diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,11005 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9998770945143187,
+  "global_step": 18306,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998196073369e-05,
+      "loss": 2.7493,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999893605206236e-05,
+      "loss": 2.6128,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999731620342936e-05,
+      "loss": 2.4239,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999949600662084e-05,
+      "loss": 2.3572,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999918676473389e-05,
+      "loss": 2.2041,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999880389559285e-05,
+      "loss": 2.0452,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999834740032535e-05,
+      "loss": 2.2267,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999781728027584e-05,
+      "loss": 2.186,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999721353700564e-05,
+      "loss": 2.0856,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999653617229287e-05,
+      "loss": 2.1413,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9995785188132515e-05,
+      "loss": 2.025,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999496058673635e-05,
+      "loss": 2.1813,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999406237053297e-05,
+      "loss": 2.0397,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9993090542167805e-05,
+      "loss": 2.1063,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999204510450306e-05,
+      "loss": 2.0306,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999092606061775e-05,
+      "loss": 2.042,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998973341380767e-05,
+      "loss": 2.0843,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9988467167585374e-05,
+      "loss": 2.0291,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9987127325680206e-05,
+      "loss": 2.1306,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.998571389203824e-05,
+      "loss": 2.1251,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99842268708223e-05,
+      "loss": 2.1679,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998266626641195e-05,
+      "loss": 2.0369,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9981032083403454e-05,
+      "loss": 2.1583,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997932432660979e-05,
+      "loss": 2.0899,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997754300106061e-05,
+      "loss": 2.023,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997568811200226e-05,
+      "loss": 1.9846,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997375966489772e-05,
+      "loss": 2.039,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.997175766542664e-05,
+      "loss": 2.056,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9969682119485274e-05,
+      "loss": 1.9968,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996753303318648e-05,
+      "loss": 2.1077,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996531041285974e-05,
+      "loss": 2.0032,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996301426505106e-05,
+      "loss": 2.0336,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996064459652303e-05,
+      "loss": 2.1477,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995820141425478e-05,
+      "loss": 2.0725,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995568472544192e-05,
+      "loss": 2.1121,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995309453749657e-05,
+      "loss": 2.0075,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.995043085804732e-05,
+      "loss": 2.017,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994769369493919e-05,
+      "loss": 2.0741,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.994488305623365e-05,
+      "loss": 2.0267,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.994199895020854e-05,
+      "loss": 2.0651,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9939041385358096e-05,
+      "loss": 1.961,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9936010370392896e-05,
+      "loss": 2.0312,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.993290591423984e-05,
+      "loss": 2.2241,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9929728026042123e-05,
+      "loss": 2.0277,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.992647671515923e-05,
+      "loss": 2.3211,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9923151991166875e-05,
+      "loss": 1.9703,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9919753863856974e-05,
+      "loss": 2.2109,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.991628234323765e-05,
+      "loss": 2.0107,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9912737439533184e-05,
+      "loss": 2.019,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990911916318398e-05,
+      "loss": 1.9858,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.990542752484653e-05,
+      "loss": 1.8274,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.990166253539338e-05,
+      "loss": 2.0249,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.989782420591314e-05,
+      "loss": 2.1272,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.98939125477104e-05,
+      "loss": 2.0735,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988992757230573e-05,
+      "loss": 2.017,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.988586929143561e-05,
+      "loss": 2.2007,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9881737717052436e-05,
+      "loss": 2.1686,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9877532861324466e-05,
+      "loss": 2.0366,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9873254736635785e-05,
+      "loss": 1.9201,
+      "step": 590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9868903355586255e-05,
+      "loss": 2.0208,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.986447873099151e-05,
+      "loss": 1.9649,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9859980875882895e-05,
+      "loss": 2.0407,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.985540980350742e-05,
+      "loss": 2.0502,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9850765527327745e-05,
+      "loss": 2.0704,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9846048061022146e-05,
+      "loss": 2.1043,
+      "step": 650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.984125741848441e-05,
+      "loss": 2.1337,
+      "step": 660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.983639361382388e-05,
+      "loss": 2.0592,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9831456661365384e-05,
+      "loss": 2.01,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.982644657564914e-05,
+      "loss": 2.0849,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9821363371430794e-05,
+      "loss": 2.0178,
+      "step": 700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.981620706368133e-05,
+      "loss": 1.9404,
+      "step": 710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9810977667587024e-05,
+      "loss": 2.0513,
+      "step": 720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.980567519854943e-05,
+      "loss": 2.0504,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.98002996721853e-05,
+      "loss": 2.0741,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9794851104326554e-05,
+      "loss": 2.0063,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.978932951102024e-05,
+      "loss": 2.0993,
+      "step": 760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.978373490852849e-05,
+      "loss": 1.9933,
+      "step": 770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.977806731332844e-05,
+      "loss": 1.9999,
+      "step": 780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.97723267421122e-05,
+      "loss": 2.1812,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9766513211786834e-05,
+      "loss": 2.0266,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.976062673947427e-05,
+      "loss": 1.9424,
+      "step": 810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.975466734251126e-05,
+      "loss": 2.1343,
+      "step": 820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.974863503844933e-05,
+      "loss": 1.9458,
+      "step": 830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.974252984505475e-05,
+      "loss": 1.9796,
+      "step": 840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.973635178030844e-05,
+      "loss": 2.0243,
+      "step": 850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.973010086240596e-05,
+      "loss": 2.0822,
+      "step": 860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.972377710975742e-05,
+      "loss": 1.9921,
+      "step": 870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.971738054098743e-05,
+      "loss": 2.0545,
+      "step": 880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.97109111749351e-05,
+      "loss": 2.0681,
+      "step": 890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.970436903065392e-05,
+      "loss": 1.9718,
+      "step": 900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.96977541274117e-05,
+      "loss": 1.9411,
+      "step": 910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9691066484690565e-05,
+      "loss": 1.9922,
+      "step": 920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.968430612218687e-05,
+      "loss": 1.9778,
+      "step": 930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9677473059811144e-05,
+      "loss": 1.9636,
+      "step": 940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.967056731768802e-05,
+      "loss": 2.0043,
+      "step": 950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.966358891615619e-05,
+      "loss": 1.936,
+      "step": 960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.965653787576836e-05,
+      "loss": 2.0057,
+      "step": 970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.964941421729115e-05,
+      "loss": 2.0494,
+      "step": 980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.964221796170504e-05,
+      "loss": 1.8848,
+      "step": 990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9634949130204355e-05,
+      "loss": 1.8391,
+      "step": 1000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.962760774419717e-05,
+      "loss": 2.0159,
+      "step": 1010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.962019382530521e-05,
+      "loss": 2.1247,
+      "step": 1020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.961270739536384e-05,
+      "loss": 2.0661,
+      "step": 1030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.960514847642201e-05,
+      "loss": 2.0714,
+      "step": 1040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.959751709074214e-05,
+      "loss": 2.081,
+      "step": 1050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.958981326080007e-05,
+      "loss": 2.0159,
+      "step": 1060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.958203700928501e-05,
+      "loss": 2.1044,
+      "step": 1070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.957418835909947e-05,
+      "loss": 2.0309,
+      "step": 1080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.95662673333592e-05,
+      "loss": 1.9821,
+      "step": 1090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.955827395539306e-05,
+      "loss": 2.0753,
+      "step": 1100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.955020824874307e-05,
+      "loss": 2.0361,
+      "step": 1110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.954207023716421e-05,
+      "loss": 2.1343,
+      "step": 1120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9533859944624456e-05,
+      "loss": 2.0185,
+      "step": 1130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9525577395304635e-05,
+      "loss": 1.9312,
+      "step": 1140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.95172226135984e-05,
+      "loss": 1.9212,
+      "step": 1150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.950879562411213e-05,
+      "loss": 2.0193,
+      "step": 1160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.950029645166488e-05,
+      "loss": 1.9428,
+      "step": 1170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.949172512128831e-05,
+      "loss": 2.0522,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9483081658226554e-05,
+      "loss": 2.0209,
+      "step": 1190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.947436608793624e-05,
+      "loss": 2.0783,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.946557843608632e-05,
+      "loss": 1.9533,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9456718728558096e-05,
+      "loss": 2.0659,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9447786991445024e-05,
+      "loss": 1.9852,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.943878325105274e-05,
+      "loss": 2.0337,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.942970753389894e-05,
+      "loss": 1.997,
+      "step": 1250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.94205598667133e-05,
+      "loss": 2.0852,
+      "step": 1260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.941134027643739e-05,
+      "loss": 1.9998,
+      "step": 1270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.940204879022462e-05,
+      "loss": 2.1454,
+      "step": 1280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9392685435440154e-05,
+      "loss": 2.0961,
+      "step": 1290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.938325023966081e-05,
+      "loss": 2.1233,
+      "step": 1300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.937374323067499e-05,
+      "loss": 1.993,
+      "step": 1310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.936416443648261e-05,
+      "loss": 1.9058,
+      "step": 1320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.935451388529499e-05,
+      "loss": 2.038,
+      "step": 1330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9344791605534805e-05,
+      "loss": 1.889,
+      "step": 1340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.933499762583598e-05,
+      "loss": 1.8135,
+      "step": 1350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.93251319750436e-05,
+      "loss": 2.096,
+      "step": 1360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9315194682213836e-05,
+      "loss": 2.0877,
+      "step": 1370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.930518577661388e-05,
+      "loss": 2.0437,
+      "step": 1380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.929510528772181e-05,
+      "loss": 1.973,
+      "step": 1390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.928495324522654e-05,
+      "loss": 2.1289,
+      "step": 1400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9274729679027734e-05,
+      "loss": 1.9313,
+      "step": 1410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9264434619235676e-05,
+      "loss": 2.0474,
+      "step": 1420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9254068096171246e-05,
+      "loss": 1.9779,
+      "step": 1430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.924363014036578e-05,
+      "loss": 2.0017,
+      "step": 1440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.923312078256099e-05,
+      "loss": 2.0076,
+      "step": 1450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9222540053708903e-05,
+      "loss": 2.0012,
+      "step": 1460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.921188798497173e-05,
+      "loss": 2.0593,
+      "step": 1470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9201164607721786e-05,
+      "loss": 2.0216,
+      "step": 1480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9190369953541416e-05,
+      "loss": 2.0021,
+      "step": 1490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.917950405422289e-05,
+      "loss": 1.9582,
+      "step": 1500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.916856694176829e-05,
+      "loss": 1.9832,
+      "step": 1510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9157558648389454e-05,
+      "loss": 2.1146,
+      "step": 1520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9146479206507856e-05,
+      "loss": 2.0394,
+      "step": 1530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.913532864875451e-05,
+      "loss": 1.9453,
+      "step": 1540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9124107007969874e-05,
+      "loss": 1.8005,
+      "step": 1550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.911281431720378e-05,
+      "loss": 2.0337,
+      "step": 1560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9101450609715295e-05,
+      "loss": 2.0409,
+      "step": 1570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9090015918972655e-05,
+      "loss": 2.0016,
+      "step": 1580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.907851027865315e-05,
+      "loss": 1.9756,
+      "step": 1590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9066933722643014e-05,
+      "loss": 2.0264,
+      "step": 1600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.905528628503738e-05,
+      "loss": 2.0501,
+      "step": 1610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.904356800014012e-05,
+      "loss": 2.046,
+      "step": 1620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.903177890246373e-05,
+      "loss": 2.028,
+      "step": 1630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.901991902672933e-05,
+      "loss": 2.1316,
+      "step": 1640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.900798840786645e-05,
+      "loss": 1.8937,
+      "step": 1650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.899598708101297e-05,
+      "loss": 2.0304,
+      "step": 1660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.898391508151504e-05,
+      "loss": 1.9616,
+      "step": 1670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.897177244492695e-05,
+      "loss": 1.9678,
+      "step": 1680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8959559207010996e-05,
+      "loss": 2.0549,
+      "step": 1690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.894727540373747e-05,
+      "loss": 1.9482,
+      "step": 1700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.893492107128443e-05,
+      "loss": 2.0119,
+      "step": 1710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8922496246037694e-05,
+      "loss": 1.9082,
+      "step": 1720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8910000964590676e-05,
+      "loss": 1.9175,
+      "step": 1730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.889743526374432e-05,
+      "loss": 2.1049,
+      "step": 1740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.888479918050693e-05,
+      "loss": 2.0112,
+      "step": 1750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8872092752094134e-05,
+      "loss": 1.9708,
+      "step": 1760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8859316015928724e-05,
+      "loss": 2.0072,
+      "step": 1770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.884646900964057e-05,
+      "loss": 2.1175,
+      "step": 1780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.883355177106649e-05,
+      "loss": 2.0776,
+      "step": 1790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.882056433825014e-05,
+      "loss": 2.0152,
+      "step": 1800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.880750674944195e-05,
+      "loss": 2.0686,
+      "step": 1810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.879437904309893e-05,
+      "loss": 2.0797,
+      "step": 1820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.878118125788462e-05,
+      "loss": 1.967,
+      "step": 1830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.876791343266896e-05,
+      "loss": 2.0605,
+      "step": 1840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8754575606528154e-05,
+      "loss": 1.9193,
+      "step": 1850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8741167818744584e-05,
+      "loss": 2.1264,
+      "step": 1860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8727690108806665e-05,
+      "loss": 1.979,
+      "step": 1870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.871414251640879e-05,
+      "loss": 1.9705,
+      "step": 1880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.870052508145111e-05,
+      "loss": 2.0058,
+      "step": 1890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.868683784403951e-05,
+      "loss": 1.9435,
+      "step": 1900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.867308084448545e-05,
+      "loss": 2.0006,
+      "step": 1910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.865925412330586e-05,
+      "loss": 2.0201,
+      "step": 1920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8645357721223e-05,
+      "loss": 1.8661,
+      "step": 1930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.863139167916435e-05,
+      "loss": 2.0768,
+      "step": 1940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.861735603826252e-05,
+      "loss": 2.0259,
+      "step": 1950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.860325083985506e-05,
+      "loss": 1.9921,
+      "step": 1960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.858907612548442e-05,
+      "loss": 1.9444,
+      "step": 1970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.857483193689777e-05,
+      "loss": 1.9617,
+      "step": 1980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.856051831604688e-05,
+      "loss": 1.9107,
+      "step": 1990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8546135305088035e-05,
+      "loss": 1.8913,
+      "step": 2000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8531682946381874e-05,
+      "loss": 1.9929,
+      "step": 2010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8517161282493284e-05,
+      "loss": 2.0572,
+      "step": 2020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.850257035619127e-05,
+      "loss": 1.9994,
+      "step": 2030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.848791021044881e-05,
+      "loss": 1.9625,
+      "step": 2040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.847318088844277e-05,
+      "loss": 1.965,
+      "step": 2050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.845838243355374e-05,
+      "loss": 1.9804,
+      "step": 2060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.844351488936593e-05,
+      "loss": 2.015,
+      "step": 2070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8428578299667e-05,
+      "loss": 1.917,
+      "step": 2080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8413572708448007e-05,
+      "loss": 2.0016,
+      "step": 2090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8398498159903194e-05,
