diff --git "a/checkpoint-12000/trainer_state.json" "b/checkpoint-12000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-12000/trainer_state.json"
@@ -0,0 +1,7216 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8493921537399798,
+  "global_step": 12000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998454568244e-05,
+      "loss": 1.3539,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999938182748876e-05,
+      "loss": 1.1833,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999870029288556e-05,
+      "loss": 1.173,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999976494017406e-05,
+      "loss": 1.0772,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999962894271507e-05,
+      "loss": 1.0715,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999462037079705e-05,
+      "loss": 1.0268,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999926422347434e-05,
+      "loss": 0.9807,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999903550214352e-05,
+      "loss": 0.9862,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999877587337004e-05,
+      "loss": 0.9725,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999848533747488e-05,
+      "loss": 0.9993,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999816389481725e-05,
+      "loss": 0.9596,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999781154579456e-05,
+      "loss": 0.979,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9997428290842444e-05,
+      "loss": 0.9748,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999701413043471e-05,
+      "loss": 0.9309,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999656906508344e-05,
+      "loss": 0.9143,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999609309533887e-05,
+      "loss": 0.9439,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999558622178947e-05,
+      "loss": 0.9286,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99950484450619e-05,
+      "loss": 0.9544,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999447976582104e-05,
+      "loss": 0.9355,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999388018476998e-05,
+      "loss": 0.9154,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999324970265001e-05,
+      "loss": 0.9326,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999258832024061e-05,
+      "loss": 0.9215,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99918960383595e-05,
+      "loss": 0.9281,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9991172857862555e-05,
+      "loss": 0.935,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99904187796439e-05,
+      "loss": 0.941,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9989633804635814e-05,
+      "loss": 0.9377,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9988817933808814e-05,
+      "loss": 0.9014,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9987971168171585e-05,
+      "loss": 0.9323,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998709350877103e-05,
+      "loss": 0.8987,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998618495669224e-05,
+      "loss": 0.8933,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9985245513058495e-05,
+      "loss": 0.893,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9984275179031276e-05,
+      "loss": 0.909,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998327395581025e-05,
+      "loss": 0.9235,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9982241844633265e-05,
+      "loss": 0.8945,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998117884677638e-05,
+      "loss": 0.9095,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998008496355382e-05,
+      "loss": 0.8919,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9978960196318006e-05,
+      "loss": 0.9088,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997780454645954e-05,
+      "loss": 0.8985,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99766180154072e-05,
+      "loss": 0.8972,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9975400604627957e-05,
+      "loss": 0.8983,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9974152315626935e-05,
+      "loss": 0.9115,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997287314994746e-05,
+      "loss": 0.8957,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.997156310917103e-05,
+      "loss": 0.8681,
+      "step": 430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9970222194917296e-05,
+      "loss": 0.894,
+      "step": 440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996885040884409e-05,
+      "loss": 0.8798,
+      "step": 450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996744775264743e-05,
+      "loss": 0.9034,
+      "step": 460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.996601422806147e-05,
+      "loss": 0.9033,
+      "step": 470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9964549836858536e-05,
+      "loss": 0.8841,
+      "step": 480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9963054580849134e-05,
+      "loss": 0.8877,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.996152846188191e-05,
+      "loss": 0.8729,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995997148184369e-05,
+      "loss": 0.8853,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9958383642659414e-05,
+      "loss": 0.8837,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995676494629221e-05,
+      "loss": 0.8833,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9955115394743354e-05,
+      "loss": 0.8843,
+      "step": 540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995343499005225e-05,
+      "loss": 0.892,
+      "step": 550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.995172373429646e-05,
+      "loss": 0.8575,
+      "step": 560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9949981629591705e-05,
+      "loss": 0.8311,
+      "step": 570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99482086780918e-05,
+      "loss": 0.8669,
+      "step": 580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994640488198874e-05,
+      "loss": 0.8388,
+      "step": 590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994457024351264e-05,
+      "loss": 0.8424,
+      "step": 600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994270476493175e-05,
+      "loss": 0.8676,
+      "step": 610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994080844855243e-05,
+      "loss": 0.8598,
+      "step": 620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.993888129671921e-05,
+      "loss": 0.824,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993692331181469e-05,
+      "loss": 0.8652,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993493449625963e-05,
+      "loss": 0.8533,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993291485251288e-05,
+      "loss": 0.8677,
+      "step": 660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993086438307143e-05,
+      "loss": 0.8459,
+      "step": 670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9928783090470365e-05,
+      "loss": 0.8626,
+      "step": 680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992667097728287e-05,
+      "loss": 0.8127,
+      "step": 690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992452804612027e-05,
+      "loss": 0.8716,
+      "step": 700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992235429963195e-05,
+      "loss": 0.8544,
+      "step": 710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992014974050542e-05,
+      "loss": 0.8562,
+      "step": 720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991791437146627e-05,
+      "loss": 0.871,
+      "step": 730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9915648195278186e-05,
+      "loss": 0.8453,
+      "step": 740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9913351214742945e-05,
+      "loss": 0.8524,
+      "step": 750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.991102343270042e-05,
+      "loss": 0.8581,
+      "step": 760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9908664852028545e-05,
+      "loss": 0.8477,
+      "step": 770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990627547564335e-05,
+      "loss": 0.8651,
+      "step": 780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990385530649891e-05,
+      "loss": 0.8453,
+      "step": 790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9901404347587404e-05,
+      "loss": 0.8586,
+      "step": 800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9898922601939056e-05,
+      "loss": 0.8746,
+      "step": 810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989641007262218e-05,
+      "loss": 0.8652,
+      "step": 820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98938667627431e-05,
+      "loss": 0.8531,
+      "step": 830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989129267544626e-05,
+      "loss": 0.8686,
+      "step": 840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988868781391408e-05,
+      "loss": 0.8692,
+      "step": 850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988605218136711e-05,
+      "loss": 0.8274,
+      "step": 860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9883385781063876e-05,
+      "loss": 0.8502,
+      "step": 870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9880688616300975e-05,
+      "loss": 0.8445,
+      "step": 880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9877960690413035e-05,
+      "loss": 0.8475,
+      "step": 890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987520200677271e-05,
+      "loss": 0.8215,
+      "step": 900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987241256879071e-05,
+      "loss": 0.8389,
+      "step": 910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.986959237991571e-05,
+      "loss": 0.8422,
+      "step": 920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9866741443634455e-05,
+      "loss": 0.8287,
+      "step": 930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.986385976347169e-05,
+      "loss": 0.8694,
+      "step": 940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.986094734299016e-05,
+      "loss": 0.847,
+      "step": 950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.985800418579063e-05,
+      "loss": 0.8191,
+      "step": 960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.985503029551184e-05,
+      "loss": 0.8419,
+      "step": 970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.985202567583057e-05,
+      "loss": 0.8517,
+      "step": 980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.984899033046155e-05,
+      "loss": 0.8653,
+      "step": 990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9845924263157526e-05,
+      "loss": 0.8349,
+      "step": 1000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.984282747770922e-05,
+      "loss": 0.8536,
+      "step": 1010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983969997794531e-05,
+      "loss": 0.8882,
+      "step": 1020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983654176773248e-05,
+      "loss": 0.8285,
+      "step": 1030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983335285097537e-05,
+      "loss": 0.8503,
+      "step": 1040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983013323161657e-05,
+      "loss": 0.8171,
+      "step": 1050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.982688291363666e-05,
+      "loss": 0.8398,
+      "step": 1060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.982360190105414e-05,
+      "loss": 0.8222,
+      "step": 1070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.982029019792548e-05,
+      "loss": 0.8333,
+      "step": 1080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.981694780834508e-05,
+      "loss": 0.8437,
+      "step": 1090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.981357473644531e-05,
+      "loss": 0.827,
+      "step": 1100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9810170986396434e-05,
+      "loss": 0.8216,
+      "step": 1110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.980673656240667e-05,
+      "loss": 0.8253,
+      "step": 1120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9803271468722146e-05,
+      "loss": 0.8195,
+      "step": 1130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9799775709626926e-05,
+      "loss": 0.8394,
+      "step": 1140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9796249289442966e-05,
+      "loss": 0.8348,
+      "step": 1150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9792692212530134e-05,
+      "loss": 0.859,
+      "step": 1160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.978910448328622e-05,
+      "loss": 0.8043,
+      "step": 1170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.97854861061469e-05,
+      "loss": 0.8433,
+      "step": 1180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.978183708558571e-05,
+      "loss": 0.8244,
+      "step": 1190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.977815742611413e-05,
+      "loss": 0.8379,
+      "step": 1200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.977444713228147e-05,
+      "loss": 0.8471,
+      "step": 1210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9770706208674946e-05,
+      "loss": 0.808,
+      "step": 1220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.976693465991963e-05,
+      "loss": 0.8384,
+      "step": 1230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9763132490678453e-05,
+      "loss": 0.856,
+      "step": 1240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.975929970565222e-05,
+      "loss": 0.8382,
+      "step": 1250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.975543630957957e-05,
+      "loss": 0.8219,
+      "step": 1260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.975154230723699e-05,
+      "loss": 0.8384,
+      "step": 1270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9747617703438824e-05,
+      "loss": 0.8276,
+      "step": 1280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.974366250303723e-05,
+      "loss": 0.8604,
+      "step": 1290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.97396767109222e-05,
+      "loss": 0.8471,
+      "step": 1300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.973566033202156e-05,
+      "loss": 0.8199,
+      "step": 1310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.973161337130094e-05,
+      "loss": 0.8243,
+      "step": 1320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.972753583376376e-05,
+      "loss": 0.7936,
+      "step": 1330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.972342772445129e-05,
+      "loss": 0.8231,
+      "step": 1340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9719289048442566e-05,
+      "loss": 0.8223,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.971511981085441e-05,
+      "loss": 0.8174,
+      "step": 1360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9710920016841455e-05,
+      "loss": 0.8088,
+      "step": 1370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9706689671596086e-05,
+      "loss": 0.8149,
+      "step": 1380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.970242878034847e-05,
+      "loss": 0.8522,
+      "step": 1390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.969813734836656e-05,
+      "loss": 0.8404,
+      "step": 1400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.969381538095602e-05,
+      "loss": 0.8608,
+      "step": 1410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.968946288346031e-05,
+      "loss": 0.8232,
+      "step": 1420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.968507986126063e-05,
+      "loss": 0.8368,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9680666319775884e-05,
+      "loss": 0.8154,
+      "step": 1440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.967622226446276e-05,
+      "loss": 0.8379,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9671747700815615e-05,
+      "loss": 0.8333,
+      "step": 1460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.966724263436658e-05,
+      "loss": 0.8542,
+      "step": 1470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9662707070685476e-05,
+      "loss": 0.8421,
+      "step": 1480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9658141015379805e-05,
+      "loss": 0.7827,
+      "step": 1490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9653544474094805e-05,
+      "loss": 0.8659,
+      "step": 1500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9648917452513384e-05,
+      "loss": 0.8166,
+      "step": 1510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.964425995635613e-05,
+      "loss": 0.8221,
+      "step": 1520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.963957199138134e-05,
+      "loss": 0.8129,
+      "step": 1530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.963485356338493e-05,
+      "loss": 0.8171,
+      "step": 1540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9630104678200526e-05,
+      "loss": 0.7984,
+      "step": 1550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.962532534169939e-05,
+      "loss": 0.8109,
+      "step": 1560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.962051555979042e-05,
+      "loss": 0.8164,
