diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12316 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9863831015734014,
+  "global_step": 20500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.9826,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.9714,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 1.9375,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.9102,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 1.8303,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 1.7734,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 1.5721,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.4128,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 1.3727,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 1.3457,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.3685,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2e-05,
+      "loss": 1.326,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.2216,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 1.3257,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-05,
+      "loss": 1.402,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.2708,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 1.1622,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-05,
+      "loss": 0.9477,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9e-05,
+      "loss": 0.959,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.9508,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1e-05,
+      "loss": 0.8677,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.9072,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 1.0147,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-05,
+      "loss": 0.8079,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 1.1345,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.8976,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.8592,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.851,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9e-05,
+      "loss": 0.9462,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 0.9441,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1e-05,
+      "loss": 0.7899,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.7144,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3e-05,
+      "loss": 0.7234,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.694,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-05,
+      "loss": 0.7667,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6e-05,
+      "loss": 0.6767,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7e-05,
+      "loss": 0.6708,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8e-05,
+      "loss": 0.5388,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 0.6132,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4e-05,
+      "loss": 0.6155,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1e-05,
+      "loss": 0.6737,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2e-05,
+      "loss": 0.6481,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3e-05,
+      "loss": 0.6223,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.7806,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5e-05,
+      "loss": 0.7097,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.7466,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7e-05,
+      "loss": 0.7344,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8e-05,
+      "loss": 0.6603,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9e-05,
+      "loss": 0.6857,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 0.7296,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997534881427797e-05,
+      "loss": 0.5692,
+      "step": 510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.995069762855594e-05,
+      "loss": 0.6488,
+      "step": 520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99260464428339e-05,
+      "loss": 0.5765,
+      "step": 530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990139525711187e-05,
+      "loss": 0.7368,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987674407138984e-05,
+      "loss": 0.8075,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98520928856678e-05,
+      "loss": 0.5601,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.982744169994577e-05,
+      "loss": 0.5784,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.980279051422374e-05,
+      "loss": 0.6215,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.97781393285017e-05,
+      "loss": 0.831,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.975348814277967e-05,
+      "loss": 0.5719,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.972883695705764e-05,
+      "loss": 0.4784,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.97041857713356e-05,
+      "loss": 0.5957,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9679534585613566e-05,
+      "loss": 0.7973,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.965488339989154e-05,
+      "loss": 0.9234,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.96302322141695e-05,
+      "loss": 0.5318,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960558102844747e-05,
+      "loss": 0.8011,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.958092984272544e-05,
+      "loss": 0.6293,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95562786570034e-05,
+      "loss": 0.5486,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.953162747128137e-05,
+      "loss": 0.6163,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9506976285559336e-05,
+      "loss": 0.6013,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.948232509983731e-05,
+      "loss": 0.3341,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.945767391411527e-05,
+      "loss": 0.4713,
+      "step": 720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9433022728393236e-05,
+      "loss": 0.622,
+      "step": 730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.940837154267121e-05,
+      "loss": 0.2656,
+      "step": 740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938372035694917e-05,
+      "loss": 0.9112,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.935906917122714e-05,
+      "loss": 0.7053,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.933441798550511e-05,
+      "loss": 0.5905,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930976679978307e-05,
+      "loss": 0.5148,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.928511561406104e-05,
+      "loss": 0.6263,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.926046442833901e-05,
+      "loss": 0.5656,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.923581324261697e-05,
+      "loss": 0.6455,
+      "step": 810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9211162056894936e-05,
+      "loss": 0.4833,
+      "step": 820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.918651087117291e-05,
+      "loss": 0.769,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.916185968545087e-05,
+      "loss": 0.4158,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9137208499728835e-05,
+      "loss": 0.6367,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9112557314006806e-05,
+      "loss": 0.7448,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.908790612828477e-05,
+      "loss": 0.75,
+      "step": 870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.906325494256274e-05,
+      "loss": 0.4781,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9038603756840706e-05,
+      "loss": 0.403,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.901395257111867e-05,
+      "loss": 0.8523,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.898930138539664e-05,
+      "loss": 0.5496,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8964650199674606e-05,
+      "loss": 0.5298,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.893999901395258e-05,
+      "loss": 0.6546,
+      "step": 930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.891534782823054e-05,
+      "loss": 0.4861,
+      "step": 940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8890696642508506e-05,
+      "loss": 0.7111,
+      "step": 950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.886604545678648e-05,
+      "loss": 0.27,
+      "step": 960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.884139427106444e-05,
+      "loss": 0.8138,
+      "step": 970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.881674308534241e-05,
+      "loss": 0.4683,
+      "step": 980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8792091899620377e-05,
+      "loss": 0.6167,
+      "step": 990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.876744071389834e-05,
+      "loss": 0.5663,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8742789528176305e-05,
+      "loss": 0.5934,
+      "step": 1010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.871813834245427e-05,
+      "loss": 0.6321,
+      "step": 1020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.869348715673224e-05,
+      "loss": 0.6397,
+      "step": 1030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8668835971010205e-05,
+      "loss": 0.5166,
+      "step": 1040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8644184785288176e-05,
+      "loss": 0.6575,
+      "step": 1050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.861953359956614e-05,
+      "loss": 0.5901,
+      "step": 1060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8594882413844105e-05,
+      "loss": 0.5692,
+      "step": 1070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8570231228122076e-05,
+      "loss": 0.5039,
+      "step": 1080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.854558004240004e-05,
+      "loss": 0.5454,
+      "step": 1090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.852092885667801e-05,
+      "loss": 0.6051,
+      "step": 1100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8496277670955976e-05,
+      "loss": 0.4466,
+      "step": 1110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.847162648523394e-05,
+      "loss": 0.4213,
+      "step": 1120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.844697529951191e-05,
+      "loss": 0.6353,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8422324113789875e-05,
+      "loss": 0.4687,
+      "step": 1140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8397672928067847e-05,
+      "loss": 0.5304,
+      "step": 1150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.837302174234581e-05,
+      "loss": 0.5071,
+      "step": 1160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8348370556623775e-05,
+      "loss": 0.6497,
+      "step": 1170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8323719370901746e-05,
+      "loss": 0.7002,
+      "step": 1180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.829906818517971e-05,
+      "loss": 0.5467,
+      "step": 1190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8274416999457675e-05,
+      "loss": 0.4599,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.824976581373564e-05,
+      "loss": 0.4324,
+      "step": 1210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.822511462801361e-05,
+      "loss": 0.3044,
+      "step": 1220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8200463442291575e-05,
+      "loss": 0.4284,
+      "step": 1230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.817581225656954e-05,
+      "loss": 0.468,
+      "step": 1240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.815116107084751e-05,
+      "loss": 0.6737,
+      "step": 1250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8126509885125474e-05,
+      "loss": 0.5412,
+      "step": 1260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8101858699403446e-05,
+      "loss": 0.6565,
+      "step": 1270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.807720751368141e-05,
+      "loss": 0.7632,
+      "step": 1280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8052556327959374e-05,
+      "loss": 0.4942,
+      "step": 1290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8027905142237345e-05,
+      "loss": 0.6173,
+      "step": 1300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.800325395651531e-05,
+      "loss": 0.4048,
+      "step": 1310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.797860277079328e-05,
+      "loss": 0.65,
+      "step": 1320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7953951585071245e-05,
+      "loss": 0.6641,
+      "step": 1330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.792930039934921e-05,
+      "loss": 0.6748,
+      "step": 1340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.790464921362718e-05,
+      "loss": 0.5054,
+      "step": 1350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7879998027905145e-05,
+      "loss": 0.4246,
+      "step": 1360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7855346842183116e-05,
+      "loss": 0.4134,
+      "step": 1370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.783069565646108e-05,
+      "loss": 0.7633,
+      "step": 1380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7806044470739045e-05,
+      "loss": 0.3475,
+      "step": 1390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.778139328501701e-05,
+      "loss": 0.3888,
+      "step": 1400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.775674209929497e-05,
+      "loss": 0.6708,
+      "step": 1410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7732090913572944e-05,
+      "loss": 0.3529,
+      "step": 1420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.770743972785091e-05,
+      "loss": 0.4623,
+      "step": 1430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.768278854212888e-05,
+      "loss": 0.5203,
+      "step": 1440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7658137356406844e-05,
+      "loss": 0.579,
+      "step": 1450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.763348617068481e-05,
+      "loss": 0.6022,
+      "step": 1460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.760883498496278e-05,
+      "loss": 0.3136,
+      "step": 1470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7584183799240744e-05,
+      "loss": 0.4059,
+      "step": 1480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7559532613518715e-05,
+      "loss": 0.4958,
+      "step": 1490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.753488142779668e-05,
+      "loss": 0.4217,
+      "step": 1500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7510230242074644e-05,
+      "loss": 0.6464,
+      "step": 1510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7485579056352615e-05,
+      "loss": 0.5194,
+      "step": 1520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.746092787063058e-05,
+      "loss": 0.6801,
+      "step": 1530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.743627668490855e-05,
+      "loss": 0.6597,
+      "step": 1540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.7411625499186515e-05,
+      "loss": 0.6397,
+      "step": 1550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.738697431346448e-05,
+      "loss": 0.3929,
+      "step": 1560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.736232312774245e-05,
+      "loss": 0.4262,
+      "step": 1570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7337671942020414e-05,
+      "loss": 0.2192,
+      "step": 1580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.731302075629838e-05,
+      "loss": 0.4398,
+      "step": 1590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.728836957057634e-05,
+      "loss": 0.6187,
+      "step": 1600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7263718384854314e-05,
+      "loss": 0.6014,
+      "step": 1610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.723906719913228e-05,
+      "loss": 0.5088,
+      "step": 1620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.721441601341024e-05,
+      "loss": 0.6424,
+      "step": 1630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7189764827688214e-05,
+      "loss": 0.5547,
+      "step": 1640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.716511364196618e-05,
+      "loss": 0.664,
+      "step": 1650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.714046245624415e-05,
+      "loss": 0.5845,
+      "step": 1660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7115811270522114e-05,
+      "loss": 0.511,
+      "step": 1670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.709116008480008e-05,
+      "loss": 0.5204,
+      "step": 1680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.706650889907805e-05,
+      "loss": 0.4804,
+      "step": 1690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7041857713356013e-05,
+      "loss": 0.5225,
+      "step": 1700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7017206527633985e-05,
+      "loss": 0.2682,
+      "step": 1710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.699255534191195e-05,
+      "loss": 0.3985,
+      "step": 1720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.696790415618991e-05,
+      "loss": 0.609,
+      "step": 1730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6943252970467884e-05,
+      "loss": 0.6113,
+      "step": 1740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.691860178474585e-05,
+      "loss": 0.5012,
+      "step": 1750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.689395059902382e-05,
+      "loss": 0.3185,
+      "step": 1760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6869299413301784e-05,
+      "loss": 0.5255,
+      "step": 1770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.684464822757975e-05,
+      "loss": 0.6918,
+      "step": 1780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.681999704185771e-05,
+      "loss": 0.6319,
+      "step": 1790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6795345856135684e-05,
+      "loss": 0.4744,
+      "step": 1800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.677069467041365e-05,
+      "loss": 0.473,
+      "step": 1810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.674604348469161e-05,
+      "loss": 0.3888,
+      "step": 1820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6721392298969584e-05,
+      "loss": 0.2423,
+      "step": 1830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.669674111324755e-05,
+      "loss": 0.4363,
+      "step": 1840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.667208992752551e-05,
+      "loss": 0.3522,
+      "step": 1850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.664743874180348e-05,
+      "loss": 0.4111,
+      "step": 1860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.662278755608145e-05,
+      "loss": 0.6048,
+      "step": 1870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.659813637035942e-05,
+      "loss": 0.4193,
+      "step": 1880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.657348518463738e-05,
+      "loss": 0.371,
+      "step": 1890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.654883399891535e-05,
+      "loss": 0.5344,
+      "step": 1900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.652418281319332e-05,
+      "loss": 0.4874,
+      "step": 1910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.649953162747128e-05,
+      "loss": 0.8494,
+      "step": 1920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6474880441749254e-05,
+      "loss": 0.4666,
+      "step": 1930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.645022925602722e-05,
+      "loss": 0.6549,
+      "step": 1940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.642557807030518e-05,
+      "loss": 0.6006,
+      "step": 1950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6400926884583154e-05,
+      "loss": 0.4878,
+      "step": 1960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.637627569886112e-05,
+      "loss": 0.7114,
+      "step": 1970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.635162451313908e-05,
+      "loss": 0.4272,
+      "step": 1980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6326973327417054e-05,
+      "loss": 0.5089,
+      "step": 1990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.630232214169502e-05,
+      "loss": 0.5786,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.627767095597298e-05,
+      "loss": 0.5152,
+      "step": 2010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.6253019770250947e-05,
+      "loss": 0.7612,
+      "step": 2020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.622836858452892e-05,
+      "loss": 0.5073,
+      "step": 2030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.620371739880688e-05,
+      "loss": 0.614,
+      "step": 2040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.617906621308485e-05,
+      "loss": 0.3983,
+      "step": 2050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.615441502736282e-05,
+      "loss": 0.4756,
+      "step": 2060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.612976384164078e-05,
+      "loss": 0.3861,
+      "step": 2070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.610511265591875e-05,
+      "loss": 0.4365,
+      "step": 2080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.608046147019672e-05,
+      "loss": 0.3928,
+      "step": 2090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.605581028447469e-05,
+      "loss": 0.5424,
+      "step": 2100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.603115909875265e-05,
+      "loss": 0.5624,
+      "step": 2110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.600650791303062e-05,
+      "loss": 0.5893,
+      "step": 2120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.598185672730859e-05,
+      "loss": 0.6344,
+      "step": 2130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.595720554158655e-05,
+      "loss": 0.3735,
+      "step": 2140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5932554355864524e-05,
+      "loss": 0.5911,
+      "step": 2150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.590790317014249e-05,
+      "loss": 0.4365,
+      "step": 2160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.588325198442046e-05,
+      "loss": 0.4161,
+      "step": 2170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.585860079869842e-05,
+      "loss": 0.5741,
+      "step": 2180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.583394961297639e-05,
+      "loss": 0.4861,
+      "step": 2190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.580929842725435e-05,
+      "loss": 0.433,
+      "step": 2200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.5784647241532316e-05,
+      "loss": 0.4433,
+      "step": 2210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.575999605581029e-05,
+      "loss": 0.3068,
+      "step": 2220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.573534487008825e-05,
+      "loss": 0.5113,
+      "step": 2230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.5710693684366216e-05,
+      "loss": 0.3644,
+      "step": 2240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.568604249864419e-05,
+      "loss": 0.3181,
+      "step": 2250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.566139131292215e-05,
+      "loss": 0.4639,
+      "step": 2260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.563674012720012e-05,
+      "loss": 0.2935,
+      "step": 2270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.561208894147809e-05,
+      "loss": 0.4554,
+      "step": 2280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.558743775575605e-05,
+      "loss": 0.4376,
+      "step": 2290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.556278657003402e-05,
+      "loss": 0.4158,
+      "step": 2300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.553813538431199e-05,
+      "loss": 0.5657,
+      "step": 2310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.551348419858996e-05,
+      "loss": 0.6733,
+      "step": 2320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.548883301286792e-05,
+      "loss": 0.458,
+      "step": 2330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.5464181827145886e-05,
+      "loss": 0.3889,
+      "step": 2340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.543953064142386e-05,
+      "loss": 0.6164,
+      "step": 2350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.541487945570182e-05,
+      "loss": 0.4811,
+      "step": 2360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.539022826997979e-05,
+      "loss": 0.4571,
+      "step": 2370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.536557708425776e-05,
+      "loss": 0.4509,
+      "step": 2380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.534092589853572e-05,
+      "loss": 0.4768,
+      "step": 2390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5316274712813686e-05,
+      "loss": 0.438,
+      "step": 2400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.529162352709165e-05,
+      "loss": 0.5153,
+      "step": 2410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.526697234136962e-05,
+      "loss": 0.5627,
+      "step": 2420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5242321155647586e-05,
+      "loss": 0.4896,
+      "step": 2430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.521766996992556e-05,
+      "loss": 0.6186,
+      "step": 2440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.519301878420352e-05,
+      "loss": 0.2618,
+      "step": 2450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5168367598481486e-05,
+      "loss": 0.7125,
+      "step": 2460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.514371641275946e-05,
+      "loss": 0.5213,
+      "step": 2470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.511906522703742e-05,
+      "loss": 0.3697,
+      "step": 2480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.509441404131539e-05,
+      "loss": 0.3848,
+      "step": 2490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5069762855593356e-05,
+      "loss": 0.4557,
+      "step": 2500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.504511166987132e-05,
+      "loss": 0.3443,
+      "step": 2510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.502046048414929e-05,
+      "loss": 0.4183,
+      "step": 2520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4995809298427256e-05,
+      "loss": 0.3765,
+      "step": 2530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.497115811270523e-05,
+      "loss": 0.6598,
+      "step": 2540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.494650692698319e-05,
+      "loss": 0.501,
+      "step": 2550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.492185574126116e-05,
+      "loss": 0.5555,
+      "step": 2560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.489720455553913e-05,
+      "loss": 0.4344,
+      "step": 2570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.487255336981709e-05,
+      "loss": 0.6524,
+      "step": 2580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4847902184095056e-05,
+      "loss": 0.2934,
+      "step": 2590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.482325099837302e-05,
+      "loss": 0.4993,
+      "step": 2600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.479859981265099e-05,
+      "loss": 0.6525,
+      "step": 2610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4773948626928955e-05,
+      "loss": 0.4719,
+      "step": 2620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.474929744120692e-05,
+      "loss": 0.7409,
+      "step": 2630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.472464625548489e-05,
+      "loss": 0.3993,
+      "step": 2640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4699995069762855e-05,
+      "loss": 0.5846,
+      "step": 2650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4675343884040826e-05,
+      "loss": 0.5105,
+      "step": 2660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.465069269831879e-05,
+      "loss": 0.4602,
+      "step": 2670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4626041512596755e-05,
+      "loss": 0.5399,
+      "step": 2680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4601390326874726e-05,
+      "loss": 0.3792,
+      "step": 2690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.457673914115269e-05,
+      "loss": 0.327,
+      "step": 2700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.455208795543066e-05,
+      "loss": 0.5713,
+      "step": 2710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4527436769708626e-05,
+      "loss": 0.5561,
+      "step": 2720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.450278558398659e-05,
+      "loss": 0.576,
+      "step": 2730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.447813439826456e-05,
+      "loss": 0.1666,
+      "step": 2740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4453483212542526e-05,
+      "loss": 0.5105,
+      "step": 2750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.44288320268205e-05,
+      "loss": 0.5274,
+      "step": 2760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.440418084109846e-05,
+      "loss": 0.3234,
+      "step": 2770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4379529655376425e-05,
+      "loss": 0.5895,
+      "step": 2780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.435487846965439e-05,
+      "loss": 0.5511,
+      "step": 2790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4330227283932354e-05,
+      "loss": 0.7844,