+      "loss": 2.0704,
+      "step": 2100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8383354698429916e-05,
+      "loss": 1.9568,
+      "step": 2110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.836814236862849e-05,
+      "loss": 1.9208,
+      "step": 2120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8352861215302035e-05,
+      "loss": 2.0346,
+      "step": 2130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.833751128345642e-05,
+      "loss": 2.046,
+      "step": 2140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.832209261830002e-05,
+      "loss": 1.9821,
+      "step": 2150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8306605265243674e-05,
+      "loss": 2.0615,
+      "step": 2160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8291049269900546e-05,
+      "loss": 1.9934,
+      "step": 2170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.82754246780859e-05,
+      "loss": 2.1143,
+      "step": 2180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.825973153581709e-05,
+      "loss": 2.2045,
+      "step": 2190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8243969889313325e-05,
+      "loss": 1.9746,
+      "step": 2200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.822813978499558e-05,
+      "loss": 1.9853,
+      "step": 2210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.821224126948648e-05,
+      "loss": 2.0529,
+      "step": 2220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8196274389610084e-05,
+      "loss": 1.9359,
+      "step": 2230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.818023919239184e-05,
+      "loss": 1.9789,
+      "step": 2240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.816413572505837e-05,
+      "loss": 1.9997,
+      "step": 2250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.81479640350374e-05,
+      "loss": 1.9251,
+      "step": 2260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.813172416995754e-05,
+      "loss": 2.0478,
+      "step": 2270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8115416177648234e-05,
+      "loss": 2.0593,
+      "step": 2280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8099040106139534e-05,
+      "loss": 2.0809,
+      "step": 2290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.808259600366204e-05,
+      "loss": 2.0221,
+      "step": 2300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.806608391864667e-05,
+      "loss": 2.107,
+      "step": 2310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.804950389972459e-05,
+      "loss": 1.9442,
+      "step": 2320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8032855995727056e-05,
+      "loss": 2.0338,
+      "step": 2330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8016140255685215e-05,
+      "loss": 2.1054,
+      "step": 2340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.799935672883005e-05,
+      "loss": 1.9695,
+      "step": 2350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.7982505464592154e-05,
+      "loss": 1.9722,
+      "step": 2360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.796558651260165e-05,
+      "loss": 2.0217,
+      "step": 2370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.794859992268798e-05,
+      "loss": 2.0171,
+      "step": 2380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.793154574487982e-05,
+      "loss": 2.0476,
+      "step": 2390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.79144240294049e-05,
+      "loss": 2.1095,
+      "step": 2400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.789723482668985e-05,
+      "loss": 2.025,
+      "step": 2410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.787997818736008e-05,
+      "loss": 1.9364,
+      "step": 2420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.786265416223957e-05,
+      "loss": 2.0866,
+      "step": 2430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.784526280235083e-05,
+      "loss": 2.0928,
+      "step": 2440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.782780415891464e-05,
+      "loss": 2.0338,
+      "step": 2450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.781027828334994e-05,
+      "loss": 1.994,
+      "step": 2460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7792685227273705e-05,
+      "loss": 1.9623,
+      "step": 2470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.777502504250075e-05,
+      "loss": 2.005,
+      "step": 2480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.775729778104361e-05,
+      "loss": 2.0083,
+      "step": 2490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.773950349511236e-05,
+      "loss": 2.0355,
+      "step": 2500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.772164223711448e-05,
+      "loss": 2.1183,
+      "step": 2510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.77037140596547e-05,
+      "loss": 1.9742,
+      "step": 2520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7685719015534844e-05,
+      "loss": 2.056,
+      "step": 2530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.766765715775366e-05,
+      "loss": 2.0144,
+      "step": 2540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7649528539506673e-05,
+      "loss": 2.0432,
+      "step": 2550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.763133321418603e-05,
+      "loss": 1.9772,
+      "step": 2560
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.761307123538036e-05,
+      "loss": 2.0117,
+      "step": 2570
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.759474265687458e-05,
+      "loss": 1.9636,
+      "step": 2580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.757634753264976e-05,
+      "loss": 2.1047,
+      "step": 2590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.755788591688297e-05,
+      "loss": 2.0909,
+      "step": 2600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.75393578639471e-05,
+      "loss": 1.9841,
+      "step": 2610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.75207634284107e-05,
+      "loss": 1.9091,
+      "step": 2620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7502102665037855e-05,
+      "loss": 2.036,
+      "step": 2630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7483375628787975e-05,
+      "loss": 2.0552,
+      "step": 2640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.746458237481567e-05,
+      "loss": 1.9515,
+      "step": 2650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.744572295847056e-05,
+      "loss": 1.9763,
+      "step": 2660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7426797435297147e-05,
+      "loss": 2.092,
+      "step": 2670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.74078058610346e-05,
+      "loss": 1.8006,
+      "step": 2680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.738874829161666e-05,
+      "loss": 2.0508,
+      "step": 2690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7369624783171386e-05,
+      "loss": 1.9731,
+      "step": 2700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7350435392021074e-05,
+      "loss": 1.9417,
+      "step": 2710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.733118017468206e-05,
+      "loss": 1.8998,
+      "step": 2720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.731185918786453e-05,
+      "loss": 1.9817,
+      "step": 2730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.729247248847239e-05,
+      "loss": 1.8598,
+      "step": 2740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.727302013360308e-05,
+      "loss": 2.0858,
+      "step": 2750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.725350218054738e-05,
+      "loss": 1.986,
+      "step": 2760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7233918686789314e-05,
+      "loss": 2.084,
+      "step": 2770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.721426971000591e-05,
+      "loss": 1.8815,
+      "step": 2780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.719455530806705e-05,
+      "loss": 1.9987,
+      "step": 2790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.717477553903532e-05,
+      "loss": 2.0184,
+      "step": 2800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7154930461165814e-05,
+      "loss": 2.0946,
+      "step": 2810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7135020132905985e-05,
+      "loss": 1.9717,
+      "step": 2820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.711504461289543e-05,
+      "loss": 2.0663,
+      "step": 2830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.709500395996581e-05,
+      "loss": 2.0342,
+      "step": 2840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.707489823314053e-05,
+      "loss": 2.1222,
+      "step": 2850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.705472749163473e-05,
+      "loss": 2.0252,
+      "step": 2860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.703449179485498e-05,
+      "loss": 2.0252,
+      "step": 2870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.701419120239917e-05,
+      "loss": 1.9319,
+      "step": 2880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.699382577405632e-05,
+      "loss": 1.9839,
+      "step": 2890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.697339556980641e-05,
+      "loss": 2.0802,
+      "step": 2900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.695290064982018e-05,
+      "loss": 2.0236,
+      "step": 2910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6932341074458984e-05,
+      "loss": 2.0469,
+      "step": 2920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6911716904274604e-05,
+      "loss": 1.9612,
+      "step": 2930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.6891028200009055e-05,
+      "loss": 2.0188,
+      "step": 2940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.687027502259441e-05,
+      "loss": 2.0128,
+      "step": 2950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6849457433152634e-05,
+      "loss": 1.9687,
+      "step": 2960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.682857549299541e-05,
+      "loss": 2.0985,
+      "step": 2970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.680762926362393e-05,
+      "loss": 2.0482,
+      "step": 2980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.6786618806728734e-05,
+      "loss": 1.9668,
+      "step": 2990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.676554418418953e-05,
+      "loss": 2.0458,
+      "step": 3000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.674440545807499e-05,
+      "loss": 1.9437,
+      "step": 3010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.672320269064261e-05,
+      "loss": 1.9303,
+      "step": 3020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.670193594433846e-05,
+      "loss": 2.1029,
+      "step": 3030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.668060528179708e-05,
+      "loss": 2.1231,
+      "step": 3040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.6659210765841244e-05,
+      "loss": 1.9148,
+      "step": 3050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.663775245948177e-05,
+      "loss": 2.0187,
+      "step": 3060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.661623042591735e-05,
+      "loss": 1.9392,
+      "step": 3070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.659464472853439e-05,
+      "loss": 2.0233,
+      "step": 3080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.6572995430906784e-05,
+      "loss": 1.957,
+      "step": 3090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.655128259679573e-05,
+      "loss": 1.9033,
+      "step": 3100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.652950629014957e-05,
+      "loss": 1.8594,
+      "step": 3110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.650766657510357e-05,
+      "loss": 2.0975,
+      "step": 3120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.648576351597976e-05,
+      "loss": 2.0428,
+      "step": 3130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.646379717728672e-05,
+      "loss": 1.9646,
+      "step": 3140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.64417676237194e-05,
+      "loss": 1.9449,
+      "step": 3150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6419674920158936e-05,
+      "loss": 1.9847,
+      "step": 3160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.639751913167244e-05,
+      "loss": 1.9585,
+      "step": 3170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.637530032351284e-05,
+      "loss": 1.969,
+      "step": 3180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6353018561118654e-05,
+      "loss": 2.0372,
+      "step": 3190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.633067391011381e-05,
+      "loss": 1.9983,
+      "step": 3200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.630826643630747e-05,
+      "loss": 1.9899,
+      "step": 3210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.628579620569381e-05,
+      "loss": 2.0894,
+      "step": 3220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.626326328445183e-05,
+      "loss": 1.9929,
+      "step": 3230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.624066773894519e-05,
+      "loss": 2.0097,
+      "step": 3240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6218009635721945e-05,
+      "loss": 1.9261,
+      "step": 3250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.619528904151445e-05,
+      "loss": 2.0682,
+      "step": 3260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.617250602323907e-05,
+      "loss": 2.0979,
+      "step": 3270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.614966064799603e-05,
+      "loss": 2.1207,
+      "step": 3280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.612675298306921e-05,
+      "loss": 2.0332,
+      "step": 3290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.610378309592594e-05,
+      "loss": 1.9067,
+      "step": 3300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.608075105421682e-05,
+      "loss": 1.9844,
+      "step": 3310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6057656925775486e-05,
+      "loss": 1.9834,
+      "step": 3320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6034500778618436e-05,
+      "loss": 2.1072,
+      "step": 3330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6011282680944845e-05,
+      "loss": 2.0828,
+      "step": 3340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.598800270113631e-05,
+      "loss": 1.9901,
+      "step": 3350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.596466090775672e-05,
+      "loss": 2.0767,
+      "step": 3360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5941257369551993e-05,
+      "loss": 1.9369,
+      "step": 3370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.591779215544989e-05,
+      "loss": 2.0302,
+      "step": 3380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5894265334559836e-05,
+      "loss": 2.1034,
+      "step": 3390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.58706769761727e-05,
+      "loss": 1.896,
+      "step": 3400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.584702714976058e-05,
+      "loss": 1.9373,
+      "step": 3410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.582331592497661e-05,
+      "loss": 1.9,
+      "step": 3420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.579954337165476e-05,
+      "loss": 2.0099,
+      "step": 3430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.577570955980962e-05,
+      "loss": 1.9277,
+      "step": 3440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.575181455963619e-05,
+      "loss": 2.1735,
+      "step": 3450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.572785844150971e-05,
+      "loss": 1.969,
+      "step": 3460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5703841275985384e-05,
+      "loss": 1.9746,
+      "step": 3470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.567976313379823e-05,
+      "loss": 2.0835,
+      "step": 3480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.5655624085862865e-05,
+      "loss": 1.9024,
+      "step": 3490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.563142420327327e-05,
+      "loss": 1.9275,
+      "step": 3500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.56071635573026e-05,
+      "loss": 1.965,
+      "step": 3510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5582842219402975e-05,
+      "loss": 1.9371,
+      "step": 3520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.555846026120525e-05,
+      "loss": 1.8545,
+      "step": 3530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.553401775451882e-05,
+      "loss": 1.8197,
+      "step": 3540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.550951477133143e-05,
+      "loss": 2.0759,
+      "step": 3550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.548495138380891e-05,
+      "loss": 1.9554,
+      "step": 3560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.546032766429501e-05,
+      "loss": 2.0326,
+      "step": 3570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.543564368531116e-05,
+      "loss": 1.982,
+      "step": 3580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.5410899519556274e-05,
+      "loss": 1.9814,
+      "step": 3590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.538609523990651e-05,
+      "loss": 2.1108,
+      "step": 3600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.53612309194151e-05,
+      "loss": 2.0795,
+      "step": 3610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.533630663131208e-05,
+      "loss": 1.8855,
+      "step": 3620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.531132244900411e-05,
+      "loss": 2.0997,
+      "step": 3630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.528627844607426e-05,
+      "loss": 2.0274,
+      "step": 3640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5261174696281786e-05,
+      "loss": 1.9669,
+      "step": 3650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.523601127356188e-05,
+      "loss": 1.8909,
+      "step": 3660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5210788252025524e-05,
+      "loss": 2.0331,
+      "step": 3670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5185505705959186e-05,
+      "loss": 1.9388,
+      "step": 3680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5160163709824676e-05,
+      "loss": 1.8146,
+      "step": 3690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.5134762338258884e-05,
+      "loss": 2.0777,
+      "step": 3700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.5109301666073575e-05,
+      "loss": 1.8371,
+      "step": 3710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.508378176825516e-05,
+      "loss": 2.0139,
+      "step": 3720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.505820271996448e-05,
+      "loss": 2.0614,
+      "step": 3730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.50325645965366e-05,
+      "loss": 1.8925,
+      "step": 3740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.5006867473480556e-05,
+      "loss": 1.9504,
+      "step": 3750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4981111426479156e-05,
+      "loss": 1.9731,
+      "step": 3760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4955296531388744e-05,
+      "loss": 1.953,
+      "step": 3770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.492942286423899e-05,
+      "loss": 2.0635,
+      "step": 3780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4903490501232655e-05,
+      "loss": 2.0147,
+      "step": 3790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.487749951874537e-05,
+      "loss": 1.9297,
+      "step": 3800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.485144999332541e-05,
+      "loss": 2.0157,
+      "step": 3810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.482534200169347e-05,
+      "loss": 2.0312,
+      "step": 3820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.479917562074246e-05,
+      "loss": 1.9649,
+      "step": 3830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4772950927537205e-05,
+      "loss": 1.9693,
+      "step": 3840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.474666799931433e-05,
+      "loss": 1.8959,
+      "step": 3850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.472032691348194e-05,
+      "loss": 1.9784,
+      "step": 3860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.4693927747619424e-05,
+      "loss": 1.9321,
+      "step": 3870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4667470579477245e-05,
+      "loss": 2.0305,
+      "step": 3880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4640955486976675e-05,
+      "loss": 2.062,
+      "step": 3890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.461438254820959e-05,
+      "loss": 1.8327,
+      "step": 3900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4587751841438254e-05,
+      "loss": 1.8706,
+      "step": 3910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.456106344509503e-05,
+      "loss": 1.9434,
+      "step": 3920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.453431743778222e-05,
+      "loss": 2.1752,
+      "step": 3930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.450751389827178e-05,
+      "loss": 1.8826,
+      "step": 3940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.448065290550514e-05,
+      "loss": 1.9663,
+      "step": 3950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.44537345385929e-05,
+      "loss": 1.79,
+      "step": 3960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.442675887681466e-05,
+      "loss": 1.8082,
+      "step": 3970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.439972599961875e-05,
+      "loss": 2.0687,
+      "step": 3980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.4372635986622044e-05,
+      "loss": 1.9169,
+      "step": 3990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.434548891760965e-05,
+      "loss": 1.9168,
+      "step": 4000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.4318284872534746e-05,
+      "loss": 1.9284,
+      "step": 4010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.429102393151829e-05,
+      "loss": 2.0423,
+      "step": 4020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.426370617484884e-05,
+      "loss": 1.994,
+      "step": 4030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.423633168298226e-05,
+      "loss": 2.0761,
+      "step": 4040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.420890053654152e-05,
+      "loss": 2.116,
+      "step": 4050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4181412816316446e-05,
+      "loss": 2.0037,