+      "step": 1570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9615675338420174e-05,
+      "loss": 0.8063,
+      "step": 1580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.961080468357284e-05,
+      "loss": 0.8123,
+      "step": 1590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9605903601270234e-05,
+      "loss": 0.8322,
+      "step": 1600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.960097209757178e-05,
+      "loss": 0.8256,
+      "step": 1610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.959601017857451e-05,
+      "loss": 0.8113,
+      "step": 1620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.959101785041309e-05,
+      "loss": 0.8323,
+      "step": 1630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.958599511925975e-05,
+      "loss": 0.7911,
+      "step": 1640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.958094199132432e-05,
+      "loss": 0.8175,
+      "step": 1650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.957585847285422e-05,
+      "loss": 0.8114,
+      "step": 1660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.957074457013442e-05,
+      "loss": 0.7619,
+      "step": 1670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.956560028948749e-05,
+      "loss": 0.7909,
+      "step": 1680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.956042563727352e-05,
+      "loss": 0.8274,
+      "step": 1690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.955522061989018e-05,
+      "loss": 0.8251,
+      "step": 1700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9549985243772664e-05,
+      "loss": 0.8129,
+      "step": 1710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.95447195153937e-05,
+      "loss": 0.8211,
+      "step": 1720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9539423441263554e-05,
+      "loss": 0.8131,
+      "step": 1730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9534097027930006e-05,
+      "loss": 0.7954,
+      "step": 1740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.952874028197833e-05,
+      "loss": 0.829,
+      "step": 1750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9523353210031325e-05,
+      "loss": 0.8021,
+      "step": 1760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9517935818749275e-05,
+      "loss": 0.8026,
+      "step": 1770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.951248811482993e-05,
+      "loss": 0.8616,
+      "step": 1780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.950701010500856e-05,
+      "loss": 0.8444,
+      "step": 1790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.950150179605785e-05,
+      "loss": 0.8206,
+      "step": 1800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9495963194787986e-05,
+      "loss": 0.7956,
+      "step": 1810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.94903943080466e-05,
+      "loss": 0.7983,
+      "step": 1820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.948479514271874e-05,
+      "loss": 0.8392,
+      "step": 1830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.947916570572693e-05,
+      "loss": 0.8538,
+      "step": 1840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.947350600403108e-05,
+      "loss": 0.7881,
+      "step": 1850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.946781604462854e-05,
+      "loss": 0.8101,
+      "step": 1860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.946209583455407e-05,
+      "loss": 0.8344,
+      "step": 1870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.945634538087983e-05,
+      "loss": 0.8239,
+      "step": 1880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.945056469071536e-05,
+      "loss": 0.8351,
+      "step": 1890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.94447537712076e-05,
+      "loss": 0.7967,
+      "step": 1900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.943891262954083e-05,
+      "loss": 0.797,
+      "step": 1910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9433041272936734e-05,
+      "loss": 0.8146,
+      "step": 1920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.942713970865435e-05,
+      "loss": 0.8237,
+      "step": 1930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.942120794399002e-05,
+      "loss": 0.7953,
+      "step": 1940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9415245986277483e-05,
+      "loss": 0.8066,
+      "step": 1950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.940925384288775e-05,
+      "loss": 0.8232,
+      "step": 1960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.940323152122921e-05,
+      "loss": 0.8156,
+      "step": 1970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.939717902874751e-05,
+      "loss": 0.8062,
+      "step": 1980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9391096372925626e-05,
+      "loss": 0.7818,
+      "step": 1990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9384983561283824e-05,
+      "loss": 0.8105,
+      "step": 2000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.937884060137966e-05,
+      "loss": 0.8112,
+      "step": 2010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9372667500807944e-05,
+      "loss": 0.8102,
+      "step": 2020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9366464267200755e-05,
+      "loss": 0.8369,
+      "step": 2030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.936023090822744e-05,
+      "loss": 0.7841,
+      "step": 2040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.935396743159459e-05,
+      "loss": 0.8299,
+      "step": 2050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.934767384504602e-05,
+      "loss": 0.8048,
+      "step": 2060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.934135015636276e-05,
+      "loss": 0.825,
+      "step": 2070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.93349963733631e-05,
+      "loss": 0.7928,
+      "step": 2080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9328612503902496e-05,
+      "loss": 0.8016,
+      "step": 2090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.932219855587362e-05,
+      "loss": 0.8134,
+      "step": 2100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.931575453720633e-05,
+      "loss": 0.8109,
+      "step": 2110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.930928045586765e-05,
+      "loss": 0.7908,
+      "step": 2120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9302776319861785e-05,
+      "loss": 0.7936,
+      "step": 2130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.92962421372301e-05,
+      "loss": 0.8008,
+      "step": 2140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.928967791605108e-05,
+      "loss": 0.8237,
+      "step": 2150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.92830836644404e-05,
+      "loss": 0.8127,
+      "step": 2160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9276459390550815e-05,
+      "loss": 0.8168,
+      "step": 2170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.926980510257222e-05,
+      "loss": 0.805,
+      "step": 2180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.926312080873161e-05,
+      "loss": 0.8125,
+      "step": 2190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9256406517293085e-05,
+      "loss": 0.8267,
+      "step": 2200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.924966223655782e-05,
+      "loss": 0.8405,
+      "step": 2210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.92428879748641e-05,
+      "loss": 0.7919,
+      "step": 2220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.923608374058721e-05,
+      "loss": 0.8398,
+      "step": 2230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9229249542139576e-05,
+      "loss": 0.8179,
+      "step": 2240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9222385387970604e-05,
+      "loss": 0.8156,
+      "step": 2250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.921549128656677e-05,
+      "loss": 0.8089,
+      "step": 2260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.920856724645155e-05,
+      "loss": 0.8244,
+      "step": 2270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.920161327618546e-05,
+      "loss": 0.8361,
+      "step": 2280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.919462938436602e-05,
+      "loss": 0.8159,
+      "step": 2290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.918761557962771e-05,
+      "loss": 0.8104,
+      "step": 2300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9180571870642034e-05,
+      "loss": 0.7877,
+      "step": 2310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.917349826611744e-05,
+      "loss": 0.7967,
+      "step": 2320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.916639477479935e-05,
+      "loss": 0.7729,
+      "step": 2330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.915926140547013e-05,
+      "loss": 0.8578,
+      "step": 2340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.915209816694908e-05,
+      "loss": 0.8219,
+      "step": 2350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.914490506809245e-05,
+      "loss": 0.8145,
+      "step": 2360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9137682117793395e-05,
+      "loss": 0.8132,
+      "step": 2370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9130429324981963e-05,
+      "loss": 0.7872,
+      "step": 2380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9123146698625134e-05,
+      "loss": 0.8177,
+      "step": 2390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.911583424772672e-05,
+      "loss": 0.8052,
+      "step": 2400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.910849198132747e-05,
+      "loss": 0.7646,
+      "step": 2410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9101119908504935e-05,
+      "loss": 0.8199,
+      "step": 2420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.909371803837355e-05,
+      "loss": 0.7819,
+      "step": 2430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.908628638008458e-05,
+      "loss": 0.7957,
+      "step": 2440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.907882494282614e-05,
+      "loss": 0.8103,
+      "step": 2450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.907133373582312e-05,
+      "loss": 0.79,
+      "step": 2460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9063812768337246e-05,
+      "loss": 0.8127,
+      "step": 2470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.905626204966705e-05,
+      "loss": 0.7915,
+      "step": 2480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.90486815891478e-05,
+      "loss": 0.8207,
+      "step": 2490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9041071396151585e-05,
+      "loss": 0.8162,
+      "step": 2500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.903343148008722e-05,
+      "loss": 0.8055,
+      "step": 2510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9025761850400283e-05,
+      "loss": 0.8019,
+      "step": 2520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9018062516573086e-05,
+      "loss": 0.801,
+      "step": 2530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.901033348812467e-05,
+      "loss": 0.7831,
+      "step": 2540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9002574774610776e-05,
+      "loss": 0.794,
+      "step": 2550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.899478638562386e-05,
+      "loss": 0.7902,
+      "step": 2560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8986968330793054e-05,
+      "loss": 0.785,
+      "step": 2570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.897912061978418e-05,
+      "loss": 0.8006,
+      "step": 2580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.897124326229972e-05,
+      "loss": 0.8208,
+      "step": 2590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.896333626807881e-05,
+      "loss": 0.7793,
+      "step": 2600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8955399646897215e-05,
+      "loss": 0.812,
+      "step": 2610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.894743340856735e-05,
+      "loss": 0.7948,
+      "step": 2620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.893943756293823e-05,
+      "loss": 0.7955,
+      "step": 2630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.893141211989549e-05,
+      "loss": 0.8363,
+      "step": 2640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.892335708936135e-05,
+      "loss": 0.7986,
+      "step": 2650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.89152724812946e-05,
+      "loss": 0.8249,
+      "step": 2660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.890715830569062e-05,
+      "loss": 0.7951,
+      "step": 2670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.889901457258133e-05,
+      "loss": 0.8098,
+      "step": 2680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.889084129203519e-05,
+      "loss": 0.7781,
+      "step": 2690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.888263847415721e-05,
+      "loss": 0.7817,
+      "step": 2700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.887440612908889e-05,
+      "loss": 0.7848,
+      "step": 2710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.886614426700826e-05,
+      "loss": 0.7965,
+      "step": 2720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8857852898129844e-05,
+      "loss": 0.8067,
+      "step": 2730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.884953203270463e-05,
+      "loss": 0.7933,
+      "step": 2740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.884118168102008e-05,
+      "loss": 0.7918,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.883280185340011e-05,
+      "loss": 0.7758,
+      "step": 2760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8824392560205085e-05,
+      "loss": 0.7765,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.88159538118318e-05,
+      "loss": 0.7848,
+      "step": 2780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8807485618713463e-05,
+      "loss": 0.7852,
+      "step": 2790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8798987991319686e-05,
+      "loss": 0.8201,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.879046094015646e-05,
+      "loss": 0.8024,
+      "step": 2810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8781904475766174e-05,
+      "loss": 0.7921,
+      "step": 2820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.877331860872758e-05,
+      "loss": 0.7541,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.876470334965576e-05,
+      "loss": 0.7689,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.875605870920217e-05,
+      "loss": 0.8107,
+      "step": 2850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8747384698054546e-05,
+      "loss": 0.7784,
+      "step": 2860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.873868132693699e-05,
+      "loss": 0.7825,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.872994860660985e-05,
+      "loss": 0.762,
+      "step": 2880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.872118654786979e-05,
+      "loss": 0.7719,
+      "step": 2890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.871239516154976e-05,
+      "loss": 0.8455,
+      "step": 2900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.870357445851893e-05,
+      "loss": 0.7819,
+      "step": 2910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.869472444968274e-05,
+      "loss": 0.7697,
+      "step": 2920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8685845145982866e-05,
+      "loss": 0.7829,
+      "step": 2930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.867693655839719e-05,
+      "loss": 0.8084,
+      "step": 2940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.866799869793979e-05,
+      "loss": 0.8239,
+      "step": 2950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8659031575660966e-05,
+      "loss": 0.7885,
+      "step": 2960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.865003520264717e-05,
+      "loss": 0.7958,
+      "step": 2970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8641009590021035e-05,
+      "loss": 0.7812,
+      "step": 2980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8631954748941327e-05,
+      "loss": 0.8139,
+      "step": 2990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.862287069060296e-05,
+      "loss": 0.7709,
+      "step": 3000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.861375742623697e-05,
+      "loss": 0.8124,
+      "step": 3010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.860461496711049e-05,
+      "loss": 0.8168,
+      "step": 3020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8595443324526765e-05,
+      "loss": 0.8055,
+      "step": 3030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.858624250982512e-05,
+      "loss": 0.7721,
+      "step": 3040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.857701253438093e-05,
+      "loss": 0.8,
+      "step": 3050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.856775340960563e-05,
+      "loss": 0.825,
+      "step": 3060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.855846514694671e-05,
+      "loss": 0.8102,
+      "step": 3070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.854914775788766e-05,