+      "step": 2800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.4305576098210325e-05,
+      "loss": 0.2844,
+      "step": 2810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.428092491248829e-05,
+      "loss": 0.3243,
+      "step": 2820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.425627372676626e-05,
+      "loss": 0.3819,
+      "step": 2830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.4231622541044225e-05,
+      "loss": 0.422,
+      "step": 2840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.420697135532219e-05,
+      "loss": 0.4656,
+      "step": 2850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.418232016960016e-05,
+      "loss": 0.3973,
+      "step": 2860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.4157668983878125e-05,
+      "loss": 0.3736,
+      "step": 2870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.4133017798156096e-05,
+      "loss": 0.484,
+      "step": 2880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.410836661243406e-05,
+      "loss": 0.4492,
+      "step": 2890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.4083715426712024e-05,
+      "loss": 0.4882,
+      "step": 2900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.4059064240989996e-05,
+      "loss": 0.3872,
+      "step": 2910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.403441305526796e-05,
+      "loss": 0.2284,
+      "step": 2920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.400976186954593e-05,
+      "loss": 0.2865,
+      "step": 2930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3985110683823895e-05,
+      "loss": 0.4237,
+      "step": 2940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.396045949810186e-05,
+      "loss": 0.4893,
+      "step": 2950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.393580831237983e-05,
+      "loss": 0.2684,
+      "step": 2960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3911157126657795e-05,
+      "loss": 0.7275,
+      "step": 2970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.388650594093576e-05,
+      "loss": 0.5442,
+      "step": 2980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3861854755213724e-05,
+      "loss": 0.5621,
+      "step": 2990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3837203569491695e-05,
+      "loss": 0.4772,
+      "step": 3000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.381255238376966e-05,
+      "loss": 0.3534,
+      "step": 3010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.3787901198047624e-05,
+      "loss": 0.605,
+      "step": 3020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.3763250012325595e-05,
+      "loss": 0.4035,
+      "step": 3030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.373859882660356e-05,
+      "loss": 0.4769,
+      "step": 3040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.371394764088153e-05,
+      "loss": 0.5327,
+      "step": 3050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.3689296455159494e-05,
+      "loss": 0.5351,
+      "step": 3060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.366464526943746e-05,
+      "loss": 0.5664,
+      "step": 3070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.363999408371543e-05,
+      "loss": 0.3872,
+      "step": 3080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.3615342897993394e-05,
+      "loss": 0.4127,
+      "step": 3090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.3590691712271365e-05,
+      "loss": 0.6673,
+      "step": 3100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.356604052654933e-05,
+      "loss": 0.2498,
+      "step": 3110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.3541389340827294e-05,
+      "loss": 0.5613,
+      "step": 3120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.3516738155105265e-05,
+      "loss": 0.4329,
+      "step": 3130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.349208696938323e-05,
+      "loss": 0.4424,
+      "step": 3140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.34674357836612e-05,
+      "loss": 0.697,
+      "step": 3150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.3442784597939165e-05,
+      "loss": 0.2307,
+      "step": 3160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.341813341221713e-05,
+      "loss": 0.351,
+      "step": 3170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.3393482226495093e-05,
+      "loss": 0.5916,
+      "step": 3180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.3368831040773065e-05,
+      "loss": 0.573,
+      "step": 3190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.334417985505103e-05,
+      "loss": 0.5308,
+      "step": 3200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.331952866932899e-05,
+      "loss": 0.412,
+      "step": 3210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.3294877483606964e-05,
+      "loss": 0.3815,
+      "step": 3220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.327022629788493e-05,
+      "loss": 0.3423,
+      "step": 3230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.324557511216289e-05,
+      "loss": 0.4478,
+      "step": 3240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.3220923926440864e-05,
+      "loss": 0.5655,
+      "step": 3250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.319627274071883e-05,
+      "loss": 0.488,
+      "step": 3260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.31716215549968e-05,
+      "loss": 0.397,
+      "step": 3270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.3146970369274764e-05,
+      "loss": 0.2258,
+      "step": 3280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.312231918355273e-05,
+      "loss": 0.4188,
+      "step": 3290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.30976679978307e-05,
+      "loss": 0.3146,
+      "step": 3300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.3073016812108664e-05,
+      "loss": 0.68,
+      "step": 3310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.3048365626386635e-05,
+      "loss": 0.3341,
+      "step": 3320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.30237144406646e-05,
+      "loss": 0.564,
+      "step": 3330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2999063254942563e-05,
+      "loss": 0.1865,
+      "step": 3340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2974412069220535e-05,
+      "loss": 0.4524,
+      "step": 3350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.29497608834985e-05,
+      "loss": 0.3583,
+      "step": 3360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.292510969777647e-05,
+      "loss": 0.3725,
+      "step": 3370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2900458512054434e-05,
+      "loss": 0.4179,
+      "step": 3380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.28758073263324e-05,
+      "loss": 0.2947,
+      "step": 3390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.285115614061036e-05,
+      "loss": 0.4268,
+      "step": 3400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.282650495488833e-05,
+      "loss": 0.4247,
+      "step": 3410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.28018537691663e-05,
+      "loss": 0.4501,
+      "step": 3420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.277720258344426e-05,
+      "loss": 0.7391,
+      "step": 3430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.2752551397722234e-05,
+      "loss": 0.4506,
+      "step": 3440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.27279002120002e-05,
+      "loss": 0.5401,
+      "step": 3450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.270324902627816e-05,
+      "loss": 0.5237,
+      "step": 3460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.2678597840556134e-05,
+      "loss": 0.3341,
+      "step": 3470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.26539466548341e-05,
+      "loss": 0.4275,
+      "step": 3480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.262929546911207e-05,
+      "loss": 0.5962,
+      "step": 3490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.2604644283390033e-05,
+      "loss": 0.2551,
+      "step": 3500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.2579993097668e-05,
+      "loss": 0.3121,
+      "step": 3510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.255534191194597e-05,
+      "loss": 0.4396,
+      "step": 3520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.253069072622393e-05,
+      "loss": 0.317,
+      "step": 3530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.2506039540501904e-05,
+      "loss": 0.2429,
+      "step": 3540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.248138835477987e-05,
+      "loss": 0.463,
+      "step": 3550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.245673716905784e-05,
+      "loss": 0.4323,
+      "step": 3560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.2432085983335804e-05,
+      "loss": 0.387,
+      "step": 3570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.240743479761377e-05,
+      "loss": 0.4828,
+      "step": 3580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.238278361189173e-05,
+      "loss": 0.5204,
+      "step": 3590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.23581324261697e-05,
+      "loss": 0.4351,
+      "step": 3600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.233348124044767e-05,
+      "loss": 0.6948,
+      "step": 3610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.230883005472563e-05,
+      "loss": 0.4538,
+      "step": 3620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.22841788690036e-05,
+      "loss": 0.4126,
+      "step": 3630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.225952768328157e-05,
+      "loss": 0.63,
+      "step": 3640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.223487649755953e-05,
+      "loss": 0.4661,
+      "step": 3650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.22102253118375e-05,
+      "loss": 0.4726,
+      "step": 3660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.218557412611547e-05,
+      "loss": 0.4252,
+      "step": 3670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.216092294039343e-05,
+      "loss": 0.3525,
+      "step": 3680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.21362717546714e-05,
+      "loss": 0.5887,
+      "step": 3690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.211162056894937e-05,
+      "loss": 0.287,
+      "step": 3700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.208696938322734e-05,
+      "loss": 0.3311,
+      "step": 3710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.20623181975053e-05,
+      "loss": 0.3756,
+      "step": 3720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.203766701178327e-05,
+      "loss": 0.4844,
+      "step": 3730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.201301582606124e-05,
+      "loss": 0.3659,
+      "step": 3740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.19883646403392e-05,
+      "loss": 0.3928,
+      "step": 3750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1963713454617174e-05,
+      "loss": 0.4392,
+      "step": 3760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.193906226889514e-05,
+      "loss": 0.3198,
+      "step": 3770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.19144110831731e-05,
+      "loss": 0.4454,
+      "step": 3780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.188975989745107e-05,
+      "loss": 0.4571,
+      "step": 3790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.186510871172903e-05,
+      "loss": 0.4873,
+      "step": 3800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1840457526007e-05,
+      "loss": 0.4197,
+      "step": 3810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1815806340284967e-05,
+      "loss": 0.5796,
+      "step": 3820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.179115515456294e-05,
+      "loss": 0.4924,
+      "step": 3830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.17665039688409e-05,
+      "loss": 0.5874,
+      "step": 3840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.1741852783118866e-05,
+      "loss": 0.4486,
+      "step": 3850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.171720159739684e-05,
+      "loss": 0.5028,
+      "step": 3860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.16925504116748e-05,
+      "loss": 0.4045,
+      "step": 3870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.166789922595277e-05,
+      "loss": 0.5503,
+      "step": 3880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.164324804023074e-05,
+      "loss": 0.2699,
+      "step": 3890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.16185968545087e-05,
+      "loss": 0.4983,
+      "step": 3900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.159394566878667e-05,
+      "loss": 0.5292,
+      "step": 3910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.156929448306464e-05,
+      "loss": 0.8901,
+      "step": 3920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.154464329734261e-05,
+      "loss": 0.3247,
+      "step": 3930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.151999211162057e-05,
+      "loss": 0.5054,
+      "step": 3940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.1495340925898543e-05,
+      "loss": 0.5793,
+      "step": 3950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.147068974017651e-05,
+      "loss": 0.2528,
+      "step": 3960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.144603855445447e-05,
+      "loss": 0.2383,
+      "step": 3970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.1421387368732436e-05,
+      "loss": 0.4946,
+      "step": 3980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.13967361830104e-05,
+      "loss": 0.5586,
+      "step": 3990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.137208499728837e-05,
+      "loss": 0.3531,
+      "step": 4000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.1347433811566336e-05,
+      "loss": 0.3558,
+      "step": 4010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.13227826258443e-05,
+      "loss": 0.4341,
+      "step": 4020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.129813144012227e-05,
+      "loss": 0.4232,
+      "step": 4030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.1273480254400236e-05,
+      "loss": 0.476,
+      "step": 4040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.124882906867821e-05,
+      "loss": 0.7422,
+      "step": 4050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.122417788295617e-05,
+      "loss": 0.5261,
+      "step": 4060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.1199526697234136e-05,
+      "loss": 0.2864,
+      "step": 4070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.117487551151211e-05,
+      "loss": 0.3779,
+      "step": 4080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.115022432579007e-05,
+      "loss": 0.6221,
+      "step": 4090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.112557314006804e-05,
+      "loss": 0.4548,
+      "step": 4100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.110092195434601e-05,
+      "loss": 0.3558,
+      "step": 4110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.107627076862397e-05,
+      "loss": 0.4273,
+      "step": 4120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.105161958290194e-05,
+      "loss": 0.5026,
+      "step": 4130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.1026968397179906e-05,
+      "loss": 0.3566,
+      "step": 4140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.100231721145788e-05,
+      "loss": 0.4205,
+      "step": 4150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.097766602573584e-05,
+      "loss": 0.5445,
+      "step": 4160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0953014840013806e-05,
+      "loss": 0.2871,
+      "step": 4170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.092836365429177e-05,
+      "loss": 0.3313,
+      "step": 4180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0903712468569735e-05,
+      "loss": 0.4918,
+      "step": 4190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0879061282847706e-05,
+      "loss": 0.389,
+      "step": 4200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.085441009712567e-05,
+      "loss": 0.1248,
+      "step": 4210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.082975891140364e-05,
+      "loss": 0.4457,
+      "step": 4220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0805107725681606e-05,
+      "loss": 0.6366,
+      "step": 4230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.078045653995957e-05,
+      "loss": 0.5966,
+      "step": 4240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.075580535423754e-05,
+      "loss": 0.4953,
+      "step": 4250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0731154168515505e-05,
+      "loss": 0.3313,
+      "step": 4260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.0706502982793477e-05,
+      "loss": 0.5055,
+      "step": 4270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.068185179707144e-05,
+      "loss": 0.5049,
+      "step": 4280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.0657200611349405e-05,
+      "loss": 0.4617,
+      "step": 4290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.0632549425627376e-05,
+      "loss": 0.5226,
+      "step": 4300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.060789823990534e-05,
+      "loss": 0.2738,
+      "step": 4310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.058324705418331e-05,
+      "loss": 0.5308,
+      "step": 4320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.0558595868461276e-05,
+      "loss": 0.4691,
+      "step": 4330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.053394468273925e-05,
+      "loss": 0.1673,
+      "step": 4340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.050929349701721e-05,
+      "loss": 0.5212,
+      "step": 4350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.0484642311295176e-05,
+      "loss": 0.4821,
+      "step": 4360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.045999112557314e-05,
+      "loss": 0.3685,
+      "step": 4370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.0435339939851105e-05,
+      "loss": 0.6277,
+      "step": 4380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.0410688754129076e-05,
+      "loss": 0.5409,
+      "step": 4390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.038603756840704e-05,
+      "loss": 0.2528,
+      "step": 4400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.0361386382685004e-05,
+      "loss": 0.3622,
+      "step": 4410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.0336735196962975e-05,
+      "loss": 0.3554,
+      "step": 4420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.031208401124094e-05,
+      "loss": 0.8313,
+      "step": 4430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.028743282551891e-05,
+      "loss": 0.5274,
+      "step": 4440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.0262781639796875e-05,
+      "loss": 0.4978,
+      "step": 4450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.023813045407484e-05,
+      "loss": 0.3827,
+      "step": 4460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.021347926835281e-05,
+      "loss": 0.4436,
+      "step": 4470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.0188828082630775e-05,
+      "loss": 0.482,
+      "step": 4480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.0164176896908746e-05,
+      "loss": 0.2945,
+      "step": 4490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.013952571118671e-05,
+      "loss": 0.3899,
+      "step": 4500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.0114874525464675e-05,
+      "loss": 0.6251,
+      "step": 4510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.0090223339742646e-05,
+      "loss": 0.3963,
+      "step": 4520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.006557215402061e-05,
+      "loss": 0.3843,
+      "step": 4530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.004092096829858e-05,
+      "loss": 0.5119,
+      "step": 4540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.0016269782576546e-05,
+      "loss": 0.5311,
+      "step": 4550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.999161859685451e-05,
+      "loss": 0.3581,
+      "step": 4560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.996696741113248e-05,
+      "loss": 0.5462,
+      "step": 4570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9942316225410445e-05,
+      "loss": 0.2145,
+      "step": 4580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.991766503968841e-05,
+      "loss": 0.4472,
+      "step": 4590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9893013853966374e-05,
+      "loss": 0.3806,
+      "step": 4600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9868362668244345e-05,
+      "loss": 0.5448,
+      "step": 4610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.984371148252231e-05,
+      "loss": 0.4161,
+      "step": 4620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9819060296800274e-05,
+      "loss": 0.3249,
+      "step": 4630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9794409111078245e-05,
+      "loss": 0.4009,
+      "step": 4640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.976975792535621e-05,
+      "loss": 0.3636,
+      "step": 4650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.974510673963418e-05,
+      "loss": 0.3853,
+      "step": 4660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9720455553912145e-05,
+      "loss": 0.2838,
+      "step": 4670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.969580436819011e-05,
+      "loss": 0.4274,
+      "step": 4680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.967115318246808e-05,
+      "loss": 0.4446,
+      "step": 4690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9646501996746044e-05,
+      "loss": 0.3325,
+      "step": 4700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9621850811024016e-05,
+      "loss": 0.4101,
+      "step": 4710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.959719962530198e-05,
+      "loss": 0.4754,
+      "step": 4720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.957254843957995e-05,
+      "loss": 0.3556,
+      "step": 4730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9547897253857915e-05,
+      "loss": 0.4635,
+      "step": 4740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.952324606813588e-05,
+      "loss": 0.4871,
+      "step": 4750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.949859488241385e-05,
+      "loss": 0.7357,
+      "step": 4760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9473943696691815e-05,
+      "loss": 0.3779,
+      "step": 4770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.944929251096978e-05,
+      "loss": 0.3729,
+      "step": 4780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9424641325247744e-05,
+      "loss": 0.5116,
+      "step": 4790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.939999013952571e-05,
+      "loss": 0.3307,
+      "step": 4800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.937533895380368e-05,
+      "loss": 0.6199,
+      "step": 4810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9350687768081644e-05,
+      "loss": 0.477,
+      "step": 4820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9326036582359615e-05,
+      "loss": 0.5573,
+      "step": 4830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.930138539663758e-05,
+      "loss": 0.5235,
+      "step": 4840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.927673421091554e-05,
+      "loss": 0.3751,
+      "step": 4850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9252083025193514e-05,
+      "loss": 0.5916,
+      "step": 4860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.922743183947148e-05,
+      "loss": 0.2718,
+      "step": 4870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.920278065374945e-05,
+      "loss": 0.3602,
+      "step": 4880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.9178129468027414e-05,
+      "loss": 0.4528,
+      "step": 4890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.915347828230538e-05,
+      "loss": 0.4506,
+      "step": 4900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.912882709658335e-05,
+      "loss": 0.3293,
+      "step": 4910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.9104175910861314e-05,
+      "loss": 0.2825,
+      "step": 4920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.9079524725139285e-05,
+      "loss": 0.7616,
+      "step": 4930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.905487353941725e-05,
+      "loss": 0.383,
+      "step": 4940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.903022235369522e-05,
+      "loss": 0.2989,
+      "step": 4950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.9005571167973185e-05,
+      "loss": 0.6415,
+      "step": 4960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.898091998225115e-05,
+      "loss": 0.4645,
+      "step": 4970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8956268796529113e-05,
+      "loss": 0.2611,
+      "step": 4980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.893161761080708e-05,
+      "loss": 0.6351,
+      "step": 4990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.890696642508505e-05,
+      "loss": 0.4235,
+      "step": 5000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.888231523936301e-05,
+      "loss": 0.4106,
+      "step": 5010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.885766405364098e-05,
+      "loss": 0.3392,
+      "step": 5020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.883301286791895e-05,
+      "loss": 0.2742,
+      "step": 5030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.880836168219691e-05,
+      "loss": 0.3874,
+      "step": 5040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8783710496474884e-05,
+      "loss": 0.3781,
+      "step": 5050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.875905931075285e-05,
+      "loss": 0.7087,
+      "step": 5060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.873440812503081e-05,
+      "loss": 0.4159,
+      "step": 5070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8709756939308784e-05,
+      "loss": 0.2988,
+      "step": 5080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.868510575358675e-05,
+      "loss": 0.7219,
+      "step": 5090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.866045456786472e-05,
+      "loss": 0.3701,
+      "step": 5100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.8635803382142684e-05,
+      "loss": 0.4571,
+      "step": 5110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.8611152196420655e-05,
+      "loss": 0.3573,
+      "step": 5120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.858650101069862e-05,
+      "loss": 0.5123,
+      "step": 5130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.8561849824976583e-05,
+      "loss": 0.3718,
+      "step": 5140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.8537198639254555e-05,
+      "loss": 0.2619,
+      "step": 5150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.851254745353252e-05,
+      "loss": 0.5275,
+      "step": 5160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.848789626781048e-05,
+      "loss": 0.4,
+      "step": 5170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.846324508208845e-05,
+      "loss": 0.2978,
+      "step": 5180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.843859389636641e-05,
+      "loss": 0.3321,
+      "step": 5190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.841394271064438e-05,
+      "loss": 0.2892,
+      "step": 5200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.838929152492235e-05,
+      "loss": 0.3292,
+      "step": 5210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.836464033920032e-05,
+      "loss": 0.2922,
+      "step": 5220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.833998915347828e-05,
+      "loss": 0.4541,
+      "step": 5230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.831533796775625e-05,
+      "loss": 0.2059,
+      "step": 5240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.829068678203422e-05,
+      "loss": 0.716,
+      "step": 5250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.826603559631218e-05,
+      "loss": 0.4391,
+      "step": 5260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.8241384410590154e-05,
+      "loss": 0.1749,
+      "step": 5270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.821673322486812e-05,
+      "loss": 0.5253,
+      "step": 5280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.819208203914608e-05,
+      "loss": 0.2852,
+      "step": 5290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.816743085342405e-05,
+      "loss": 0.4735,
+      "step": 5300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.814277966770202e-05,
+      "loss": 0.2623,
+      "step": 5310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.811812848197999e-05,
+      "loss": 0.3793,
+      "step": 5320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.809347729625795e-05,
+      "loss": 0.5528,
+      "step": 5330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.8068826110535924e-05,
+      "loss": 0.4821,