+      "step": 4060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4153868603263496e-05,
+      "loss": 1.967,
+      "step": 4070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4126267978505486e-05,
+      "loss": 1.8838,
+      "step": 4080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4098611023331395e-05,
+      "loss": 1.9042,
+      "step": 4090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.40708978191961e-05,
+      "loss": 2.0116,
+      "step": 4100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.404312844772014e-05,
+      "loss": 1.9652,
+      "step": 4110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.4015302990689475e-05,
+      "loss": 2.0005,
+      "step": 4120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.398742153005526e-05,
+      "loss": 2.1425,
+      "step": 4130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.395948414793357e-05,
+      "loss": 1.9169,
+      "step": 4140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.39314909266052e-05,
+      "loss": 2.0102,
+      "step": 4150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3903441948515385e-05,
+      "loss": 2.031,
+      "step": 4160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.387533729627359e-05,
+      "loss": 2.002,
+      "step": 4170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3847177052653235e-05,
+      "loss": 1.9693,
+      "step": 4180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3818961300591476e-05,
+      "loss": 2.0389,
+      "step": 4190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3790690123188955e-05,
+      "loss": 1.9288,
+      "step": 4200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3762363603709545e-05,
+      "loss": 1.9471,
+      "step": 4210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.37339818255801e-05,
+      "loss": 2.0029,
+      "step": 4220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3705544872390265e-05,
+      "loss": 2.0622,
+      "step": 4230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.367705282789213e-05,
+      "loss": 2.0739,
+      "step": 4240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.364850577600007e-05,
+      "loss": 1.9894,
+      "step": 4250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3619903800790465e-05,
+      "loss": 2.0259,
+      "step": 4260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359124698650146e-05,
+      "loss": 2.055,
+      "step": 4270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356253541753267e-05,
+      "loss": 2.0733,
+      "step": 4280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353376917844504e-05,
+      "loss": 2.0299,
+      "step": 4290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.350494835396048e-05,
+      "loss": 2.0819,
+      "step": 4300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3476073028961684e-05,
+      "loss": 1.8574,
+      "step": 4310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.344714328849185e-05,
+      "loss": 1.917,
+      "step": 4320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.341815921775444e-05,
+      "loss": 1.9747,
+      "step": 4330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.338912090211293e-05,
+      "loss": 2.1776,
+      "step": 4340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.336002842709057e-05,
+      "loss": 2.0182,
+      "step": 4350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.33308818783701e-05,
+      "loss": 1.9635,
+      "step": 4360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.330168134179353e-05,
+      "loss": 1.9474,
+      "step": 4370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.327535477033421e-05,
+      "loss": 1.9531,
+      "step": 4380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.324605189389505e-05,
+      "loss": 1.8636,
+      "step": 4390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3216695279439864e-05,
+      "loss": 1.9228,
+      "step": 4400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.318728501342936e-05,
+      "loss": 2.056,
+      "step": 4410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.315782118248223e-05,
+      "loss": 1.9998,
+      "step": 4420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.312830387337492e-05,
+      "loss": 1.9555,
+      "step": 4430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.3098733173041416e-05,
+      "loss": 2.0842,
+      "step": 4440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.306910916857292e-05,
+      "loss": 1.9577,
+      "step": 4450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.303943194721764e-05,
+      "loss": 1.9089,
+      "step": 4460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.30097015963805e-05,
+      "loss": 2.0344,
+      "step": 4470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.297991820362293e-05,
+      "loss": 2.0412,
+      "step": 4480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.295008185666254e-05,
+      "loss": 2.0576,
+      "step": 4490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.292019264337294e-05,
+      "loss": 1.9196,
+      "step": 4500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.289025065178341e-05,
+      "loss": 1.9516,
+      "step": 4510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.286025597007869e-05,
+      "loss": 1.9636,
+      "step": 4520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.283020868659869e-05,
+      "loss": 2.003,
+      "step": 4530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2800108889838244e-05,
+      "loss": 2.1171,
+      "step": 4540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2769956668446864e-05,
+      "loss": 1.9659,
+      "step": 4550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.273975211122844e-05,
+      "loss": 1.9895,
+      "step": 4560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.270949530714101e-05,
+      "loss": 1.9716,
+      "step": 4570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2679186345296496e-05,
+      "loss": 2.0097,
+      "step": 4580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.264882531496042e-05,
+      "loss": 1.949,
+      "step": 4590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.261841230555166e-05,
+      "loss": 2.0087,
+      "step": 4600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.258794740664219e-05,
+      "loss": 1.9034,
+      "step": 4610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.255743070795681e-05,
+      "loss": 1.8005,
+      "step": 4620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.252686229937286e-05,
+      "loss": 1.914,
+      "step": 4630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2496242270919995e-05,
+      "loss": 2.027,
+      "step": 4640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.246557071277989e-05,
+      "loss": 1.9616,
+      "step": 4650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.243484771528599e-05,
+      "loss": 1.8854,
+      "step": 4660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2404073368923224e-05,
+      "loss": 1.966,
+      "step": 4670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.237324776432779e-05,
+      "loss": 2.1098,
+      "step": 4680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2342370992286807e-05,
+      "loss": 1.9611,
+      "step": 4690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.231144314373812e-05,
+      "loss": 1.8736,
+      "step": 4700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.228046430976998e-05,
+      "loss": 1.94,
+      "step": 4710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2249434581620834e-05,
+      "loss": 2.0265,
+      "step": 4720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.221835405067899e-05,
+      "loss": 1.9346,
+      "step": 4730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.21872228084824e-05,
+      "loss": 1.8302,
+      "step": 4740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.215604094671835e-05,
+      "loss": 2.0056,
+      "step": 4750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2124808557223215e-05,
+      "loss": 1.98,
+      "step": 4760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.209352573198222e-05,
+      "loss": 2.0143,
+      "step": 4770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.206219256312908e-05,
+      "loss": 2.0,
+      "step": 4780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.203080914294582e-05,
+      "loss": 2.0888,
+      "step": 4790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.199937556386243e-05,
+      "loss": 1.9942,
+      "step": 4800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1967891918456646e-05,
+      "loss": 2.0879,
+      "step": 4810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1936358299453685e-05,
+      "loss": 1.808,
+      "step": 4820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.190477479972589e-05,
+      "loss": 2.0908,
+      "step": 4830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1873141512292556e-05,
+      "loss": 1.8843,
+      "step": 4840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.18414585303196e-05,
+      "loss": 1.9642,
+      "step": 4850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1809725947119284e-05,
+      "loss": 2.0837,
+      "step": 4860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.177794385614998e-05,
+      "loss": 1.891,
+      "step": 4870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.174611235101584e-05,
+      "loss": 2.0398,
+      "step": 4880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1714231525466585e-05,
+      "loss": 1.911,
+      "step": 4890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1682301473397155e-05,
+      "loss": 1.9827,
+      "step": 4900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.165032228884751e-05,
+      "loss": 2.0538,
+      "step": 4910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.161829406600229e-05,
+      "loss": 1.9562,
+      "step": 4920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.158621689919056e-05,
+      "loss": 1.9036,
+      "step": 4930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1554090882885557e-05,
+      "loss": 1.9053,
+      "step": 4940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.152191611170435e-05,
+      "loss": 2.0952,
+      "step": 4950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.148969268040764e-05,
+      "loss": 1.9669,
+      "step": 4960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.145742068389943e-05,
+      "loss": 1.9376,
+      "step": 4970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.142510021722673e-05,
+      "loss": 1.9656,
+      "step": 4980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1392731375579355e-05,
+      "loss": 1.984,
+      "step": 4990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.136031425428953e-05,
+      "loss": 1.9876,
+      "step": 5000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1327848948831724e-05,
+      "loss": 1.8261,
+      "step": 5010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.12953355548223e-05,
+      "loss": 1.971,
+      "step": 5020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1262774168019254e-05,
+      "loss": 1.9736,
+      "step": 5030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.123016488432191e-05,
+      "loss": 2.0716,
+      "step": 5040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.119750779977068e-05,
+      "loss": 2.0165,
+      "step": 5050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.116480301054676e-05,
+      "loss": 1.7702,
+      "step": 5060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.113205061297183e-05,
+      "loss": 1.9245,
+      "step": 5070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1099250703507786e-05,
+      "loss": 2.1085,
+      "step": 5080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1066403378756466e-05,
+      "loss": 1.955,
+      "step": 5090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.103350873545935e-05,
+      "loss": 2.0126,
+      "step": 5100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.100056687049728e-05,
+      "loss": 1.9846,
+      "step": 5110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0967577880890177e-05,
+      "loss": 1.9775,
+      "step": 5120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.093454186379674e-05,
+      "loss": 2.082,
+      "step": 5130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.090145891651419e-05,
+      "loss": 1.9719,
+      "step": 5140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.086832913647796e-05,
+      "loss": 2.0042,
+      "step": 5150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.083515262126142e-05,
+      "loss": 1.9684,
+      "step": 5160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.080192946857556e-05,
+      "loss": 1.9915,
+      "step": 5170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.076865977626876e-05,
+      "loss": 1.9712,
+      "step": 5180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.073534364232644e-05,
+      "loss": 2.013,
+      "step": 5190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0701981164870815e-05,
+      "loss": 1.8408,
+      "step": 5200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.066857244216059e-05,
+      "loss": 1.7848,
+      "step": 5210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.063511757259066e-05,
+      "loss": 2.028,
+      "step": 5220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0601616654691834e-05,
+      "loss": 1.9126,
+      "step": 5230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0568069787130544e-05,
+      "loss": 1.929,
+      "step": 5240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0534477068708575e-05,
+      "loss": 1.9715,
+      "step": 5250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.05008385983627e-05,
+      "loss": 1.9898,
+      "step": 5260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0467154475164496e-05,
+      "loss": 1.9725,
+      "step": 5270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.043342479831995e-05,
+      "loss": 1.9426,
+      "step": 5280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.039964966716924e-05,
+      "loss": 1.9315,
+      "step": 5290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.036582918118642e-05,
+      "loss": 2.04,
+      "step": 5300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.033196343997909e-05,
+      "loss": 1.9925,
+      "step": 5310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.029805254328817e-05,
+      "loss": 2.1308,
+      "step": 5320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0264096590987554e-05,
+      "loss": 1.9229,
+      "step": 5330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.023009568308383e-05,
+      "loss": 2.0006,
+      "step": 5340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.019945651169198e-05,
+      "loss": 1.983,
+      "step": 5350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.016537046413449e-05,
+      "loss": 2.0312,
+      "step": 5360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.013123975174071e-05,
+      "loss": 1.977,
+      "step": 5370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.009706447503192e-05,
+      "loss": 2.0191,
+      "step": 5380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0062844734660677e-05,
+      "loss": 2.0966,
+      "step": 5390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0028580631410475e-05,
+      "loss": 1.889,
+      "step": 5400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.999427226619546e-05,
+      "loss": 1.9679,
+      "step": 5410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.995991974006016e-05,
+      "loss": 1.8825,
+      "step": 5420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.992552315417916e-05,
+      "loss": 1.9696,
+      "step": 5430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.989108260985678e-05,
+      "loss": 1.8736,
+      "step": 5440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.985659820852685e-05,
+      "loss": 1.9988,
+      "step": 5450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9822070051752335e-05,
+      "loss": 2.0182,
+      "step": 5460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.978749824122508e-05,
+      "loss": 2.0026,
+      "step": 5470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.975288287876549e-05,
+      "loss": 2.0005,
+      "step": 5480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9718224066322254e-05,
+      "loss": 1.8963,
+      "step": 5490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.968352190597201e-05,
+      "loss": 1.9234,
+      "step": 5500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.964877649991908e-05,
+      "loss": 1.9644,
+      "step": 5510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.961398795049516e-05,
+      "loss": 1.8932,
+      "step": 5520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9579156360158967e-05,
+      "loss": 1.8613,
+      "step": 5530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9544281831496034e-05,
+      "loss": 2.0588,
+      "step": 5540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.950936446721833e-05,
+      "loss": 1.94,
+      "step": 5550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.947440437016398e-05,
+      "loss": 2.0391,
+      "step": 5560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.943940164329698e-05,
+      "loss": 2.0112,
+      "step": 5570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.940435638970687e-05,
+      "loss": 1.998,
+      "step": 5580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.936926871260842e-05,
+      "loss": 1.8842,
+      "step": 5590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.933413871534139e-05,
+      "loss": 2.0282,
+      "step": 5600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.929896650137014e-05,
+      "loss": 1.9234,
+      "step": 5610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.926375217428339e-05,
+      "loss": 2.0473,
+      "step": 5620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9228495837793876e-05,
+      "loss": 1.9804,
+      "step": 5630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.919319759573806e-05,
+      "loss": 1.9598,
+      "step": 5640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9157857552075843e-05,
+      "loss": 1.9973,
+      "step": 5650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9122475810890216e-05,
+      "loss": 1.9633,
+      "step": 5660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.908705247638699e-05,
+      "loss": 1.8992,
+      "step": 5670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9051587652894475e-05,
+      "loss": 2.1263,
+      "step": 5680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.901608144486317e-05,
+      "loss": 2.0531,
+      "step": 5690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.898053395686545e-05,
+      "loss": 1.7571,
+      "step": 5700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.89449452935953e-05,
+      "loss": 2.0408,
+      "step": 5710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8909315559867926e-05,
+      "loss": 1.9259,
+      "step": 5720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.887364486061954e-05,
+      "loss": 1.9477,
+      "step": 5730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.883793330090698e-05,
+      "loss": 1.9546,
+      "step": 5740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.880218098590742e-05,
+      "loss": 2.1015,
+      "step": 5750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8766388020918084e-05,
+      "loss": 1.8506,
+      "step": 5760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8730554511355924e-05,
+      "loss": 2.0107,
+      "step": 5770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.869468056275727e-05,
+      "loss": 1.9434,
+      "step": 5780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8658766280777575e-05,
+      "loss": 2.0205,
+      "step": 5790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.862281177119108e-05,
+      "loss": 2.0349,
+      "step": 5800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8586817139890515e-05,
+      "loss": 1.8608,
+      "step": 5810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8550782492886745e-05,
+      "loss": 1.8905,
+      "step": 5820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.851470793630852e-05,
+      "loss": 1.8698,
+      "step": 5830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.84785935764021e-05,
+      "loss": 1.8985,
+      "step": 5840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.844243951953101e-05,
+      "loss": 2.0068,
+      "step": 5850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.840624587217566e-05,
+      "loss": 2.0469,
+      "step": 5860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.837001274093308e-05,
+      "loss": 2.1191,
+      "step": 5870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.833374023251657e-05,
+      "loss": 2.0286,
+      "step": 5880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.829742845375542e-05,
+      "loss": 1.9234,
+      "step": 5890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.826107751159458e-05,
+      "loss": 1.8821,
+      "step": 5900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.822468751309432e-05,
+      "loss": 2.0205,
+      "step": 5910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.818825856542996e-05,
+      "loss": 2.0593,
+      "step": 5920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.815179077589151e-05,
+      "loss": 1.9954,
+      "step": 5930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.811528425188342e-05,
+      "loss": 1.9752,
+      "step": 5940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.807873910092418e-05,
+      "loss": 1.8811,
+      "step": 5950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.804215543064604e-05,
+      "loss": 1.928,
+      "step": 5960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8005533348794735e-05,
+      "loss": 1.9776,
+      "step": 5970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7968872963229095e-05,
+      "loss": 2.1684,
+      "step": 5980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.793217438192078e-05,
+      "loss": 1.9877,
+      "step": 5990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7895437712953924e-05,
+      "loss": 2.0868,
+      "step": 6000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.785866306452486e-05,
+      "loss": 1.9248,
+      "step": 6010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7821850544941765e-05,
+      "loss": 1.9546,
+      "step": 6020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.778500026262436e-05,
+      "loss": 2.0194,