+      "loss": 0.8078,
+      "step": 3080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.853980125394799e-05,
+      "loss": 0.7921,
+      "step": 3090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.853042564668321e-05,
+      "loss": 0.772,
+      "step": 3100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8521020947684815e-05,
+      "loss": 0.8153,
+      "step": 3110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8511587168580254e-05,
+      "loss": 0.7686,
+      "step": 3120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.850212432103294e-05,
+      "loss": 0.7748,
+      "step": 3130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8492632416742214e-05,
+      "loss": 0.7876,
+      "step": 3140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.848311146744335e-05,
+      "loss": 0.8033,
+      "step": 3150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.847356148490755e-05,
+      "loss": 0.7947,
+      "step": 3160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8463982480941865e-05,
+      "loss": 0.7956,
+      "step": 3170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.845437446738926e-05,
+      "loss": 0.8006,
+      "step": 3180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.844473745612857e-05,
+      "loss": 0.8075,
+      "step": 3190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8435071459074456e-05,
+      "loss": 0.795,
+      "step": 3200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.842537648817743e-05,
+      "loss": 0.7916,
+      "step": 3210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.841565255542384e-05,
+      "loss": 0.7825,
+      "step": 3220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.84058996728358e-05,
+      "loss": 0.8057,
+      "step": 3230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.839611785247125e-05,
+      "loss": 0.7943,
+      "step": 3240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8386307106423924e-05,
+      "loss": 0.8024,
+      "step": 3250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8376467446823266e-05,
+      "loss": 0.7555,
+      "step": 3260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8366598885834496e-05,
+      "loss": 0.7957,
+      "step": 3270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.835670143565857e-05,
+      "loss": 0.7763,
+      "step": 3280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.834677510853216e-05,
+      "loss": 0.8111,
+      "step": 3290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8336819916727624e-05,
+      "loss": 0.764,
+      "step": 3300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.832683587255302e-05,
+      "loss": 0.7501,
+      "step": 3310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.831682298835208e-05,
+      "loss": 0.8185,
+      "step": 3320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8306781276504186e-05,
+      "loss": 0.7918,
+      "step": 3330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8296710749424355e-05,
+      "loss": 0.8076,
+      "step": 3340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.828661141956325e-05,
+      "loss": 0.8178,
+      "step": 3350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8276483299407124e-05,
+      "loss": 0.8239,
+      "step": 3360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.826632640147783e-05,
+      "loss": 0.7565,
+      "step": 3370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.82561407383328e-05,
+      "loss": 0.8099,
+      "step": 3380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.824592632256504e-05,
+      "loss": 0.7945,
+      "step": 3390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.823568316680309e-05,
+      "loss": 0.7583,
+      "step": 3400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.822541128371104e-05,
+      "loss": 0.8081,
+      "step": 3410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.821511068598846e-05,
+      "loss": 0.7955,
+      "step": 3420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.820478138637048e-05,
+      "loss": 0.7948,
+      "step": 3430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8194423397627654e-05,
+      "loss": 0.7969,
+      "step": 3440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.818403673256604e-05,
+      "loss": 0.7719,
+      "step": 3450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.817362140402716e-05,
+      "loss": 0.7689,
+      "step": 3460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.816317742488794e-05,
+      "loss": 0.7976,
+      "step": 3470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.815270480806075e-05,
+      "loss": 0.7869,
+      "step": 3480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.814220356649336e-05,
+      "loss": 0.8099,
+      "step": 3490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.813167371316894e-05,
+      "loss": 0.8057,
+      "step": 3500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.812111526110602e-05,
+      "loss": 0.764,
+      "step": 3510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.811052822335849e-05,
+      "loss": 0.7714,
+      "step": 3520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8099912613015596e-05,
+      "loss": 0.8108,
+      "step": 3530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.808926844320189e-05,
+      "loss": 0.772,
+      "step": 3540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.807859572707725e-05,
+      "loss": 0.8022,
+      "step": 3550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.806789447783683e-05,
+      "loss": 0.7885,
+      "step": 3560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8057164708711064e-05,
+      "loss": 0.7847,
+      "step": 3570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.804640643296568e-05,
+      "loss": 0.7756,
+      "step": 3580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.80356196639016e-05,
+      "loss": 0.7849,
+      "step": 3590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8024804414855e-05,
+      "loss": 0.8072,
+      "step": 3600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.801396069919727e-05,
+      "loss": 0.7894,
+      "step": 3610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.800308853033498e-05,
+      "loss": 0.8029,
+      "step": 3620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7992187921709895e-05,
+      "loss": 0.8059,
+      "step": 3630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.798125888679893e-05,
+      "loss": 0.7736,
+      "step": 3640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7970301439114145e-05,
+      "loss": 0.7819,
+      "step": 3650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.795931559220273e-05,
+      "loss": 0.8138,
+      "step": 3660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.794830135964698e-05,
+      "loss": 0.7952,
+      "step": 3670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.79372587550643e-05,
+      "loss": 0.7933,
+      "step": 3680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.792618779210716e-05,
+      "loss": 0.7588,
+      "step": 3690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.79150884844631e-05,
+      "loss": 0.788,
+      "step": 3700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.790396084585469e-05,
+      "loss": 0.7668,
+      "step": 3710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7892804890039535e-05,
+      "loss": 0.7863,
+      "step": 3720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.788162063081025e-05,
+      "loss": 0.8216,
+      "step": 3730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.787040808199445e-05,
+      "loss": 0.7619,
+      "step": 3740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.785916725745471e-05,
+      "loss": 0.7967,
+      "step": 3750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.784789817108858e-05,
+      "loss": 0.793,
+      "step": 3760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.783660083682853e-05,
+      "loss": 0.7863,
+      "step": 3770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7825275268641984e-05,
+      "loss": 0.7362,
+      "step": 3780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.781392148053124e-05,
+      "loss": 0.7477,
+      "step": 3790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.780253948653352e-05,
+      "loss": 0.7581,
+      "step": 3800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.779112930072087e-05,
+      "loss": 0.7883,
+      "step": 3810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7779690937200254e-05,
+      "loss": 0.7659,
+      "step": 3820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7768224410113424e-05,
+      "loss": 0.7475,
+      "step": 3830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7756729733636976e-05,
+      "loss": 0.7468,
+      "step": 3840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.774520692198228e-05,
+      "loss": 0.7625,
+      "step": 3850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7733655989395533e-05,
+      "loss": 0.7745,
+      "step": 3860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.772207695015767e-05,
+      "loss": 0.7741,
+      "step": 3870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.771046981858439e-05,
+      "loss": 0.7774,
+      "step": 3880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.76988346090261e-05,
+      "loss": 0.7632,
+      "step": 3890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.768717133586795e-05,
+      "loss": 0.7729,
+      "step": 3900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.767548001352978e-05,
+      "loss": 0.7626,
+      "step": 3910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7663760656466085e-05,
+      "loss": 0.771,
+      "step": 3920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.765201327916605e-05,
+      "loss": 0.7865,
+      "step": 3930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.764023789615349e-05,
+      "loss": 0.7758,
+      "step": 3940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7628434521986845e-05,
+      "loss": 0.7699,
+      "step": 3950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.761660317125917e-05,
+      "loss": 0.7967,
+      "step": 3960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.760474385859808e-05,
+      "loss": 0.767,
+      "step": 3970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.75928565986658e-05,
+      "loss": 0.8021,
+      "step": 3980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7580941406159084e-05,
+      "loss": 0.7811,
+      "step": 3990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.756899829580923e-05,
+      "loss": 0.773,
+      "step": 4000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.755702728238204e-05,
+      "loss": 0.7848,
+      "step": 4010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.754502838067782e-05,
+      "loss": 0.7723,
+      "step": 4020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.753300160553136e-05,
+      "loss": 0.7581,
+      "step": 4030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.752094697181192e-05,
+      "loss": 0.8092,
+      "step": 4040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.750886449442318e-05,
+      "loss": 0.7962,
+      "step": 4050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.749675418830325e-05,
+      "loss": 0.7947,
+      "step": 4060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7484616068424656e-05,
+      "loss": 0.7743,
+      "step": 4070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7472450149794314e-05,
+      "loss": 0.7677,
+      "step": 4080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7460256447453486e-05,
+      "loss": 0.7854,
+      "step": 4090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.744803497647782e-05,
+      "loss": 0.7867,
+      "step": 4100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.743578575197726e-05,
+      "loss": 0.7568,
+      "step": 4110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.742350878909608e-05,
+      "loss": 0.7739,
+      "step": 4120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.741120410301286e-05,
+      "loss": 0.8267,
+      "step": 4130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7398871708940426e-05,
+      "loss": 0.7795,
+      "step": 4140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.738651162212589e-05,
+      "loss": 0.7619,
+      "step": 4150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7374123857850575e-05,
+      "loss": 0.7704,
+      "step": 4160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.736170843143004e-05,
+      "loss": 0.7591,
+      "step": 4170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7349265358214043e-05,
+      "loss": 0.7845,
+      "step": 4180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7336794653586534e-05,
+      "loss": 0.7719,
+      "step": 4190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.732429633296558e-05,
+      "loss": 0.7608,
+      "step": 4200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.731177041180346e-05,
+      "loss": 0.758,
+      "step": 4210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7299216905586505e-05,
+      "loss": 0.7861,
+      "step": 4220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.72866358298352e-05,
+      "loss": 0.7758,
+      "step": 4230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.72740272001041e-05,
+      "loss": 0.7504,
+      "step": 4240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.726139103198183e-05,
+      "loss": 0.7682,
+      "step": 4250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.724872734109106e-05,
+      "loss": 0.7687,
+      "step": 4260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.723603614308847e-05,
+      "loss": 0.7583,
+      "step": 4270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7223317453664774e-05,
+      "loss": 0.8159,
+      "step": 4280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.721057128854467e-05,
+      "loss": 0.7985,
+      "step": 4290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.719779766348682e-05,
+      "loss": 0.7919,
+      "step": 4300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7184996594283824e-05,
+      "loss": 0.7549,
+      "step": 4310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.717216809676224e-05,
+      "loss": 0.76,
+      "step": 4320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.715931218678251e-05,
+      "loss": 0.7879,
+      "step": 4330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.714642888023899e-05,
+      "loss": 0.7934,
+      "step": 4340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.71335181930599e-05,
+      "loss": 0.7648,
+      "step": 4350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.712058014120729e-05,
+      "loss": 0.758,
+      "step": 4360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.710761474067707e-05,
+      "loss": 0.8095,
+      "step": 4370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.709462200749897e-05,
+      "loss": 0.7676,
+      "step": 4380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.708160195773648e-05,
+      "loss": 0.7818,
+      "step": 4390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7068554607486866e-05,
+      "loss": 0.7766,
+      "step": 4400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.705547997288118e-05,
+      "loss": 0.7824,
+      "step": 4410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.704237807008418e-05,
+      "loss": 0.7713,
+      "step": 4420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.702924891529434e-05,
+      "loss": 0.7972,
+      "step": 4430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.701609252474384e-05,
+      "loss": 0.766,
+      "step": 4440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7002908914698505e-05,
+      "loss": 0.7817,
+      "step": 4450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.698969810145786e-05,
+      "loss": 0.7626,
+      "step": 4460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6976460101355004e-05,
+      "loss": 0.8012,
+      "step": 4470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.696319493075668e-05,
+      "loss": 0.7746,
+      "step": 4480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.694990260606324e-05,
+      "loss": 0.8053,
+      "step": 4490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6936583143708586e-05,
+      "loss": 0.7903,
+      "step": 4500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.692323656016016e-05,
+      "loss": 0.7562,
+      "step": 4510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.690986287191895e-05,
+      "loss": 0.7919,
+      "step": 4520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.689646209551947e-05,
+      "loss": 0.7616,
+      "step": 4530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.688303424752969e-05,
+      "loss": 0.7718,
+      "step": 4540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6869579344551073e-05,
+      "loss": 0.7858,
+      "step": 4550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6856097403218534e-05,
+      "loss": 0.7657,
+      "step": 4560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6842588440200405e-05,
+      "loss": 0.7698,