+      "step": 5340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.804417492481389e-05,
+      "loss": 0.2751,
+      "step": 5350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.801952373909185e-05,
+      "loss": 0.5512,
+      "step": 5360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.799487255336982e-05,
+      "loss": 0.2212,
+      "step": 5370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.797022136764778e-05,
+      "loss": 0.3166,
+      "step": 5380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.794557018192575e-05,
+      "loss": 0.5707,
+      "step": 5390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.792091899620372e-05,
+      "loss": 0.399,
+      "step": 5400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.789626781048168e-05,
+      "loss": 0.4221,
+      "step": 5410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.787161662475965e-05,
+      "loss": 0.5271,
+      "step": 5420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.784696543903762e-05,
+      "loss": 0.2978,
+      "step": 5430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.782231425331559e-05,
+      "loss": 0.4731,
+      "step": 5440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.779766306759355e-05,
+      "loss": 0.3468,
+      "step": 5450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7773011881871517e-05,
+      "loss": 0.6613,
+      "step": 5460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.774836069614949e-05,
+      "loss": 0.3035,
+      "step": 5470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.772370951042745e-05,
+      "loss": 0.2596,
+      "step": 5480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.769905832470542e-05,
+      "loss": 0.6131,
+      "step": 5490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.767440713898339e-05,
+      "loss": 0.3935,
+      "step": 5500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.764975595326136e-05,
+      "loss": 0.3636,
+      "step": 5510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.762510476753932e-05,
+      "loss": 0.4786,
+      "step": 5520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.760045358181729e-05,
+      "loss": 0.3205,
+      "step": 5530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.757580239609526e-05,
+      "loss": 0.3827,
+      "step": 5540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.755115121037322e-05,
+      "loss": 0.1741,
+      "step": 5550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.752650002465119e-05,
+      "loss": 0.3244,
+      "step": 5560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.750184883892915e-05,
+      "loss": 0.4751,
+      "step": 5570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.7477197653207116e-05,
+      "loss": 0.3489,
+      "step": 5580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.745254646748509e-05,
+      "loss": 0.5568,
+      "step": 5590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.742789528176305e-05,
+      "loss": 0.3504,
+      "step": 5600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.740324409604102e-05,
+      "loss": 0.3172,
+      "step": 5610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.7378592910318986e-05,
+      "loss": 0.4326,
+      "step": 5620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.735394172459695e-05,
+      "loss": 0.2253,
+      "step": 5630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.732929053887492e-05,
+      "loss": 0.519,
+      "step": 5640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.7304639353152886e-05,
+      "loss": 0.4458,
+      "step": 5650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.727998816743086e-05,
+      "loss": 0.4013,
+      "step": 5660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.725533698170882e-05,
+      "loss": 0.5565,
+      "step": 5670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.7230685795986786e-05,
+      "loss": 0.2501,
+      "step": 5680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.720603461026476e-05,
+      "loss": 0.5052,
+      "step": 5690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.718138342454272e-05,
+      "loss": 0.3676,
+      "step": 5700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.715673223882069e-05,
+      "loss": 0.5062,
+      "step": 5710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.713208105309866e-05,
+      "loss": 0.3744,
+      "step": 5720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.710742986737663e-05,
+      "loss": 0.3217,
+      "step": 5730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.708277868165459e-05,
+      "loss": 0.4339,
+      "step": 5740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.705812749593256e-05,
+      "loss": 0.3789,
+      "step": 5750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.703347631021052e-05,
+      "loss": 0.546,
+      "step": 5760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.700882512448849e-05,
+      "loss": 0.5278,
+      "step": 5770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6984173938766456e-05,
+      "loss": 0.3825,
+      "step": 5780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.695952275304442e-05,
+      "loss": 0.4407,
+      "step": 5790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6934871567322385e-05,
+      "loss": 0.1947,
+      "step": 5800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6910220381600356e-05,
+      "loss": 0.4573,
+      "step": 5810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.688556919587832e-05,
+      "loss": 0.4531,
+      "step": 5820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.686091801015629e-05,
+      "loss": 0.4556,
+      "step": 5830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6836266824434256e-05,
+      "loss": 0.2921,
+      "step": 5840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.681161563871222e-05,
+      "loss": 0.378,
+      "step": 5850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.678696445299019e-05,
+      "loss": 0.2369,
+      "step": 5860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6762313267268156e-05,
+      "loss": 0.5602,
+      "step": 5870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.673766208154613e-05,
+      "loss": 0.3904,
+      "step": 5880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.671301089582409e-05,
+      "loss": 0.3861,
+      "step": 5890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.668835971010206e-05,
+      "loss": 0.5805,
+      "step": 5900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6663708524380027e-05,
+      "loss": 0.2908,
+      "step": 5910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.663905733865799e-05,
+      "loss": 0.709,
+      "step": 5920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.661440615293596e-05,
+      "loss": 0.4353,
+      "step": 5930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6589754967213926e-05,
+      "loss": 0.2691,
+      "step": 5940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.656510378149189e-05,
+      "loss": 0.5299,
+      "step": 5950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.654045259576986e-05,
+      "loss": 0.4158,
+      "step": 5960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6515801410047826e-05,
+      "loss": 0.609,
+      "step": 5970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.649115022432579e-05,
+      "loss": 0.3064,
+      "step": 5980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6466499038603755e-05,
+      "loss": 0.4364,
+      "step": 5990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6441847852881726e-05,
+      "loss": 0.3817,
+      "step": 6000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.641719666715969e-05,
+      "loss": 0.52,
+      "step": 6010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6392545481437655e-05,
+      "loss": 0.2914,
+      "step": 6020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6367894295715626e-05,
+      "loss": 0.4154,
+      "step": 6030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.634324310999359e-05,
+      "loss": 0.2134,
+      "step": 6040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.631859192427156e-05,
+      "loss": 0.2253,
+      "step": 6050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6293940738549525e-05,
+      "loss": 0.494,
+      "step": 6060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.626928955282749e-05,
+      "loss": 0.4535,
+      "step": 6070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.624463836710546e-05,
+      "loss": 0.1217,
+      "step": 6080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6219987181383425e-05,
+      "loss": 0.5292,
+      "step": 6090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6195335995661396e-05,
+      "loss": 0.2099,
+      "step": 6100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.617068480993936e-05,
+      "loss": 0.4723,
+      "step": 6110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.614603362421733e-05,
+      "loss": 0.3341,
+      "step": 6120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6121382438495296e-05,
+      "loss": 0.5677,
+      "step": 6130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.609673125277326e-05,
+      "loss": 0.2799,
+      "step": 6140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.607208006705123e-05,
+      "loss": 0.4294,
+      "step": 6150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.6047428881329196e-05,
+      "loss": 0.1886,
+      "step": 6160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.602277769560716e-05,
+      "loss": 0.6114,
+      "step": 6170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5998126509885124e-05,
+      "loss": 0.4859,
+      "step": 6180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.597347532416309e-05,
+      "loss": 0.541,
+      "step": 6190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.594882413844106e-05,
+      "loss": 0.6173,
+      "step": 6200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5924172952719024e-05,
+      "loss": 0.2135,
+      "step": 6210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5899521766996995e-05,
+      "loss": 0.7649,
+      "step": 6220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.587487058127496e-05,
+      "loss": 0.2958,
+      "step": 6230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5850219395552924e-05,
+      "loss": 0.2355,
+      "step": 6240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5825568209830895e-05,
+      "loss": 0.4471,
+      "step": 6250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.580091702410886e-05,
+      "loss": 0.3203,
+      "step": 6260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.577626583838683e-05,
+      "loss": 0.7445,
+      "step": 6270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5751614652664795e-05,
+      "loss": 0.5971,
+      "step": 6280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5726963466942766e-05,
+      "loss": 0.3467,
+      "step": 6290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.570231228122073e-05,
+      "loss": 0.5079,
+      "step": 6300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5677661095498695e-05,
+      "loss": 0.3399,
+      "step": 6310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5653009909776666e-05,
+      "loss": 0.4564,
+      "step": 6320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.562835872405463e-05,
+      "loss": 0.3879,
+      "step": 6330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.56037075383326e-05,
+      "loss": 0.2319,
+      "step": 6340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.5579056352610566e-05,
+      "loss": 0.4022,
+      "step": 6350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.555440516688853e-05,
+      "loss": 0.6756,
+      "step": 6360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.5529753981166494e-05,
+      "loss": 0.2236,
+      "step": 6370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.550510279544446e-05,
+      "loss": 0.4342,
+      "step": 6380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.548045160972243e-05,
+      "loss": 0.5317,
+      "step": 6390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.5455800424000394e-05,
+      "loss": 0.3527,
+      "step": 6400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.543114923827836e-05,
+      "loss": 0.464,
+      "step": 6410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.540649805255633e-05,
+      "loss": 0.4065,
+      "step": 6420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.5381846866834294e-05,
+      "loss": 0.2413,
+      "step": 6430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.5357195681112265e-05,
+      "loss": 0.2495,
+      "step": 6440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.533254449539023e-05,
+      "loss": 0.3118,
+      "step": 6450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.5307893309668194e-05,
+      "loss": 0.4255,
+      "step": 6460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.5283242123946165e-05,
+      "loss": 0.3428,
+      "step": 6470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.525859093822413e-05,
+      "loss": 0.3611,
+      "step": 6480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.52339397525021e-05,
+      "loss": 0.3655,
+      "step": 6490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.5209288566780064e-05,
+      "loss": 0.2349,
+      "step": 6500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.5184637381058035e-05,
+      "loss": 0.1839,
+      "step": 6510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.5159986195336e-05,
+      "loss": 0.3761,
+      "step": 6520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.5135335009613964e-05,
+      "loss": 0.2626,
+      "step": 6530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.5110683823891935e-05,
+      "loss": 0.229,
+      "step": 6540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.50860326381699e-05,
+      "loss": 0.4279,
+      "step": 6550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5061381452447864e-05,
+      "loss": 0.3816,
+      "step": 6560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.503673026672583e-05,
+      "loss": 0.2296,
+      "step": 6570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.501207908100379e-05,
+      "loss": 0.386,
+      "step": 6580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4987427895281764e-05,
+      "loss": 0.4793,
+      "step": 6590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.496277670955973e-05,
+      "loss": 0.4584,
+      "step": 6600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.49381255238377e-05,
+      "loss": 0.4182,
+      "step": 6610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4913474338115663e-05,
+      "loss": 0.5153,
+      "step": 6620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.488882315239363e-05,
+      "loss": 0.3682,
+      "step": 6630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.48641719666716e-05,
+      "loss": 0.1635,
+      "step": 6640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.483952078094956e-05,
+      "loss": 0.2291,
+      "step": 6650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4814869595227534e-05,
+      "loss": 0.4349,
+      "step": 6660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.47902184095055e-05,
+      "loss": 0.6211,
+      "step": 6670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.476556722378347e-05,
+      "loss": 0.2548,
+      "step": 6680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4740916038061434e-05,
+      "loss": 0.4494,
+      "step": 6690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.47162648523394e-05,
+      "loss": 0.48,
+      "step": 6700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.469161366661737e-05,
+      "loss": 0.449,
+      "step": 6710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4666962480895334e-05,
+      "loss": 0.3775,
+      "step": 6720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4642311295173305e-05,
+      "loss": 0.6433,
+      "step": 6730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.461766010945127e-05,
+      "loss": 0.6543,
+      "step": 6740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4593008923729234e-05,
+      "loss": 0.3437,
+      "step": 6750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.45683577380072e-05,
+      "loss": 0.5718,
+      "step": 6760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.454370655228516e-05,
+      "loss": 0.3025,
+      "step": 6770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.4519055366563133e-05,
+      "loss": 0.5036,
+      "step": 6780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.44944041808411e-05,
+      "loss": 0.2662,
+      "step": 6790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.446975299511906e-05,
+      "loss": 0.165,
+      "step": 6800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.444510180939703e-05,
+      "loss": 0.412,
+      "step": 6810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.4420450623675e-05,
+      "loss": 0.2364,
+      "step": 6820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.439579943795297e-05,
+      "loss": 0.2675,
+      "step": 6830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.437114825223093e-05,
+      "loss": 0.4334,
+      "step": 6840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.43464970665089e-05,
+      "loss": 0.4931,
+      "step": 6850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.432184588078687e-05,
+      "loss": 0.427,
+      "step": 6860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.429719469506483e-05,
+      "loss": 0.4686,
+      "step": 6870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.4272543509342804e-05,
+      "loss": 0.3212,
+      "step": 6880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.424789232362077e-05,
+      "loss": 0.4078,
+      "step": 6890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.422324113789874e-05,
+      "loss": 0.3772,
+      "step": 6900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.4198589952176704e-05,
+      "loss": 0.5817,
+      "step": 6910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.417393876645467e-05,
+      "loss": 0.3895,
+      "step": 6920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.414928758073264e-05,
+      "loss": 0.7611,
+      "step": 6930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.41246363950106e-05,
+      "loss": 0.2929,
+      "step": 6940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.409998520928857e-05,
+      "loss": 0.4246,
+      "step": 6950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.407533402356653e-05,
+      "loss": 0.3569,
+      "step": 6960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.40506828378445e-05,
+      "loss": 0.4494,
+      "step": 6970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.402603165212247e-05,
+      "loss": 0.2869,
+      "step": 6980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.400138046640043e-05,
+      "loss": 0.2508,
+      "step": 6990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.39767292806784e-05,
+      "loss": 0.4217,
+      "step": 7000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.395207809495637e-05,
+      "loss": 0.4069,
+      "step": 7010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.392742690923433e-05,
+      "loss": 0.4392,
+      "step": 7020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.39027757235123e-05,
+      "loss": 0.3353,
+      "step": 7030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.387812453779027e-05,
+      "loss": 0.4315,
+      "step": 7040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.385347335206824e-05,
+      "loss": 0.3902,
+      "step": 7050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.38288221663462e-05,
+      "loss": 0.3844,
+      "step": 7060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.380417098062417e-05,
+      "loss": 0.3009,
+      "step": 7070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.377951979490214e-05,
+      "loss": 0.3553,
+      "step": 7080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.37548686091801e-05,
+      "loss": 0.271,
+      "step": 7090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.373021742345807e-05,
+      "loss": 0.5679,
+      "step": 7100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.370556623773604e-05,
+      "loss": 0.5226,
+      "step": 7110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.368091505201401e-05,
+      "loss": 0.4013,
+      "step": 7120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.365626386629197e-05,
+      "loss": 0.3866,
+      "step": 7130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.363161268056994e-05,
+      "loss": 0.3358,
+      "step": 7140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.36069614948479e-05,
+      "loss": 0.4495,
+      "step": 7150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.358231030912587e-05,
+      "loss": 0.4724,
+      "step": 7160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.355765912340384e-05,
+      "loss": 0.3475,
+      "step": 7170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.35330079376818e-05,
+      "loss": 0.2321,
+      "step": 7180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.3508356751959766e-05,
+      "loss": 0.3047,
+      "step": 7190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.348370556623774e-05,
+      "loss": 0.2528,
+      "step": 7200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.34590543805157e-05,
+      "loss": 0.4118,
+      "step": 7210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.343440319479367e-05,
+      "loss": 0.4468,
+      "step": 7220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.340975200907164e-05,
+      "loss": 0.6296,
+      "step": 7230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.33851008233496e-05,
+      "loss": 0.5661,
+      "step": 7240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.336044963762757e-05,
+      "loss": 0.4491,
+      "step": 7250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.3335798451905536e-05,
+      "loss": 0.5508,
+      "step": 7260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.331114726618351e-05,
+      "loss": 0.3847,
+      "step": 7270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.328649608046147e-05,
+      "loss": 0.2936,
+      "step": 7280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.326184489473944e-05,
+      "loss": 0.1914,
+      "step": 7290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.323719370901741e-05,
+      "loss": 0.6391,
+      "step": 7300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.321254252329537e-05,
+      "loss": 0.3712,
+      "step": 7310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.318789133757334e-05,
+      "loss": 0.4174,
+      "step": 7320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.316324015185131e-05,
+      "loss": 0.4568,
+      "step": 7330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.313858896612927e-05,
+      "loss": 0.5235,
+      "step": 7340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.311393778040724e-05,
+      "loss": 0.4026,
+      "step": 7350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.308928659468521e-05,
+      "loss": 0.3248,
+      "step": 7360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.306463540896317e-05,
+      "loss": 0.3618,
+      "step": 7370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.3039984223241136e-05,
+      "loss": 0.2773,
+      "step": 7380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.301533303751911e-05,
+      "loss": 0.3587,
+      "step": 7390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.299068185179707e-05,
+      "loss": 0.3789,
+      "step": 7400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2966030666075035e-05,
+      "loss": 0.2265,
+      "step": 7410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2941379480353006e-05,
+      "loss": 0.5042,
+      "step": 7420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.291672829463097e-05,
+      "loss": 0.5727,
+      "step": 7430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.289207710890894e-05,
+      "loss": 0.4672,
+      "step": 7440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2867425923186906e-05,
+      "loss": 0.6851,
+      "step": 7450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.284277473746487e-05,
+      "loss": 0.3069,
+      "step": 7460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.281812355174284e-05,
+      "loss": 0.3177,
+      "step": 7470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2793472366020806e-05,
+      "loss": 0.3126,
+      "step": 7480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.276882118029878e-05,
+      "loss": 0.4276,
+      "step": 7490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.274416999457674e-05,
+      "loss": 0.2453,
+      "step": 7500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.271951880885471e-05,
+      "loss": 0.571,
+      "step": 7510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.269486762313268e-05,
+      "loss": 0.8,
+      "step": 7520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.267021643741064e-05,
+      "loss": 0.2241,
+      "step": 7530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.264556525168861e-05,
+      "loss": 0.5563,
+      "step": 7540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2620914065966577e-05,
+      "loss": 0.4021,
+      "step": 7550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.259626288024454e-05,
+      "loss": 0.4412,
+      "step": 7560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2571611694522505e-05,
+      "loss": 0.608,
+      "step": 7570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.254696050880047e-05,
+      "loss": 0.2765,
+      "step": 7580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.252230932307844e-05,
+      "loss": 0.5993,
+      "step": 7590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2497658137356405e-05,
+      "loss": 0.4767,
+      "step": 7600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2473006951634376e-05,
+      "loss": 0.4281,
+      "step": 7610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.244835576591234e-05,
+      "loss": 0.5228,
+      "step": 7620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2423704580190305e-05,
+      "loss": 0.3558,
+      "step": 7630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2399053394468276e-05,
+      "loss": 0.3752,
+      "step": 7640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.237440220874624e-05,
+      "loss": 0.3974,
+      "step": 7650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.234975102302421e-05,
+      "loss": 0.4491,
+      "step": 7660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2325099837302176e-05,
+      "loss": 0.3503,
+      "step": 7670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.230044865158015e-05,
+      "loss": 0.3208,
+      "step": 7680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.227579746585811e-05,
+      "loss": 0.3439,
+      "step": 7690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2251146280136075e-05,
+      "loss": 0.4509,
+      "step": 7700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2226495094414047e-05,
+      "loss": 0.6971,
+      "step": 7710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.220184390869201e-05,
+      "loss": 0.3765,
+      "step": 7720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.217719272296998e-05,
+      "loss": 0.4668,
+      "step": 7730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2152541537247946e-05,
+      "loss": 0.338,
+      "step": 7740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.212789035152591e-05,
+      "loss": 0.2418,
+      "step": 7750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2103239165803875e-05,
+      "loss": 0.2801,
+      "step": 7760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.207858798008184e-05,
+      "loss": 0.5561,
+      "step": 7770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.205393679435981e-05,
+      "loss": 0.5979,
+      "step": 7780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2029285608637775e-05,
+      "loss": 0.408,
+      "step": 7790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.200463442291574e-05,
+      "loss": 0.3394,
+      "step": 7800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.197998323719371e-05,
+      "loss": 0.3674,
+      "step": 7810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1955332051471674e-05,
+      "loss": 0.1881,
+      "step": 7820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1930680865749646e-05,
+      "loss": 0.3687,
+      "step": 7830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.190602968002761e-05,
+      "loss": 0.2179,
+      "step": 7840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1881378494305574e-05,
+      "loss": 0.4377,
+      "step": 7850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1856727308583545e-05,
+      "loss": 0.3917,
+      "step": 7860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.183207612286151e-05,
+      "loss": 0.7002,
+      "step": 7870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.180742493713948e-05,
+      "loss": 0.5206,
+      "step": 7880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1782773751417445e-05,