+      "step": 6030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.774811232610356e-05,
+      "loss": 1.9202,
+      "step": 6040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7711186844021224e-05,
+      "loss": 1.9292,
+      "step": 6050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7674223925129746e-05,
+      "loss": 1.8941,
+      "step": 6060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.76372236782918e-05,
+      "loss": 2.0049,
+      "step": 6070
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.760018621248e-05,
+      "loss": 2.022,
+      "step": 6080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.756311163677656e-05,
+      "loss": 1.994,
+      "step": 6090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.752600006037301e-05,
+      "loss": 1.9079,
+      "step": 6100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.748885159256984e-05,
+      "loss": 1.8415,
+      "step": 6110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7451666342776195e-05,
+      "loss": 1.9982,
+      "step": 6120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7414444420509546e-05,
+      "loss": 1.9564,
+      "step": 6130
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7377185935395395e-05,
+      "loss": 1.8164,
+      "step": 6140
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7339890997166885e-05,
+      "loss": 1.9376,
+      "step": 6150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7302559715664567e-05,
+      "loss": 1.9859,
+      "step": 6160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7265192200835986e-05,
+      "loss": 1.8536,
+      "step": 6170
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.7227788562735436e-05,
+      "loss": 1.8531,
+      "step": 6180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.719034891152359e-05,
+      "loss": 1.9031,
+      "step": 6190
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7152873357467176e-05,
+      "loss": 1.9184,
+      "step": 6200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.711536201093867e-05,
+      "loss": 1.9665,
+      "step": 6210
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.707781498241596e-05,
+      "loss": 1.8841,
+      "step": 6220
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.704023238248204e-05,
+      "loss": 1.8692,
+      "step": 6230
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7002614321824655e-05,
+      "loss": 1.8995,
+      "step": 6240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.696496091123597e-05,
+      "loss": 1.9404,
+      "step": 6250
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.69272722616123e-05,
+      "loss": 1.7922,
+      "step": 6260
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.688954848395372e-05,
+      "loss": 2.0401,
+      "step": 6270
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.685178968936378e-05,
+      "loss": 1.7741,
+      "step": 6280
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.681399598904914e-05,
+      "loss": 1.8793,
+      "step": 6290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.677616749431928e-05,
+      "loss": 1.9928,
+      "step": 6300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.673830431658615e-05,
+      "loss": 1.9869,
+      "step": 6310
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.670040656736386e-05,
+      "loss": 1.8293,
+      "step": 6320
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.666247435826832e-05,
+      "loss": 1.8803,
+      "step": 6330
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.6624507801016934e-05,
+      "loss": 1.9685,
+      "step": 6340
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.658650700742828e-05,
+      "loss": 1.8484,
+      "step": 6350
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.654847208942177e-05,
+      "loss": 2.0776,
+      "step": 6360
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.65104031590173e-05,
+      "loss": 1.8762,
+      "step": 6370
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6472300328334954e-05,
+      "loss": 2.1146,
+      "step": 6380
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.643416370959466e-05,
+      "loss": 1.9551,
+      "step": 6390
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.6395993415115834e-05,
+      "loss": 2.0127,
+      "step": 6400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.635778955731712e-05,
+      "loss": 1.9544,
+      "step": 6410
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.631955224871596e-05,
+      "loss": 1.8784,
+      "step": 6420
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.628128160192834e-05,
+      "loss": 2.0609,
+      "step": 6430
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.624297772966845e-05,
+      "loss": 2.0819,
+      "step": 6440
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6204640744748296e-05,
+      "loss": 1.8544,
+      "step": 6450
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6166270760077444e-05,
+      "loss": 1.8568,
+      "step": 6460
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6127867888662624e-05,
+      "loss": 1.9933,
+      "step": 6470
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.608943224360744e-05,
+      "loss": 1.9074,
+      "step": 6480
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.605096393811202e-05,
+      "loss": 1.9247,
+      "step": 6490
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.601246308547266e-05,
+      "loss": 1.9073,
+      "step": 6500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.597392979908154e-05,
+      "loss": 1.9841,
+      "step": 6510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.593536419242636e-05,
+      "loss": 1.9886,
+      "step": 6520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.589676637909001e-05,
+      "loss": 1.8148,
+      "step": 6530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.58581364727502e-05,
+      "loss": 1.8943,
+      "step": 6540
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.581947458717921e-05,
+      "loss": 1.9289,
+      "step": 6550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.578078083624349e-05,
+      "loss": 1.9069,
+      "step": 6560
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.574205533390331e-05,
+      "loss": 1.8945,
+      "step": 6570
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.5703298194212493e-05,
+      "loss": 1.6912,
+      "step": 6580
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.566450953131801e-05,
+      "loss": 1.9443,
+      "step": 6590
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.56256894594597e-05,
+      "loss": 1.9717,
+      "step": 6600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.558683809296989e-05,
+      "loss": 1.968,
+      "step": 6610
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.554795554627307e-05,
+      "loss": 2.102,
+      "step": 6620
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.550904193388558e-05,
+      "loss": 1.9143,
+      "step": 6630
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.547009737041524e-05,
+      "loss": 1.915,
+      "step": 6640
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.5431121970561045e-05,
+      "loss": 2.0739,
+      "step": 6650
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.5392115849112776e-05,
+      "loss": 1.9887,
+      "step": 6660
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.5353079120950736e-05,
+      "loss": 1.9973,
+      "step": 6670
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.531401190104533e-05,
+      "loss": 1.9086,
+      "step": 6680
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.527491430445679e-05,
+      "loss": 1.8369,
+      "step": 6690
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.523578644633482e-05,
+      "loss": 1.9956,
+      "step": 6700
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.519662844191822e-05,
+      "loss": 2.0041,
+      "step": 6710
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.515744040653462e-05,
+      "loss": 1.9738,
+      "step": 6720
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.511822245560005e-05,
+      "loss": 1.9102,
+      "step": 6730
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.507897470461868e-05,
+      "loss": 1.9079,
+      "step": 6740
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.503969726918243e-05,
+      "loss": 1.9866,
+      "step": 6750
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.500039026497065e-05,
+      "loss": 2.0397,
+      "step": 6760
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.496105380774979e-05,
+      "loss": 2.0346,
+      "step": 6770
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.492168801337302e-05,
+      "loss": 2.0524,
+      "step": 6780
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.488229299777993e-05,
+      "loss": 1.9803,
+      "step": 6790
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.4842868876996166e-05,
+      "loss": 2.004,
+      "step": 6800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.480341576713308e-05,
+      "loss": 1.8878,
+      "step": 6810
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.476393378438745e-05,
+      "loss": 1.7989,
+      "step": 6820
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.4724423045041043e-05,
+      "loss": 2.0179,
+      "step": 6830
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.468488366546032e-05,
+      "loss": 1.7946,
+      "step": 6840
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.464531576209613e-05,
+      "loss": 2.0855,
+      "step": 6850
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.4605719451483284e-05,
+      "loss": 2.007,
+      "step": 6860
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.456609485024029e-05,
+      "loss": 1.9238,
+      "step": 6870
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.452644207506898e-05,
+      "loss": 1.9564,
+      "step": 6880
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.448676124275414e-05,
+      "loss": 1.9127,
+      "step": 6890
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.4447052470163194e-05,
+      "loss": 1.991,
+      "step": 6900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.440731587424587e-05,
+      "loss": 1.892,
+      "step": 6910
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.4367551572033827e-05,
+      "loss": 1.9243,
+      "step": 6920
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.432775968064034e-05,
+      "loss": 1.9049,
+      "step": 6930
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.428794031725992e-05,
+      "loss": 1.9647,
+      "step": 6940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.4248093599168e-05,
+      "loss": 1.9019,
+      "step": 6950
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.4208219643720585e-05,
+      "loss": 1.8408,
+      "step": 6960
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.4168318568353875e-05,
+      "loss": 1.9868,
+      "step": 6970
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.412839049058396e-05,
+      "loss": 1.9153,
+      "step": 6980
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.408843552800646e-05,
+      "loss": 1.9551,
+      "step": 6990
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.404845379829616e-05,
+      "loss": 1.9874,
+      "step": 7000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.40084454192067e-05,
+      "loss": 1.7889,
+      "step": 7010
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.3968410508570195e-05,
+      "loss": 2.1259,
+      "step": 7020
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.39283491842969e-05,
+      "loss": 1.8387,
+      "step": 7030
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.388826156437486e-05,
+      "loss": 1.828,
+      "step": 7040
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.384814776686958e-05,
+      "loss": 1.8782,
+      "step": 7050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.380800790992365e-05,
+      "loss": 1.9089,
+      "step": 7060
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.376784211175641e-05,
+      "loss": 1.9197,
+      "step": 7070
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.372765049066361e-05,
+      "loss": 2.0088,
+      "step": 7080
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.3687433165017043e-05,
+      "loss": 1.852,
+      "step": 7090
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.364719025326422e-05,
+      "loss": 1.8726,
+      "step": 7100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.3606921873927994e-05,
+      "loss": 2.0017,
+      "step": 7110
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.3566628145606225e-05,
+      "loss": 1.91,
+      "step": 7120
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.352630918697145e-05,
+      "loss": 1.9897,
+      "step": 7130
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.34859651167705e-05,
+      "loss": 1.8563,
+      "step": 7140
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.344559605382415e-05,
+      "loss": 1.982,
+      "step": 7150
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.340520211702681e-05,
+      "loss": 1.9946,
+      "step": 7160
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.336478342534613e-05,
+      "loss": 1.9853,
+      "step": 7170
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.3324340097822674e-05,
+      "loss": 1.8857,
+      "step": 7180
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.328387225356957e-05,
+      "loss": 2.0665,
+      "step": 7190
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.324338001177214e-05,
+      "loss": 1.8744,
+      "step": 7200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.320286349168756e-05,
+      "loss": 1.8437,
+      "step": 7210
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.316232281264453e-05,
+      "loss": 1.8133,
+      "step": 7220
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.3121758094042873e-05,
+      "loss": 1.9322,
+      "step": 7230
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.3081169455353236e-05,
+      "loss": 1.9957,
+      "step": 7240
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.304055701611671e-05,
+      "loss": 1.9397,
+      "step": 7250
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.299992089594449e-05,
+      "loss": 2.0457,
+      "step": 7260
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.295926121451749e-05,
+      "loss": 1.9236,
+      "step": 7270
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.291857809158605e-05,
+      "loss": 1.869,
+      "step": 7280
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.287787164696951e-05,
+      "loss": 1.9491,
+      "step": 7290
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.2837142000555944e-05,
+      "loss": 1.9828,
+      "step": 7300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.279638927230171e-05,
+      "loss": 2.0047,
+      "step": 7310
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.27556135822312e-05,
+      "loss": 1.9448,
+      "step": 7320
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.271481505043639e-05,
+      "loss": 1.8932,
+      "step": 7330
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.2673993797076537e-05,
+      "loss": 2.0415,
+      "step": 7340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.2633149942377834e-05,
+      "loss": 1.8329,
+      "step": 7350
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.259228360663302e-05,
+      "loss": 1.9159,
+      "step": 7360
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.255139491020106e-05,
+      "loss": 1.8783,
+      "step": 7370
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.251048397350677e-05,
+      "loss": 1.9245,
+      "step": 7380
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.2469550917040466e-05,
+      "loss": 1.9007,
+      "step": 7390
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.2428595861357605e-05,
+      "loss": 1.8904,
+      "step": 7400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.238761892707845e-05,
+      "loss": 1.8759,
+      "step": 7410
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.2346620234887685e-05,
+      "loss": 1.8945,
+      "step": 7420
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.230559990553409e-05,
+      "loss": 2.141,
+      "step": 7430
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.226455805983017e-05,
+      "loss": 1.8849,
+      "step": 7440
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.222349481865177e-05,
+      "loss": 2.006,
+      "step": 7450
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.2182410302937796e-05,
+      "loss": 1.9785,
+      "step": 7460
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.214130463368977e-05,
+      "loss": 2.0398,
+      "step": 7470
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.210017793197154e-05,
+      "loss": 1.917,
+      "step": 7480
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.20590303189089e-05,
+      "loss": 1.9382,
+      "step": 7490
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.201786191568921e-05,
+      "loss": 1.8686,
+      "step": 7500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.197667284356107e-05,
+      "loss": 1.9509,
+      "step": 7510
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.193546322383397e-05,
+      "loss": 1.8872,
+      "step": 7520
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.1894233177877895e-05,
+      "loss": 2.108,
+      "step": 7530
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.1852982827123e-05,
+      "loss": 1.8493,
+      "step": 7540
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.181171229305924e-05,
+      "loss": 1.9917,
+      "step": 7550
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.177042169723602e-05,
+      "loss": 2.0789,
+      "step": 7560
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.17291111612618e-05,
+      "loss": 1.8568,
+      "step": 7570
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.1687780806803826e-05,
+      "loss": 1.9191,
+      "step": 7580
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.164643075558765e-05,
+      "loss": 2.091,
+      "step": 7590
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.1605061129396866e-05,
+      "loss": 2.0078,
+      "step": 7600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.1563672050072726e-05,
+      "loss": 1.943,
+      "step": 7610
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.1522263639513765e-05,
+      "loss": 1.8967,
+      "step": 7620
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.1480836019675454e-05,
+      "loss": 1.9349,
+      "step": 7630
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.143938931256982e-05,
+      "loss": 2.0013,
+      "step": 7640
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.139792364026515e-05,
+      "loss": 1.8863,
+      "step": 7650
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.135643912488554e-05,
+      "loss": 1.8622,
+      "step": 7660
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.1314935888610606e-05,
+      "loss": 1.8247,
+      "step": 7670
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.127341405367511e-05,
+      "loss": 1.8877,
+      "step": 7680
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.123187374236856e-05,
+      "loss": 1.9251,
+      "step": 7690
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.119031507703491e-05,
+      "loss": 2.0492,
+      "step": 7700
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.114873818007215e-05,
+      "loss": 1.8882,
+      "step": 7710
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.1107143173931985e-05,
+      "loss": 1.9262,
+      "step": 7720
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.106553018111943e-05,
+      "loss": 1.9726,
+      "step": 7730
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.102389932419249e-05,
+      "loss": 1.9772,
+      "step": 7740
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.0982250725761797e-05,
+      "loss": 1.905,
+      "step": 7750
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.09405845084902e-05,
+      "loss": 1.7988,
+      "step": 7760
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.089890079509246e-05,
+      "loss": 1.959,
+      "step": 7770
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.085719970833487e-05,
+      "loss": 1.9042,
+      "step": 7780
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.0815481371034886e-05,
+      "loss": 1.8039,
+      "step": 7790
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.077374590606077e-05,
+      "loss": 2.0714,
+      "step": 7800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.073199343633122e-05,
+      "loss": 1.9137,
+      "step": 7810
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.069022408481503e-05,
+      "loss": 2.0302,
+      "step": 7820
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.064843797453071e-05,
+      "loss": 1.916,
+      "step": 7830
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.060663522854612e-05,
+      "loss": 1.9138,
+      "step": 7840
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.056481596997813e-05,
+      "loss": 1.9282,
+      "step": 7850
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.052298032199222e-05,
+      "loss": 1.9297,
+      "step": 7860
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.048112840780216e-05,
+      "loss": 1.6586,
+      "step": 7870
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.043926035066963e-05,
+      "loss": 1.9164,
+      "step": 7880
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.0397376273903834e-05,
+      "loss": 1.9542,
+      "step": 7890
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.035547630086116e-05,
+      "loss": 2.1619,
+      "step": 7900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.031356055494484e-05,
+      "loss": 1.904,
+      "step": 7910
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.027162915960453e-05,
+      "loss": 1.9112,
+      "step": 7920
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.0229682238335983e-05,
+      "loss": 1.9247,
+      "step": 7930
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.018771991468069e-05,
+      "loss": 1.948,
+      "step": 7940
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.0145742312225496e-05,
+      "loss": 1.948,
+      "step": 7950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.0103749554602244e-05,
+      "loss": 2.052,
+      "step": 7960
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.0061741765487418e-05,
+      "loss": 1.939,
+      "step": 7970
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.0019719068601766e-05,
+      "loss": 1.8501,
+      "step": 7980
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.9977681587709945e-05,