+      "step": 4570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.682905247219843e-05,
+      "loss": 0.7716,
+      "step": 4580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.681548951594774e-05,
+      "loss": 0.7889,
+      "step": 4590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.680189958821683e-05,
+      "loss": 0.8046,
+      "step": 4600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.678828270580756e-05,
+      "loss": 0.7613,
+      "step": 4610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.677463888555508e-05,
+      "loss": 0.7745,
+      "step": 4620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6760968144327876e-05,
+      "loss": 0.7697,
+      "step": 4630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.674727049902771e-05,
+      "loss": 0.7795,
+      "step": 4640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6733545966589587e-05,
+      "loss": 0.7851,
+      "step": 4650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.671979456398179e-05,
+      "loss": 0.7905,
+      "step": 4660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.670601630820578e-05,
+      "loss": 0.7617,
+      "step": 4670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6692211216296257e-05,
+      "loss": 0.7769,
+      "step": 4680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.667837930532108e-05,
+      "loss": 0.7952,
+      "step": 4690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.666452059238127e-05,
+      "loss": 0.803,
+      "step": 4700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.665063509461097e-05,
+      "loss": 0.7749,
+      "step": 4710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6636722829177466e-05,
+      "loss": 0.7641,
+      "step": 4720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6622783813281114e-05,
+      "loss": 0.7548,
+      "step": 4730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6608818064155356e-05,
+      "loss": 0.7696,
+      "step": 4740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.659482559906669e-05,
+      "loss": 0.8007,
+      "step": 4750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.658080643531462e-05,
+      "loss": 0.7548,
+      "step": 4760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.656676059023169e-05,
+      "loss": 0.7572,
+      "step": 4770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6552688081183405e-05,
+      "loss": 0.7546,
+      "step": 4780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.653858892556825e-05,
+      "loss": 0.771,
+      "step": 4790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.652446314081765e-05,
+      "loss": 0.7633,
+      "step": 4800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.651031074439596e-05,
+      "loss": 0.7614,
+      "step": 4810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.649613175380043e-05,
+      "loss": 0.7694,
+      "step": 4820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.648192618656118e-05,
+      "loss": 0.7628,
+      "step": 4830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6467694060241206e-05,
+      "loss": 0.7782,
+      "step": 4840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.645343539243633e-05,
+      "loss": 0.7816,
+      "step": 4850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.643915020077519e-05,
+      "loss": 0.7886,
+      "step": 4860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.642483850291922e-05,
+      "loss": 0.7335,
+      "step": 4870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.641050031656262e-05,
+      "loss": 0.7666,
+      "step": 4880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.639613565943233e-05,
+      "loss": 0.7764,
+      "step": 4890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.638174454928805e-05,
+      "loss": 0.7386,
+      "step": 4900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.636732700392215e-05,
+      "loss": 0.7629,
+      "step": 4910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.635288304115969e-05,
+      "loss": 0.7725,
+      "step": 4920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.633841267885841e-05,
+      "loss": 0.7857,
+      "step": 4930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6323915934908665e-05,
+      "loss": 0.7632,
+      "step": 4940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.630939282723344e-05,
+      "loss": 0.7667,
+      "step": 4950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.629484337378832e-05,
+      "loss": 0.7853,
+      "step": 4960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.628026759256145e-05,
+      "loss": 0.7849,
+      "step": 4970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.626566550157353e-05,
+      "loss": 0.7754,
+      "step": 4980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6251037118877784e-05,
+      "loss": 0.7892,
+      "step": 4990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.623638246255996e-05,
+      "loss": 0.7652,
+      "step": 5000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.622170155073825e-05,
+      "loss": 0.7959,
+      "step": 5010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6206994401563355e-05,
+      "loss": 0.7871,
+      "step": 5020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6192261033218384e-05,
+      "loss": 0.7697,
+      "step": 5030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.617750146391887e-05,
+      "loss": 0.7742,
+      "step": 5040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.616271571191273e-05,
+      "loss": 0.775,
+      "step": 5050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.614790379548027e-05,
+      "loss": 0.745,
+      "step": 5060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.613306573293413e-05,
+      "loss": 0.7829,
+      "step": 5070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6118201542619285e-05,
+      "loss": 0.7785,
+      "step": 5080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6103311242913016e-05,
+      "loss": 0.8053,
+      "step": 5090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.608839485222486e-05,
+      "loss": 0.7801,
+      "step": 5100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.607345238899663e-05,
+      "loss": 0.8004,
+      "step": 5110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.605848387170238e-05,
+      "loss": 0.7903,
+      "step": 5120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6043489318848365e-05,
+      "loss": 0.7794,
+      "step": 5130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.602846874897303e-05,
+      "loss": 0.7509,
+      "step": 5140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6013422180646983e-05,
+      "loss": 0.7748,
+      "step": 5150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5998349632472994e-05,
+      "loss": 0.762,
+      "step": 5160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5983251123085925e-05,
+      "loss": 0.7515,
+      "step": 5170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.596812667115275e-05,
+      "loss": 0.7714,
+      "step": 5180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.595297629537252e-05,
+      "loss": 0.7723,
+      "step": 5190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5937800014476334e-05,
+      "loss": 0.7754,
+      "step": 5200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5922597847227316e-05,
+      "loss": 0.7633,
+      "step": 5210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5907369812420595e-05,
+      "loss": 0.7812,
+      "step": 5220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5892115928883274e-05,
+      "loss": 0.7358,
+      "step": 5230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5876836215474434e-05,
+      "loss": 0.7895,
+      "step": 5240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.586153069108507e-05,
+      "loss": 0.7751,
+      "step": 5250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.58461993746381e-05,
+      "loss": 0.7407,
+      "step": 5260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.583084228508833e-05,
+      "loss": 0.7787,
+      "step": 5270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.581545944142243e-05,
+      "loss": 0.7861,
+      "step": 5280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.580005086265888e-05,
+      "loss": 0.7661,
+      "step": 5290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.578461656784805e-05,
+      "loss": 0.7507,
+      "step": 5300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.576915657607202e-05,
+      "loss": 0.7674,
+      "step": 5310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.575367090644471e-05,
+      "loss": 0.7532,
+      "step": 5320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.573815957811174e-05,
+      "loss": 0.7624,
+      "step": 5330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.5722622610250466e-05,
+      "loss": 0.8019,
+      "step": 5340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.570706002206996e-05,
+      "loss": 0.7635,
+      "step": 5350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.569147183281095e-05,
+      "loss": 0.762,
+      "step": 5360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.5675858061745814e-05,
+      "loss": 0.756,
+      "step": 5370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.566021872817858e-05,
+      "loss": 0.7495,
+      "step": 5380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.564455385144486e-05,
+      "loss": 0.761,
+      "step": 5390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.562886345091185e-05,
+      "loss": 0.753,
+      "step": 5400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.561314754597831e-05,
+      "loss": 0.76,
+      "step": 5410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.559740615607453e-05,
+      "loss": 0.7307,
+      "step": 5420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.558163930066229e-05,
+      "loss": 0.7455,
+      "step": 5430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.556584699923488e-05,
+      "loss": 0.7863,
+      "step": 5440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.555002927131704e-05,
+      "loss": 0.7518,
+      "step": 5450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.553418613646494e-05,
+      "loss": 0.735,
+      "step": 5460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.551831761426617e-05,
+      "loss": 0.7715,
+      "step": 5470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5502423724339706e-05,
+      "loss": 0.7423,
+      "step": 5480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5486504486335876e-05,
+      "loss": 0.7504,
+      "step": 5490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.547055991993638e-05,
+      "loss": 0.7598,
+      "step": 5500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5454590044854185e-05,
+      "loss": 0.7517,
+      "step": 5510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5438594880833586e-05,
+      "loss": 0.7533,
+      "step": 5520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5422574447650126e-05,
+      "loss": 0.7872,
+      "step": 5530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.540652876511059e-05,
+      "loss": 0.7777,
+      "step": 5540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5390457853052994e-05,
+      "loss": 0.7838,
+      "step": 5550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5374361731346526e-05,
+      "loss": 0.7678,
+      "step": 5560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.535824041989156e-05,
+      "loss": 0.7444,
+      "step": 5570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.534209393861959e-05,
+      "loss": 0.7691,
+      "step": 5580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5325922307493274e-05,
+      "loss": 0.7975,
+      "step": 5590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.530972554650631e-05,
+      "loss": 0.7718,
+      "step": 5600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.529350367568349e-05,
+      "loss": 0.7626,
+      "step": 5610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.527725671508066e-05,
+      "loss": 0.7574,
+      "step": 5620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5260984684784656e-05,
+      "loss": 0.7403,
+      "step": 5630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.524468760491336e-05,
+      "loss": 0.7511,
+      "step": 5640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.522836549561556e-05,
+      "loss": 0.7649,
+      "step": 5650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5212018377071044e-05,
+      "loss": 0.7782,
+      "step": 5660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5195646269490475e-05,
+      "loss": 0.784,
+      "step": 5670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.517924919311545e-05,
+      "loss": 0.7662,
+      "step": 5680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5162827168218413e-05,
+      "loss": 0.761,
+      "step": 5690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5146380215102666e-05,
+      "loss": 0.7609,
+      "step": 5700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.512990835410231e-05,
+      "loss": 0.7946,
+      "step": 5710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.5113411605582266e-05,
+      "loss": 0.7226,
+      "step": 5720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.509688998993821e-05,
+      "loss": 0.7565,
+      "step": 5730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5080343527596555e-05,
+      "loss": 0.776,
+      "step": 5740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.506377223901447e-05,
+      "loss": 0.779,
+      "step": 5750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.504717614467977e-05,
+      "loss": 0.7387,
+      "step": 5760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5030555265110964e-05,
+      "loss": 0.7812,
+      "step": 5770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.50139096208572e-05,
+      "loss": 0.7568,
+      "step": 5780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.499723923249824e-05,
+      "loss": 0.7773,
+      "step": 5790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.4980544120644456e-05,
+      "loss": 0.7523,
+      "step": 5800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.4963824305936764e-05,
+      "loss": 0.748,
+      "step": 5810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.494707980904662e-05,
+      "loss": 0.7493,
+      "step": 5820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.4930310650676026e-05,
+      "loss": 0.7691,
+      "step": 5830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.491351685155744e-05,
+      "loss": 0.7611,
+      "step": 5840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.4896698432453804e-05,
+      "loss": 0.7332,
+      "step": 5850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.487985541415849e-05,
+      "loss": 0.7486,
+      "step": 5860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.486298781749528e-05,
+      "loss": 0.7807,
+      "step": 5870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.484609566331837e-05,
+      "loss": 0.7707,
+      "step": 5880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.482917897251227e-05,
+      "loss": 0.7831,
+      "step": 5890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.481223776599188e-05,
+      "loss": 0.7667,
+      "step": 5900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.479527206470238e-05,
+      "loss": 0.7681,
+      "step": 5910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.47782818896192e-05,
+      "loss": 0.7836,
+      "step": 5920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.4761267261748106e-05,
+      "loss": 0.7464,
+      "step": 5930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.474422820212504e-05,
+      "loss": 0.7858,
+      "step": 5940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.472716473181617e-05,
+      "loss": 0.7458,
+      "step": 5950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.4710076871917825e-05,
+      "loss": 0.7579,
+      "step": 5960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.4692964643556526e-05,
+      "loss": 0.7861,
+      "step": 5970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.467582806788887e-05,
+      "loss": 0.7688,
+      "step": 5980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.4658667166101605e-05,
+      "loss": 0.7387,
+      "step": 5990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.464148195941152e-05,
+      "loss": 0.7929,
+      "step": 6000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.462427246906548e-05,
+      "loss": 0.7441,
+      "step": 6010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.460703871634035e-05,
+      "loss": 0.746,
+      "step": 6020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.4589780722542994e-05,
+      "loss": 0.7437,
+      "step": 6030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.4572498509010275e-05,
+      "loss": 0.7837,
+      "step": 6040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.4555192097108954e-05,
+      "loss": 0.7534,
+      "step": 6050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.4537861508235746e-05,
+      "loss": 0.7585,