+      "loss": 0.402,
+      "step": 7890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1758122565695416e-05,
+      "loss": 0.3917,
+      "step": 7900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.173347137997338e-05,
+      "loss": 0.3866,
+      "step": 7910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1708820194251345e-05,
+      "loss": 0.4261,
+      "step": 7920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1684169008529316e-05,
+      "loss": 0.3319,
+      "step": 7930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.165951782280728e-05,
+      "loss": 0.6121,
+      "step": 7940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1634866637085245e-05,
+      "loss": 0.3743,
+      "step": 7950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.161021545136321e-05,
+      "loss": 0.361,
+      "step": 7960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.158556426564117e-05,
+      "loss": 0.3438,
+      "step": 7970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1560913079919144e-05,
+      "loss": 0.5814,
+      "step": 7980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.153626189419711e-05,
+      "loss": 0.4318,
+      "step": 7990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.151161070847508e-05,
+      "loss": 0.1905,
+      "step": 8000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.1486959522753044e-05,
+      "loss": 0.2555,
+      "step": 8010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.146230833703101e-05,
+      "loss": 0.4865,
+      "step": 8020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.143765715130898e-05,
+      "loss": 0.2169,
+      "step": 8030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.1413005965586944e-05,
+      "loss": 0.2861,
+      "step": 8040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.1388354779864915e-05,
+      "loss": 0.3536,
+      "step": 8050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.136370359414288e-05,
+      "loss": 0.2458,
+      "step": 8060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.133905240842085e-05,
+      "loss": 0.6201,
+      "step": 8070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.1314401222698815e-05,
+      "loss": 0.3208,
+      "step": 8080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.128975003697678e-05,
+      "loss": 0.3672,
+      "step": 8090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.126509885125475e-05,
+      "loss": 0.4408,
+      "step": 8100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.1240447665532715e-05,
+      "loss": 0.2765,
+      "step": 8110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.1215796479810686e-05,
+      "loss": 0.2976,
+      "step": 8120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.119114529408865e-05,
+      "loss": 0.4782,
+      "step": 8130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.1166494108366614e-05,
+      "loss": 0.3813,
+      "step": 8140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.114184292264458e-05,
+      "loss": 0.7383,
+      "step": 8150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.111719173692254e-05,
+      "loss": 0.5339,
+      "step": 8160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.1092540551200514e-05,
+      "loss": 0.3959,
+      "step": 8170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.106788936547848e-05,
+      "loss": 0.4924,
+      "step": 8180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.104323817975644e-05,
+      "loss": 0.355,
+      "step": 8190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.1018586994034414e-05,
+      "loss": 0.3508,
+      "step": 8200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.099393580831238e-05,
+      "loss": 0.4097,
+      "step": 8210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.096928462259035e-05,
+      "loss": 0.4753,
+      "step": 8220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0944633436868314e-05,
+      "loss": 0.4855,
+      "step": 8230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.091998225114628e-05,
+      "loss": 0.3868,
+      "step": 8240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.089533106542425e-05,
+      "loss": 0.5793,
+      "step": 8250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0870679879702213e-05,
+      "loss": 0.363,
+      "step": 8260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0846028693980185e-05,
+      "loss": 0.509,
+      "step": 8270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.082137750825815e-05,
+      "loss": 0.2247,
+      "step": 8280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.079672632253612e-05,
+      "loss": 0.3051,
+      "step": 8290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0772075136814084e-05,
+      "loss": 0.3603,
+      "step": 8300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.074742395109205e-05,
+      "loss": 0.2864,
+      "step": 8310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.072277276537002e-05,
+      "loss": 0.3383,
+      "step": 8320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0698121579647984e-05,
+      "loss": 0.6133,
+      "step": 8330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.067347039392595e-05,
+      "loss": 0.2454,
+      "step": 8340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.064881920820391e-05,
+      "loss": 0.4744,
+      "step": 8350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0624168022481884e-05,
+      "loss": 0.4114,
+      "step": 8360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.059951683675985e-05,
+      "loss": 0.3021,
+      "step": 8370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.057486565103781e-05,
+      "loss": 0.4181,
+      "step": 8380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0550214465315784e-05,
+      "loss": 0.3719,
+      "step": 8390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.052556327959375e-05,
+      "loss": 0.4693,
+      "step": 8400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0500912093871716e-05,
+      "loss": 0.2873,
+      "step": 8410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.0476260908149683e-05,
+      "loss": 0.5997,
+      "step": 8420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.0451609722427648e-05,
+      "loss": 0.302,
+      "step": 8430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.042695853670562e-05,
+      "loss": 0.448,
+      "step": 8440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.0402307350983583e-05,
+      "loss": 0.4114,
+      "step": 8450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.0377656165261554e-05,
+      "loss": 0.1906,
+      "step": 8460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.035300497953952e-05,
+      "loss": 0.5432,
+      "step": 8470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.0328353793817483e-05,
+      "loss": 0.3962,
+      "step": 8480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.030370260809545e-05,
+      "loss": 0.4202,
+      "step": 8490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.027905142237342e-05,
+      "loss": 0.3017,
+      "step": 8500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.0254400236651386e-05,
+      "loss": 0.5239,
+      "step": 8510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.022974905092935e-05,
+      "loss": 0.3419,
+      "step": 8520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.0205097865207315e-05,
+      "loss": 0.4099,
+      "step": 8530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.0180446679485286e-05,
+      "loss": 0.3235,
+      "step": 8540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.015579549376325e-05,
+      "loss": 0.2392,
+      "step": 8550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.013114430804122e-05,
+      "loss": 0.6259,
+      "step": 8560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.0106493122319186e-05,
+      "loss": 0.4078,
+      "step": 8570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.008184193659715e-05,
+      "loss": 0.3856,
+      "step": 8580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.0057190750875118e-05,
+      "loss": 0.2923,
+      "step": 8590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.0032539565153085e-05,
+      "loss": 0.1937,
+      "step": 8600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.0007888379431053e-05,
+      "loss": 0.2599,
+      "step": 8610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9983237193709017e-05,
+      "loss": 0.3317,
+      "step": 8620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9958586007986982e-05,
+      "loss": 0.3491,
+      "step": 8630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9933934822264953e-05,
+      "loss": 0.4013,
+      "step": 8640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9909283636542917e-05,
+      "loss": 0.4898,
+      "step": 8650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.988463245082089e-05,
+      "loss": 0.4177,
+      "step": 8660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9859981265098853e-05,
+      "loss": 0.5963,
+      "step": 8670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.983533007937682e-05,
+      "loss": 0.2722,
+      "step": 8680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9810678893654788e-05,
+      "loss": 0.3851,
+      "step": 8690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9786027707932752e-05,
+      "loss": 0.5974,
+      "step": 8700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.976137652221072e-05,
+      "loss": 0.3484,
+      "step": 8710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9736725336488684e-05,
+      "loss": 0.3152,
+      "step": 8720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9712074150766656e-05,
+      "loss": 0.316,
+      "step": 8730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.968742296504462e-05,
+      "loss": 0.7165,
+      "step": 8740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9662771779322584e-05,
+      "loss": 0.2202,
+      "step": 8750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9638120593600555e-05,
+      "loss": 0.2998,
+      "step": 8760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.961346940787852e-05,
+      "loss": 0.2975,
+      "step": 8770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.958881822215649e-05,
+      "loss": 0.5051,
+      "step": 8780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9564167036434455e-05,
+      "loss": 0.2036,
+      "step": 8790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.953951585071242e-05,
+      "loss": 0.4892,
+      "step": 8800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9514864664990387e-05,
+      "loss": 0.5566,
+      "step": 8810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.949021347926835e-05,
+      "loss": 0.2568,
+      "step": 8820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9465562293546323e-05,
+      "loss": 0.4512,
+      "step": 8830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9440911107824287e-05,
+      "loss": 0.2557,
+      "step": 8840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9416259922102258e-05,
+      "loss": 0.3948,
+      "step": 8850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9391608736380222e-05,
+      "loss": 0.2354,
+      "step": 8860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9366957550658187e-05,
+      "loss": 0.3821,
+      "step": 8870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9342306364936158e-05,
+      "loss": 0.2988,
+      "step": 8880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9317655179214122e-05,
+      "loss": 0.3471,
+      "step": 8890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.929300399349209e-05,
+      "loss": 0.3016,
+      "step": 8900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9268352807770054e-05,
+      "loss": 0.3516,
+      "step": 8910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.924370162204802e-05,
+      "loss": 0.5237,
+      "step": 8920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.921905043632599e-05,
+      "loss": 0.3405,
+      "step": 8930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9194399250603954e-05,
+      "loss": 0.2709,
+      "step": 8940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9169748064881925e-05,
+      "loss": 0.3269,
+      "step": 8950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.914509687915989e-05,
+      "loss": 0.4213,
+      "step": 8960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9120445693437854e-05,
+      "loss": 0.1926,
+      "step": 8970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9095794507715825e-05,
+      "loss": 0.281,
+      "step": 8980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.907114332199379e-05,
+      "loss": 0.7733,
+      "step": 8990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9046492136271757e-05,
+      "loss": 0.4117,
+      "step": 9000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.902184095054972e-05,
+      "loss": 0.5475,
+      "step": 9010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8997189764827686e-05,
+      "loss": 0.358,
+      "step": 9020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8972538579105657e-05,
+      "loss": 0.4073,
+      "step": 9030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.894788739338362e-05,
+      "loss": 0.5033,
+      "step": 9040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8923236207661592e-05,
+      "loss": 0.4783,
+      "step": 9050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8898585021939556e-05,
+      "loss": 0.6663,
+      "step": 9060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8873933836217528e-05,
+      "loss": 0.2404,
+      "step": 9070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8849282650495492e-05,
+      "loss": 0.2374,
+      "step": 9080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8824631464773456e-05,
+      "loss": 0.2459,
+      "step": 9090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8799980279051424e-05,
+      "loss": 0.4372,
+      "step": 9100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8775329093329388e-05,
+      "loss": 0.1949,
+      "step": 9110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.875067790760736e-05,
+      "loss": 0.3766,
+      "step": 9120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8726026721885324e-05,
+      "loss": 0.393,
+      "step": 9130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8701375536163288e-05,
+      "loss": 0.2297,
+      "step": 9140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.867672435044126e-05,
+      "loss": 0.1835,
+      "step": 9150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8652073164719223e-05,
+      "loss": 0.5488,
+      "step": 9160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8627421978997195e-05,
+      "loss": 0.5796,
+      "step": 9170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.860277079327516e-05,
+      "loss": 0.5971,
+      "step": 9180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8578119607553123e-05,
+      "loss": 0.5223,
+      "step": 9190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.855346842183109e-05,
+      "loss": 0.354,
+      "step": 9200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8528817236109055e-05,
+      "loss": 0.27,
+      "step": 9210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8504166050387026e-05,
+      "loss": 0.2594,
+      "step": 9220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.847951486466499e-05,
+      "loss": 0.1832,
+      "step": 9230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8454863678942962e-05,
+      "loss": 0.3514,
+      "step": 9240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8430212493220926e-05,
+      "loss": 0.3991,
+      "step": 9250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.840556130749889e-05,
+      "loss": 0.3952,
+      "step": 9260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.838091012177686e-05,
+      "loss": 0.3737,
+      "step": 9270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8356258936054826e-05,
+      "loss": 0.3367,
+      "step": 9280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8331607750332794e-05,
+      "loss": 0.3731,
+      "step": 9290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8306956564610758e-05,
+      "loss": 0.2723,
+      "step": 9300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8282305378888722e-05,
+      "loss": 0.3247,
+      "step": 9310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8257654193166693e-05,
+      "loss": 0.3352,
+      "step": 9320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8233003007444658e-05,
+      "loss": 0.3247,
+      "step": 9330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.820835182172263e-05,
+      "loss": 0.2581,
+      "step": 9340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8183700636000593e-05,
+      "loss": 0.3546,
+      "step": 9350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8159049450278557e-05,
+      "loss": 0.3488,
+      "step": 9360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.813439826455653e-05,
+      "loss": 0.2634,
+      "step": 9370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8109747078834493e-05,
+      "loss": 0.2363,
+      "step": 9380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.808509589311246e-05,
+      "loss": 0.1794,
+      "step": 9390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8060444707390425e-05,
+      "loss": 0.1704,
+      "step": 9400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.803579352166839e-05,
+      "loss": 0.4069,
+      "step": 9410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.801114233594636e-05,
+      "loss": 0.5863,
+      "step": 9420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7986491150224325e-05,
+      "loss": 0.3216,
+      "step": 9430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7961839964502296e-05,
+      "loss": 0.7227,
+      "step": 9440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.793718877878026e-05,
+      "loss": 0.4447,
+      "step": 9450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.791253759305823e-05,
+      "loss": 0.295,
+      "step": 9460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7887886407336196e-05,
+      "loss": 0.2135,
+      "step": 9470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.786323522161416e-05,
+      "loss": 0.5438,
+      "step": 9480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7838584035892128e-05,
+      "loss": 0.4155,
+      "step": 9490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7813932850170092e-05,
+      "loss": 0.513,
+      "step": 9500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7789281664448063e-05,
+      "loss": 0.3628,
+      "step": 9510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7764630478726027e-05,
+      "loss": 0.5421,
+      "step": 9520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7739979293003992e-05,
+      "loss": 0.2448,
+      "step": 9530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7715328107281963e-05,
+      "loss": 0.4539,
+      "step": 9540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7690676921559927e-05,
+      "loss": 0.2674,
+      "step": 9550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7666025735837898e-05,
+      "loss": 0.3451,
+      "step": 9560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7641374550115863e-05,
+      "loss": 0.391,
+      "step": 9570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7616723364393827e-05,
+      "loss": 0.1397,
+      "step": 9580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7592072178671795e-05,
+      "loss": 0.239,
+      "step": 9590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.756742099294976e-05,
+      "loss": 0.6772,
+      "step": 9600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.754276980722773e-05,
+      "loss": 0.3906,
+      "step": 9610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7518118621505694e-05,
+      "loss": 0.3148,
+      "step": 9620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7493467435783666e-05,
+      "loss": 0.4008,
+      "step": 9630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.746881625006163e-05,
+      "loss": 0.4055,
+      "step": 9640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7444165064339594e-05,
+      "loss": 0.3117,
+      "step": 9650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7419513878617565e-05,
+      "loss": 0.2951,
+      "step": 9660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.739486269289553e-05,
+      "loss": 0.3202,
+      "step": 9670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7370211507173497e-05,
+      "loss": 0.3442,
+      "step": 9680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7345560321451462e-05,
+      "loss": 0.4642,
+      "step": 9690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.732090913572943e-05,
+      "loss": 0.2655,
+      "step": 9700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7296257950007397e-05,
+      "loss": 0.3481,
+      "step": 9710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.727160676428536e-05,
+      "loss": 0.42,
+      "step": 9720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7246955578563333e-05,
+      "loss": 0.2694,
+      "step": 9730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7222304392841297e-05,
+      "loss": 0.4237,
+      "step": 9740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.719765320711926e-05,
+      "loss": 0.2607,
+      "step": 9750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7173002021397232e-05,
+      "loss": 0.4269,
+      "step": 9760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7148350835675197e-05,
+      "loss": 0.4721,
+      "step": 9770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7123699649953164e-05,
+      "loss": 0.5805,
+      "step": 9780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.709904846423113e-05,
+      "loss": 0.7721,
+      "step": 9790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7074397278509096e-05,
+      "loss": 0.3813,
+      "step": 9800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7049746092787064e-05,
+      "loss": 0.3882,
+      "step": 9810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.702509490706503e-05,
+      "loss": 0.2411,
+      "step": 9820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7000443721343e-05,
+      "loss": 0.3997,
+      "step": 9830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6975792535620964e-05,
+      "loss": 0.3869,
+      "step": 9840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6951141349898935e-05,
+      "loss": 0.38,
+      "step": 9850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.69264901641769e-05,
+      "loss": 0.2164,
+      "step": 9860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6901838978454864e-05,
+      "loss": 0.2705,
+      "step": 9870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.687718779273283e-05,
+      "loss": 0.3977,
+      "step": 9880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.68525366070108e-05,
+      "loss": 0.4178,
+      "step": 9890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6827885421288767e-05,
+      "loss": 0.6109,
+      "step": 9900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.680323423556673e-05,
+      "loss": 0.1557,
+      "step": 9910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6778583049844696e-05,
+      "loss": 0.6049,
+      "step": 9920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6753931864122667e-05,
+      "loss": 0.3627,
+      "step": 9930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.672928067840063e-05,
+      "loss": 0.5592,
+      "step": 9940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6704629492678602e-05,
+      "loss": 0.3895,
+      "step": 9950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6679978306956566e-05,
+      "loss": 0.5813,
+      "step": 9960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.665532712123453e-05,
+      "loss": 0.4155,
+      "step": 9970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6630675935512502e-05,
+      "loss": 0.4507,
+      "step": 9980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6606024749790466e-05,
+      "loss": 0.4305,
+      "step": 9990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6581373564068434e-05,
+      "loss": 0.1999,
+      "step": 10000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6556722378346398e-05,
+      "loss": 0.2565,
+      "step": 10010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.653207119262437e-05,
+      "loss": 0.3819,
+      "step": 10020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6507420006902334e-05,
+      "loss": 0.2462,
+      "step": 10030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6482768821180298e-05,
+      "loss": 0.3143,
+      "step": 10040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.645811763545827e-05,
+      "loss": 0.39,
+      "step": 10050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6433466449736233e-05,
+      "loss": 0.3682,
+      "step": 10060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.64088152640142e-05,
+      "loss": 0.3964,
+      "step": 10070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.638416407829217e-05,
+      "loss": 0.3052,
+      "step": 10080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6359512892570133e-05,
+      "loss": 0.4053,
+      "step": 10090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.63348617068481e-05,
+      "loss": 0.3836,
+      "step": 10100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6310210521126065e-05,
+      "loss": 0.3624,
+      "step": 10110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6285559335404036e-05,
+      "loss": 0.2914,
+      "step": 10120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6260908149682e-05,
+      "loss": 0.1379,
+      "step": 10130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6236256963959965e-05,
+      "loss": 0.2266,
+      "step": 10140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6211605778237936e-05,
+      "loss": 0.3573,
+      "step": 10150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.61869545925159e-05,
+      "loss": 0.3021,
+      "step": 10160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.616230340679387e-05,
+      "loss": 0.5404,
+      "step": 10170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6137652221071836e-05,
+      "loss": 0.3562,
+      "step": 10180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.61130010353498e-05,
+      "loss": 0.4048,
+      "step": 10190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6088349849627768e-05,
+      "loss": 0.3189,
+      "step": 10200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6063698663905732e-05,
+      "loss": 0.2895,
+      "step": 10210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6039047478183703e-05,
+      "loss": 0.5148,
+      "step": 10220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6014396292461668e-05,
+      "loss": 0.3265,
+      "step": 10230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.598974510673964e-05,
+      "loss": 0.5211,
+      "step": 10240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5965093921017603e-05,
+      "loss": 0.2398,
+      "step": 10250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5940442735295567e-05,
+      "loss": 0.4808,
+      "step": 10260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.591579154957354e-05,
+      "loss": 0.3594,
+      "step": 10270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5891140363851503e-05,
+      "loss": 0.2884,
+      "step": 10280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.586648917812947e-05,
+      "loss": 0.4529,
+      "step": 10290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5841837992407435e-05,
+      "loss": 0.5053,
+      "step": 10300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.58171868066854e-05,
+      "loss": 0.5038,
+      "step": 10310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.579253562096337e-05,
+      "loss": 0.3015,
+      "step": 10320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5767884435241335e-05,
+      "loss": 0.2832,
+      "step": 10330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5743233249519306e-05,
+      "loss": 0.2772,
+      "step": 10340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.571858206379727e-05,
+      "loss": 0.374,
+      "step": 10350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5693930878075234e-05,
+      "loss": 0.3431,
+      "step": 10360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5669279692353206e-05,
+      "loss": 0.4676,
+      "step": 10370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.564462850663117e-05,
+      "loss": 0.3615,
+      "step": 10380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5619977320909138e-05,
+      "loss": 0.2765,
+      "step": 10390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5595326135187102e-05,
+      "loss": 0.6806,
+      "step": 10400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5570674949465073e-05,
+      "loss": 0.6347,
+      "step": 10410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5546023763743037e-05,
+      "loss": 0.383,