+      "loss": 2.0538,
+      "step": 7990
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.9935629446620156e-05,
+      "loss": 1.8272,
+      "step": 8000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.989356276918378e-05,
+      "loss": 1.9738,
+      "step": 8010
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.9851481679295e-05,
+      "loss": 1.8888,
+      "step": 8020
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.9809386300890445e-05,
+      "loss": 1.8957,
+      "step": 8030
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.976727675794885e-05,
+      "loss": 2.0007,
+      "step": 8040
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.9725153174490626e-05,
+      "loss": 1.8566,
+      "step": 8050
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.9683015674577586e-05,
+      "loss": 1.8144,
+      "step": 8060
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.9640864382312482e-05,
+      "loss": 2.0015,
+      "step": 8070
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.959869942183872e-05,
+      "loss": 1.881,
+      "step": 8080
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.9556520917339947e-05,
+      "loss": 1.9322,
+      "step": 8090
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.9514328993039702e-05,
+      "loss": 1.8597,
+      "step": 8100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.9472123773201054e-05,
+      "loss": 2.0391,
+      "step": 8110
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.9429905382126217e-05,
+      "loss": 1.9099,
+      "step": 8120
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.93876739441562e-05,
+      "loss": 1.8762,
+      "step": 8130
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.934542958367046e-05,
+      "loss": 1.961,
+      "step": 8140
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.9303172425086467e-05,
+      "loss": 2.0124,
+      "step": 8150
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.9260902592859436e-05,
+      "loss": 1.9177,
+      "step": 8160
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.921862021148188e-05,
+      "loss": 1.9627,
+      "step": 8170
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.9176325405483263e-05,
+      "loss": 1.9842,
+      "step": 8180
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.9134018299429666e-05,
+      "loss": 1.8967,
+      "step": 8190
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.909169901792338e-05,
+      "loss": 2.0008,
+      "step": 8200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.9049367685602552e-05,
+      "loss": 1.9733,
+      "step": 8210
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.9007024427140844e-05,
+      "loss": 1.7992,
+      "step": 8220
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.896466936724701e-05,
+      "loss": 2.0379,
+      "step": 8230
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.892230263066459e-05,
+      "loss": 1.9733,
+      "step": 8240
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.88799243421715e-05,
+      "loss": 1.9041,
+      "step": 8250
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.883753462657967e-05,
+      "loss": 2.0478,
+      "step": 8260
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8795133608734714e-05,
+      "loss": 1.865,
+      "step": 8270
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.87527214135155e-05,
+      "loss": 1.9392,
+      "step": 8280
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.871029816583383e-05,
+      "loss": 1.927,
+      "step": 8290
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.866786399063407e-05,
+      "loss": 2.083,
+      "step": 8300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.862541901289275e-05,
+      "loss": 2.0775,
+      "step": 8310
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.858296335761822e-05,
+      "loss": 1.8612,
+      "step": 8320
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.854049714985028e-05,
+      "loss": 1.9326,
+      "step": 8330
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.849802051465982e-05,
+      "loss": 2.0342,
+      "step": 8340
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.845553357714842e-05,
+      "loss": 1.8879,
+      "step": 8350
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.841303646244801e-05,
+      "loss": 2.0243,
+      "step": 8360
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.837052929572051e-05,
+      "loss": 1.9649,
+      "step": 8370
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.8328012202157417e-05,
+      "loss": 1.9839,
+      "step": 8380
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.828548530697948e-05,
+      "loss": 1.9964,
+      "step": 8390
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.824294873543632e-05,
+      "loss": 1.8394,
+      "step": 8400
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.8200402612806044e-05,
+      "loss": 2.0053,
+      "step": 8410
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.8157847064394906e-05,
+      "loss": 1.9461,
+      "step": 8420
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.8115282215536897e-05,
+      "loss": 1.9126,
+      "step": 8430
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.8072708191593423e-05,
+      "loss": 1.8972,
+      "step": 8440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.8030125117952895e-05,
+      "loss": 1.888,
+      "step": 8450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7987533120030386e-05,
+      "loss": 1.8331,
+      "step": 8460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.794493232326725e-05,
+      "loss": 1.9081,
+      "step": 8470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7902322853130757e-05,
+      "loss": 1.8716,
+      "step": 8480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7859704835113724e-05,
+      "loss": 1.8939,
+      "step": 8490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7817078394734136e-05,
+      "loss": 1.9546,
+      "step": 8500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7774443657534788e-05,
+      "loss": 2.0163,
+      "step": 8510
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.773180074908291e-05,
+      "loss": 1.9137,
+      "step": 8520
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.7689149794969792e-05,
+      "loss": 1.8041,
+      "step": 8530
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.7646490920810435e-05,
+      "loss": 2.0293,
+      "step": 8540
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.760382425224315e-05,
+      "loss": 1.9179,
+      "step": 8550
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.7561149914929214e-05,
+      "loss": 1.8952,
+      "step": 8560
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.751846803455248e-05,
+      "loss": 1.9326,
+      "step": 8570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.7475778736819035e-05,
+      "loss": 1.8933,
+      "step": 8580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.743308214745679e-05,
+      "loss": 2.0392,
+      "step": 8590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.7390378392215145e-05,
+      "loss": 1.8614,
+      "step": 8600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.73476675968646e-05,
+      "loss": 1.8287,
+      "step": 8610
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.730494988719639e-05,
+      "loss": 1.8142,
+      "step": 8620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.726222538902211e-05,
+      "loss": 1.9309,
+      "step": 8630
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.7219494228173365e-05,
+      "loss": 1.9795,
+      "step": 8640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.7176756530501356e-05,
+      "loss": 2.1641,
+      "step": 8650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.7134012421876558e-05,
+      "loss": 2.025,
+      "step": 8660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.7091262028188313e-05,
+      "loss": 1.9185,
+      "step": 8670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.7048505475344492e-05,
+      "loss": 2.0523,
+      "step": 8680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.7005742889271092e-05,
+      "loss": 1.9776,
+      "step": 8690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6962974395911868e-05,
+      "loss": 1.9455,
+      "step": 8700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6920200121227996e-05,
+      "loss": 1.922,
+      "step": 8710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6877420191197677e-05,
+      "loss": 2.1397,
+      "step": 8720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6834634731815757e-05,
+      "loss": 2.0048,
+      "step": 8730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.679184386909336e-05,
+      "loss": 1.9587,
+      "step": 8740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.674904772905754e-05,
+      "loss": 1.9571,
+      "step": 8750
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6706246437750897e-05,
+      "loss": 1.799,
+      "step": 8760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6663440121231186e-05,
+      "loss": 2.0012,
+      "step": 8770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6620628905570964e-05,
+      "loss": 1.8956,
+      "step": 8780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6577812916857235e-05,
+      "loss": 1.9246,
+      "step": 8790
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6534992281191046e-05,
+      "loss": 1.9071,
+      "step": 8800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6492167124687122e-05,
+      "loss": 1.9827,
+      "step": 8810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6449337573473522e-05,
+      "loss": 1.9296,
+      "step": 8820
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6406503753691235e-05,
+      "loss": 1.9471,
+      "step": 8830
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.636366579149383e-05,
+      "loss": 1.8902,
+      "step": 8840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6320823813047068e-05,
+      "loss": 2.0053,
+      "step": 8850
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.627797794452855e-05,
+      "loss": 1.848,
+      "step": 8860
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6235128312127317e-05,
+      "loss": 1.9529,
+      "step": 8870
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.619227504204351e-05,
+      "loss": 1.991,
+      "step": 8880
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6149418260487978e-05,
+      "loss": 1.8716,
+      "step": 8890
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6106558093681915e-05,
+      "loss": 2.0044,
+      "step": 8900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.606369466785648e-05,
+      "loss": 1.9489,
+      "step": 8910
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.602082810925243e-05,
+      "loss": 1.8745,
+      "step": 8920
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.597795854411976e-05,
+      "loss": 1.9142,
+      "step": 8930
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5935086098717316e-05,
+      "loss": 1.8823,
+      "step": 8940
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5892210899312403e-05,
+      "loss": 1.9581,
+      "step": 8950
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5849333072180475e-05,
+      "loss": 1.9509,
+      "step": 8960
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5806452743604698e-05,
+      "loss": 1.8871,
+      "step": 8970
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5763570039875617e-05,
+      "loss": 1.9067,
+      "step": 8980
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.572068508729077e-05,
+      "loss": 2.0189,
+      "step": 8990
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5677798012154314e-05,
+      "loss": 2.0078,
+      "step": 9000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5634908940776658e-05,
+      "loss": 1.9555,
+      "step": 9010
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5592017999474105e-05,
+      "loss": 1.9384,
+      "step": 9020
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5549125314568444e-05,
+      "loss": 1.9055,
+      "step": 9030
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5506231012386618e-05,
+      "loss": 1.8214,
+      "step": 9040
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.546333521926031e-05,
+      "loss": 1.8583,
+      "step": 9050
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5420438061525616e-05,
+      "loss": 1.9694,
+      "step": 9060
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5377539665522642e-05,
+      "loss": 1.8255,
+      "step": 9070
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.533464015759514e-05,
+      "loss": 2.0016,
+      "step": 9080
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.529173966409015e-05,
+      "loss": 1.9666,
+      "step": 9090
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5248838311357594e-05,
+      "loss": 1.9809,
+      "step": 9100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5205936225749927e-05,
+      "loss": 1.9496,
+      "step": 9110
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.516303353362178e-05,
+      "loss": 1.9599,
+      "step": 9120
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5120130361329563e-05,
+      "loss": 1.9269,
+      "step": 9130
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5077226835231087e-05,
+      "loss": 1.8869,
+      "step": 9140
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5034323081685223e-05,
+      "loss": 1.9668,
+      "step": 9150
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.49914192270515e-05,
+      "loss": 2.0014,
+      "step": 9160
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.4948515397689746e-05,
+      "loss": 1.9867,
+      "step": 9170
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.490561171995972e-05,
+      "loss": 1.7927,
+      "step": 9180
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4862708320220735e-05,
+      "loss": 1.9955,
+      "step": 9190
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4819805324831277e-05,
+      "loss": 1.9986,
+      "step": 9200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4776902860148648e-05,
+      "loss": 1.9717,
+      "step": 9210
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4734001052528586e-05,
+      "loss": 1.9631,
+      "step": 9220
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.469110002832489e-05,
+      "loss": 1.9488,
+      "step": 9230
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4648199913889057e-05,
+      "loss": 1.9523,
+      "step": 9240
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4605300835569904e-05,
+      "loss": 1.9669,
+      "step": 9250
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.456240291971319e-05,
+      "loss": 2.0942,
+      "step": 9260
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4519506292661258e-05,
+      "loss": 1.9066,
+      "step": 9270
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4476611080752647e-05,
+      "loss": 1.9441,
+      "step": 9280
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.443371741032174e-05,
+      "loss": 1.8257,
+      "step": 9290
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4390825407698366e-05,
+      "loss": 2.0184,
+      "step": 9300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4347935199207462e-05,
+      "loss": 2.0188,
+      "step": 9310
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4305046911168653e-05,
+      "loss": 1.9646,
+      "step": 9320
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4262160669895932e-05,
+      "loss": 1.7424,
+      "step": 9330
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4219276601697252e-05,
+      "loss": 1.8991,
+      "step": 9340
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4176394832874173e-05,
+      "loss": 1.8611,
+      "step": 9350
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4133515489721473e-05,
+      "loss": 1.995,
+      "step": 9360
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4090638698526798e-05,
+      "loss": 1.9856,
+      "step": 9370
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.404776458557026e-05,
+      "loss": 1.9171,
+      "step": 9380
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4004893277124107e-05,
+      "loss": 1.8741,
+      "step": 9390
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.3962024899452314e-05,
+      "loss": 2.0754,
+      "step": 9400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.3919159578810224e-05,
+      "loss": 1.9571,
+      "step": 9410
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.3876297441444176e-05,
+      "loss": 1.9676,
+      "step": 9420
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.3833438613591136e-05,
+      "loss": 1.9653,
+      "step": 9430
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.3790583221478322e-05,
+      "loss": 1.8635,
+      "step": 9440
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.3747731391322836e-05,
+      "loss": 2.0375,
+      "step": 9450
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.3704883249331284e-05,
+      "loss": 1.9624,
+      "step": 9460
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.3662038921699415e-05,
+      "loss": 1.9342,
+      "step": 9470
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.3619198534611736e-05,
+      "loss": 1.9666,
+      "step": 9480
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3576362214241164e-05,
+      "loss": 1.9232,
+      "step": 9490
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3533530086748618e-05,
+      "loss": 1.9101,
+      "step": 9500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3490702278282687e-05,
+      "loss": 1.9451,
+      "step": 9510
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3447878914979226e-05,
+      "loss": 1.8747,
+      "step": 9520
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.340506012296101e-05,
+      "loss": 1.9355,
+      "step": 9530
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3362246028337335e-05,
+      "loss": 1.901,
+      "step": 9540
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3319436757203684e-05,
+      "loss": 1.9962,
+      "step": 9550
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.327663243564131e-05,
+      "loss": 1.8469,
+      "step": 9560
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3233833189716906e-05,
+      "loss": 2.04,
+      "step": 9570
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.319103914548221e-05,
+      "loss": 1.9402,
+      "step": 9580
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.314825042897364e-05,
+      "loss": 1.8907,
+      "step": 9590
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3105467166211924e-05,
+      "loss": 1.777,
+      "step": 9600
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.306268948320173e-05,
+      "loss": 1.8906,
+      "step": 9610
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.301991750593128e-05,
+      "loss": 1.91,
+      "step": 9620
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2977151360372014e-05,
+      "loss": 2.0309,
+      "step": 9630
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.293439117247818e-05,
+      "loss": 1.8602,
+      "step": 9640
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2891637068186478e-05,
+      "loss": 1.8182,
+      "step": 9650
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2848889173415695e-05,
+      "loss": 1.9348,
+      "step": 9660
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.280614761406634e-05,
+      "loss": 1.8705,
+      "step": 9670
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.276341251602025e-05,
+      "loss": 1.9411,
+      "step": 9680
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.2720684005140237e-05,
+      "loss": 1.8468,
+      "step": 9690
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.2677962207269706e-05,
+      "loss": 1.8006,
+      "step": 9700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.26352472482323e-05,
+      "loss": 1.9849,
+      "step": 9710
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.259253925383152e-05,
+      "loss": 1.9391,
+      "step": 9720
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.254983834985035e-05,
+      "loss": 1.8878,
+      "step": 9730
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.250714466205089e-05,
+      "loss": 1.9733,
+      "step": 9740
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.2464458316173994e-05,
+      "loss": 2.0766,
+      "step": 9750
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.242177943793889e-05,
+      "loss": 1.9123,
+      "step": 9760
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.2379108153042806e-05,
+      "loss": 1.9031,
+      "step": 9770
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.2336444587160615e-05,
+      "loss": 1.7736,
+      "step": 9780
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.2293788865944456e-05,
+      "loss": 1.939,
+      "step": 9790
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.2251141115023354e-05,
+      "loss": 1.9272,
+      "step": 9800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.220850146000287e-05,
+      "loss": 1.9113,
+      "step": 9810
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.216587002646472e-05,
+      "loss": 1.9534,
+      "step": 9820
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.2123246939966398e-05,
+      "loss": 2.0214,
+      "step": 9830
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.2080632326040824e-05,
+      "loss": 2.1044,
+      "step": 9840
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.203802631019596e-05,
+      "loss": 1.9333,
+      "step": 9850
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1995429017914444e-05,
+      "loss": 2.0227,
+      "step": 9860
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1952840574653234e-05,
+      "loss": 1.8902,
+      "step": 9870
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.19102611058432e-05,
+      "loss": 2.0004,
+      "step": 9880
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1867690736888808e-05,
+      "loss": 2.0526,
+      "step": 9890
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1825129593167705e-05,
+      "loss": 1.9531,
+      "step": 9900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1782577800030375e-05,
+      "loss": 1.9861,
+      "step": 9910
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1740035482799765e-05,
+      "loss": 1.9593,
+      "step": 9920
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1697502766770903e-05,
+      "loss": 1.9101,
+      "step": 9930
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.165497977721056e-05,