+      "step": 6060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.452050676381725e-05,
+      "loss": 0.7431,
+      "step": 6070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.450312788530991e-05,
+      "loss": 0.769,
+      "step": 6080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.448572489420003e-05,
+      "loss": 0.7781,
+      "step": 6090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.4468297812003724e-05,
+      "loss": 0.7682,
+      "step": 6100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.445084666026688e-05,
+      "loss": 0.8062,
+      "step": 6110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.443337146056515e-05,
+      "loss": 0.7512,
+      "step": 6120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.441587223450391e-05,
+      "loss": 0.7637,
+      "step": 6130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.4398349003718257e-05,
+      "loss": 0.7575,
+      "step": 6140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.438080178987296e-05,
+      "loss": 0.7549,
+      "step": 6150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.436323061466242e-05,
+      "loss": 0.7705,
+      "step": 6160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.434739608795997e-05,
+      "loss": 0.7726,
+      "step": 6170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.432977944602969e-05,
+      "loss": 0.7431,
+      "step": 6180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.431390403463827e-05,
+      "loss": 0.7338,
+      "step": 6190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.429624200461494e-05,
+      "loss": 0.7498,
+      "step": 6200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4278556117771474e-05,
+      "loss": 0.7325,
+      "step": 6210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4260846395973755e-05,
+      "loss": 0.7703,
+      "step": 6220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.424311286111709e-05,
+      "loss": 0.7717,
+      "step": 6230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.422535553512627e-05,
+      "loss": 0.7324,
+      "step": 6240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.420757443995548e-05,
+      "loss": 0.7564,
+      "step": 6250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4189769597588294e-05,
+      "loss": 0.7186,
+      "step": 6260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.417194103003765e-05,
+      "loss": 0.7419,
+      "step": 6270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4154088759345805e-05,
+      "loss": 0.7456,
+      "step": 6280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.4136212807584345e-05,
+      "loss": 0.7672,
+      "step": 6290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.411831319685412e-05,
+      "loss": 0.7548,
+      "step": 6300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.410038994928522e-05,
+      "loss": 0.7847,
+      "step": 6310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.408244308703699e-05,
+      "loss": 0.7269,
+      "step": 6320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.406447263229792e-05,
+      "loss": 0.7509,
+      "step": 6330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.4046478607285725e-05,
+      "loss": 0.749,
+      "step": 6340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.402846103424722e-05,
+      "loss": 0.74,
+      "step": 6350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.401041993545837e-05,
+      "loss": 0.7405,
+      "step": 6360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.399235533322419e-05,
+      "loss": 0.7815,
+      "step": 6370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.397426724987876e-05,
+      "loss": 0.7583,
+      "step": 6380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.3956155707785204e-05,
+      "loss": 0.7438,
+      "step": 6390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.393802072933566e-05,
+      "loss": 0.7448,
+      "step": 6400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.39198623369512e-05,
+      "loss": 0.7583,
+      "step": 6410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.390168055308189e-05,
+      "loss": 0.7528,
+      "step": 6420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.388347540020669e-05,
+      "loss": 0.7568,
+      "step": 6430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.386524690083343e-05,
+      "loss": 0.7638,
+      "step": 6440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.3846995077498875e-05,
+      "loss": 0.7391,
+      "step": 6450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.382871995276856e-05,
+      "loss": 0.7421,
+      "step": 6460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.3810421549236845e-05,
+      "loss": 0.7869,
+      "step": 6470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.37920998895269e-05,
+      "loss": 0.7767,
+      "step": 6480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.37737549962906e-05,
+      "loss": 0.7687,
+      "step": 6490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.375538689220858e-05,
+      "loss": 0.7374,
+      "step": 6500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.373699559999017e-05,
+      "loss": 0.7617,
+      "step": 6510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.371858114237335e-05,
+      "loss": 0.7686,
+      "step": 6520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.3700143542124745e-05,
+      "loss": 0.739,
+      "step": 6530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.36816828220396e-05,
+      "loss": 0.7728,
+      "step": 6540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.3663199004941756e-05,
+      "loss": 0.7622,
+      "step": 6550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.364469211368358e-05,
+      "loss": 0.7655,
+      "step": 6560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.362616217114599e-05,
+      "loss": 0.7227,
+      "step": 6570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.360760920023839e-05,
+      "loss": 0.7899,
+      "step": 6580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.3589033223898654e-05,
+      "loss": 0.7411,
+      "step": 6590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.357043426509312e-05,
+      "loss": 0.7544,
+      "step": 6600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.3551812346816514e-05,
+      "loss": 0.7661,
+      "step": 6610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.3533167492091965e-05,
+      "loss": 0.7741,
+      "step": 6620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.351449972397095e-05,
+      "loss": 0.7939,
+      "step": 6630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.3495809065533275e-05,
+      "loss": 0.7487,
+      "step": 6640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.347709553988707e-05,
+      "loss": 0.7369,
+      "step": 6650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.345835917016869e-05,
+      "loss": 0.74,
+      "step": 6660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.3439599979542775e-05,
+      "loss": 0.7471,
+      "step": 6670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.342081799120216e-05,
+      "loss": 0.7852,
+      "step": 6680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.3402013228367866e-05,
+      "loss": 0.7979,
+      "step": 6690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.3383185714289075e-05,
+      "loss": 0.766,
+      "step": 6700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.336433547224311e-05,
+      "loss": 0.7547,
+      "step": 6710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.334546252553537e-05,
+      "loss": 0.7385,
+      "step": 6720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.332656689749933e-05,
+      "loss": 0.7328,
+      "step": 6730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.3307648611496534e-05,
+      "loss": 0.8058,
+      "step": 6740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.32887076909165e-05,
+      "loss": 0.7683,
+      "step": 6750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.326974415917675e-05,
+      "loss": 0.772,
+      "step": 6760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.325075803972277e-05,
+      "loss": 0.769,
+      "step": 6770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.3231749356027953e-05,
+      "loss": 0.7472,
+      "step": 6780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.32127181315936e-05,
+      "loss": 0.7345,
+      "step": 6790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.319366438994887e-05,
+      "loss": 0.753,
+      "step": 6800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.3174588154650786e-05,
+      "loss": 0.7583,
+      "step": 6810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.3155489449284145e-05,
+      "loss": 0.758,
+      "step": 6820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.313636829746155e-05,
+      "loss": 0.7883,
+      "step": 6830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.311722472282336e-05,
+      "loss": 0.7471,
+      "step": 6840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.309805874903764e-05,
+      "loss": 0.7488,
+      "step": 6850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.307887039980014e-05,
+      "loss": 0.7445,
+      "step": 6860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.30596596988343e-05,
+      "loss": 0.7558,
+      "step": 6870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.3040426669891185e-05,
+      "loss": 0.7653,
+      "step": 6880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.3021171336749456e-05,
+      "loss": 0.7492,
+      "step": 6890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.3001893723215345e-05,
+      "loss": 0.7834,
+      "step": 6900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2982593853122665e-05,
+      "loss": 0.7641,
+      "step": 6910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2963271750332715e-05,
+      "loss": 0.7951,
+      "step": 6920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.294392743873427e-05,
+      "loss": 0.7493,
+      "step": 6930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2924560942243594e-05,
+      "loss": 0.7314,
+      "step": 6940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2905172284804366e-05,
+      "loss": 0.7427,
+      "step": 6950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.288576149038767e-05,
+      "loss": 0.7733,
+      "step": 6960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.286632858299193e-05,
+      "loss": 0.717,
+      "step": 6970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.284687358664296e-05,
+      "loss": 0.7715,
+      "step": 6980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2827396525393834e-05,
+      "loss": 0.7389,
+      "step": 6990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.280789742332494e-05,
+      "loss": 0.7324,
+      "step": 7000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.27883763045439e-05,
+      "loss": 0.7295,
+      "step": 7010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.2768833193185555e-05,
+      "loss": 0.7567,
+      "step": 7020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.2749268113411945e-05,
+      "loss": 0.7474,
+      "step": 7030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.272968108941226e-05,
+      "loss": 0.7627,
+      "step": 7040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.2710072145402834e-05,
+      "loss": 0.7624,
+      "step": 7050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.269044130562709e-05,
+      "loss": 0.7408,
+      "step": 7060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.267078859435554e-05,
+      "loss": 0.7312,
+      "step": 7070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.265111403588571e-05,
+      "loss": 0.728,
+      "step": 7080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.263141765454215e-05,
+      "loss": 0.7289,
+      "step": 7090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.261169947467639e-05,
+      "loss": 0.7292,
+      "step": 7100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.259195952066693e-05,
+      "loss": 0.745,
+      "step": 7110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.257219781691914e-05,
+      "loss": 0.7376,
+      "step": 7120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.255241438786533e-05,
+      "loss": 0.7655,
+      "step": 7130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.253260925796465e-05,
+      "loss": 0.7414,
+      "step": 7140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.251278245170308e-05,
+      "loss": 0.7371,
+      "step": 7150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.249293399359341e-05,
+      "loss": 0.7798,
+      "step": 7160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.247306390817518e-05,
+      "loss": 0.7531,
+      "step": 7170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.245317222001467e-05,
+      "loss": 0.7621,
+      "step": 7180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.243325895370489e-05,
+      "loss": 0.7582,
+      "step": 7190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.2413324133865516e-05,
+      "loss": 0.7491,
+      "step": 7200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.239336778514287e-05,
+      "loss": 0.7751,
+      "step": 7210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.237338993220988e-05,
+      "loss": 0.7497,
+      "step": 7220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.23533905997661e-05,
+      "loss": 0.7692,
+      "step": 7230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.2333369812537583e-05,
+      "loss": 0.7796,
+      "step": 7240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.231332759527695e-05,
+      "loss": 0.7387,
+      "step": 7250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.2293263972763295e-05,
+      "loss": 0.7472,
+      "step": 7260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.227317896980221e-05,
+      "loss": 0.7488,
+      "step": 7270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.225307261122568e-05,
+      "loss": 0.7418,
+      "step": 7280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.223294492189209e-05,
+      "loss": 0.7462,
+      "step": 7290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.2212795926686255e-05,
+      "loss": 0.7761,
+      "step": 7300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.2192625650519265e-05,
+      "loss": 0.7454,
+      "step": 7310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.217243411832856e-05,
+      "loss": 0.7579,
+      "step": 7320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.215222135507784e-05,
+      "loss": 0.773,
+      "step": 7330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.2131987385757066e-05,
+      "loss": 0.7655,
+      "step": 7340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.211173223538242e-05,
+      "loss": 0.7359,
+      "step": 7350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.209145592899625e-05,
+      "loss": 0.7741,
+      "step": 7360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.207115849166709e-05,
+      "loss": 0.7681,
+      "step": 7370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.2050839948489565e-05,
+      "loss": 0.7548,
+      "step": 7380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.203050032458443e-05,
+      "loss": 0.7798,
+      "step": 7390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.2010139645098476e-05,
+      "loss": 0.7405,
+      "step": 7400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1989757935204535e-05,
+      "loss": 0.7491,
+      "step": 7410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1969355220101446e-05,
+      "loss": 0.7777,
+      "step": 7420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.194893152501401e-05,
+      "loss": 0.7521,
+      "step": 7430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.192848687519296e-05,
+      "loss": 0.7891,
+      "step": 7440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.190802129591496e-05,
+      "loss": 0.768,
+      "step": 7450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.188753481248253e-05,
+      "loss": 0.7514,
+      "step": 7460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.186702745022403e-05,
+      "loss": 0.7322,
+      "step": 7470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1846499234493655e-05,
+      "loss": 0.7411,
+      "step": 7480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.182595019067136e-05,
+      "loss": 0.743,
+      "step": 7490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.180538034416287e-05,
+      "loss": 0.7602,
+      "step": 7500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.178478972039961e-05,
+      "loss": 0.7293,
+      "step": 7510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1764178344838716e-05,
+      "loss": 0.763,
+      "step": 7520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.174354624296296e-05,
+      "loss": 0.7368,
+      "step": 7530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.172289344028075e-05,
+      "loss": 0.7689,
+      "step": 7540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.170221996232607e-05,
+      "loss": 0.79,
+      "step": 7550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.16815258346585e-05,