+      "step": 10420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5521372578021002e-05,
+      "loss": 0.3478,
+      "step": 10430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5496721392298973e-05,
+      "loss": 0.4347,
+      "step": 10440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5472070206576937e-05,
+      "loss": 0.3468,
+      "step": 10450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5447419020854908e-05,
+      "loss": 0.4449,
+      "step": 10460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5422767835132873e-05,
+      "loss": 0.3736,
+      "step": 10470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5398116649410837e-05,
+      "loss": 0.4212,
+      "step": 10480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5373465463688805e-05,
+      "loss": 0.6261,
+      "step": 10490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.534881427796677e-05,
+      "loss": 0.3702,
+      "step": 10500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.532416309224474e-05,
+      "loss": 0.3271,
+      "step": 10510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5299511906522704e-05,
+      "loss": 0.3905,
+      "step": 10520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.527486072080067e-05,
+      "loss": 0.4685,
+      "step": 10530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.525020953507864e-05,
+      "loss": 0.3701,
+      "step": 10540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5225558349356604e-05,
+      "loss": 0.6473,
+      "step": 10550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5200907163634575e-05,
+      "loss": 0.3593,
+      "step": 10560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.517625597791254e-05,
+      "loss": 0.555,
+      "step": 10570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5151604792190504e-05,
+      "loss": 0.358,
+      "step": 10580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.512695360646847e-05,
+      "loss": 0.5423,
+      "step": 10590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5102302420746436e-05,
+      "loss": 0.3276,
+      "step": 10600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5077651235024407e-05,
+      "loss": 0.4244,
+      "step": 10610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.505300004930237e-05,
+      "loss": 0.5,
+      "step": 10620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5028348863580343e-05,
+      "loss": 0.4416,
+      "step": 10630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5003697677858307e-05,
+      "loss": 0.3694,
+      "step": 10640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4979046492136275e-05,
+      "loss": 0.2158,
+      "step": 10650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4954395306414242e-05,
+      "loss": 0.357,
+      "step": 10660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4929744120692207e-05,
+      "loss": 0.412,
+      "step": 10670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4905092934970174e-05,
+      "loss": 0.4961,
+      "step": 10680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.488044174924814e-05,
+      "loss": 0.5135,
+      "step": 10690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4855790563526106e-05,
+      "loss": 0.3493,
+      "step": 10700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4831139377804074e-05,
+      "loss": 0.266,
+      "step": 10710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.480648819208204e-05,
+      "loss": 0.3125,
+      "step": 10720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4781837006360006e-05,
+      "loss": 0.2915,
+      "step": 10730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4757185820637974e-05,
+      "loss": 0.4856,
+      "step": 10740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.473253463491594e-05,
+      "loss": 0.2195,
+      "step": 10750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.470788344919391e-05,
+      "loss": 0.3942,
+      "step": 10760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4683232263471877e-05,
+      "loss": 0.3433,
+      "step": 10770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.465858107774984e-05,
+      "loss": 0.4046,
+      "step": 10780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4633929892027806e-05,
+      "loss": 0.3905,
+      "step": 10790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4609278706305773e-05,
+      "loss": 0.5868,
+      "step": 10800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.458462752058374e-05,
+      "loss": 0.2739,
+      "step": 10810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.455997633486171e-05,
+      "loss": 0.2768,
+      "step": 10820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4535325149139673e-05,
+      "loss": 0.1595,
+      "step": 10830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.451067396341764e-05,
+      "loss": 0.4146,
+      "step": 10840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.448602277769561e-05,
+      "loss": 0.3069,
+      "step": 10850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4461371591973576e-05,
+      "loss": 0.5316,
+      "step": 10860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4436720406251544e-05,
+      "loss": 0.379,
+      "step": 10870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.441206922052951e-05,
+      "loss": 0.4796,
+      "step": 10880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4387418034807473e-05,
+      "loss": 0.3809,
+      "step": 10890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.436276684908544e-05,
+      "loss": 0.2193,
+      "step": 10900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4338115663363408e-05,
+      "loss": 0.4733,
+      "step": 10910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4313464477641376e-05,
+      "loss": 0.5891,
+      "step": 10920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4288813291919344e-05,
+      "loss": 0.4608,
+      "step": 10930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4264162106197308e-05,
+      "loss": 0.5599,
+      "step": 10940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4239510920475276e-05,
+      "loss": 0.2986,
+      "step": 10950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4214859734753243e-05,
+      "loss": 0.4286,
+      "step": 10960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.419020854903121e-05,
+      "loss": 0.6394,
+      "step": 10970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4165557363309175e-05,
+      "loss": 0.3005,
+      "step": 10980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4140906177587143e-05,
+      "loss": 0.2178,
+      "step": 10990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4116254991865107e-05,
+      "loss": 0.4905,
+      "step": 11000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4091603806143075e-05,
+      "loss": 0.4865,
+      "step": 11010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4066952620421043e-05,
+      "loss": 0.6146,
+      "step": 11020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.404230143469901e-05,
+      "loss": 0.3905,
+      "step": 11030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.401765024897698e-05,
+      "loss": 0.5403,
+      "step": 11040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3992999063254946e-05,
+      "loss": 0.3083,
+      "step": 11050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.396834787753291e-05,
+      "loss": 0.6038,
+      "step": 11060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3943696691810878e-05,
+      "loss": 0.4522,
+      "step": 11070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3919045506088842e-05,
+      "loss": 0.2951,
+      "step": 11080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.389439432036681e-05,
+      "loss": 0.4535,
+      "step": 11090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3869743134644778e-05,
+      "loss": 0.2842,
+      "step": 11100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3845091948922742e-05,
+      "loss": 0.599,
+      "step": 11110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.382044076320071e-05,
+      "loss": 0.4948,
+      "step": 11120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3795789577478678e-05,
+      "loss": 0.2516,
+      "step": 11130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3771138391756645e-05,
+      "loss": 0.5886,
+      "step": 11140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3746487206034613e-05,
+      "loss": 0.3552,
+      "step": 11150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.372183602031258e-05,
+      "loss": 0.3326,
+      "step": 11160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3697184834590545e-05,
+      "loss": 0.3615,
+      "step": 11170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3672533648868513e-05,
+      "loss": 0.2969,
+      "step": 11180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3647882463146477e-05,
+      "loss": 0.5317,
+      "step": 11190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3623231277424445e-05,
+      "loss": 0.2708,
+      "step": 11200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3598580091702413e-05,
+      "loss": 0.2124,
+      "step": 11210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3573928905980377e-05,
+      "loss": 0.4673,
+      "step": 11220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3549277720258345e-05,
+      "loss": 0.2567,
+      "step": 11230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3524626534536312e-05,
+      "loss": 0.3621,
+      "step": 11240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.349997534881428e-05,
+      "loss": 0.379,
+      "step": 11250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3475324163092248e-05,
+      "loss": 0.4306,
+      "step": 11260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3450672977370212e-05,
+      "loss": 0.3438,
+      "step": 11270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.342602179164818e-05,
+      "loss": 0.2015,
+      "step": 11280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3401370605926144e-05,
+      "loss": 0.6039,
+      "step": 11290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3376719420204112e-05,
+      "loss": 0.3251,
+      "step": 11300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.335206823448208e-05,
+      "loss": 0.3646,
+      "step": 11310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3327417048760047e-05,
+      "loss": 0.488,
+      "step": 11320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3302765863038012e-05,
+      "loss": 0.2994,
+      "step": 11330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.327811467731598e-05,
+      "loss": 0.4827,
+      "step": 11340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3253463491593947e-05,
+      "loss": 0.4243,
+      "step": 11350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3228812305871915e-05,
+      "loss": 0.4482,
+      "step": 11360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3204161120149883e-05,
+      "loss": 0.4008,
+      "step": 11370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3179509934427847e-05,
+      "loss": 0.433,
+      "step": 11380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.315485874870581e-05,
+      "loss": 0.2891,
+      "step": 11390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.313020756298378e-05,
+      "loss": 0.3614,
+      "step": 11400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3105556377261747e-05,
+      "loss": 0.3351,
+      "step": 11410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3080905191539714e-05,
+      "loss": 0.4242,
+      "step": 11420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3056254005817682e-05,
+      "loss": 0.4562,
+      "step": 11430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.303160282009565e-05,
+      "loss": 0.1212,
+      "step": 11440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.3006951634373614e-05,
+      "loss": 0.4438,
+      "step": 11450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2982300448651582e-05,
+      "loss": 0.32,
+      "step": 11460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.295764926292955e-05,
+      "loss": 0.3369,
+      "step": 11470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2932998077207514e-05,
+      "loss": 0.5295,
+      "step": 11480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.290834689148548e-05,
+      "loss": 0.2031,
+      "step": 11490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2883695705763446e-05,
+      "loss": 0.3,
+      "step": 11500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2859044520041414e-05,
+      "loss": 0.245,
+      "step": 11510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.283439333431938e-05,
+      "loss": 0.4095,
+      "step": 11520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.280974214859735e-05,
+      "loss": 0.4302,
+      "step": 11530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2785090962875317e-05,
+      "loss": 0.5369,
+      "step": 11540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2760439777153285e-05,
+      "loss": 0.3804,
+      "step": 11550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.273578859143125e-05,
+      "loss": 0.3271,
+      "step": 11560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2711137405709217e-05,
+      "loss": 0.4611,
+      "step": 11570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.268648621998718e-05,
+      "loss": 0.4213,
+      "step": 11580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.266183503426515e-05,
+      "loss": 0.2125,
+      "step": 11590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2637183848543116e-05,
+      "loss": 0.4314,
+      "step": 11600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.261253266282108e-05,
+      "loss": 0.3224,
+      "step": 11610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.258788147709905e-05,
+      "loss": 0.2868,
+      "step": 11620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2563230291377016e-05,
+      "loss": 0.459,
+      "step": 11630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2538579105654984e-05,
+      "loss": 0.3281,
+      "step": 11640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.251392791993295e-05,
+      "loss": 0.4368,
+      "step": 11650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.248927673421092e-05,
+      "loss": 0.1928,
+      "step": 11660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2464625548488884e-05,
+      "loss": 0.3083,
+      "step": 11670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2439974362766848e-05,
+      "loss": 0.3846,
+      "step": 11680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2415323177044816e-05,
+      "loss": 0.5223,
+      "step": 11690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2390671991322783e-05,
+      "loss": 0.1233,
+      "step": 11700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.236602080560075e-05,
+      "loss": 0.3608,
+      "step": 11710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2341369619878715e-05,
+      "loss": 0.5109,
+      "step": 11720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2316718434156683e-05,
+      "loss": 0.452,
+      "step": 11730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.229206724843465e-05,
+      "loss": 0.2085,
+      "step": 11740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.226741606271262e-05,
+      "loss": 0.2834,
+      "step": 11750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2242764876990586e-05,
+      "loss": 0.2736,
+      "step": 11760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.221811369126855e-05,
+      "loss": 0.1662,
+      "step": 11770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.219346250554652e-05,
+      "loss": 0.4305,
+      "step": 11780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2168811319824483e-05,
+      "loss": 0.2479,
+      "step": 11790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.214416013410245e-05,
+      "loss": 0.3548,
+      "step": 11800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2119508948380418e-05,
+      "loss": 0.3846,
+      "step": 11810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2094857762658386e-05,
+      "loss": 0.2706,
+      "step": 11820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2070206576936354e-05,
+      "loss": 0.4998,
+      "step": 11830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2045555391214318e-05,
+      "loss": 0.1841,
+      "step": 11840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.2020904205492286e-05,
+      "loss": 0.435,
+      "step": 11850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1996253019770253e-05,
+      "loss": 0.3946,
+      "step": 11860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1971601834048218e-05,
+      "loss": 0.3709,
+      "step": 11870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1946950648326185e-05,
+      "loss": 0.424,
+      "step": 11880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.192229946260415e-05,
+      "loss": 0.4833,
+      "step": 11890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1897648276882117e-05,
+      "loss": 0.5251,
+      "step": 11900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1872997091160085e-05,
+      "loss": 0.3553,
+      "step": 11910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1848345905438053e-05,
+      "loss": 0.5274,
+      "step": 11920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.182369471971602e-05,
+      "loss": 0.2116,
+      "step": 11930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.179904353399399e-05,
+      "loss": 0.5285,
+      "step": 11940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1774392348271953e-05,
+      "loss": 0.2884,
+      "step": 11950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.174974116254992e-05,
+      "loss": 0.4101,
+      "step": 11960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1725089976827888e-05,
+      "loss": 0.3401,
+      "step": 11970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1700438791105852e-05,
+      "loss": 0.3079,
+      "step": 11980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.167578760538382e-05,
+      "loss": 0.4898,
+      "step": 11990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1651136419661784e-05,
+      "loss": 0.3352,
+      "step": 12000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1626485233939752e-05,
+      "loss": 0.5145,
+      "step": 12010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.160183404821772e-05,
+      "loss": 0.6432,
+      "step": 12020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1577182862495688e-05,
+      "loss": 0.5132,
+      "step": 12030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1552531676773655e-05,
+      "loss": 0.6068,
+      "step": 12040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1527880491051623e-05,
+      "loss": 0.2758,
+      "step": 12050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1503229305329587e-05,
+      "loss": 0.3288,
+      "step": 12060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1478578119607555e-05,
+      "loss": 0.3173,
+      "step": 12070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.145392693388552e-05,
+      "loss": 0.3998,
+      "step": 12080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1429275748163487e-05,
+      "loss": 0.274,
+      "step": 12090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1404624562441455e-05,
+      "loss": 0.4498,
+      "step": 12100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.137997337671942e-05,
+      "loss": 0.3469,
+      "step": 12110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1355322190997387e-05,
+      "loss": 0.4546,
+      "step": 12120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1330671005275355e-05,
+      "loss": 0.2439,
+      "step": 12130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1306019819553322e-05,
+      "loss": 0.4102,
+      "step": 12140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.128136863383129e-05,
+      "loss": 0.3021,
+      "step": 12150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1256717448109258e-05,
+      "loss": 0.4878,
+      "step": 12160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1232066262387222e-05,
+      "loss": 0.424,
+      "step": 12170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1207415076665186e-05,
+      "loss": 0.4337,
+      "step": 12180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1182763890943154e-05,
+      "loss": 0.5545,
+      "step": 12190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1158112705221122e-05,
+      "loss": 0.2055,
+      "step": 12200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.113346151949909e-05,
+      "loss": 0.3117,
+      "step": 12210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1108810333777054e-05,
+      "loss": 0.339,
+      "step": 12220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.108415914805502e-05,
+      "loss": 0.3873,
+      "step": 12230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.105950796233299e-05,
+      "loss": 0.2526,
+      "step": 12240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1034856776610957e-05,
+      "loss": 0.5629,
+      "step": 12250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1010205590888925e-05,
+      "loss": 0.3571,
+      "step": 12260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.098555440516689e-05,
+      "loss": 0.3509,
+      "step": 12270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0960903219444854e-05,
+      "loss": 0.2768,
+      "step": 12280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.093625203372282e-05,
+      "loss": 0.174,
+      "step": 12290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.091160084800079e-05,
+      "loss": 0.2614,
+      "step": 12300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0886949662278757e-05,
+      "loss": 0.3888,
+      "step": 12310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0862298476556724e-05,
+      "loss": 0.2136,
+      "step": 12320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0837647290834692e-05,
+      "loss": 0.3879,
+      "step": 12330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0812996105112656e-05,
+      "loss": 0.379,
+      "step": 12340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0788344919390624e-05,
+      "loss": 0.3293,
+      "step": 12350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0763693733668592e-05,
+      "loss": 0.2004,
+      "step": 12360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0739042547946556e-05,
+      "loss": 0.2802,
+      "step": 12370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0714391362224524e-05,
+      "loss": 0.4183,
+      "step": 12380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0689740176502488e-05,
+      "loss": 0.3064,
+      "step": 12390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0665088990780456e-05,
+      "loss": 0.2741,
+      "step": 12400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0640437805058424e-05,
+      "loss": 0.2975,
+      "step": 12410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.061578661933639e-05,
+      "loss": 0.5264,
+      "step": 12420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.059113543361436e-05,
+      "loss": 0.2412,
+      "step": 12430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0566484247892327e-05,
+      "loss": 0.4583,
+      "step": 12440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.054183306217029e-05,
+      "loss": 0.6049,
+      "step": 12450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.051718187644826e-05,
+      "loss": 0.3883,
+      "step": 12460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0492530690726223e-05,
+      "loss": 0.3947,
+      "step": 12470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.046787950500419e-05,
+      "loss": 0.5501,
+      "step": 12480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.044322831928216e-05,
+      "loss": 0.4407,
+      "step": 12490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0418577133560123e-05,
+      "loss": 0.3461,
+      "step": 12500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.039392594783809e-05,
+      "loss": 0.2866,
+      "step": 12510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.036927476211606e-05,
+      "loss": 0.3218,
+      "step": 12520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0344623576394026e-05,
+      "loss": 0.191,
+      "step": 12530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0319972390671994e-05,
+      "loss": 0.8112,
+      "step": 12540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.029532120494996e-05,
+      "loss": 0.3474,
+      "step": 12550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0270670019227926e-05,
+      "loss": 0.65,
+      "step": 12560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0246018833505894e-05,
+      "loss": 0.1614,
+      "step": 12570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0221367647783858e-05,
+      "loss": 0.3274,
+      "step": 12580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0196716462061826e-05,
+      "loss": 0.5221,
+      "step": 12590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0172065276339793e-05,
+      "loss": 0.1818,
+      "step": 12600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0147414090617758e-05,
+      "loss": 0.2716,
+      "step": 12610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0122762904895725e-05,
+      "loss": 0.1628,
+      "step": 12620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0098111719173693e-05,
+      "loss": 0.4291,
+      "step": 12630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.007346053345166e-05,
+      "loss": 0.4349,
+      "step": 12640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.004880934772963e-05,
+      "loss": 0.3482,
+      "step": 12650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0024158162007596e-05,
+      "loss": 0.441,
+      "step": 12660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999950697628556e-05,
+      "loss": 0.5999,
+      "step": 12670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9974855790563525e-05,
+      "loss": 0.2458,
+      "step": 12680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9950204604841493e-05,
+      "loss": 0.5562,
+      "step": 12690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.992555341911946e-05,
+      "loss": 0.3325,
+      "step": 12700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9900902233397428e-05,
+      "loss": 0.3052,
+      "step": 12710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9876251047675396e-05,
+      "loss": 0.471,
+      "step": 12720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.985159986195336e-05,
+      "loss": 0.2917,
+      "step": 12730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9826948676231328e-05,
+      "loss": 0.3196,
+      "step": 12740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9802297490509296e-05,
+      "loss": 0.2436,
+      "step": 12750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9777646304787263e-05,
+      "loss": 0.5109,
+      "step": 12760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9752995119065228e-05,
+      "loss": 0.6373,
+      "step": 12770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9728343933343192e-05,
+      "loss": 0.3441,
+      "step": 12780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.970369274762116e-05,
+      "loss": 0.367,
+      "step": 12790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9679041561899127e-05,
+      "loss": 0.1593,
+      "step": 12800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9654390376177095e-05,
+      "loss": 0.1967,
+      "step": 12810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9629739190455063e-05,
+      "loss": 0.4012,
+      "step": 12820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.960508800473303e-05,
+      "loss": 0.2573,
+      "step": 12830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9580436819010995e-05,
+      "loss": 0.2843,
+      "step": 12840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9555785633288963e-05,
+      "loss": 0.3308,
+      "step": 12850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953113444756693e-05,
+      "loss": 0.341,
+      "step": 12860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9506483261844895e-05,
+      "loss": 0.2586,
+      "step": 12870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9481832076122862e-05,
+      "loss": 0.4286,
+      "step": 12880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9457180890400827e-05,
+      "loss": 0.2428,
+      "step": 12890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9432529704678794e-05,
+      "loss": 0.1486,
+      "step": 12900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9407878518956762e-05,
+      "loss": 0.4281,
+      "step": 12910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.938322733323473e-05,
+      "loss": 0.2588,
+      "step": 12920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9358576147512698e-05,
+      "loss": 0.2877,
+      "step": 12930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9333924961790665e-05,
+      "loss": 0.2958,
+      "step": 12940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.930927377606863e-05,