+      "loss": 1.9489,
+      "step": 9940
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1612466639356837e-05,
+      "loss": 1.8974,
+      "step": 9950
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1569963478418843e-05,
+      "loss": 1.9795,
+      "step": 9960
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1527470419576278e-05,
+      "loss": 1.7991,
+      "step": 9970
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.148498758797911e-05,
+      "loss": 2.184,
+      "step": 9980
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.144251510874718e-05,
+      "loss": 1.8311,
+      "step": 9990
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.1400053106969833e-05,
+      "loss": 1.8869,
+      "step": 10000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.136184636695538e-05,
+      "loss": 1.9989,
+      "step": 10010
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.131940461685209e-05,
+      "loss": 2.0471,
+      "step": 10020
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.1276973706786676e-05,
+      "loss": 1.9202,
+      "step": 10030
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.1234553761726063e-05,
+      "loss": 1.9259,
+      "step": 10040
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.1192144906604876e-05,
+      "loss": 1.9674,
+      "step": 10050
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.1149747266325082e-05,
+      "loss": 1.9238,
+      "step": 10060
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.1107360965755614e-05,
+      "loss": 1.9393,
+      "step": 10070
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.1064986129732007e-05,
+      "loss": 1.863,
+      "step": 10080
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.1022622883056043e-05,
+      "loss": 1.9715,
+      "step": 10090
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.098027135049535e-05,
+      "loss": 1.9518,
+      "step": 10100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.093793165678308e-05,
+      "loss": 1.8673,
+      "step": 10110
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.0895603926617488e-05,
+      "loss": 1.9282,
+      "step": 10120
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.0853288284661627e-05,
+      "loss": 1.9571,
+      "step": 10130
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.0810984855542922e-05,
+      "loss": 1.9179,
+      "step": 10140
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.076869376385285e-05,
+      "loss": 1.8714,
+      "step": 10150
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.0726415134146525e-05,
+      "loss": 1.9576,
+      "step": 10160
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0684149090942385e-05,
+      "loss": 2.0512,
+      "step": 10170
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0641895758721786e-05,
+      "loss": 1.8169,
+      "step": 10180
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0599655261928648e-05,
+      "loss": 1.9669,
+      "step": 10190
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0557427724969087e-05,
+      "loss": 2.0943,
+      "step": 10200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0515213272211055e-05,
+      "loss": 1.9366,
+      "step": 10210
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.0473012027983968e-05,
+      "loss": 1.9655,
+      "step": 10220
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0430824116578334e-05,
+      "loss": 1.898,
+      "step": 10230
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0388649662245402e-05,
+      "loss": 2.0214,
+      "step": 10240
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0346488789196778e-05,
+      "loss": 1.8397,
+      "step": 10250
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.030434162160408e-05,
+      "loss": 1.9689,
+      "step": 10260
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.026220828359855e-05,
+      "loss": 1.9507,
+      "step": 10270
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.0220088899270706e-05,
+      "loss": 1.9769,
+      "step": 10280
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.0177983592669962e-05,
+      "loss": 1.8119,
+      "step": 10290
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.0135892487804285e-05,
+      "loss": 2.0847,
+      "step": 10300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.00938157086398e-05,
+      "loss": 1.8294,
+      "step": 10310
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.0051753379100456e-05,
+      "loss": 1.9204,
+      "step": 10320
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.0009705623067624e-05,
+      "loss": 1.9993,
+      "step": 10330
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.996767256437977e-05,
+      "loss": 1.8716,
+      "step": 10340
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9925654326832066e-05,
+      "loss": 1.9045,
+      "step": 10350
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.988365103417604e-05,
+      "loss": 1.8363,
+      "step": 10360
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.98416628101192e-05,
+      "loss": 1.9373,
+      "step": 10370
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9799689778324673e-05,
+      "loss": 1.8191,
+      "step": 10380
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.975773206241084e-05,
+      "loss": 1.9998,
+      "step": 10390
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9715789785950984e-05,
+      "loss": 1.8703,
+      "step": 10400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9673863072472905e-05,
+      "loss": 1.8561,
+      "step": 10410
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9631952045458567e-05,
+      "loss": 2.0243,
+      "step": 10420
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.959005682834375e-05,
+      "loss": 1.885,
+      "step": 10430
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9548177544517638e-05,
+      "loss": 1.9944,
+      "step": 10440
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.950631431732253e-05,
+      "loss": 1.8742,
+      "step": 10450
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9464467270053422e-05,
+      "loss": 1.9065,
+      "step": 10460
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9422636525957632e-05,
+      "loss": 1.8486,
+      "step": 10470
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9380822208234488e-05,
+      "loss": 1.8087,
+      "step": 10480
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9339024440034932e-05,
+      "loss": 2.0835,
+      "step": 10490
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9297243344461164e-05,
+      "loss": 1.9516,
+      "step": 10500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.925547904456628e-05,
+      "loss": 2.0303,
+      "step": 10510
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9213731663353907e-05,
+      "loss": 1.9044,
+      "step": 10520
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.917200132377785e-05,
+      "loss": 1.9487,
+      "step": 10530
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9130288148741722e-05,
+      "loss": 1.8156,
+      "step": 10540
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9088592261098577e-05,
+      "loss": 1.9664,
+      "step": 10550
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.904691378365056e-05,
+      "loss": 1.8768,
+      "step": 10560
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9005252839148535e-05,
+      "loss": 1.9158,
+      "step": 10570
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8963609550291736e-05,
+      "loss": 1.8299,
+      "step": 10580
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8921984039727386e-05,
+      "loss": 1.9149,
+      "step": 10590
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.888037643005036e-05,
+      "loss": 2.0225,
+      "step": 10600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.883878684380281e-05,
+      "loss": 2.0248,
+      "step": 10610
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.879721540347379e-05,
+      "loss": 1.9839,
+      "step": 10620
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8755662231498936e-05,
+      "loss": 2.0023,
+      "step": 10630
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8714127450260056e-05,
+      "loss": 1.8994,
+      "step": 10640
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.867261118208481e-05,
+      "loss": 1.787,
+      "step": 10650
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8631113549246332e-05,
+      "loss": 1.9655,
+      "step": 10660
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.858963467396286e-05,
+      "loss": 1.9958,
+      "step": 10670
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.85481746783974e-05,
+      "loss": 1.861,
+      "step": 10680
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8506733684657352e-05,
+      "loss": 1.832,
+      "step": 10690
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8465311814794146e-05,
+      "loss": 2.0125,
+      "step": 10700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.842390919080289e-05,
+      "loss": 1.8956,
+      "step": 10710
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.838252593462201e-05,
+      "loss": 2.0167,
+      "step": 10720
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.83411621681329e-05,
+      "loss": 1.9799,
+      "step": 10730
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8299818013159538e-05,
+      "loss": 1.9343,
+      "step": 10740
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8258493591468144e-05,
+      "loss": 1.7918,
+      "step": 10750
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8217189024766836e-05,
+      "loss": 1.8932,
+      "step": 10760
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.817590443470524e-05,
+      "loss": 2.1068,
+      "step": 10770
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8134639942874153e-05,
+      "loss": 1.9724,
+      "step": 10780
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8093395670805176e-05,
+      "loss": 1.7987,
+      "step": 10790
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8052171739970355e-05,
+      "loss": 2.0056,
+      "step": 10800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.801096827178184e-05,
+      "loss": 2.0165,
+      "step": 10810
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7969785387591507e-05,
+      "loss": 2.0446,
+      "step": 10820
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7928623208690605e-05,
+      "loss": 1.9494,
+      "step": 10830
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7887481856309407e-05,
+      "loss": 1.9438,
+      "step": 10840
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7846361451616845e-05,
+      "loss": 1.8628,
+      "step": 10850
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.780526211572016e-05,
+      "loss": 1.9008,
+      "step": 10860
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7764183969664543e-05,
+      "loss": 1.8994,
+      "step": 10870
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7723127134432765e-05,
+      "loss": 1.94,
+      "step": 10880
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.768209173094485e-05,
+      "loss": 1.9206,
+      "step": 10890
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7641077880057688e-05,
+      "loss": 1.8399,
+      "step": 10900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7600085702564695e-05,
+      "loss": 1.8322,
+      "step": 10910
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7559115319195456e-05,
+      "loss": 2.0276,
+      "step": 10920
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.751816685061538e-05,
+      "loss": 1.8318,
+      "step": 10930
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.7477240417425313e-05,
+      "loss": 1.7686,
+      "step": 10940
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.743633614016121e-05,
+      "loss": 1.9835,
+      "step": 10950
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7395454139293783e-05,
+      "loss": 1.999,
+      "step": 10960
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7354594535228117e-05,
+      "loss": 1.9484,
+      "step": 10970
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7313757448303354e-05,
+      "loss": 2.1417,
+      "step": 10980
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7272942998792303e-05,
+      "loss": 1.9128,
+      "step": 10990
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.7232151306901108e-05,
+      "loss": 2.0156,
+      "step": 11000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.719138249276889e-05,
+      "loss": 1.9242,
+      "step": 11010
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.715063667646739e-05,
+      "loss": 1.9097,
+      "step": 11020
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.7109913978000615e-05,
+      "loss": 1.9543,
+      "step": 11030
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.706921451730448e-05,
+      "loss": 1.9684,
+      "step": 11040
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.702853841424647e-05,
+      "loss": 1.8991,
+      "step": 11050
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6987885788625268e-05,
+      "loss": 1.7819,
+      "step": 11060
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6947256760170422e-05,
+      "loss": 1.9105,
+      "step": 11070
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.690665144854198e-05,
+      "loss": 1.9893,
+      "step": 11080
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.686606997333012e-05,
+      "loss": 1.9429,
+      "step": 11090
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6825512454054847e-05,
+      "loss": 1.9329,
+      "step": 11100
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6784979010165597e-05,
+      "loss": 1.9689,
+      "step": 11110
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.67444697610409e-05,
+      "loss": 1.9106,
+      "step": 11120
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.670398482598802e-05,
+      "loss": 2.0134,
+      "step": 11130
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.666352432424263e-05,
+      "loss": 1.8691,
+      "step": 11140
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6623088374968425e-05,
+      "loss": 1.9457,
+      "step": 11150
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6582677097256797e-05,
+      "loss": 1.8367,
+      "step": 11160
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.654229061012648e-05,
+      "loss": 1.9116,
+      "step": 11170
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6501929032523176e-05,
+      "loss": 1.9444,
+      "step": 11180
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6461592483319256e-05,
+      "loss": 1.716,
+      "step": 11190
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.642128108131334e-05,
+      "loss": 1.9249,
+      "step": 11200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6380994945230014e-05,
+      "loss": 1.9622,
+      "step": 11210
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6340734193719438e-05,
+      "loss": 2.0361,
+      "step": 11220
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.630049894535701e-05,
+      "loss": 1.962,
+      "step": 11230
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6260289318643028e-05,
+      "loss": 1.8252,
+      "step": 11240
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.622010543200231e-05,
+      "loss": 1.8004,
+      "step": 11250
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6179947403783878e-05,
+      "loss": 1.9624,
+      "step": 11260
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6139815352260595e-05,
+      "loss": 1.921,
+      "step": 11270
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6099709395628817e-05,
+      "loss": 1.9816,
+      "step": 11280
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.605962965200804e-05,
+      "loss": 1.826,
+      "step": 11290
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.601957623944057e-05,
+      "loss": 2.0038,
+      "step": 11300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5979549275891142e-05,
+      "loss": 1.885,
+      "step": 11310
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5939548879246623e-05,
+      "loss": 1.8514,
+      "step": 11320
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.589957516731562e-05,
+      "loss": 1.8615,
+      "step": 11330
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5859628257828135e-05,
+      "loss": 1.8731,
+      "step": 11340
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5819708268435253e-05,
+      "loss": 1.9081,
+      "step": 11350
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.577981531670877e-05,
+      "loss": 1.9615,
+      "step": 11360
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.573994952014084e-05,
+      "loss": 1.9939,
+      "step": 11370
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.570011099614365e-05,
+      "loss": 1.8591,
+      "step": 11380
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.566029986204907e-05,
+      "loss": 1.9059,
+      "step": 11390
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.5620516235108283e-05,
+      "loss": 1.8914,
+      "step": 11400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.558076023249147e-05,
+      "loss": 1.9376,
+      "step": 11410
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.5541031971287457e-05,
+      "loss": 1.909,
+      "step": 11420
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.550133156850336e-05,
+      "loss": 2.027,
+      "step": 11430
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.546165914106425e-05,
+      "loss": 1.8476,
+      "step": 11440
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.5422014805812794e-05,
+      "loss": 1.9253,
+      "step": 11450
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.5382398679508948e-05,
+      "loss": 1.8983,
+      "step": 11460
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.5342810878829557e-05,
+      "loss": 1.8594,
+      "step": 11470
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.5303251520368063e-05,
+      "loss": 1.9555,
+      "step": 11480
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.526372072063413e-05,
+      "loss": 1.892,
+      "step": 11490
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.5224218596053314e-05,
+      "loss": 1.9785,
+      "step": 11500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5184745262966727e-05,
+      "loss": 1.9193,
+      "step": 11510
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5145300837630665e-05,
+      "loss": 2.0285,
+      "step": 11520
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5105885436216302e-05,
+      "loss": 1.9913,
+      "step": 11530
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5066499174809322e-05,
+      "loss": 1.8473,
+      "step": 11540
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5027142169409595e-05,
+      "loss": 2.0129,
+      "step": 11550
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4987814535930814e-05,
+      "loss": 2.0078,
+      "step": 11560
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4948516390200176e-05,
+      "loss": 1.8992,
+      "step": 11570
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.490924784795803e-05,
+      "loss": 1.8327,
+      "step": 11580
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4870009024857535e-05,
+      "loss": 1.9823,
+      "step": 11590
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4830800036464322e-05,
+      "loss": 1.9335,
+      "step": 11600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.479162099825615e-05,
+      "loss": 1.8784,
+      "step": 11610
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4752472025622577e-05,
+      "loss": 1.9835,
+      "step": 11620
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4713353233864608e-05,
+      "loss": 1.9121,
+      "step": 11630
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.467426473819436e-05,
+      "loss": 1.8449,
+      "step": 11640
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4635206653734714e-05,
+      "loss": 1.9675,
+      "step": 11650
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4596179095519003e-05,
+      "loss": 1.9685,
+      "step": 11660
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4557182178490636e-05,
+      "loss": 1.9834,
+      "step": 11670
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4518216017502796e-05,
+      "loss": 2.0739,
+      "step": 11680
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4479280727318056e-05,
+      "loss": 2.0029,
+      "step": 11690
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.444037642260811e-05,
+      "loss": 2.0393,
+      "step": 11700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4401503217953343e-05,
+      "loss": 1.8968,
+      "step": 11710
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.43626612278426e-05,
+      "loss": 1.7695,
+      "step": 11720
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4323850566672741e-05,
+      "loss": 1.9631,
+      "step": 11730
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4285071348748402e-05,
+      "loss": 2.019,
+      "step": 11740
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4246323688281569e-05,
+      "loss": 2.024,
+      "step": 11750
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4207607699391329e-05,
+      "loss": 1.9119,
+      "step": 11760
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4168923496103454e-05,
+      "loss": 1.9272,
+      "step": 11770
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.413027119235013e-05,
+      "loss": 1.9799,
+      "step": 11780
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.409165090196956e-05,
+      "loss": 2.1668,
+      "step": 11790
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4053062738705708e-05,
+      "loss": 1.9532,
+      "step": 11800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4014506816207864e-05,
+      "loss": 1.8707,
+      "step": 11810
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3975983248030417e-05,
+      "loss": 1.9681,
+      "step": 11820
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3937492147632408e-05,
+      "loss": 1.7505,
+      "step": 11830
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3899033628377317e-05,
+      "loss": 1.7208,
+      "step": 11840
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.386060780353261e-05,
+      "loss": 1.9784,
+      "step": 11850
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.382221478626951e-05,
+      "loss": 2.0716,
+      "step": 11860
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3783854689662573e-05,
+      "loss": 2.0195,