+      "loss": 0.7563,
+      "step": 7560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1660811082863115e-05,
+      "loss": 0.7594,
+      "step": 7570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.164007573255052e-05,
+      "loss": 0.7512,
+      "step": 7580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.161931980935675e-05,
+      "loss": 0.7693,
+      "step": 7590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.15985433389433e-05,
+      "loss": 0.7577,
+      "step": 7600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.157774634699707e-05,
+      "loss": 0.7549,
+      "step": 7610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.155692885923033e-05,
+      "loss": 0.7464,
+      "step": 7620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1536090901380664e-05,
+      "loss": 0.7663,
+      "step": 7630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.151523249921101e-05,
+      "loss": 0.7683,
+      "step": 7640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.149435367850955e-05,
+      "loss": 0.7438,
+      "step": 7650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.14734544650897e-05,
+      "loss": 0.7332,
+      "step": 7660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.145253488479013e-05,
+      "loss": 0.7226,
+      "step": 7670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.143159496347466e-05,
+      "loss": 0.7398,
+      "step": 7680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1410634727032264e-05,
+      "loss": 0.784,
+      "step": 7690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.138965420137704e-05,
+      "loss": 0.7534,
+      "step": 7700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.136865341244815e-05,
+      "loss": 0.746,
+      "step": 7710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1347632386209834e-05,
+      "loss": 0.7369,
+      "step": 7720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.132659114865134e-05,
+      "loss": 0.7417,
+      "step": 7730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.13055297257869e-05,
+      "loss": 0.7658,
+      "step": 7740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1284448143655716e-05,
+      "loss": 0.7414,
+      "step": 7750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.126334642832189e-05,
+      "loss": 0.7202,
+      "step": 7760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1242224605874456e-05,
+      "loss": 0.7547,
+      "step": 7770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.122108270242726e-05,
+      "loss": 0.7254,
+      "step": 7780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.119992074411901e-05,
+      "loss": 0.7217,
+      "step": 7790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1178738757113186e-05,
+      "loss": 0.7806,
+      "step": 7800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.115753676759805e-05,
+      "loss": 0.7418,
+      "step": 7810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.113631480178657e-05,
+      "loss": 0.7323,
+      "step": 7820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.111507288591645e-05,
+      "loss": 0.7351,
+      "step": 7830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.109381104625001e-05,
+      "loss": 0.7437,
+      "step": 7840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1072529309074235e-05,
+      "loss": 0.7061,
+      "step": 7850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.105122770070071e-05,
+      "loss": 0.7358,
+      "step": 7860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1029906247465576e-05,
+      "loss": 0.7275,
+      "step": 7870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1008564975729514e-05,
+      "loss": 0.8013,
+      "step": 7880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.098720391187771e-05,
+      "loss": 0.7475,
+      "step": 7890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.096582308231981e-05,
+      "loss": 0.7264,
+      "step": 7900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.094442251348991e-05,
+      "loss": 0.7853,
+      "step": 7910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.092300223184651e-05,
+      "loss": 0.7747,
+      "step": 7920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0901562263872465e-05,
+      "loss": 0.7651,
+      "step": 7930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.088010263607499e-05,
+      "loss": 0.7529,
+      "step": 7940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.08586233749856e-05,
+      "loss": 0.7526,
+      "step": 7950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0837124507160064e-05,
+      "loss": 0.7322,
+      "step": 7960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0815606059178423e-05,
+      "loss": 0.757,
+      "step": 7970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0794068057644904e-05,
+      "loss": 0.7799,
+      "step": 7980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0772510529187924e-05,
+      "loss": 0.7197,
+      "step": 7990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0750933500460025e-05,
+      "loss": 0.7224,
+      "step": 8000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.072933699813788e-05,
+      "loss": 0.7208,
+      "step": 8010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.070772104892221e-05,
+      "loss": 0.7544,
+      "step": 8020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.068608567953781e-05,
+      "loss": 0.7631,
+      "step": 8030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.066443091673345e-05,
+      "loss": 0.7584,
+      "step": 8040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.064275678728191e-05,
+      "loss": 0.7454,
+      "step": 8050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0621063317979904e-05,
+      "loss": 0.7882,
+      "step": 8060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059935053564805e-05,
+      "loss": 0.7521,
+      "step": 8070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057761846713084e-05,
+      "loss": 0.7452,
+      "step": 8080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055586713929662e-05,
+      "loss": 0.7729,
+      "step": 8090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053409657903755e-05,
+      "loss": 0.7471,
+      "step": 8100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0512306813269555e-05,
+      "loss": 0.7553,
+      "step": 8110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0490497868932306e-05,
+      "loss": 0.7342,
+      "step": 8120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.046866977298921e-05,
+      "loss": 0.7419,
+      "step": 8130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.044682255242732e-05,
+      "loss": 0.7688,
+      "step": 8140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042495623425735e-05,
+      "loss": 0.7387,
+      "step": 8150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040307084551362e-05,
+      "loss": 0.7394,
+      "step": 8160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.038116641325403e-05,
+      "loss": 0.7233,
+      "step": 8170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035924296456003e-05,
+      "loss": 0.7869,
+      "step": 8180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033730052653656e-05,
+      "loss": 0.7391,
+      "step": 8190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031533912631207e-05,
+      "loss": 0.7531,
+      "step": 8200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0293358791038426e-05,
+      "loss": 0.7616,
+      "step": 8210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.027135954789093e-05,
+      "loss": 0.7474,
+      "step": 8220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.024934142406822e-05,
+      "loss": 0.7436,
+      "step": 8230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0227304446792313e-05,
+      "loss": 0.7671,
+      "step": 8240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.020524864330854e-05,
+      "loss": 0.7358,
+      "step": 8250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.018317404088546e-05,
+      "loss": 0.7542,
+      "step": 8260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.016108066681494e-05,
+      "loss": 0.7609,
+      "step": 8270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0138968548412006e-05,
+      "loss": 0.7676,
+      "step": 8280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.011683771301486e-05,
+      "loss": 0.7197,
+      "step": 8290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.009468818798488e-05,
+      "loss": 0.7711,
+      "step": 8300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.007252000070653e-05,
+      "loss": 0.7477,
+      "step": 8310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.005033317858734e-05,
+      "loss": 0.7677,
+      "step": 8320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.002812774905788e-05,
+      "loss": 0.739,
+      "step": 8330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0005903739571725e-05,
+      "loss": 0.7243,
+      "step": 8340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.998366117760545e-05,
+      "loss": 0.7648,
+      "step": 8350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9961400090658526e-05,
+      "loss": 0.721,
+      "step": 8360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.993912050625336e-05,
+      "loss": 0.7516,
+      "step": 8370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991682245193519e-05,
+      "loss": 0.7644,
+      "step": 8380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.989450595527214e-05,
+      "loss": 0.7364,
+      "step": 8390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.987217104385509e-05,
+      "loss": 0.7517,
+      "step": 8400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.984981774529771e-05,
+      "loss": 0.7686,
+      "step": 8410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.982744608723641e-05,
+      "loss": 0.7526,
+      "step": 8420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.980505609733027e-05,
+      "loss": 0.7468,
+      "step": 8430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.978264780326105e-05,
+      "loss": 0.7765,
+      "step": 8440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.976022123273316e-05,
+      "loss": 0.7367,
+      "step": 8450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.973777641347357e-05,
+      "loss": 0.732,
+      "step": 8460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.971531337323183e-05,
+      "loss": 0.7508,
+      "step": 8470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.969283213978003e-05,
+      "loss": 0.739,
+      "step": 8480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.967033274091273e-05,
+      "loss": 0.7511,
+      "step": 8490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.964781520444696e-05,
+      "loss": 0.7497,
+      "step": 8500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.962527955822217e-05,
+      "loss": 0.7393,
+      "step": 8510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.96027258301002e-05,
+      "loss": 0.7489,
+      "step": 8520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.958015404796526e-05,
+      "loss": 0.7484,
+      "step": 8530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.955756423972385e-05,
+      "loss": 0.7324,
+      "step": 8540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9534956433304806e-05,
+      "loss": 0.7289,
+      "step": 8550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9512330656659155e-05,
+      "loss": 0.7621,
+      "step": 8560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9489686937760195e-05,
+      "loss": 0.7426,
+      "step": 8570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.946702530460337e-05,
+      "loss": 0.7531,
+      "step": 8580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9444345785206285e-05,
+      "loss": 0.7292,
+      "step": 8590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.942164840760866e-05,
+      "loss": 0.7191,
+      "step": 8600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.93989331998723e-05,
+      "loss": 0.7325,
+      "step": 8610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.937620019008105e-05,
+      "loss": 0.7309,
+      "step": 8620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9353449406340755e-05,
+      "loss": 0.7346,
+      "step": 8630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.933068087677924e-05,
+      "loss": 0.7604,
+      "step": 8640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.930789462954628e-05,
+      "loss": 0.7602,
+      "step": 8650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9285090692813544e-05,
+      "loss": 0.7238,
+      "step": 8660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9262269094774564e-05,
+      "loss": 0.7481,
+      "step": 8670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9239429863644736e-05,
+      "loss": 0.7412,
+      "step": 8680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.921657302766123e-05,
+      "loss": 0.7643,
+      "step": 8690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9193698615082995e-05,
+      "loss": 0.7115,
+      "step": 8700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9170806654190695e-05,
+      "loss": 0.77,
+      "step": 8710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.914789717328671e-05,
+      "loss": 0.7304,
+      "step": 8720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.912497020069505e-05,
+      "loss": 0.7337,
+      "step": 8730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.910202576476142e-05,
+      "loss": 0.7589,
+      "step": 8740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.907906389385302e-05,
+      "loss": 0.733,
+      "step": 8750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9056084616358666e-05,
+      "loss": 0.7525,
+      "step": 8760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.90330879606887e-05,
+      "loss": 0.7483,
+      "step": 8770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9010073955274915e-05,
+      "loss": 0.7159,
+      "step": 8780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.898704262857057e-05,
+      "loss": 0.7235,
+      "step": 8790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8963994009050356e-05,
+      "loss": 0.7327,
+      "step": 8800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.894092812521031e-05,
+      "loss": 0.7502,
+      "step": 8810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.891784500556784e-05,
+      "loss": 0.7344,
+      "step": 8820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8894744678661655e-05,
+      "loss": 0.7401,
+      "step": 8830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.887162717305173e-05,
+      "loss": 0.7561,
+      "step": 8840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.88484925173193e-05,
+      "loss": 0.7565,
+      "step": 8850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.882534074006678e-05,
+      "loss": 0.7528,
+      "step": 8860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8802171869917765e-05,
+      "loss": 0.7342,
+      "step": 8870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8778985935516985e-05,
+      "loss": 0.7542,
+      "step": 8880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8755782965530265e-05,
+      "loss": 0.7435,
+      "step": 8890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.873256298864448e-05,
+      "loss": 0.7558,
+      "step": 8900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.870932603356755e-05,
+      "loss": 0.7552,
+      "step": 8910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8686072129028385e-05,
+      "loss": 0.7223,
+      "step": 8920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.866280130377682e-05,
+      "loss": 0.7385,
+      "step": 8930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8639513586583656e-05,
+      "loss": 0.7372,
+      "step": 8940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.861620900624054e-05,
+      "loss": 0.7408,
+      "step": 8950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.859288759156e-05,
+      "loss": 0.7633,
+      "step": 8960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8569549371375346e-05,
+      "loss": 0.7412,
+      "step": 8970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.854619437454068e-05,
+      "loss": 0.7195,
+      "step": 8980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.8522822629930844e-05,
+      "loss": 0.7281,
+      "step": 8990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.849943416644139e-05,
+      "loss": 0.7029,
+      "step": 9000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.847602901298854e-05,
+      "loss": 0.7543,
+      "step": 9010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.845260719850915e-05,
+      "loss": 0.7569,
+      "step": 9020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.842916875196066e-05,
+      "loss": 0.7212,
+      "step": 9030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.84057137023211e-05,
+      "loss": 0.734,
+      "step": 9040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.8382242078589006e-05,