+      "loss": 0.5915,
+      "step": 12950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9284622590346597e-05,
+      "loss": 0.2602,
+      "step": 12960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9259971404624562e-05,
+      "loss": 0.2184,
+      "step": 12970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.923532021890253e-05,
+      "loss": 0.2096,
+      "step": 12980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9210669033180497e-05,
+      "loss": 0.349,
+      "step": 12990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.918601784745846e-05,
+      "loss": 0.5189,
+      "step": 13000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.916136666173643e-05,
+      "loss": 0.2695,
+      "step": 13010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9136715476014397e-05,
+      "loss": 0.5485,
+      "step": 13020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9112064290292365e-05,
+      "loss": 0.2635,
+      "step": 13030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9087413104570332e-05,
+      "loss": 0.2592,
+      "step": 13040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.90627619188483e-05,
+      "loss": 0.1576,
+      "step": 13050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9038110733126264e-05,
+      "loss": 0.5118,
+      "step": 13060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.901345954740423e-05,
+      "loss": 0.4693,
+      "step": 13070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8988808361682196e-05,
+      "loss": 0.4258,
+      "step": 13080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8964157175960164e-05,
+      "loss": 0.2251,
+      "step": 13090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8939505990238132e-05,
+      "loss": 0.3187,
+      "step": 13100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.89148548045161e-05,
+      "loss": 0.6121,
+      "step": 13110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8890203618794064e-05,
+      "loss": 0.4171,
+      "step": 13120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.886555243307203e-05,
+      "loss": 0.4736,
+      "step": 13130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.884090124735e-05,
+      "loss": 0.6989,
+      "step": 13140
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8816250061627967e-05,
+      "loss": 0.3552,
+      "step": 13150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.879159887590593e-05,
+      "loss": 0.3787,
+      "step": 13160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.87669476901839e-05,
+      "loss": 0.4889,
+      "step": 13170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8742296504461863e-05,
+      "loss": 0.3474,
+      "step": 13180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.871764531873983e-05,
+      "loss": 0.3398,
+      "step": 13190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.86929941330178e-05,
+      "loss": 0.4574,
+      "step": 13200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8668342947295767e-05,
+      "loss": 0.4,
+      "step": 13210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8643691761573734e-05,
+      "loss": 0.3016,
+      "step": 13220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.86190405758517e-05,
+      "loss": 0.2138,
+      "step": 13230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8594389390129666e-05,
+      "loss": 0.6396,
+      "step": 13240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8569738204407634e-05,
+      "loss": 0.356,
+      "step": 13250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8545087018685602e-05,
+      "loss": 0.4072,
+      "step": 13260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8520435832963566e-05,
+      "loss": 0.4433,
+      "step": 13270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.849578464724153e-05,
+      "loss": 0.2822,
+      "step": 13280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8471133461519498e-05,
+      "loss": 0.2526,
+      "step": 13290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8446482275797466e-05,
+      "loss": 0.2534,
+      "step": 13300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8421831090075434e-05,
+      "loss": 0.4129,
+      "step": 13310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.83971799043534e-05,
+      "loss": 0.4607,
+      "step": 13320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.837252871863137e-05,
+      "loss": 0.4179,
+      "step": 13330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8347877532909333e-05,
+      "loss": 0.4393,
+      "step": 13340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.83232263471873e-05,
+      "loss": 0.4608,
+      "step": 13350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829857516146527e-05,
+      "loss": 0.3207,
+      "step": 13360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8273923975743233e-05,
+      "loss": 0.4117,
+      "step": 13370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82492727900212e-05,
+      "loss": 0.4252,
+      "step": 13380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8224621604299165e-05,
+      "loss": 0.494,
+      "step": 13390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8199970418577133e-05,
+      "loss": 0.4933,
+      "step": 13400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.81753192328551e-05,
+      "loss": 0.4079,
+      "step": 13410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.815066804713307e-05,
+      "loss": 0.3995,
+      "step": 13420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8126016861411036e-05,
+      "loss": 0.5378,
+      "step": 13430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8101365675689004e-05,
+      "loss": 0.2483,
+      "step": 13440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8076714489966968e-05,
+      "loss": 0.1336,
+      "step": 13450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8052063304244936e-05,
+      "loss": 0.5042,
+      "step": 13460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.80274121185229e-05,
+      "loss": 0.3575,
+      "step": 13470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8002760932800868e-05,
+      "loss": 0.4602,
+      "step": 13480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7978109747078836e-05,
+      "loss": 0.3178,
+      "step": 13490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7953458561356803e-05,
+      "loss": 0.28,
+      "step": 13500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7928807375634768e-05,
+      "loss": 0.3413,
+      "step": 13510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7904156189912735e-05,
+      "loss": 0.3754,
+      "step": 13520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7879505004190703e-05,
+      "loss": 0.2949,
+      "step": 13530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.785485381846867e-05,
+      "loss": 0.5453,
+      "step": 13540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.783020263274664e-05,
+      "loss": 0.3991,
+      "step": 13550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7805551447024603e-05,
+      "loss": 0.2848,
+      "step": 13560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7780900261302567e-05,
+      "loss": 0.2857,
+      "step": 13570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7756249075580535e-05,
+      "loss": 0.1661,
+      "step": 13580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7731597889858503e-05,
+      "loss": 0.2878,
+      "step": 13590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.770694670413647e-05,
+      "loss": 0.1108,
+      "step": 13600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7682295518414438e-05,
+      "loss": 0.3388,
+      "step": 13610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7657644332692402e-05,
+      "loss": 0.3631,
+      "step": 13620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.763299314697037e-05,
+      "loss": 0.3831,
+      "step": 13630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7608341961248338e-05,
+      "loss": 0.4041,
+      "step": 13640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7583690775526306e-05,
+      "loss": 0.3253,
+      "step": 13650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.755903958980427e-05,
+      "loss": 0.3236,
+      "step": 13660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7534388404082234e-05,
+      "loss": 0.2762,
+      "step": 13670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7509737218360202e-05,
+      "loss": 0.2728,
+      "step": 13680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.748508603263817e-05,
+      "loss": 0.2454,
+      "step": 13690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7460434846916137e-05,
+      "loss": 0.4133,
+      "step": 13700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7435783661194105e-05,
+      "loss": 0.5144,
+      "step": 13710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7411132475472073e-05,
+      "loss": 0.3197,
+      "step": 13720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7386481289750037e-05,
+      "loss": 0.4292,
+      "step": 13730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7361830104028005e-05,
+      "loss": 0.4785,
+      "step": 13740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7337178918305973e-05,
+      "loss": 0.2871,
+      "step": 13750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7312527732583937e-05,
+      "loss": 0.3349,
+      "step": 13760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7287876546861905e-05,
+      "loss": 0.4314,
+      "step": 13770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.726322536113987e-05,
+      "loss": 0.7642,
+      "step": 13780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7238574175417837e-05,
+      "loss": 0.4122,
+      "step": 13790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7213922989695804e-05,
+      "loss": 0.2957,
+      "step": 13800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7189271803973772e-05,
+      "loss": 0.2526,
+      "step": 13810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.716462061825174e-05,
+      "loss": 0.377,
+      "step": 13820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7139969432529708e-05,
+      "loss": 0.5431,
+      "step": 13830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7115318246807672e-05,
+      "loss": 0.4303,
+      "step": 13840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.709066706108564e-05,
+      "loss": 0.4037,
+      "step": 13850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7066015875363607e-05,
+      "loss": 0.436,
+      "step": 13860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7041364689641572e-05,
+      "loss": 0.3631,
+      "step": 13870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.701671350391954e-05,
+      "loss": 0.2844,
+      "step": 13880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6992062318197507e-05,
+      "loss": 0.2653,
+      "step": 13890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.696741113247547e-05,
+      "loss": 0.4654,
+      "step": 13900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.694275994675344e-05,
+      "loss": 0.199,
+      "step": 13910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6918108761031407e-05,
+      "loss": 0.5463,
+      "step": 13920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6893457575309375e-05,
+      "loss": 0.4209,
+      "step": 13930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6868806389587342e-05,
+      "loss": 0.3791,
+      "step": 13940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6844155203865307e-05,
+      "loss": 0.3089,
+      "step": 13950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6819504018143274e-05,
+      "loss": 0.2555,
+      "step": 13960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.679485283242124e-05,
+      "loss": 0.3116,
+      "step": 13970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6770201646699206e-05,
+      "loss": 0.539,
+      "step": 13980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6745550460977174e-05,
+      "loss": 0.3283,
+      "step": 13990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6720899275255142e-05,
+      "loss": 0.1728,
+      "step": 14000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6696248089533106e-05,
+      "loss": 0.3464,
+      "step": 14010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6671596903811074e-05,
+      "loss": 0.1692,
+      "step": 14020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.664694571808904e-05,
+      "loss": 0.3559,
+      "step": 14030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.662229453236701e-05,
+      "loss": 0.445,
+      "step": 14040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6597643346644977e-05,
+      "loss": 0.1334,
+      "step": 14050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.657299216092294e-05,
+      "loss": 0.3464,
+      "step": 14060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6548340975200906e-05,
+      "loss": 0.2893,
+      "step": 14070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6523689789478873e-05,
+      "loss": 0.5104,
+      "step": 14080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.649903860375684e-05,
+      "loss": 0.1885,
+      "step": 14090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.647438741803481e-05,
+      "loss": 0.4262,
+      "step": 14100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6449736232312777e-05,
+      "loss": 0.121,
+      "step": 14110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.642508504659074e-05,
+      "loss": 0.3456,
+      "step": 14120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.640043386086871e-05,
+      "loss": 0.4471,
+      "step": 14130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6375782675146676e-05,
+      "loss": 0.4457,
+      "step": 14140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6351131489424644e-05,
+      "loss": 0.4693,
+      "step": 14150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.632648030370261e-05,
+      "loss": 0.3575,
+      "step": 14160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6301829117980573e-05,
+      "loss": 0.6242,
+      "step": 14170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.627717793225854e-05,
+      "loss": 0.4741,
+      "step": 14180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6252526746536508e-05,
+      "loss": 0.2356,
+      "step": 14190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6227875560814476e-05,
+      "loss": 0.2205,
+      "step": 14200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6203224375092444e-05,
+      "loss": 0.4903,
+      "step": 14210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.617857318937041e-05,
+      "loss": 0.3181,
+      "step": 14220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6153922003648376e-05,
+      "loss": 0.424,
+      "step": 14230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6129270817926343e-05,
+      "loss": 0.255,
+      "step": 14240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.610461963220431e-05,
+      "loss": 0.1645,
+      "step": 14250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6079968446482275e-05,
+      "loss": 0.5152,
+      "step": 14260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6055317260760243e-05,
+      "loss": 0.3821,
+      "step": 14270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6030666075038208e-05,
+      "loss": 0.3482,
+      "step": 14280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.6006014889316175e-05,
+      "loss": 0.2749,
+      "step": 14290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5981363703594143e-05,
+      "loss": 0.3779,
+      "step": 14300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.595671251787211e-05,
+      "loss": 0.2502,
+      "step": 14310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.593206133215008e-05,
+      "loss": 0.1874,
+      "step": 14320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5907410146428046e-05,
+      "loss": 0.3963,
+      "step": 14330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.588275896070601e-05,
+      "loss": 0.2877,
+      "step": 14340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5858107774983978e-05,
+      "loss": 0.1091,
+      "step": 14350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5833456589261942e-05,
+      "loss": 0.2092,
+      "step": 14360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.580880540353991e-05,
+      "loss": 0.4488,
+      "step": 14370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5784154217817878e-05,
+      "loss": 0.3025,
+      "step": 14380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5759503032095846e-05,
+      "loss": 0.2318,
+      "step": 14390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.573485184637381e-05,
+      "loss": 0.3959,
+      "step": 14400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5710200660651778e-05,
+      "loss": 0.4022,
+      "step": 14410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5685549474929745e-05,
+      "loss": 0.4767,
+      "step": 14420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5660898289207713e-05,
+      "loss": 0.288,
+      "step": 14430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.563624710348568e-05,
+      "loss": 0.2802,
+      "step": 14440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5611595917763645e-05,
+      "loss": 0.2535,
+      "step": 14450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5586944732041613e-05,
+      "loss": 0.3506,
+      "step": 14460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5562293546319577e-05,
+      "loss": 0.3707,
+      "step": 14470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5537642360597545e-05,
+      "loss": 0.2658,
+      "step": 14480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5512991174875513e-05,
+      "loss": 0.3972,
+      "step": 14490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.548833998915348e-05,
+      "loss": 0.4008,
+      "step": 14500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5463688803431445e-05,
+      "loss": 0.3977,
+      "step": 14510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5439037617709412e-05,
+      "loss": 0.3872,
+      "step": 14520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.541438643198738e-05,
+      "loss": 0.3243,
+      "step": 14530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5389735246265348e-05,
+      "loss": 0.438,
+      "step": 14540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5365084060543312e-05,
+      "loss": 0.3489,
+      "step": 14550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.534043287482128e-05,
+      "loss": 0.4832,
+      "step": 14560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5315781689099244e-05,
+      "loss": 0.2555,
+      "step": 14570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5291130503377212e-05,
+      "loss": 0.3954,
+      "step": 14580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.526647931765518e-05,
+      "loss": 0.4425,
+      "step": 14590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5241828131933147e-05,
+      "loss": 0.4202,
+      "step": 14600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5217176946211115e-05,
+      "loss": 0.413,
+      "step": 14610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.519252576048908e-05,
+      "loss": 0.326,
+      "step": 14620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5167874574767047e-05,
+      "loss": 0.3954,
+      "step": 14630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5143223389045013e-05,
+      "loss": 0.2909,
+      "step": 14640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5118572203322981e-05,
+      "loss": 0.1953,
+      "step": 14650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5093921017600949e-05,
+      "loss": 0.3769,
+      "step": 14660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5069269831878913e-05,
+      "loss": 0.3306,
+      "step": 14670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.504461864615688e-05,
+      "loss": 0.3595,
+      "step": 14680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5019967460434847e-05,
+      "loss": 0.3402,
+      "step": 14690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4995316274712814e-05,
+      "loss": 0.3082,
+      "step": 14700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4970665088990782e-05,
+      "loss": 0.5243,
+      "step": 14710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.494601390326875e-05,
+      "loss": 0.3326,
+      "step": 14720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4921362717546714e-05,
+      "loss": 0.2748,
+      "step": 14730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.489671153182468e-05,
+      "loss": 0.4686,
+      "step": 14740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4872060346102648e-05,
+      "loss": 0.3653,
+      "step": 14750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4847409160380616e-05,
+      "loss": 0.4206,
+      "step": 14760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4822757974658583e-05,
+      "loss": 0.36,
+      "step": 14770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.479810678893655e-05,
+      "loss": 0.3486,
+      "step": 14780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4773455603214514e-05,
+      "loss": 0.3479,
+      "step": 14790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4748804417492481e-05,
+      "loss": 0.7005,
+      "step": 14800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.472415323177045e-05,
+      "loss": 0.329,
+      "step": 14810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4699502046048417e-05,
+      "loss": 0.229,
+      "step": 14820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4674850860326383e-05,
+      "loss": 0.3416,
+      "step": 14830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4650199674604347e-05,
+      "loss": 0.3974,
+      "step": 14840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4625548488882315e-05,
+      "loss": 0.2928,
+      "step": 14850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4600897303160283e-05,
+      "loss": 0.5281,
+      "step": 14860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.457624611743825e-05,
+      "loss": 0.3885,
+      "step": 14870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4551594931716216e-05,
+      "loss": 0.2475,
+      "step": 14880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4526943745994184e-05,
+      "loss": 0.2729,
+      "step": 14890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4502292560272148e-05,
+      "loss": 0.1794,
+      "step": 14900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4477641374550116e-05,
+      "loss": 0.2436,
+      "step": 14910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4452990188828084e-05,
+      "loss": 0.2076,
+      "step": 14920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.442833900310605e-05,
+      "loss": 0.4524,
+      "step": 14930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4403687817384018e-05,
+      "loss": 0.1966,
+      "step": 14940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4379036631661982e-05,
+      "loss": 0.5351,
+      "step": 14950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.435438544593995e-05,
+      "loss": 0.4736,
+      "step": 14960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4329734260217917e-05,
+      "loss": 0.408,
+      "step": 14970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4305083074495883e-05,
+      "loss": 0.2242,
+      "step": 14980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4280431888773851e-05,
+      "loss": 0.084,
+      "step": 14990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4255780703051819e-05,
+      "loss": 0.3657,
+      "step": 15000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4231129517329783e-05,
+      "loss": 0.445,
+      "step": 15010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4206478331607751e-05,
+      "loss": 0.3277,
+      "step": 15020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4181827145885717e-05,
+      "loss": 0.2648,
+      "step": 15030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4157175960163685e-05,
+      "loss": 0.4005,
+      "step": 15040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4132524774441652e-05,
+      "loss": 0.2147,
+      "step": 15050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4107873588719617e-05,
+      "loss": 0.2822,
+      "step": 15060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4083222402997584e-05,
+      "loss": 0.3629,
+      "step": 15070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.405857121727555e-05,
+      "loss": 0.4072,
+      "step": 15080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4033920031553518e-05,
+      "loss": 0.2418,
+      "step": 15090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4009268845831486e-05,
+      "loss": 0.3424,
+      "step": 15100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3984617660109454e-05,
+      "loss": 0.388,
+      "step": 15110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3959966474387418e-05,
+      "loss": 0.3001,
+      "step": 15120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3935315288665386e-05,
+      "loss": 0.5886,
+      "step": 15130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3910664102943352e-05,
+      "loss": 0.1943,
+      "step": 15140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.388601291722132e-05,
+      "loss": 0.2503,
+      "step": 15150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3861361731499287e-05,
+      "loss": 0.3945,
+      "step": 15160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3836710545777253e-05,
+      "loss": 0.548,
+      "step": 15170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.381205936005522e-05,
+      "loss": 0.3524,
+      "step": 15180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3787408174333185e-05,
+      "loss": 0.2996,
+      "step": 15190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3762756988611153e-05,
+      "loss": 0.2119,
+      "step": 15200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.373810580288912e-05,
+      "loss": 0.4872,
+      "step": 15210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3713454617167087e-05,
+      "loss": 0.5026,
+      "step": 15220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3688803431445053e-05,
+      "loss": 0.323,
+      "step": 15230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3664152245723019e-05,
+      "loss": 0.3392,
+      "step": 15240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3639501060000986e-05,
+      "loss": 0.347,
+      "step": 15250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3614849874278954e-05,
+      "loss": 0.3396,
+      "step": 15260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3590198688556922e-05,
+      "loss": 0.3511,
+      "step": 15270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3565547502834888e-05,
+      "loss": 0.6511,
+      "step": 15280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3540896317112852e-05,
+      "loss": 0.2951,
+      "step": 15290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.351624513139082e-05,
+      "loss": 0.4393,
+      "step": 15300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3491593945668788e-05,
+      "loss": 0.1996,
+      "step": 15310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3466942759946755e-05,
+      "loss": 0.2178,
+      "step": 15320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3442291574224721e-05,
+      "loss": 0.2665,
+      "step": 15330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3417640388502686e-05,
+      "loss": 0.2148,
+      "step": 15340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3392989202780653e-05,
+      "loss": 0.362,
+      "step": 15350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3368338017058621e-05,
+      "loss": 0.4888,
+      "step": 15360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3343686831336589e-05,
+      "loss": 0.2035,
+      "step": 15370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3319035645614555e-05,
+      "loss": 0.2338,
+      "step": 15380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3294384459892523e-05,
+      "loss": 0.1553,
+      "step": 15390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3269733274170487e-05,
+      "loss": 0.2495,
+      "step": 15400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3245082088448455e-05,
+      "loss": 0.1438,
+      "step": 15410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3220430902726422e-05,
+      "loss": 0.2665,
+      "step": 15420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3195779717004388e-05,
+      "loss": 0.4553,
+      "step": 15430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3171128531282356e-05,
+      "loss": 0.2854,
+      "step": 15440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.314647734556032e-05,
+      "loss": 0.2407,
+      "step": 15450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3121826159838288e-05,
+      "loss": 0.3534,
+      "step": 15460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3097174974116256e-05,