+      "step": 11870
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3745527626689431e-05,
+      "loss": 1.9062,
+      "step": 11880
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.37072337102304e-05,
+      "loss": 2.0449,
+      "step": 11890
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3668973053068197e-05,
+      "loss": 1.874,
+      "step": 11900
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3630745767887559e-05,
+      "loss": 1.9891,
+      "step": 11910
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3592551967274971e-05,
+      "loss": 1.9407,
+      "step": 11920
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3554391763718255e-05,
+      "loss": 1.9498,
+      "step": 11930
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3516265269606329e-05,
+      "loss": 2.0592,
+      "step": 11940
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3478172597228787e-05,
+      "loss": 2.023,
+      "step": 11950
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3440113858775656e-05,
+      "loss": 1.944,
+      "step": 11960
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3402089166336972e-05,
+      "loss": 1.8695,
+      "step": 11970
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.336409863190255e-05,
+      "loss": 1.7808,
+      "step": 11980
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.332614236736156e-05,
+      "loss": 2.009,
+      "step": 11990
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.328822048450228e-05,
+      "loss": 1.9562,
+      "step": 12000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3250333095011685e-05,
+      "loss": 2.0046,
+      "step": 12010
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3212480310475207e-05,
+      "loss": 2.0237,
+      "step": 12020
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3174662242376307e-05,
+      "loss": 1.8996,
+      "step": 12030
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3136879002096258e-05,
+      "loss": 1.918,
+      "step": 12040
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3099130700913703e-05,
+      "loss": 1.9592,
+      "step": 12050
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3061417450004434e-05,
+      "loss": 1.8684,
+      "step": 12060
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3023739360440964e-05,
+      "loss": 2.0278,
+      "step": 12070
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2986096543192297e-05,
+      "loss": 1.9092,
+      "step": 12080
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2948489109123507e-05,
+      "loss": 1.8514,
+      "step": 12090
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2910917168995504e-05,
+      "loss": 1.8742,
+      "step": 12100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2873380833464616e-05,
+      "loss": 1.8696,
+      "step": 12110
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2835880213082355e-05,
+      "loss": 1.926,
+      "step": 12120
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2798415418295e-05,
+      "loss": 1.9938,
+      "step": 12130
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.276098655944335e-05,
+      "loss": 1.851,
+      "step": 12140
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2723593746762363e-05,
+      "loss": 1.9653,
+      "step": 12150
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.268623709038081e-05,
+      "loss": 2.03,
+      "step": 12160
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2648916700321004e-05,
+      "loss": 1.9661,
+      "step": 12170
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.261163268649842e-05,
+      "loss": 2.0107,
+      "step": 12180
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2574385158721433e-05,
+      "loss": 1.9078,
+      "step": 12190
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2537174226690917e-05,
+      "loss": 1.8743,
+      "step": 12200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2500000000000006e-05,
+      "loss": 1.9264,
+      "step": 12210
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2462862588133691e-05,
+      "loss": 1.8712,
+      "step": 12220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.242576210046858e-05,
+      "loss": 1.8131,
+      "step": 12230
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.2388698646272485e-05,
+      "loss": 1.8272,
+      "step": 12240
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.2351672334704192e-05,
+      "loss": 1.9069,
+      "step": 12250
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.2314683274813054e-05,
+      "loss": 1.8907,
+      "step": 12260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.2277731575538753e-05,
+      "loss": 1.8932,
+      "step": 12270
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.2240817345710895e-05,
+      "loss": 1.8835,
+      "step": 12280
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.2203940694048768e-05,
+      "loss": 1.9287,
+      "step": 12290
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.2167101729160954e-05,
+      "loss": 1.909,
+      "step": 12300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.2130300559545074e-05,
+      "loss": 1.8343,
+      "step": 12310
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.2093537293587393e-05,
+      "loss": 1.9186,
+      "step": 12320
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.205681203956259e-05,
+      "loss": 1.8223,
+      "step": 12330
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.2020124905633346e-05,
+      "loss": 1.9455,
+      "step": 12340
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1983475999850113e-05,
+      "loss": 1.7888,
+      "step": 12350
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1946865430150717e-05,
+      "loss": 2.0018,
+      "step": 12360
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.191029330436011e-05,
+      "loss": 1.9438,
+      "step": 12370
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1873759730189988e-05,
+      "loss": 1.7908,
+      "step": 12380
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1837264815238536e-05,
+      "loss": 1.74,
+      "step": 12390
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1800808666990049e-05,
+      "loss": 1.8098,
+      "step": 12400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1764391392814681e-05,
+      "loss": 1.9655,
+      "step": 12410
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.172801309996806e-05,
+      "loss": 2.0308,
+      "step": 12420
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.169167389559104e-05,
+      "loss": 1.9189,
+      "step": 12430
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.165537388670932e-05,
+      "loss": 1.9241,
+      "step": 12440
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1619113180233193e-05,
+      "loss": 1.9124,
+      "step": 12450
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1582891882957167e-05,
+      "loss": 1.7544,
+      "step": 12460
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1546710101559713e-05,
+      "loss": 2.1265,
+      "step": 12470
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.151056794260289e-05,
+      "loss": 1.931,
+      "step": 12480
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1474465512532095e-05,
+      "loss": 1.7336,
+      "step": 12490
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1438402917675684e-05,
+      "loss": 1.9121,
+      "step": 12500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1402380264244716e-05,
+      "loss": 1.9345,
+      "step": 12510
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1366397658332592e-05,
+      "loss": 1.8121,
+      "step": 12520
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1330455205914795e-05,
+      "loss": 1.8702,
+      "step": 12530
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1294553012848514e-05,
+      "loss": 1.9788,
+      "step": 12540
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.12586911848724e-05,
+      "loss": 2.0244,
+      "step": 12550
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1222869827606188e-05,
+      "loss": 1.9206,
+      "step": 12560
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.118708904655045e-05,
+      "loss": 1.9594,
+      "step": 12570
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1151348947086235e-05,
+      "loss": 1.7068,
+      "step": 12580
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1115649634474794e-05,
+      "loss": 1.9706,
+      "step": 12590
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1079991213857224e-05,
+      "loss": 1.9291,
+      "step": 12600
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.104437379025422e-05,
+      "loss": 1.8426,
+      "step": 12610
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1008797468565705e-05,
+      "loss": 1.7654,
+      "step": 12620
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0973262353570576e-05,
+      "loss": 1.9344,
+      "step": 12630
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0937768549926339e-05,
+      "loss": 1.7826,
+      "step": 12640
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0902316162168865e-05,
+      "loss": 1.8331,
+      "step": 12650
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.086690529471201e-05,
+      "loss": 1.8746,
+      "step": 12660
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0831536051847383e-05,
+      "loss": 1.929,
+      "step": 12670
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0796208537743963e-05,
+      "loss": 2.0229,
+      "step": 12680
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0760922856447867e-05,
+      "loss": 1.9805,
+      "step": 12690
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.072567911188197e-05,
+      "loss": 2.0412,
+      "step": 12700
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0690477407845672e-05,
+      "loss": 1.8469,
+      "step": 12710
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0655317848014524e-05,
+      "loss": 1.8281,
+      "step": 12720
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0620200535939977e-05,
+      "loss": 2.0146,
+      "step": 12730
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0585125575049038e-05,
+      "loss": 1.8658,
+      "step": 12740
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0550093068644e-05,
+      "loss": 1.9931,
+      "step": 12750
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0515103119902095e-05,
+      "loss": 1.9768,
+      "step": 12760
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0480155831875246e-05,
+      "loss": 1.8891,
+      "step": 12770
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0445251307489698e-05,
+      "loss": 1.8495,
+      "step": 12780
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0410389649545785e-05,
+      "loss": 1.9119,
+      "step": 12790
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0375570960717555e-05,
+      "loss": 1.9386,
+      "step": 12800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0340795343552534e-05,
+      "loss": 1.7968,
+      "step": 12810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0306062900471372e-05,
+      "loss": 1.8991,
+      "step": 12820
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0271373733767587e-05,
+      "loss": 1.939,
+      "step": 12830
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0236727945607208e-05,
+      "loss": 1.9135,
+      "step": 12840
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.0202125638028542e-05,
+      "loss": 1.9383,
+      "step": 12850
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.0167566912941803e-05,
+      "loss": 1.8411,
+      "step": 12860
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.0133051872128882e-05,
+      "loss": 1.8738,
+      "step": 12870
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.0098580617242972e-05,
+      "loss": 1.9741,
+      "step": 12880
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.006415324980835e-05,
+      "loss": 1.8719,
+      "step": 12890
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.0029769871219995e-05,
+      "loss": 1.8857,
+      "step": 12900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.995430582743373e-06,
+      "loss": 1.7881,
+      "step": 12910
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.96113548551405e-06,
+      "loss": 1.9407,
+      "step": 12920
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.926884680537482e-06,
+      "loss": 1.979,
+      "step": 12930
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.892678268688662e-06,
+      "loss": 1.9309,
+      "step": 12940
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.85851635071182e-06,
+      "loss": 1.9391,
+      "step": 12950
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.824399027220172e-06,
+      "loss": 1.8407,
+      "step": 12960
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.790326398695565e-06,
+      "loss": 1.8616,
+      "step": 12970
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.756298565488245e-06,
+      "loss": 1.9572,
+      "step": 12980
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.722315627816488e-06,
+      "loss": 1.6884,
+      "step": 12990
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.68837768576639e-06,
+      "loss": 1.8643,
+      "step": 13000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.654484839291474e-06,
+      "loss": 1.9333,
+      "step": 13010
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.6206371882125e-06,
+      "loss": 2.0164,
+      "step": 13020
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.586834832217076e-06,
+      "loss": 1.9204,
+      "step": 13030
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.553077870859444e-06,
+      "loss": 1.9232,
+      "step": 13040
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.519366403560114e-06,
+      "loss": 1.9398,
+      "step": 13050
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.485700529605649e-06,
+      "loss": 1.8918,
+      "step": 13060
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.452080348148284e-06,
+      "loss": 2.0889,
+      "step": 13070
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.418505958205729e-06,
+      "loss": 1.9136,
+      "step": 13080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.384977458660785e-06,
+      "loss": 1.7499,
+      "step": 13090
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.351494948261139e-06,
+      "loss": 2.036,
+      "step": 13100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.318058525618989e-06,
+      "loss": 1.8337,
+      "step": 13110
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.28466828921084e-06,
+      "loss": 1.8465,
+      "step": 13120
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.251324337377126e-06,
+      "loss": 1.9427,
+      "step": 13130
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.218026768322007e-06,
+      "loss": 1.9434,
+      "step": 13140
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.184775680113e-06,
+      "loss": 2.0562,
+      "step": 13150
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.151571170680755e-06,
+      "loss": 1.8532,
+      "step": 13160
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.118413337818716e-06,
+      "loss": 1.8191,
+      "step": 13170
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.085302279182877e-06,
+      "loss": 1.8,
+      "step": 13180
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.052238092291455e-06,
+      "loss": 1.8312,
+      "step": 13190
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.01922087452464e-06,
+      "loss": 1.9003,
+      "step": 13200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.986250723124266e-06,
+      "loss": 2.1251,
+      "step": 13210
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.95332773519357e-06,
+      "loss": 1.8011,
+      "step": 13220
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.920452007696864e-06,
+      "loss": 2.0219,
+      "step": 13230
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.887623637459297e-06,
+      "loss": 1.9489,
+      "step": 13240
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.854842721166501e-06,
+      "loss": 1.7814,
+      "step": 13250
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.822109355364394e-06,
+      "loss": 1.8679,
+      "step": 13260
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.789423636458804e-06,
+      "loss": 1.9127,
+      "step": 13270
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.756785660715272e-06,
+      "loss": 1.8789,
+      "step": 13280
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.724195524258688e-06,
+      "loss": 1.8896,
+      "step": 13290
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.691653323073082e-06,
+      "loss": 1.7484,
+      "step": 13300
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.659159153001267e-06,
+      "loss": 1.8104,
+      "step": 13310
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.626713109744642e-06,
+      "loss": 1.9525,
+      "step": 13320
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.594315288862814e-06,
+      "loss": 1.8442,
+      "step": 13330
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.561965785773413e-06,
+      "loss": 1.8327,
+      "step": 13340
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.529664695751722e-06,
+      "loss": 1.9843,
+      "step": 13350
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.497412113930472e-06,
+      "loss": 1.8873,
+      "step": 13360
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.465208135299502e-06,
+      "loss": 1.9601,
+      "step": 13370
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.433052854705534e-06,
+      "loss": 1.7818,
+      "step": 13380
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.400946366851833e-06,
+      "loss": 1.9288,
+      "step": 13390
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.36888876629799e-06,
+      "loss": 1.9425,
+      "step": 13400
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.336880147459586e-06,
+      "loss": 1.9015,
+      "step": 13410
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.304920604607975e-06,
+      "loss": 1.9082,
+      "step": 13420
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.273010231869932e-06,
+      "loss": 1.8731,
+      "step": 13430
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.24114912322746e-06,
+      "loss": 1.9551,
+      "step": 13440
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.20933737251743e-06,
+      "loss": 1.8492,
+      "step": 13450
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.177575073431381e-06,
+      "loss": 1.8778,
+      "step": 13460
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.145862319515177e-06,
+      "loss": 1.8834,
+      "step": 13470
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.114199204168783e-06,
+      "loss": 1.7894,
+      "step": 13480
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.082585820645965e-06,
+      "loss": 1.9025,
+      "step": 13490
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.051022262054006e-06,
+      "loss": 1.8592,
+      "step": 13500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.019508621353472e-06,
+      "loss": 1.9892,
+      "step": 13510
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.988044991357877e-06,
+      "loss": 1.8222,
+      "step": 13520
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.956631464733477e-06,
+      "loss": 1.9866,
+      "step": 13530
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.925268133998936e-06,
+      "loss": 2.0084,
+      "step": 13540
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.893955091525115e-06,
+      "loss": 2.0195,
+      "step": 13550
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.862692429534727e-06,
+      "loss": 1.9685,
+      "step": 13560
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.831480240102141e-06,
+      "loss": 1.9169,
+      "step": 13570
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.800318615153046e-06,
+      "loss": 1.8868,
+      "step": 13580
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.769207646464235e-06,
+      "loss": 1.8398,
+      "step": 13590
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.738147425663287e-06,
+      "loss": 1.9281,
+      "step": 13600
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.707138044228337e-06,
+      "loss": 1.9223,
+      "step": 13610
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.676179593487767e-06,
+      "loss": 1.7944,
+      "step": 13620
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.64527216461999e-06,
+      "loss": 1.9501,
+      "step": 13630
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.614415848653114e-06,
+      "loss": 1.9483,
+      "step": 13640
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.583610736464744e-06,
+      "loss": 2.017,
+      "step": 13650
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.552856918781648e-06,
+      "loss": 1.9811,
+      "step": 13660
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.522154486179553e-06,
+      "loss": 1.9875,
+      "step": 13670
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.491503529082819e-06,
+      "loss": 1.9164,
+      "step": 13680
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.46090413776423e-06,
+      "loss": 1.8927,
+      "step": 13690
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.430356402344665e-06,
+      "loss": 1.9834,
+      "step": 13700
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.3998604127929084e-06,
+      "loss": 1.9289,
+      "step": 13710
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.369416258925299e-06,
+      "loss": 2.0492,
+      "step": 13720
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.339024030405547e-06,
+      "loss": 1.9417,
+      "step": 13730
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.308683816744408e-06,
+      "loss": 1.8852,
+      "step": 13740
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.278395707299462e-06,
+      "loss": 1.9363,
+      "step": 13750
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.248159791274814e-06,
+      "loss": 1.7293,
+      "step": 13760
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.217976157720871e-06,
+      "loss": 1.8648,
+      "step": 13770
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.187844895534035e-06,
+      "loss": 1.9317,
+      "step": 13780
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.157766093456489e-06,
+      "loss": 1.8735,
+      "step": 13790
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.127739840075884e-06,
+      "loss": 1.8825,
+      "step": 13800
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.097766223825136e-06,