+      "loss": 0.7038,
+      "step": 9050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.8358753909783405e-05,
+      "loss": 0.7444,
+      "step": 9060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.83352492249438e-05,
+      "loss": 0.7663,
+      "step": 9070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.831172805313009e-05,
+      "loss": 0.7659,
+      "step": 9080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.8288190423422585e-05,
+      "loss": 0.7406,
+      "step": 9090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.8264636364921904e-05,
+      "loss": 0.7292,
+      "step": 9100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.824106590674901e-05,
+      "loss": 0.7383,
+      "step": 9110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.821747907804513e-05,
+      "loss": 0.7222,
+      "step": 9120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.819387590797172e-05,
+      "loss": 0.7535,
+      "step": 9130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.817025642571046e-05,
+      "loss": 0.7512,
+      "step": 9140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.814662066046319e-05,
+      "loss": 0.7285,
+      "step": 9150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.81229686414519e-05,
+      "loss": 0.7604,
+      "step": 9160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.8099300397918606e-05,
+      "loss": 0.7449,
+      "step": 9170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.8075615959125465e-05,
+      "loss": 0.7395,
+      "step": 9180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.805191535435463e-05,
+      "loss": 0.7444,
+      "step": 9190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.802819861290822e-05,
+      "loss": 0.7471,
+      "step": 9200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.800446576410831e-05,
+      "loss": 0.7874,
+      "step": 9210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.7980716837296924e-05,
+      "loss": 0.7581,
+      "step": 9220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.795695186183592e-05,
+      "loss": 0.7719,
+      "step": 9230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.793317086710703e-05,
+      "loss": 0.7324,
+      "step": 9240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.790937388251176e-05,
+      "loss": 0.752,
+      "step": 9250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.788556093747142e-05,
+      "loss": 0.7395,
+      "step": 9260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.7861732061427024e-05,
+      "loss": 0.7337,
+      "step": 9270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.783788728383929e-05,
+      "loss": 0.7559,
+      "step": 9280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.7814026634188616e-05,
+      "loss": 0.7456,
+      "step": 9290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.779015014197499e-05,
+      "loss": 0.7293,
+      "step": 9300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.776625783671802e-05,
+      "loss": 0.7386,
+      "step": 9310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.774234974795683e-05,
+      "loss": 0.711,
+      "step": 9320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.771842590525008e-05,
+      "loss": 0.7369,
+      "step": 9330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.769448633817591e-05,
+      "loss": 0.7446,
+      "step": 9340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.7670531076331895e-05,
+      "loss": 0.7554,
+      "step": 9350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.7646560149334995e-05,
+      "loss": 0.7632,
+      "step": 9360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.762257358682158e-05,
+      "loss": 0.7249,
+      "step": 9370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.759857141844732e-05,
+      "loss": 0.7343,
+      "step": 9380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.7574553673887164e-05,
+      "loss": 0.747,
+      "step": 9390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7550520382835365e-05,
+      "loss": 0.7378,
+      "step": 9400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.752647157500536e-05,
+      "loss": 0.7587,
+      "step": 9410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.750240728012979e-05,
+      "loss": 0.7305,
+      "step": 9420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7478327527960424e-05,
+      "loss": 0.7188,
+      "step": 9430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.745423234826817e-05,
+      "loss": 0.7295,
+      "step": 9440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7430121770842974e-05,
+      "loss": 0.7137,
+      "step": 9450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7405995825493855e-05,
+      "loss": 0.7619,
+      "step": 9460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.73818545420488e-05,
+      "loss": 0.7388,
+      "step": 9470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.735769795035477e-05,
+      "loss": 0.7496,
+      "step": 9480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.733352608027768e-05,
+      "loss": 0.7716,
+      "step": 9490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.730933896170229e-05,
+      "loss": 0.7513,
+      "step": 9500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7285136624532244e-05,
+      "loss": 0.7472,
+      "step": 9510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.726091909868998e-05,
+      "loss": 0.726,
+      "step": 9520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.7236686414116736e-05,
+      "loss": 0.728,
+      "step": 9530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.721243860077247e-05,
+      "loss": 0.7283,
+      "step": 9540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.718817568863586e-05,
+      "loss": 0.7674,
+      "step": 9550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7163897707704244e-05,
+      "loss": 0.738,
+      "step": 9560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.71396046879936e-05,
+      "loss": 0.7461,
+      "step": 9570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.711529665953847e-05,
+      "loss": 0.7427,
+      "step": 9580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7090973652392e-05,
+      "loss": 0.7268,
+      "step": 9590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.706663569662581e-05,
+      "loss": 0.7508,
+      "step": 9600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.704228282233003e-05,
+      "loss": 0.7623,
+      "step": 9610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.7017915059613214e-05,
+      "loss": 0.7626,
+      "step": 9620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.699353243860235e-05,
+      "loss": 0.7394,
+      "step": 9630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.696913498944276e-05,
+      "loss": 0.7422,
+      "step": 9640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6944722742298135e-05,
+      "loss": 0.7552,
+      "step": 9650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.692029572735042e-05,
+      "loss": 0.6867,
+      "step": 9660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.6895853974799876e-05,
+      "loss": 0.7644,
+      "step": 9670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6871397514864924e-05,
+      "loss": 0.7547,
+      "step": 9680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6846926377782216e-05,
+      "loss": 0.7313,
+      "step": 9690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.682244059380651e-05,
+      "loss": 0.7643,
+      "step": 9700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6797940193210714e-05,
+      "loss": 0.7561,
+      "step": 9710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6773425206285765e-05,
+      "loss": 0.7326,
+      "step": 9720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.674889566334067e-05,
+      "loss": 0.7435,
+      "step": 9730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6724351594702404e-05,
+      "loss": 0.7259,
+      "step": 9740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6699793030715933e-05,
+      "loss": 0.7106,
+      "step": 9750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.66752200017441e-05,
+      "loss": 0.7552,
+      "step": 9760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.6650632538167674e-05,
+      "loss": 0.7305,
+      "step": 9770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.662603067038524e-05,
+      "loss": 0.7236,
+      "step": 9780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.660141442881322e-05,
+      "loss": 0.7464,
+      "step": 9790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.657678384388578e-05,
+      "loss": 0.7186,
+      "step": 9800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.655213894605483e-05,
+      "loss": 0.7587,
+      "step": 9810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.652747976578998e-05,
+      "loss": 0.7431,
+      "step": 9820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.650280633357849e-05,
+      "loss": 0.7776,
+      "step": 9830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.6478118679925254e-05,
+      "loss": 0.7266,
+      "step": 9840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.6453416835352725e-05,
+      "loss": 0.7521,
+      "step": 9850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.642870083040093e-05,
+      "loss": 0.7532,
+      "step": 9860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.6403970695627384e-05,
+      "loss": 0.7215,
+      "step": 9870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.637922646160706e-05,
+      "loss": 0.7475,
+      "step": 9880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.6354468158932395e-05,
+      "loss": 0.757,
+      "step": 9890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.632969581821321e-05,
+      "loss": 0.7066,
+      "step": 9900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.6304909470076645e-05,
+      "loss": 0.7627,
+      "step": 9910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.628010914516723e-05,
+      "loss": 0.7341,
+      "step": 9920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.6255294874146684e-05,
+      "loss": 0.7256,
+      "step": 9930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.6230466687694054e-05,
+      "loss": 0.7241,
+      "step": 9940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.620562461650553e-05,
+      "loss": 0.7269,
+      "step": 9950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.618076869129452e-05,
+      "loss": 0.7487,
+      "step": 9960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.61558989427915e-05,
+      "loss": 0.735,
+      "step": 9970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.61310154017441e-05,
+      "loss": 0.7476,
+      "step": 9980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.6106118098916954e-05,
+      "loss": 0.7394,
+      "step": 9990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.608120706509173e-05,
+      "loss": 0.7288,
+      "step": 10000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.605628233106707e-05,
+      "loss": 0.7491,
+      "step": 10010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.6031343927658564e-05,
+      "loss": 0.7687,
+      "step": 10020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.600639188569868e-05,
+      "loss": 0.7579,
+      "step": 10030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.598142623603676e-05,
+      "loss": 0.7054,
+      "step": 10040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.595644700953898e-05,
+      "loss": 0.7501,
+      "step": 10050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.5931454237088283e-05,
+      "loss": 0.713,
+      "step": 10060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.590644794958438e-05,
+      "loss": 0.735,
+      "step": 10070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.5881428177943674e-05,
+      "loss": 0.7051,
+      "step": 10080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.5856394953099234e-05,
+      "loss": 0.75,
+      "step": 10090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.583134830600079e-05,
+      "loss": 0.7514,
+      "step": 10100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5806288267614636e-05,
+      "loss": 0.7233,
+      "step": 10110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5781214868923633e-05,
+      "loss": 0.7099,
+      "step": 10120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.575612814092718e-05,
+      "loss": 0.7144,
+      "step": 10130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5731028114641116e-05,
+      "loss": 0.7626,
+      "step": 10140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.570591482109777e-05,
+      "loss": 0.7193,
+      "step": 10150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.568078829134582e-05,
+      "loss": 0.737,
+      "step": 10160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5655648556450356e-05,
+      "loss": 0.7606,
+      "step": 10170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.563049564749275e-05,
+      "loss": 0.7435,
+      "step": 10180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5605329595570714e-05,
+      "loss": 0.7496,
+      "step": 10190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.558015043179816e-05,
+      "loss": 0.7282,
+      "step": 10200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.555495818730524e-05,
+      "loss": 0.7563,
+      "step": 10210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5529752893238264e-05,
+      "loss": 0.7196,
+      "step": 10220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5504534580759695e-05,
+      "loss": 0.761,
+      "step": 10230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.547930328104806e-05,
+      "loss": 0.7364,
+      "step": 10240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.545405902529797e-05,
+      "loss": 0.7307,
+      "step": 10250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.542880184472004e-05,
+      "loss": 0.7517,
+      "step": 10260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.540353177054088e-05,
+      "loss": 0.7236,
+      "step": 10270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5378248834003017e-05,
+      "loss": 0.73,
+      "step": 10280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.535295306636489e-05,
+      "loss": 0.7336,
+      "step": 10290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5327644498900824e-05,
+      "loss": 0.7248,
+      "step": 10300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.530232316290094e-05,
+      "loss": 0.7291,
+      "step": 10310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5276989089671154e-05,
+      "loss": 0.7609,
+      "step": 10320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5251642310533135e-05,
+      "loss": 0.7445,
+      "step": 10330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.522628285682425e-05,
+      "loss": 0.7711,
+      "step": 10340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.520091075989755e-05,
+      "loss": 0.7469,
+      "step": 10350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.517552605112171e-05,
+      "loss": 0.7453,
+      "step": 10360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.515012876188099e-05,
+      "loss": 0.726,
+      "step": 10370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.512471892357522e-05,
+      "loss": 0.7439,
+      "step": 10380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.509929656761973e-05,
+      "loss": 0.7299,
+      "step": 10390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.507386172544534e-05,
+      "loss": 0.7795,
+      "step": 10400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.50484144284983e-05,
+      "loss": 0.7389,
+      "step": 10410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.502295470824026e-05,
+      "loss": 0.7409,
+      "step": 10420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4997482596148215e-05,
+      "loss": 0.7453,
+      "step": 10430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.497199812371451e-05,
+      "loss": 0.7331,
+      "step": 10440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4946501322446745e-05,
+      "loss": 0.7345,
+      "step": 10450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4920992223867784e-05,
+      "loss": 0.7448,
+      "step": 10460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.489547085951567e-05,
+      "loss": 0.7118,
+      "step": 10470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.486993726094363e-05,
+      "loss": 0.741,
+      "step": 10480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4844391459720014e-05,
+      "loss": 0.708,
+      "step": 10490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.481883348742826e-05,
+      "loss": 0.7703,
+      "step": 10500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.479326337566683e-05,
+      "loss": 0.7467,
+      "step": 10510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4767681156049236e-05,
+      "loss": 0.7501,
+      "step": 10520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4742086860203926e-05,
+      "loss": 0.764,
+      "step": 10530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.47164805197743e-05,