+      "loss": 0.4359,
+      "step": 15470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3072523788394222e-05,
+      "loss": 0.41,
+      "step": 15480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.304787260267219e-05,
+      "loss": 0.2851,
+      "step": 15490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3023221416950157e-05,
+      "loss": 0.4676,
+      "step": 15500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2998570231228122e-05,
+      "loss": 0.4529,
+      "step": 15510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.297391904550609e-05,
+      "loss": 0.1634,
+      "step": 15520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2949267859784055e-05,
+      "loss": 0.6176,
+      "step": 15530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2924616674062023e-05,
+      "loss": 0.2873,
+      "step": 15540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2899965488339991e-05,
+      "loss": 0.6365,
+      "step": 15550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2875314302617959e-05,
+      "loss": 0.4621,
+      "step": 15560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2850663116895923e-05,
+      "loss": 0.4184,
+      "step": 15570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2826011931173889e-05,
+      "loss": 0.1762,
+      "step": 15580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2801360745451857e-05,
+      "loss": 0.2635,
+      "step": 15590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2776709559729824e-05,
+      "loss": 0.047,
+      "step": 15600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2752058374007792e-05,
+      "loss": 0.5074,
+      "step": 15610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2727407188285756e-05,
+      "loss": 0.4086,
+      "step": 15620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2702756002563722e-05,
+      "loss": 0.4381,
+      "step": 15630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.267810481684169e-05,
+      "loss": 0.8051,
+      "step": 15640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2653453631119658e-05,
+      "loss": 0.2921,
+      "step": 15650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2628802445397626e-05,
+      "loss": 0.348,
+      "step": 15660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2604151259675592e-05,
+      "loss": 0.2885,
+      "step": 15670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2579500073953556e-05,
+      "loss": 0.315,
+      "step": 15680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2554848888231524e-05,
+      "loss": 0.3684,
+      "step": 15690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2530197702509491e-05,
+      "loss": 0.1829,
+      "step": 15700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2505546516787459e-05,
+      "loss": 0.3954,
+      "step": 15710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2480895331065425e-05,
+      "loss": 0.3242,
+      "step": 15720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2456244145343391e-05,
+      "loss": 0.2162,
+      "step": 15730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2431592959621359e-05,
+      "loss": 0.4469,
+      "step": 15740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2406941773899325e-05,
+      "loss": 0.2894,
+      "step": 15750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2382290588177293e-05,
+      "loss": 0.3716,
+      "step": 15760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2357639402455259e-05,
+      "loss": 0.4017,
+      "step": 15770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2332988216733225e-05,
+      "loss": 0.3156,
+      "step": 15780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2308337031011192e-05,
+      "loss": 0.4801,
+      "step": 15790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2283685845289158e-05,
+      "loss": 0.3212,
+      "step": 15800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2259034659567126e-05,
+      "loss": 0.2353,
+      "step": 15810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2234383473845092e-05,
+      "loss": 0.4434,
+      "step": 15820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2209732288123058e-05,
+      "loss": 0.141,
+      "step": 15830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2185081102401026e-05,
+      "loss": 0.461,
+      "step": 15840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2160429916678994e-05,
+      "loss": 0.4557,
+      "step": 15850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.213577873095696e-05,
+      "loss": 0.5482,
+      "step": 15860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2111127545234927e-05,
+      "loss": 0.2542,
+      "step": 15870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2086476359512893e-05,
+      "loss": 0.3232,
+      "step": 15880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.206182517379086e-05,
+      "loss": 0.3333,
+      "step": 15890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2037173988068827e-05,
+      "loss": 0.4741,
+      "step": 15900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2012522802346793e-05,
+      "loss": 0.4282,
+      "step": 15910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1987871616624761e-05,
+      "loss": 0.4474,
+      "step": 15920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1963220430902727e-05,
+      "loss": 0.3238,
+      "step": 15930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1938569245180693e-05,
+      "loss": 0.4022,
+      "step": 15940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.191391805945866e-05,
+      "loss": 0.5169,
+      "step": 15950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1889266873736628e-05,
+      "loss": 0.1826,
+      "step": 15960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1864615688014594e-05,
+      "loss": 0.2385,
+      "step": 15970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.183996450229256e-05,
+      "loss": 0.5243,
+      "step": 15980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1815313316570528e-05,
+      "loss": 0.3732,
+      "step": 15990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1790662130848494e-05,
+      "loss": 0.3113,
+      "step": 16000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1766010945126462e-05,
+      "loss": 0.1953,
+      "step": 16010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1741359759404428e-05,
+      "loss": 0.128,
+      "step": 16020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1716708573682394e-05,
+      "loss": 0.3069,
+      "step": 16030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1692057387960362e-05,
+      "loss": 0.4214,
+      "step": 16040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1667406202238328e-05,
+      "loss": 0.3067,
+      "step": 16050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1642755016516295e-05,
+      "loss": 0.292,
+      "step": 16060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1618103830794261e-05,
+      "loss": 0.3178,
+      "step": 16070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1593452645072227e-05,
+      "loss": 0.2644,
+      "step": 16080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1568801459350195e-05,
+      "loss": 0.3853,
+      "step": 16090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1544150273628163e-05,
+      "loss": 0.2958,
+      "step": 16100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1519499087906129e-05,
+      "loss": 0.4036,
+      "step": 16110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1494847902184095e-05,
+      "loss": 0.155,
+      "step": 16120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1470196716462063e-05,
+      "loss": 0.6171,
+      "step": 16130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1445545530740029e-05,
+      "loss": 0.3909,
+      "step": 16140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1420894345017996e-05,
+      "loss": 0.5967,
+      "step": 16150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1396243159295962e-05,
+      "loss": 0.1722,
+      "step": 16160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.137159197357393e-05,
+      "loss": 0.3,
+      "step": 16170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1346940787851896e-05,
+      "loss": 0.329,
+      "step": 16180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1322289602129862e-05,
+      "loss": 0.4152,
+      "step": 16190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.129763841640783e-05,
+      "loss": 0.2645,
+      "step": 16200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1272987230685798e-05,
+      "loss": 0.342,
+      "step": 16210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1248336044963764e-05,
+      "loss": 0.4332,
+      "step": 16220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.122368485924173e-05,
+      "loss": 0.1442,
+      "step": 16230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1199033673519697e-05,
+      "loss": 0.4385,
+      "step": 16240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1174382487797663e-05,
+      "loss": 0.395,
+      "step": 16250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1149731302075631e-05,
+      "loss": 0.3714,
+      "step": 16260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1125080116353597e-05,
+      "loss": 0.2193,
+      "step": 16270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1100428930631563e-05,
+      "loss": 0.5178,
+      "step": 16280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1075777744909531e-05,
+      "loss": 0.2744,
+      "step": 16290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1051126559187497e-05,
+      "loss": 0.2637,
+      "step": 16300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1026475373465465e-05,
+      "loss": 0.4899,
+      "step": 16310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.100182418774343e-05,
+      "loss": 0.1632,
+      "step": 16320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0977173002021397e-05,
+      "loss": 0.4569,
+      "step": 16330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0952521816299364e-05,
+      "loss": 0.3033,
+      "step": 16340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0927870630577332e-05,
+      "loss": 0.5535,
+      "step": 16350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0903219444855298e-05,
+      "loss": 0.2474,
+      "step": 16360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0878568259133264e-05,
+      "loss": 0.3185,
+      "step": 16370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0853917073411232e-05,
+      "loss": 0.43,
+      "step": 16380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0829265887689198e-05,
+      "loss": 0.3472,
+      "step": 16390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0804614701967166e-05,
+      "loss": 0.419,
+      "step": 16400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0779963516245132e-05,
+      "loss": 0.2338,
+      "step": 16410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0755312330523098e-05,
+      "loss": 0.2215,
+      "step": 16420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0730661144801065e-05,
+      "loss": 0.3096,
+      "step": 16430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0706009959079031e-05,
+      "loss": 0.1324,
+      "step": 16440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0681358773357e-05,
+      "loss": 0.2103,
+      "step": 16450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0656707587634967e-05,
+      "loss": 0.182,
+      "step": 16460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0632056401912933e-05,
+      "loss": 0.3384,
+      "step": 16470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0607405216190899e-05,
+      "loss": 0.4207,
+      "step": 16480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0582754030468867e-05,
+      "loss": 0.2617,
+      "step": 16490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0558102844746833e-05,
+      "loss": 0.3457,
+      "step": 16500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.05334516590248e-05,
+      "loss": 0.1975,
+      "step": 16510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0508800473302766e-05,
+      "loss": 0.4073,
+      "step": 16520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0484149287580732e-05,
+      "loss": 0.3414,
+      "step": 16530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.04594981018587e-05,
+      "loss": 0.5428,
+      "step": 16540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0434846916136666e-05,
+      "loss": 0.3908,
+      "step": 16550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0410195730414634e-05,
+      "loss": 0.1762,
+      "step": 16560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.03855445446926e-05,
+      "loss": 0.2366,
+      "step": 16570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0360893358970566e-05,
+      "loss": 0.435,
+      "step": 16580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0336242173248534e-05,
+      "loss": 0.5432,
+      "step": 16590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0311590987526501e-05,
+      "loss": 0.3632,
+      "step": 16600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0286939801804467e-05,
+      "loss": 0.4321,
+      "step": 16610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0262288616082433e-05,
+      "loss": 0.4733,
+      "step": 16620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0237637430360401e-05,
+      "loss": 0.3217,
+      "step": 16630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0212986244638367e-05,
+      "loss": 0.4969,
+      "step": 16640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0188335058916335e-05,
+      "loss": 0.5666,
+      "step": 16650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0163683873194303e-05,
+      "loss": 0.2907,
+      "step": 16660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0139032687472267e-05,
+      "loss": 0.2781,
+      "step": 16670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0114381501750235e-05,
+      "loss": 0.3673,
+      "step": 16680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.00897303160282e-05,
+      "loss": 0.3415,
+      "step": 16690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0065079130306168e-05,
+      "loss": 0.6866,
+      "step": 16700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0040427944584136e-05,
+      "loss": 0.4865,
+      "step": 16710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.00157767588621e-05,
+      "loss": 0.5147,
+      "step": 16720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.991125573140068e-06,
+      "loss": 0.3001,
+      "step": 16730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.966474387418036e-06,
+      "loss": 0.2011,
+      "step": 16740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.941823201696002e-06,
+      "loss": 0.1761,
+      "step": 16750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.91717201597397e-06,
+      "loss": 0.5812,
+      "step": 16760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.892520830251936e-06,
+      "loss": 0.2515,
+      "step": 16770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.867869644529902e-06,
+      "loss": 0.2977,
+      "step": 16780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.84321845880787e-06,
+      "loss": 0.3696,
+      "step": 16790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.818567273085835e-06,
+      "loss": 0.2966,
+      "step": 16800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.793916087363803e-06,
+      "loss": 0.3694,
+      "step": 16810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.76926490164177e-06,
+      "loss": 0.3266,
+      "step": 16820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.744613715919735e-06,
+      "loss": 0.2245,
+      "step": 16830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.719962530197703e-06,
+      "loss": 0.5022,
+      "step": 16840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.69531134447567e-06,
+      "loss": 0.4172,
+      "step": 16850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.670660158753637e-06,
+      "loss": 0.4143,
+      "step": 16860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.646008973031603e-06,
+      "loss": 0.2138,
+      "step": 16870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.62135778730957e-06,
+      "loss": 0.3111,
+      "step": 16880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.596706601587536e-06,
+      "loss": 0.3647,
+      "step": 16890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.572055415865504e-06,
+      "loss": 0.18,
+      "step": 16900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.54740423014347e-06,
+      "loss": 0.3669,
+      "step": 16910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.522753044421436e-06,
+      "loss": 0.2295,
+      "step": 16920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.498101858699404e-06,
+      "loss": 0.5543,
+      "step": 16930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.47345067297737e-06,
+      "loss": 0.4085,
+      "step": 16940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.448799487255338e-06,
+      "loss": 0.4724,
+      "step": 16950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.424148301533305e-06,
+      "loss": 0.281,
+      "step": 16960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.39949711581127e-06,
+      "loss": 0.1761,
+      "step": 16970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.374845930089237e-06,
+      "loss": 0.3814,
+      "step": 16980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.350194744367205e-06,
+      "loss": 0.3777,
+      "step": 16990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.325543558645171e-06,
+      "loss": 0.2603,
+      "step": 17000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.300892372923139e-06,
+      "loss": 0.18,
+      "step": 17010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.276241187201105e-06,
+      "loss": 0.2452,
+      "step": 17020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.251590001479071e-06,
+      "loss": 0.2379,
+      "step": 17030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.226938815757039e-06,
+      "loss": 0.1876,
+      "step": 17040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.202287630035005e-06,
+      "loss": 0.5002,
+      "step": 17050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.177636444312972e-06,
+      "loss": 0.421,
+      "step": 17060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.152985258590938e-06,
+      "loss": 0.4006,
+      "step": 17070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.128334072868904e-06,
+      "loss": 0.2653,
+      "step": 17080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.103682887146872e-06,
+      "loss": 0.3808,
+      "step": 17090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.07903170142484e-06,
+      "loss": 0.3144,
+      "step": 17100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.054380515702806e-06,
+      "loss": 0.6526,
+      "step": 17110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.029729329980772e-06,
+      "loss": 0.3268,
+      "step": 17120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.00507814425874e-06,
+      "loss": 0.4002,
+      "step": 17130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.980426958536706e-06,
+      "loss": 0.3769,
+      "step": 17140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.955775772814673e-06,
+      "loss": 0.2565,
+      "step": 17150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.93112458709264e-06,
+      "loss": 0.2972,
+      "step": 17160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.906473401370605e-06,
+      "loss": 0.5313,
+      "step": 17170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.881822215648573e-06,
+      "loss": 0.4699,
+      "step": 17180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.85717102992654e-06,
+      "loss": 0.3197,
+      "step": 17190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.832519844204507e-06,
+      "loss": 0.5474,
+      "step": 17200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.807868658482473e-06,
+      "loss": 0.409,
+      "step": 17210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.783217472760439e-06,
+      "loss": 0.3291,
+      "step": 17220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.758566287038407e-06,
+      "loss": 0.5911,
+      "step": 17230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.733915101316374e-06,
+      "loss": 0.2967,
+      "step": 17240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.70926391559434e-06,
+      "loss": 0.5989,
+      "step": 17250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.684612729872308e-06,
+      "loss": 0.3953,
+      "step": 17260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.659961544150274e-06,
+      "loss": 0.2315,
+      "step": 17270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.63531035842824e-06,
+      "loss": 0.3014,
+      "step": 17280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.610659172706208e-06,
+      "loss": 0.4268,
+      "step": 17290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.586007986984176e-06,
+      "loss": 0.2072,
+      "step": 17300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.561356801262142e-06,
+      "loss": 0.193,
+      "step": 17310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.536705615540108e-06,
+      "loss": 0.3697,
+      "step": 17320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.512054429818074e-06,
+      "loss": 0.2965,
+      "step": 17330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.487403244096041e-06,
+      "loss": 0.285,
+      "step": 17340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.462752058374009e-06,
+      "loss": 0.4329,
+      "step": 17350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.438100872651975e-06,
+      "loss": 0.4804,
+      "step": 17360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.413449686929941e-06,
+      "loss": 0.4024,
+      "step": 17370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.388798501207909e-06,
+      "loss": 0.4419,
+      "step": 17380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.364147315485875e-06,
+      "loss": 0.3274,
+      "step": 17390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.339496129763843e-06,
+      "loss": 0.2917,
+      "step": 17400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.314844944041809e-06,
+      "loss": 0.2537,
+      "step": 17410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.290193758319775e-06,
+      "loss": 0.1533,
+      "step": 17420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.265542572597742e-06,
+      "loss": 0.3195,
+      "step": 17430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.240891386875708e-06,
+      "loss": 0.4452,
+      "step": 17440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.216240201153676e-06,
+      "loss": 0.1809,
+      "step": 17450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.191589015431642e-06,
+      "loss": 0.3352,
+      "step": 17460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.166937829709608e-06,
+      "loss": 0.23,
+      "step": 17470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.142286643987576e-06,
+      "loss": 0.1711,
+      "step": 17480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.117635458265544e-06,
+      "loss": 0.3563,
+      "step": 17490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.09298427254351e-06,
+      "loss": 0.335,
+      "step": 17500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.068333086821476e-06,
+      "loss": 0.5397,
+      "step": 17510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.043681901099443e-06,
+      "loss": 0.1569,
+      "step": 17520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.01903071537741e-06,
+      "loss": 0.2066,
+      "step": 17530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.994379529655377e-06,
+      "loss": 0.1875,
+      "step": 17540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.969728343933345e-06,
+      "loss": 0.3202,
+      "step": 17550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.945077158211311e-06,
+      "loss": 0.2324,
+      "step": 17560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.920425972489277e-06,
+      "loss": 0.499,
+      "step": 17570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.895774786767243e-06,
+      "loss": 0.3339,
+      "step": 17580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.87112360104521e-06,
+      "loss": 0.4461,
+      "step": 17590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.846472415323178e-06,
+      "loss": 0.3335,
+      "step": 17600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.821821229601144e-06,
+      "loss": 0.2427,
+      "step": 17610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.79717004387911e-06,
+      "loss": 0.2833,
+      "step": 17620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.772518858157078e-06,
+      "loss": 0.2557,
+      "step": 17630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.747867672435044e-06,
+      "loss": 0.2858,
+      "step": 17640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.723216486713012e-06,
+      "loss": 0.5751,
+      "step": 17650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.698565300990978e-06,
+      "loss": 0.2811,
+      "step": 17660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.673914115268944e-06,
+      "loss": 0.3062,
+      "step": 17670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.649262929546912e-06,
+      "loss": 0.1707,
+      "step": 17680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.624611743824879e-06,
+      "loss": 0.2329,
+      "step": 17690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.5999605581028446e-06,
+      "loss": 0.3514,
+      "step": 17700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.575309372380812e-06,
+      "loss": 0.4602,
+      "step": 17710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.550658186658778e-06,
+      "loss": 0.3923,
+      "step": 17720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.526007000936745e-06,
+      "loss": 0.4334,
+      "step": 17730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.501355815214713e-06,
+      "loss": 0.3539,
+      "step": 17740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.476704629492678e-06,
+      "loss": 0.3573,
+      "step": 17750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.452053443770646e-06,
+      "loss": 0.3838,
+      "step": 17760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.427402258048613e-06,
+      "loss": 0.433,
+      "step": 17770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.402751072326579e-06,
+      "loss": 0.4811,
+      "step": 17780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.378099886604546e-06,
+      "loss": 0.6158,
+      "step": 17790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.353448700882513e-06,
+      "loss": 0.2819,
+      "step": 17800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.328797515160479e-06,
+      "loss": 0.4859,
+      "step": 17810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.304146329438446e-06,
+      "loss": 0.4382,
+      "step": 17820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.279495143716412e-06,
+      "loss": 0.3561,
+      "step": 17830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.25484395799438e-06,
+      "loss": 0.4649,
+      "step": 17840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.230192772272347e-06,
+      "loss": 0.2133,
+      "step": 17850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.205541586550313e-06,
+      "loss": 0.3836,
+      "step": 17860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.1808904008282805e-06,
+      "loss": 0.3938,
+      "step": 17870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.156239215106247e-06,
+      "loss": 0.361,
+      "step": 17880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.1315880293842134e-06,
+      "loss": 0.3087,
+      "step": 17890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.10693684366218e-06,
+      "loss": 0.5121,
+      "step": 17900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.082285657940148e-06,
+      "loss": 0.2565,
+      "step": 17910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.057634472218114e-06,
+      "loss": 0.4888,
+      "step": 17920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.032983286496081e-06,
+      "loss": 0.4914,
+      "step": 17930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.008332100774049e-06,
+      "loss": 0.2137,
+      "step": 17940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.983680915052014e-06,
+      "loss": 0.3332,
+      "step": 17950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.9590297293299815e-06,
+      "loss": 0.2159,
+      "step": 17960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.9343785436079476e-06,
+      "loss": 0.2132,
+      "step": 17970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.9097273578859144e-06,
+      "loss": 0.3394,
+      "step": 17980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.885076172163882e-06,
+      "loss": 0.4361,
