0.6845,
+      "step": 37200
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.931719355976026e-07,
+      "loss": 0.6181,
+      "step": 37210
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.78481607709484e-07,
+      "loss": 0.6262,
+      "step": 37220
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.637912798213656e-07,
+      "loss": 0.6708,
+      "step": 37230
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.491009519332473e-07,
+      "loss": 0.6327,
+      "step": 37240
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.344106240451287e-07,
+      "loss": 0.6463,
+      "step": 37250
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.197202961570103e-07,
+      "loss": 0.6703,
+      "step": 37260
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.050299682688919e-07,
+      "loss": 0.6312,
+      "step": 37270
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.903396403807734e-07,
+      "loss": 0.617,
+      "step": 37280
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.756493124926549e-07,
+      "loss": 0.6226,
+      "step": 37290
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.609589846045364e-07,
+      "loss": 0.6339,
+      "step": 37300
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.462686567164179e-07,
+      "loss": 0.606,
+      "step": 37310
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.315783288282995e-07,
+      "loss": 0.6789,
+      "step": 37320
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.16888000940181e-07,
+      "loss": 0.6131,
+      "step": 37330
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.021976730520625e-07,
+      "loss": 0.6702,
+      "step": 37340
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.875073451639441e-07,
+      "loss": 0.5906,
+      "step": 37350
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.728170172758256e-07,
+      "loss": 0.6364,
+      "step": 37360
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.581266893877071e-07,
+      "loss": 0.6156,
+      "step": 37370
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.434363614995887e-07,
+      "loss": 0.586,
+      "step": 37380
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.287460336114703e-07,
+      "loss": 0.6294,
+      "step": 37390
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.140557057233519e-07,
+      "loss": 0.6601,
+      "step": 37400
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.993653778352334e-07,
+      "loss": 0.6303,
+      "step": 37410
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.846750499471149e-07,
+      "loss": 0.6055,
+      "step": 37420
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.699847220589964e-07,
+      "loss": 0.6239,
+      "step": 37430
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.552943941708779e-07,
+      "loss": 0.6571,
+      "step": 37440
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.406040662827594e-07,
+      "loss": 0.6228,
+      "step": 37450
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.25913738394641e-07,
+      "loss": 0.6136,
+      "step": 37460
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.112234105065225e-07,
+      "loss": 0.6361,
+      "step": 37470
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.96533082618404e-07,
+      "loss": 0.6663,
+      "step": 37480
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.818427547302856e-07,
+      "loss": 0.6748,
+      "step": 37490
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.671524268421671e-07,
+      "loss": 0.5639,
+      "step": 37500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.5246209895404873e-07,
+      "loss": 0.623,
+      "step": 37510
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.3777177106593024e-07,
+      "loss": 0.6324,
+      "step": 37520
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.2308144317781174e-07,
+      "loss": 0.6734,
+      "step": 37530
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.083911152896933e-07,
+      "loss": 0.6736,
+      "step": 37540
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.937007874015748e-07,
+      "loss": 0.6828,
+      "step": 37550
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.7901045951345637e-07,
+      "loss": 0.5967,
+      "step": 37560
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.643201316253379e-07,
+      "loss": 0.6455,
+      "step": 37570
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.4962980373721943e-07,
+      "loss": 0.6171,
+      "step": 37580
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.34939475849101e-07,
+      "loss": 0.6688,
+      "step": 37590
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.202491479609825e-07,
+      "loss": 0.591,
+      "step": 37600
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0555882007286405e-07,
+      "loss": 0.6656,
+      "step": 37610
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9086849218474555e-07,
+      "loss": 0.6576,
+      "step": 37620
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.761781642966271e-07,
+      "loss": 0.6347,
+      "step": 37630
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.6148783640850867e-07,
+      "loss": 0.6317,
+      "step": 37640
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.4679750852039023e-07,
+      "loss": 0.5829,
+      "step": 37650
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.321071806322717e-07,
+      "loss": 0.6135,
+      "step": 37660
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1741685274415326e-07,
+      "loss": 0.7097,
+      "step": 37670
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.027265248560348e-07,
+      "loss": 0.6206,
+      "step": 37680
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8803619696791633e-07,
+      "loss": 0.6686,
+      "step": 37690
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7334586907979786e-07,
+      "loss": 0.624,
+      "step": 37700
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5865554119167942e-07,
+      "loss": 0.6411,
+      "step": 37710
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4396521330356092e-07,
+      "loss": 0.6251,
+      "step": 37720
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2927488541544248e-07,
+      "loss": 0.6162,
+      "step": 37730
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1458455752732401e-07,
+      "loss": 0.648,
+      "step": 37740
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.989422963920556e-08,
+      "loss": 0.6385,
+      "step": 37750
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.520390175108709e-08,
+      "loss": 0.6032,
+      "step": 37760
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.051357386296863e-08,
+      "loss": 0.6015,
+      "step": 37770
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.5823245974850164e-08,
+      "loss": 0.6486,
+      "step": 37780
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.1132918086731696e-08,
+      "loss": 0.6525,
+      "step": 37790
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.6442590198613234e-08,
+      "loss": 0.6148,
+      "step": 37800
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.175226231049477e-08,
+      "loss": 0.6174,
+      "step": 37810
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8097558194444616,
+      "eval_loss": 0.7372148036956787,
+      "eval_runtime": 1971.7641,
+      "eval_samples_per_second": 409.187,
+      "eval_steps_per_second": 6.394,
+      "step": 37818
+    },
+    {
+      "epoch": 3.0,
+      "step": 37818,
+      "total_flos": 7.523739101107678e+20,
+      "train_loss": 0.47222403545925346,
+      "train_runtime": 46332.5983,
+      "train_samples_per_second": 208.964,
+      "train_steps_per_second": 0.816
+    }
+  ],
+  "max_steps": 37818,
+  "num_train_epochs": 3,
+  "total_flos": 7.523739101107678e+20,
+  "trial_name": null,
+  "trial_params": null
+}