+      "loss": 0.7412,
+      "step": 10540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.469086216641863e-05,
+      "loss": 0.7403,
+      "step": 10550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.466523183181005e-05,
+      "loss": 0.7317,
+      "step": 10560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.463958954763652e-05,
+      "loss": 0.7539,
+      "step": 10570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.461393534560073e-05,
+      "loss": 0.7554,
+      "step": 10580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.458826925742017e-05,
+      "loss": 0.7161,
+      "step": 10590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.456259131482696e-05,
+      "loss": 0.7023,
+      "step": 10600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.453690154956793e-05,
+      "loss": 0.7644,
+      "step": 10610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4511199993404496e-05,
+      "loss": 0.7552,
+      "step": 10620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.448548667811265e-05,
+      "loss": 0.7156,
+      "step": 10630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.445976163548294e-05,
+      "loss": 0.7464,
+      "step": 10640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.443402489732041e-05,
+      "loss": 0.7252,
+      "step": 10650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4408276495444534e-05,
+      "loss": 0.7355,
+      "step": 10660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.438251646168926e-05,
+      "loss": 0.7304,
+      "step": 10670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.435674482790287e-05,
+      "loss": 0.7544,
+      "step": 10680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.433096162594801e-05,
+      "loss": 0.7299,
+      "step": 10690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.430516688770161e-05,
+      "loss": 0.7387,
+      "step": 10700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4279360645054905e-05,
+      "loss": 0.7235,
+      "step": 10710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.425354292991329e-05,
+      "loss": 0.7559,
+      "step": 10720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4227713774196415e-05,
+      "loss": 0.7226,
+      "step": 10730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4201873209838e-05,
+      "loss": 0.7245,
+      "step": 10740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.417602126878593e-05,
+      "loss": 0.7257,
+      "step": 10750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.415015798300214e-05,
+      "loss": 0.7327,
+      "step": 10760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.412428338446257e-05,
+      "loss": 0.7503,
+      "step": 10770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.409839750515717e-05,
+      "loss": 0.7504,
+      "step": 10780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.407250037708982e-05,
+      "loss": 0.716,
+      "step": 10790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.404659203227832e-05,
+      "loss": 0.7614,
+      "step": 10800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.4020672502754333e-05,
+      "loss": 0.7691,
+      "step": 10810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.3994741820563344e-05,
+      "loss": 0.7403,
+      "step": 10820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.3968800017764645e-05,
+      "loss": 0.7404,
+      "step": 10830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.394284712643126e-05,
+      "loss": 0.7394,
+      "step": 10840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.391688317864992e-05,
+      "loss": 0.7452,
+      "step": 10850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.389090820652104e-05,
+      "loss": 0.7121,
+      "step": 10860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.386492224215865e-05,
+      "loss": 0.7231,
+      "step": 10870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.383892531769039e-05,
+      "loss": 0.7617,
+      "step": 10880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.381291746525742e-05,
+      "loss": 0.7573,
+      "step": 10890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.378689871701445e-05,
+      "loss": 0.7483,
+      "step": 10900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.376086910512962e-05,
+      "loss": 0.742,
+      "step": 10910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.3734828661784535e-05,
+      "loss": 0.7302,
+      "step": 10920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.370877741917418e-05,
+      "loss": 0.6999,
+      "step": 10930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.368271540950687e-05,
+      "loss": 0.7196,
+      "step": 10940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.365664266500426e-05,
+      "loss": 0.7372,
+      "step": 10950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.363055921790128e-05,
+      "loss": 0.768,
+      "step": 10960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.3604465100446064e-05,
+      "loss": 0.7356,
+      "step": 10970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.3578360344899965e-05,
+      "loss": 0.7345,
+      "step": 10980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.355224498353747e-05,
+      "loss": 0.708,
+      "step": 10990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.3526119048646196e-05,
+      "loss": 0.7387,
+      "step": 11000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.349998257252681e-05,
+      "loss": 0.7346,
+      "step": 11010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.347383558749303e-05,
+      "loss": 0.7535,
+      "step": 11020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.344767812587157e-05,
+      "loss": 0.7271,
+      "step": 11030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.342151022000207e-05,
+      "loss": 0.7259,
+      "step": 11040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.339533190223711e-05,
+      "loss": 0.7319,
+      "step": 11050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.3369143204942125e-05,
+      "loss": 0.7324,
+      "step": 11060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.3342944160495406e-05,
+      "loss": 0.7375,
+      "step": 11070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.331673480128801e-05,
+      "loss": 0.7354,
+      "step": 11080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.329051515972376e-05,
+      "loss": 0.7361,
+      "step": 11090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.326428526821919e-05,
+      "loss": 0.7464,
+      "step": 11100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.3238045159203494e-05,
+      "loss": 0.7313,
+      "step": 11110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.321179486511853e-05,
+      "loss": 0.7223,
+      "step": 11120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.318553441841872e-05,
+      "loss": 0.7402,
+      "step": 11130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.315926385157105e-05,
+      "loss": 0.7253,
+      "step": 11140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.313298319705501e-05,
+      "loss": 0.726,
+      "step": 11150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.3106692487362555e-05,
+      "loss": 0.7543,
+      "step": 11160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.3080391754998106e-05,
+      "loss": 0.728,
+      "step": 11170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.305408103247845e-05,
+      "loss": 0.7323,
+      "step": 11180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.3027760352332705e-05,
+      "loss": 0.7665,
+      "step": 11190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.300142974710234e-05,
+      "loss": 0.7486,
+      "step": 11200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.297508924934108e-05,
+      "loss": 0.7451,
+      "step": 11210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.2948738891614876e-05,
+      "loss": 0.7647,
+      "step": 11220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.292237870650187e-05,
+      "loss": 0.7415,
+      "step": 11230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.289600872659235e-05,
+      "loss": 0.746,
+      "step": 11240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.286962898448873e-05,
+      "loss": 0.7256,
+      "step": 11250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.284323951280547e-05,
+      "loss": 0.745,
+      "step": 11260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.281684034416909e-05,
+      "loss": 0.7154,
+      "step": 11270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.2790431511218064e-05,
+      "loss": 0.7422,
+      "step": 11280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.276401304660284e-05,
+      "loss": 0.7168,
+      "step": 11290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.2737584982985766e-05,
+      "loss": 0.7441,
+      "step": 11300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.271114735304105e-05,
+      "loss": 0.7541,
+      "step": 11310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.2684700189454744e-05,
+      "loss": 0.7001,
+      "step": 11320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.265824352492467e-05,
+      "loss": 0.7379,
+      "step": 11330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.2631777392160403e-05,
+      "loss": 0.72,
+      "step": 11340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.2605301823883226e-05,
+      "loss": 0.7386,
+      "step": 11350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.257881685282609e-05,
+      "loss": 0.7074,
+      "step": 11360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.255232251173357e-05,
+      "loss": 0.7308,
+      "step": 11370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.252581883336181e-05,
+      "loss": 0.7069,
+      "step": 11380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.249930585047852e-05,
+      "loss": 0.7334,
+      "step": 11390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2472783595862896e-05,
+      "loss": 0.7444,
+      "step": 11400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2446252102305625e-05,
+      "loss": 0.7503,
+      "step": 11410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2419711402608774e-05,
+      "loss": 0.7331,
+      "step": 11420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2393161529585836e-05,
+      "loss": 0.7449,
+      "step": 11430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.236660251606161e-05,
+      "loss": 0.7125,
+      "step": 11440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2340034394872217e-05,
+      "loss": 0.7201,
+      "step": 11450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.231345719886502e-05,
+      "loss": 0.7293,
+      "step": 11460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.228687096089863e-05,
+      "loss": 0.7301,
+      "step": 11470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.226027571384281e-05,
+      "loss": 0.7094,
+      "step": 11480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.2233671490578474e-05,
+      "loss": 0.7153,
+      "step": 11490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.220705832399763e-05,
+      "loss": 0.7271,
+      "step": 11500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.218043624700335e-05,
+      "loss": 0.731,
+      "step": 11510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.215380529250971e-05,
+      "loss": 0.7227,
+      "step": 11520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.212716549344177e-05,
+      "loss": 0.7455,
+      "step": 11530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.210051688273552e-05,
+      "loss": 0.7609,
+      "step": 11540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.207385949333785e-05,
+      "loss": 0.7306,
+      "step": 11550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.204719335820651e-05,
+      "loss": 0.7132,
+      "step": 11560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.202051851031004e-05,
+      "loss": 0.735,
+      "step": 11570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.199383498262777e-05,
+      "loss": 0.7182,
+      "step": 11580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.196714280814976e-05,
+      "loss": 0.7235,
+      "step": 11590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.194044201987675e-05,
+      "loss": 0.7094,
+      "step": 11600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.191373265082015e-05,
+      "loss": 0.7078,
+      "step": 11610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.188701473400195e-05,
+      "loss": 0.7232,
+      "step": 11620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.1860288302454735e-05,
+      "loss": 0.7361,
+      "step": 11630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.18335533892216e-05,
+      "loss": 0.7037,
+      "step": 11640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.180681002735614e-05,
+      "loss": 0.7403,
+      "step": 11650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.178005824992237e-05,
+      "loss": 0.7395,
+      "step": 11660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.175329808999475e-05,
+      "loss": 0.738,
+      "step": 11670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.172652958065806e-05,
+      "loss": 0.7386,
+      "step": 11680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.169975275500743e-05,
+      "loss": 0.6953,
+      "step": 11690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1672967646148285e-05,
+      "loss": 0.7369,
+      "step": 11700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.164617428719624e-05,
+      "loss": 0.737,
+      "step": 11710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.161937271127717e-05,
+      "loss": 0.7133,
+      "step": 11720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.159256295152705e-05,
+      "loss": 0.7289,
+      "step": 11730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.156574504109203e-05,
+      "loss": 0.7018,
+      "step": 11740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1538919013128295e-05,
+      "loss": 0.7293,
+      "step": 11750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.151208490080209e-05,
+      "loss": 0.7382,
+      "step": 11760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.148524273728964e-05,
+      "loss": 0.7483,
+      "step": 11770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.145839255577714e-05,
+      "loss": 0.7483,
+      "step": 11780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.1431534389460665e-05,
+      "loss": 0.7278,
+      "step": 11790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.140466827154622e-05,
+      "loss": 0.7551,
+      "step": 11800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.137779423524958e-05,
+      "loss": 0.7652,
+      "step": 11810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1350912313796336e-05,
+      "loss": 0.7296,
+      "step": 11820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.132402254042185e-05,
+      "loss": 0.722,
+      "step": 11830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.129712494837115e-05,
+      "loss": 0.6992,
+      "step": 11840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.127021957089896e-05,
+      "loss": 0.7204,
+      "step": 11850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.124330644126962e-05,
+      "loss": 0.7393,
+      "step": 11860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1216385592757045e-05,
+      "loss": 0.7287,
+      "step": 11870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.118945705864471e-05,
+      "loss": 0.7548,
+      "step": 11880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1162520872225584e-05,
+      "loss": 0.7513,
+      "step": 11890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.11355770668021e-05,
+      "loss": 0.724,
+      "step": 11900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.11086256756861e-05,
+      "loss": 0.7224,
+      "step": 11910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1081666732198805e-05,
+      "loss": 0.7403,
+      "step": 11920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1054700269670814e-05,
+      "loss": 0.7338,
+      "step": 11930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.102772632144195e-05,
+      "loss": 0.69,
+      "step": 11940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.100074492086136e-05,
+      "loss": 0.725,
+      "step": 11950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0973756101287344e-05,
+      "loss": 0.7465,
+      "step": 11960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.094675989608744e-05,
+      "loss": 0.7249,
+      "step": 11970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.091975633863826e-05,
+      "loss": 0.7192,
+      "step": 11980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.089274546232554e-05,
+      "loss": 0.7273,
+      "step": 11990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0865727300544026e-05,
+      "loss": 0.7629,
+      "step": 12000
+    }
+  ],
+  "max_steps": 28254,
+  "num_train_epochs": 2,
+  "total_flos": 2.9208226729316844e+18,
+  "trial_name": null,
+  "trial_params": null
+}