+      "step": 17990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.860424986441847e-06,
+      "loss": 0.2987,
+      "step": 18000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.835773800719815e-06,
+      "loss": 0.4295,
+      "step": 18010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.811122614997782e-06,
+      "loss": 0.2778,
+      "step": 18020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.786471429275748e-06,
+      "loss": 0.4894,
+      "step": 18030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.761820243553716e-06,
+      "loss": 0.5762,
+      "step": 18040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.7371690578316825e-06,
+      "loss": 0.3306,
+      "step": 18050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.7125178721096486e-06,
+      "loss": 0.5304,
+      "step": 18060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.6878666863876154e-06,
+      "loss": 0.4459,
+      "step": 18070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.6632155006655815e-06,
+      "loss": 0.4393,
+      "step": 18080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.638564314943549e-06,
+      "loss": 0.232,
+      "step": 18090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.613913129221516e-06,
+      "loss": 0.1991,
+      "step": 18100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.589261943499482e-06,
+      "loss": 0.4982,
+      "step": 18110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.564610757777449e-06,
+      "loss": 0.1679,
+      "step": 18120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.539959572055417e-06,
+      "loss": 0.268,
+      "step": 18130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.515308386333383e-06,
+      "loss": 0.28,
+      "step": 18140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.4906572006113496e-06,
+      "loss": 0.265,
+      "step": 18150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.466006014889317e-06,
+      "loss": 0.3037,
+      "step": 18160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.441354829167283e-06,
+      "loss": 0.4575,
+      "step": 18170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.41670364344525e-06,
+      "loss": 0.4269,
+      "step": 18180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.392052457723217e-06,
+      "loss": 0.4635,
+      "step": 18190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.367401272001183e-06,
+      "loss": 0.4057,
+      "step": 18200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.342750086279151e-06,
+      "loss": 0.1586,
+      "step": 18210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.318098900557117e-06,
+      "loss": 0.2082,
+      "step": 18220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.293447714835084e-06,
+      "loss": 0.4236,
+      "step": 18230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.268796529113051e-06,
+      "loss": 0.2193,
+      "step": 18240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.244145343391017e-06,
+      "loss": 0.4059,
+      "step": 18250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.219494157668984e-06,
+      "loss": 0.6304,
+      "step": 18260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.19484297194695e-06,
+      "loss": 0.3164,
+      "step": 18270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.170191786224918e-06,
+      "loss": 0.3164,
+      "step": 18280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.145540600502885e-06,
+      "loss": 0.2126,
+      "step": 18290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.120889414780851e-06,
+      "loss": 0.3119,
+      "step": 18300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.096238229058818e-06,
+      "loss": 0.1626,
+      "step": 18310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.071587043336785e-06,
+      "loss": 0.1983,
+      "step": 18320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.0469358576147515e-06,
+      "loss": 0.3398,
+      "step": 18330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.022284671892718e-06,
+      "loss": 0.312,
+      "step": 18340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.997633486170685e-06,
+      "loss": 0.4185,
+      "step": 18350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.972982300448652e-06,
+      "loss": 0.3027,
+      "step": 18360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.948331114726618e-06,
+      "loss": 0.2054,
+      "step": 18370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.923679929004585e-06,
+      "loss": 0.4365,
+      "step": 18380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.899028743282553e-06,
+      "loss": 0.193,
+      "step": 18390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.874377557560519e-06,
+      "loss": 0.3897,
+      "step": 18400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.849726371838486e-06,
+      "loss": 0.4807,
+      "step": 18410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8250751861164525e-06,
+      "loss": 0.239,
+      "step": 18420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.800424000394419e-06,
+      "loss": 0.141,
+      "step": 18430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.775772814672386e-06,
+      "loss": 0.1387,
+      "step": 18440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.751121628950352e-06,
+      "loss": 0.2759,
+      "step": 18450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.72647044322832e-06,
+      "loss": 0.5047,
+      "step": 18460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.701819257506286e-06,
+      "loss": 0.2187,
+      "step": 18470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.677168071784253e-06,
+      "loss": 0.3091,
+      "step": 18480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.65251688606222e-06,
+      "loss": 0.327,
+      "step": 18490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.627865700340187e-06,
+      "loss": 0.4562,
+      "step": 18500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.6032145146181535e-06,
+      "loss": 0.3327,
+      "step": 18510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5785633288961196e-06,
+      "loss": 0.3244,
+      "step": 18520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.553912143174087e-06,
+      "loss": 0.4432,
+      "step": 18530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.529260957452054e-06,
+      "loss": 0.6273,
+      "step": 18540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.50460977173002e-06,
+      "loss": 0.4407,
+      "step": 18550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.479958586007988e-06,
+      "loss": 0.1739,
+      "step": 18560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.455307400285954e-06,
+      "loss": 0.473,
+      "step": 18570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.430656214563921e-06,
+      "loss": 0.4215,
+      "step": 18580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.406005028841888e-06,
+      "loss": 0.5417,
+      "step": 18590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3813538431198545e-06,
+      "loss": 0.2057,
+      "step": 18600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.356702657397821e-06,
+      "loss": 0.6125,
+      "step": 18610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3320514716757874e-06,
+      "loss": 0.2203,
+      "step": 18620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.307400285953754e-06,
+      "loss": 0.1484,
+      "step": 18630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.282749100231721e-06,
+      "loss": 0.386,
+      "step": 18640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.258097914509688e-06,
+      "loss": 0.4012,
+      "step": 18650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.233446728787655e-06,
+      "loss": 0.3153,
+      "step": 18660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.208795543065622e-06,
+      "loss": 0.4928,
+      "step": 18670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.184144357343589e-06,
+      "loss": 0.2107,
+      "step": 18680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1594931716215555e-06,
+      "loss": 0.3793,
+      "step": 18690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1348419858995216e-06,
+      "loss": 0.2515,
+      "step": 18700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.110190800177489e-06,
+      "loss": 0.2509,
+      "step": 18710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.085539614455455e-06,
+      "loss": 0.3661,
+      "step": 18720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.060888428733422e-06,
+      "loss": 0.3081,
+      "step": 18730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.036237243011389e-06,
+      "loss": 0.2897,
+      "step": 18740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.011586057289356e-06,
+      "loss": 0.3627,
+      "step": 18750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.986934871567323e-06,
+      "loss": 0.1679,
+      "step": 18760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.962283685845289e-06,
+      "loss": 0.1891,
+      "step": 18770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9376325001232565e-06,
+      "loss": 0.3761,
+      "step": 18780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9129813144012226e-06,
+      "loss": 0.3076,
+      "step": 18790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8883301286791894e-06,
+      "loss": 0.2772,
+      "step": 18800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.863678942957157e-06,
+      "loss": 0.4072,
+      "step": 18810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.839027757235123e-06,
+      "loss": 0.2922,
+      "step": 18820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.81437657151309e-06,
+      "loss": 0.4012,
+      "step": 18830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.789725385791057e-06,
+      "loss": 0.3339,
+      "step": 18840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.765074200069024e-06,
+      "loss": 0.5243,
+      "step": 18850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.740423014346991e-06,
+      "loss": 0.1583,
+      "step": 18860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.715771828624957e-06,
+      "loss": 0.5343,
+      "step": 18870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.691120642902924e-06,
+      "loss": 0.1761,
+      "step": 18880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6664694571808904e-06,
+      "loss": 0.5082,
+      "step": 18890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.641818271458857e-06,
+      "loss": 0.2565,
+      "step": 18900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.617167085736824e-06,
+      "loss": 0.3583,
+      "step": 18910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.592515900014791e-06,
+      "loss": 0.3533,
+      "step": 18920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.567864714292758e-06,
+      "loss": 0.2816,
+      "step": 18930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.543213528570724e-06,
+      "loss": 0.1755,
+      "step": 18940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.518562342848691e-06,
+      "loss": 0.1361,
+      "step": 18950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4939111571266585e-06,
+      "loss": 0.2598,
+      "step": 18960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4692599714046246e-06,
+      "loss": 0.271,
+      "step": 18970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4446087856825914e-06,
+      "loss": 0.3031,
+      "step": 18980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.419957599960558e-06,
+      "loss": 0.3661,
+      "step": 18990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.395306414238525e-06,
+      "loss": 0.5415,
+      "step": 19000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.370655228516492e-06,
+      "loss": 0.3253,
+      "step": 19010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.346004042794458e-06,
+      "loss": 0.3956,
+      "step": 19020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.321352857072426e-06,
+      "loss": 0.409,
+      "step": 19030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.296701671350392e-06,
+      "loss": 0.2334,
+      "step": 19040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.272050485628359e-06,
+      "loss": 0.3086,
+      "step": 19050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.247399299906326e-06,
+      "loss": 0.3581,
+      "step": 19060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2227481141842924e-06,
+      "loss": 0.7027,
+      "step": 19070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.198096928462259e-06,
+      "loss": 0.3876,
+      "step": 19080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.173445742740225e-06,
+      "loss": 0.4764,
+      "step": 19090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.148794557018193e-06,
+      "loss": 0.3436,
+      "step": 19100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.12414337129616e-06,
+      "loss": 0.5726,
+      "step": 19110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.099492185574126e-06,
+      "loss": 0.4525,
+      "step": 19120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.074840999852094e-06,
+      "loss": 0.4989,
+      "step": 19130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.05018981413006e-06,
+      "loss": 0.4916,
+      "step": 19140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0255386284080265e-06,
+      "loss": 0.4521,
+      "step": 19150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.000887442685993e-06,
+      "loss": 0.1961,
+      "step": 19160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.97623625696396e-06,
+      "loss": 0.3356,
+      "step": 19170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.951585071241927e-06,
+      "loss": 0.4268,
+      "step": 19180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.926933885519893e-06,
+      "loss": 0.4095,
+      "step": 19190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.902282699797861e-06,
+      "loss": 0.2113,
+      "step": 19200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.877631514075828e-06,
+      "loss": 0.3674,
+      "step": 19210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.852980328353794e-06,
+      "loss": 0.4417,
+      "step": 19220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.828329142631761e-06,
+      "loss": 0.5142,
+      "step": 19230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.803677956909728e-06,
+      "loss": 0.8249,
+      "step": 19240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7790267711876944e-06,
+      "loss": 0.4452,
+      "step": 19250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.754375585465661e-06,
+      "loss": 0.4036,
+      "step": 19260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7297243997436273e-06,
+      "loss": 0.2404,
+      "step": 19270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7050732140215946e-06,
+      "loss": 0.1921,
+      "step": 19280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6804220282995615e-06,
+      "loss": 0.3323,
+      "step": 19290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.655770842577528e-06,
+      "loss": 0.2904,
+      "step": 19300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6311196568554952e-06,
+      "loss": 0.3149,
+      "step": 19310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6064684711334617e-06,
+      "loss": 0.476,
+      "step": 19320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.581817285411428e-06,
+      "loss": 0.375,
+      "step": 19330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.557166099689395e-06,
+      "loss": 0.4899,
+      "step": 19340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5325149139673623e-06,
+      "loss": 0.3565,
+      "step": 19350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5078637282453287e-06,
+      "loss": 0.4295,
+      "step": 19360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.483212542523295e-06,
+      "loss": 0.4022,
+      "step": 19370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4585613568012625e-06,
+      "loss": 0.3183,
+      "step": 19380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4339101710792294e-06,
+      "loss": 0.5083,
+      "step": 19390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.409258985357196e-06,
+      "loss": 0.2479,
+      "step": 19400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3846077996351622e-06,
+      "loss": 0.4415,
+      "step": 19410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3599566139131295e-06,
+      "loss": 0.6481,
+      "step": 19420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.335305428191096e-06,
+      "loss": 0.4261,
+      "step": 19430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.310654242469063e-06,
+      "loss": 0.1803,
+      "step": 19440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.28600305674703e-06,
+      "loss": 0.2892,
+      "step": 19450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2613518710249966e-06,
+      "loss": 0.3315,
+      "step": 19460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.236700685302963e-06,
+      "loss": 0.2591,
+      "step": 19470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.2120494995809295e-06,
+      "loss": 0.4469,
+      "step": 19480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.187398313858897e-06,
+      "loss": 0.247,
+      "step": 19490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1627471281368637e-06,
+      "loss": 0.1789,
+      "step": 19500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.13809594241483e-06,
+      "loss": 0.2931,
+      "step": 19510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.113444756692797e-06,
+      "loss": 0.278,
+      "step": 19520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.088793570970764e-06,
+      "loss": 0.4345,
+      "step": 19530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0641423852487307e-06,
+      "loss": 0.3808,
+      "step": 19540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0394911995266976e-06,
+      "loss": 0.23,
+      "step": 19550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.014840013804664e-06,
+      "loss": 0.2244,
+      "step": 19560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.990188828082631e-06,
+      "loss": 0.5542,
+      "step": 19570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9655376423605974e-06,
+      "loss": 0.4606,
+      "step": 19580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9408864566385642e-06,
+      "loss": 0.3286,
+      "step": 19590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9162352709165315e-06,
+      "loss": 0.3421,
+      "step": 19600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.891584085194498e-06,
+      "loss": 0.18,
+      "step": 19610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.866932899472465e-06,
+      "loss": 0.3044,
+      "step": 19620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8422817137504313e-06,
+      "loss": 0.2459,
+      "step": 19630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.817630528028398e-06,
+      "loss": 0.558,
+      "step": 19640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.792979342306365e-06,
+      "loss": 0.3365,
+      "step": 19650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.768328156584332e-06,
+      "loss": 0.2115,
+      "step": 19660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.743676970862299e-06,
+      "loss": 0.2301,
+      "step": 19670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7190257851402652e-06,
+      "loss": 0.2473,
+      "step": 19680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.694374599418232e-06,
+      "loss": 0.4009,
+      "step": 19690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.669723413696199e-06,
+      "loss": 0.3702,
+      "step": 19700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.645072227974166e-06,
+      "loss": 0.6113,
+      "step": 19710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6204210422521323e-06,
+      "loss": 0.3281,
+      "step": 19720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.595769856530099e-06,
+      "loss": 0.3892,
+      "step": 19730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5711186708080656e-06,
+      "loss": 0.2427,
+      "step": 19740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.546467485086033e-06,
+      "loss": 0.5016,
+      "step": 19750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.521816299364e-06,
+      "loss": 0.4293,
+      "step": 19760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4971651136419662e-06,
+      "loss": 0.3405,
+      "step": 19770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.472513927919933e-06,
+      "loss": 0.5096,
+      "step": 19780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4478627421978996e-06,
+      "loss": 0.2084,
+      "step": 19790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.423211556475867e-06,
+      "loss": 0.2582,
+      "step": 19800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3985603707538333e-06,
+      "loss": 0.2504,
+      "step": 19810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3739091850318e-06,
+      "loss": 0.3199,
+      "step": 19820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.349257999309767e-06,
+      "loss": 0.3321,
+      "step": 19830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3246068135877335e-06,
+      "loss": 0.3011,
+      "step": 19840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2999556278657004e-06,
+      "loss": 0.5959,
+      "step": 19850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2753044421436672e-06,
+      "loss": 0.2567,
+      "step": 19860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.250653256421634e-06,
+      "loss": 0.2896,
+      "step": 19870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2260020706996006e-06,
+      "loss": 0.2773,
+      "step": 19880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2013508849775674e-06,
+      "loss": 0.2607,
+      "step": 19890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1766996992555343e-06,
+      "loss": 0.2719,
+      "step": 19900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.152048513533501e-06,
+      "loss": 0.5321,
+      "step": 19910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.127397327811468e-06,
+      "loss": 0.3946,
+      "step": 19920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1027461420894345e-06,
+      "loss": 0.5159,
+      "step": 19930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0780949563674014e-06,
+      "loss": 0.2976,
+      "step": 19940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0534437706453682e-06,
+      "loss": 0.4534,
+      "step": 19950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.028792584923335e-06,
+      "loss": 0.3106,
+      "step": 19960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0041413992013016e-06,
+      "loss": 0.3205,
+      "step": 19970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9794902134792684e-06,
+      "loss": 0.2273,
+      "step": 19980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9548390277572353e-06,
+      "loss": 0.4238,
+      "step": 19990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9301878420352017e-06,
+      "loss": 0.4192,
+      "step": 20000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9055366563131688e-06,
+      "loss": 0.3962,
+      "step": 20010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8808854705911355e-06,
+      "loss": 0.2994,
+      "step": 20020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8562342848691024e-06,
+      "loss": 0.4449,
+      "step": 20030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.831583099147069e-06,
+      "loss": 0.448,
+      "step": 20040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8069319134250359e-06,
+      "loss": 0.431,
+      "step": 20050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7822807277030023e-06,
+      "loss": 0.3851,
+      "step": 20060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7576295419809694e-06,
+      "loss": 0.4497,
+      "step": 20070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7329783562589363e-06,
+      "loss": 0.3158,
+      "step": 20080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7083271705369027e-06,
+      "loss": 0.3032,
+      "step": 20090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6836759848148698e-06,
+      "loss": 0.4129,
+      "step": 20100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6590247990928363e-06,
+      "loss": 0.3285,
+      "step": 20110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6343736133708034e-06,
+      "loss": 0.2217,
+      "step": 20120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6097224276487698e-06,
+      "loss": 0.29,
+      "step": 20130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5850712419267367e-06,
+      "loss": 0.3063,
+      "step": 20140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5604200562047035e-06,
+      "loss": 0.1756,
+      "step": 20150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5357688704826702e-06,
+      "loss": 0.264,
+      "step": 20160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.511117684760637e-06,
+      "loss": 0.2421,
+      "step": 20170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4864664990386037e-06,
+      "loss": 0.511,
+      "step": 20180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4618153133165706e-06,
+      "loss": 0.3218,
+      "step": 20190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4371641275945375e-06,
+      "loss": 0.1969,
+      "step": 20200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4125129418725041e-06,
+      "loss": 0.3551,
+      "step": 20210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3878617561504708e-06,
+      "loss": 0.3892,
+      "step": 20220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3632105704284377e-06,
+      "loss": 0.2168,
+      "step": 20230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3385593847064043e-06,
+      "loss": 0.2363,
+      "step": 20240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3139081989843712e-06,
+      "loss": 0.3253,
+      "step": 20250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2892570132623379e-06,
+      "loss": 0.3846,
+      "step": 20260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2646058275403047e-06,
+      "loss": 0.3673,
+      "step": 20270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2399546418182716e-06,
+      "loss": 0.5782,
+      "step": 20280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2153034560962383e-06,
+      "loss": 0.5202,
+      "step": 20290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1906522703742051e-06,
+      "loss": 0.1608,
+      "step": 20300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1660010846521718e-06,
+      "loss": 0.3432,
+      "step": 20310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1413498989301385e-06,
+      "loss": 0.3098,
+      "step": 20320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1166987132081053e-06,
+      "loss": 0.1572,
+      "step": 20330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0920475274860722e-06,
+      "loss": 0.2841,
+      "step": 20340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0673963417640389e-06,
+      "loss": 0.2672,
+      "step": 20350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0427451560420057e-06,
+      "loss": 0.436,
+      "step": 20360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0180939703199724e-06,
+      "loss": 0.4984,
+      "step": 20370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.934427845979393e-07,
+      "loss": 0.3363,
+      "step": 20380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.68791598875906e-07,
+      "loss": 0.5749,
+      "step": 20390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.441404131538727e-07,
+      "loss": 0.333,
+      "step": 20400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.194892274318395e-07,
+      "loss": 0.2594,
+      "step": 20410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.948380417098063e-07,
+      "loss": 0.6961,
+      "step": 20420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.701868559877731e-07,
+      "loss": 0.322,
+      "step": 20430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.455356702657399e-07,
+      "loss": 0.4001,
+      "step": 20440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.208844845437066e-07,
+      "loss": 0.4529,
+      "step": 20450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.962332988216734e-07,
+      "loss": 0.3841,
+      "step": 20460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.7158211309964e-07,
+      "loss": 0.351,
+      "step": 20470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.469309273776069e-07,
+      "loss": 0.366,
+      "step": 20480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.222797416555737e-07,
+      "loss": 0.337,
+      "step": 20490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.976285559335405e-07,
+      "loss": 0.4363,
+      "step": 20500
+    }
+  ],
+  "max_steps": 20783,
+  "num_train_epochs": 1,
+  "total_flos": 1400346654696000.0,
+  "trial_name": null,
+  "trial_params": null
+}