+      "loss": 1.9425,
+      "step": 13810
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.067845332982103e-06,
+      "loss": 1.846,
+      "step": 13820
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.037977255669393e-06,
+      "loss": 1.9604,
+      "step": 13830
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.008162079854025e-06,
+      "loss": 1.8816,
+      "step": 13840
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.978399893347257e-06,
+      "loss": 1.7719,
+      "step": 13850
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.948690783804249e-06,
+      "loss": 1.9361,
+      "step": 13860
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.919034838723862e-06,
+      "loss": 1.9281,
+      "step": 13870
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.8894321454483605e-06,
+      "loss": 1.9012,
+      "step": 13880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.859882791163194e-06,
+      "loss": 1.8314,
+      "step": 13890
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.830386862896693e-06,
+      "loss": 1.9059,
+      "step": 13900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.8009444475198635e-06,
+      "loss": 1.8494,
+      "step": 13910
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.771555631746082e-06,
+      "loss": 1.9318,
+      "step": 13920
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.742220502130889e-06,
+      "loss": 1.8882,
+      "step": 13930
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.712939145071684e-06,
+      "loss": 1.8446,
+      "step": 13940
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.68371164680752e-06,
+      "loss": 1.9169,
+      "step": 13950
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.654538093418805e-06,
+      "loss": 1.9489,
+      "step": 13960
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.625418570827091e-06,
+      "loss": 1.7729,
+      "step": 13970
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.596353164794772e-06,
+      "loss": 2.0623,
+      "step": 13980
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.567341960924892e-06,
+      "loss": 1.9759,
+      "step": 13990
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.5383850446608215e-06,
+      "loss": 1.9727,
+      "step": 14000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.509482501286085e-06,
+      "loss": 2.0384,
+      "step": 14010
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.480634415924028e-06,
+      "loss": 1.9534,
+      "step": 14020
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.451840873537635e-06,
+      "loss": 1.8118,
+      "step": 14030
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.423101958929254e-06,
+      "loss": 1.9228,
+      "step": 14040
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.394417756740312e-06,
+      "loss": 1.9676,
+      "step": 14050
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.365788351451141e-06,
+      "loss": 1.8344,
+      "step": 14060
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.337213827380642e-06,
+      "loss": 2.0527,
+      "step": 14070
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.3086942686861215e-06,
+      "loss": 1.9575,
+      "step": 14080
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.2802297593629685e-06,
+      "loss": 1.7797,
+      "step": 14090
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.25182038324447e-06,
+      "loss": 1.9235,
+      "step": 14100
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.223466224001506e-06,
+      "loss": 2.003,
+      "step": 14110
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.1951673651423656e-06,
+      "loss": 1.8902,
+      "step": 14120
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.1669238900124355e-06,
+      "loss": 1.9974,
+      "step": 14130
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.1387358817940165e-06,
+      "loss": 1.9061,
+      "step": 14140
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.11060342350602e-06,
+      "loss": 1.8801,
+      "step": 14150
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.085331774720657e-06,
+      "loss": 1.9623,
+      "step": 14160
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.057305089431472e-06,
+      "loss": 1.8564,
+      "step": 14170
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.029334193901554e-06,
+      "loss": 2.0141,
+      "step": 14180
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.001419170510373e-06,
+      "loss": 1.8377,
+      "step": 14190
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.9735601014728806e-06,
+      "loss": 1.8277,
+      "step": 14200
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.9457570688392026e-06,
+      "loss": 1.8967,
+      "step": 14210
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.918010154494455e-06,
+      "loss": 1.9961,
+      "step": 14220
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.890319440158443e-06,
+      "loss": 1.9461,
+      "step": 14230
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.862685007385487e-06,
+      "loss": 1.7879,
+      "step": 14240
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.835106937564119e-06,
+      "loss": 1.9553,
+      "step": 14250
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.8075853119168935e-06,
+      "loss": 1.9797,
+      "step": 14260
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.780120211500103e-06,
+      "loss": 1.9029,
+      "step": 14270
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.752711717203588e-06,
+      "loss": 1.8274,
+      "step": 14280
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.725359909750452e-06,
+      "loss": 1.8903,
+      "step": 14290
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.698064869696865e-06,
+      "loss": 1.7643,
+      "step": 14300
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.670826677431781e-06,
+      "loss": 1.8665,
+      "step": 14310
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.64364541317676e-06,
+      "loss": 1.7783,
+      "step": 14320
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.616521156985663e-06,
+      "loss": 1.8444,
+      "step": 14330
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.5894539887444835e-06,
+      "loss": 1.9887,
+      "step": 14340
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.562443988171051e-06,
+      "loss": 1.9218,
+      "step": 14350
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.535491234814855e-06,
+      "loss": 1.9454,
+      "step": 14360
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.508595808056752e-06,
+      "loss": 1.9652,
+      "step": 14370
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.48175778710879e-06,
+      "loss": 1.8258,
+      "step": 14380
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.454977251013918e-06,
+      "loss": 1.9759,
+      "step": 14390
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.428254278645808e-06,
+      "loss": 1.8744,
+      "step": 14400
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.4015889487085695e-06,
+      "loss": 1.919,
+      "step": 14410
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.374981339736571e-06,
+      "loss": 2.079,
+      "step": 14420
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.348431530094158e-06,
+      "loss": 1.9218,
+      "step": 14430
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.32193959797547e-06,
+      "loss": 1.885,
+      "step": 14440
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.2955056214041596e-06,
+      "loss": 1.9626,
+      "step": 14450
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.269129678233217e-06,
+      "loss": 1.7693,
+      "step": 14460
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.242811846144688e-06,
+      "loss": 1.8157,
+      "step": 14470
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.216552202649494e-06,
+      "loss": 2.0238,
+      "step": 14480
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.190350825087148e-06,
+      "loss": 1.9121,
+      "step": 14490
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.1642077906256085e-06,
+      "loss": 1.8987,
+      "step": 14500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.138123176260951e-06,
+      "loss": 1.91,
+      "step": 14510
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.112097058817233e-06,
+      "loss": 1.755,
+      "step": 14520
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.086129514946195e-06,
+      "loss": 1.7913,
+      "step": 14530
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.060220621127093e-06,
+      "loss": 1.9376,
+      "step": 14540
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.0343704536664285e-06,
+      "loss": 1.8013,
+      "step": 14550
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.0085790886977656e-06,
+      "loss": 1.9334,
+      "step": 14560
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.982846602181454e-06,
+      "loss": 1.8958,
+      "step": 14570
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.95717306990447e-06,
+      "loss": 1.8037,
+      "step": 14580
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.931558567480122e-06,
+      "loss": 1.8731,
+      "step": 14590
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.906003170347897e-06,
+      "loss": 1.9277,
+      "step": 14600
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.880506953773181e-06,
+      "loss": 1.9439,
+      "step": 14610
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.855069992847089e-06,
+      "loss": 1.8471,
+      "step": 14620
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.829692362486182e-06,
+      "loss": 1.8144,
+      "step": 14630
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.804374137432319e-06,
+      "loss": 1.8278,
+      "step": 14640
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.779115392252365e-06,
+      "loss": 1.753,
+      "step": 14650
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.753916201338035e-06,
+      "loss": 1.9973,
+      "step": 14660
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.728776638905622e-06,
+      "loss": 1.929,
+      "step": 14670
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.703696778995826e-06,
+      "loss": 1.9177,
+      "step": 14680
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.67867669547348e-06,
+      "loss": 2.0693,
+      "step": 14690
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.653716462027402e-06,
+      "loss": 1.9268,
+      "step": 14700
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.628816152170104e-06,
+      "loss": 1.963,
+      "step": 14710
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.603975839237648e-06,
+      "loss": 1.8768,
+      "step": 14720
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.579195596389355e-06,
+      "loss": 1.9078,
+      "step": 14730
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.554475496607666e-06,
+      "loss": 1.9057,
+      "step": 14740
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.5298156126978574e-06,
+      "loss": 1.8076,
+      "step": 14750
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.505216017287886e-06,
+      "loss": 2.0492,
+      "step": 14760
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.480676782828122e-06,
+      "loss": 1.8181,
+      "step": 14770
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.45619798159119e-06,
+      "loss": 1.7664,
+      "step": 14780
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.431779685671697e-06,
+      "loss": 1.7688,
+      "step": 14790
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.407421966986077e-06,
+      "loss": 1.9146,
+      "step": 14800
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.383124897272331e-06,
+      "loss": 1.8565,
+      "step": 14810
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.3588885480898595e-06,
+      "loss": 1.8486,
+      "step": 14820
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.334712990819206e-06,
+      "loss": 1.944,
+      "step": 14830
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.310598296661894e-06,
+      "loss": 1.8339,
+      "step": 14840
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.286544536640169e-06,
+      "loss": 1.9796,
+      "step": 14850
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.2625517815968355e-06,
+      "loss": 1.9414,
+      "step": 14860
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.238620102195026e-06,
+      "loss": 1.9255,
+      "step": 14870
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.214749568917972e-06,
+      "loss": 1.8358,
+      "step": 14880
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.1909402520688476e-06,
+      "loss": 1.8538,
+      "step": 14890
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.167192221770508e-06,
+      "loss": 1.9295,
+      "step": 14900
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.143505547965326e-06,
+      "loss": 1.8649,
+      "step": 14910
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.119880300414952e-06,
+      "loss": 1.7826,
+      "step": 14920
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.096316548700143e-06,
+      "loss": 1.9212,
+      "step": 14930
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.072814362220523e-06,
+      "loss": 1.8635,
+      "step": 14940
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.0493738101944095e-06,
+      "loss": 1.8372,
+      "step": 14950
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.025994961658577e-06,
+      "loss": 1.8654,
+      "step": 14960
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.002677885468095e-06,
+      "loss": 2.0274,
+      "step": 14970
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.979422650296075e-06,
+      "loss": 2.0229,
+      "step": 14980
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.956229324633523e-06,
+      "loss": 1.8554,
+      "step": 14990
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.933097976789082e-06,
+      "loss": 1.8497,
+      "step": 15000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.910028674888891e-06,
+      "loss": 1.934,
+      "step": 15010
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.887021486876314e-06,
+      "loss": 1.7985,
+      "step": 15020
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8640764805118165e-06,
+      "loss": 2.0005,
+      "step": 15030
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.841193723372694e-06,
+      "loss": 2.1335,
+      "step": 15040
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.818373282852936e-06,
+      "loss": 1.9744,
+      "step": 15050
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7956152261629684e-06,
+      "loss": 1.941,
+      "step": 15060
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.772919620329518e-06,
+      "loss": 1.7884,
+      "step": 15070
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7502865321953463e-06,
+      "loss": 1.7926,
+      "step": 15080
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7277160284191234e-06,
+      "loss": 1.7868,
+      "step": 15090
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7052081754751667e-06,
+      "loss": 1.9935,
+      "step": 15100
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.6827630396533004e-06,
+      "loss": 1.9245,
+      "step": 15110
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.6603806870586147e-06,
+      "loss": 1.7544,
+      "step": 15120
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.6380611836113078e-06,
+      "loss": 1.9619,
+      "step": 15130
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.615804595046457e-06,
+      "loss": 1.8413,
+      "step": 15140
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5936109869138702e-06,
+      "loss": 1.8472,
+      "step": 15150
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.571480424577836e-06,
+      "loss": 1.9637,
+      "step": 15160
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5494129732169868e-06,
+      "loss": 1.8977,
+      "step": 15170
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.527408697824061e-06,
+      "loss": 1.9526,
+      "step": 15180
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5054676632057556e-06,
+      "loss": 1.8644,
+      "step": 15190
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4835899339824784e-06,
+      "loss": 1.9649,
+      "step": 15200
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4617755745882302e-06,
+      "loss": 1.9672,
+      "step": 15210
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.440024649270343e-06,
+      "loss": 1.906,
+      "step": 15220
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.418337222089346e-06,
+      "loss": 1.9428,
+      "step": 15230
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.396713356918735e-06,
+      "loss": 1.8653,
+      "step": 15240
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3751531174448236e-06,
+      "loss": 2.0333,
+      "step": 15250
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.353656567166519e-06,
+      "loss": 1.8261,
+      "step": 15260
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3322237693951707e-06,
+      "loss": 1.905,
+      "step": 15270
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.310854787254339e-06,
+      "loss": 1.9595,
+      "step": 15280
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.289549683679663e-06,
+      "loss": 1.8179,
+      "step": 15290
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.2683085214186253e-06,
+      "loss": 1.8491,
+      "step": 15300
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.247131363030409e-06,
+      "loss": 1.8802,
+      "step": 15310
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.226018270885675e-06,
+      "loss": 1.9408,
+      "step": 15320
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.204969307166411e-06,
+      "loss": 1.9344,
+      "step": 15330
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.183984533865725e-06,
+      "loss": 1.9085,
+      "step": 15340
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1630640127876836e-06,
+      "loss": 1.993,
+      "step": 15350
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1422078055471037e-06,
+      "loss": 1.7518,
+      "step": 15360
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1214159735694028e-06,
+      "loss": 1.8783,
+      "step": 15370
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.1006885780903853e-06,
+      "loss": 1.6631,
+      "step": 15380
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.0800256801560926e-06,
+      "loss": 2.0445,
+      "step": 15390
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.059427340622592e-06,
+      "loss": 1.9366,
+      "step": 15400
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.0388936201558303e-06,
+      "loss": 2.0305,
+      "step": 15410
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.018424579231441e-06,
+      "loss": 1.8622,
+      "step": 15420
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.998020278134539e-06,
+      "loss": 1.846,
+      "step": 15430
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.9776807769596015e-06,
+      "loss": 1.8682,
+      "step": 15440
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.9574061356102273e-06,
+      "loss": 1.9536,
+      "step": 15450
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.9371964137990206e-06,
+      "loss": 1.8773,
+      "step": 15460
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.9170516710473577e-06,
+      "loss": 1.8857,
+      "step": 15470
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.8969719666852633e-06,
+      "loss": 1.6682,
+      "step": 15480
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.8769573598511874e-06,
+      "loss": 1.8611,
+      "step": 15490
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.857007909491885e-06,
+      "loss": 1.8346,
+      "step": 15500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.8371236743621824e-06,
+      "loss": 1.9589,
+      "step": 15510
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.8173047130248652e-06,
+      "loss": 1.8968,
+      "step": 15520
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.797551083850444e-06,
+      "loss": 1.9241,
+      "step": 15530
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.7778628450170456e-06,
+      "loss": 1.9223,
+      "step": 15540
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.7582400545101806e-06,
+      "loss": 2.018,
+      "step": 15550
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.73868277012263e-06,
+      "loss": 1.9078,
+      "step": 15560
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.71919104945422e-06,
+      "loss": 1.9974,
+      "step": 15570
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.6997649499117066e-06,
+      "loss": 1.8279,
+      "step": 15580
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.680404528708552e-06,
+      "loss": 1.8841,
+      "step": 15590
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.6611098428648064e-06,
+      "loss": 1.9593,
+      "step": 15600
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.641880949206904e-06,
+      "loss": 1.8068,
+      "step": 15610
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.6227179043675137e-06,
+      "loss": 1.7997,
+      "step": 15620
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.6036207647853637e-06,
+      "loss": 1.8606,
+      "step": 15630
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5845895867050883e-06,
+      "loss": 1.8571,
+      "step": 15640
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.565624426177038e-06,
+      "loss": 1.9085,
+      "step": 15650
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.546725339057146e-06,
+      "loss": 1.8827,
+      "step": 15660
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5278923810067273e-06,
+      "loss": 1.9845,
+      "step": 15670
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5091256074923575e-06,
+      "loss": 1.9278,
+      "step": 15680
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4904250737856626e-06,
+      "loss": 1.8214,
+      "step": 15690
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.471790834963206e-06,
+      "loss": 1.9502,
+      "step": 15700
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.453222945906275e-06,
+      "loss": 1.7207,
+      "step": 15710
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4347214613007623e-06,
+      "loss": 1.8848,
+      "step": 15720
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.416286435636975e-06,
+      "loss": 2.1009,
+      "step": 15730
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3979179232095e-06,
+      "loss": 1.9554,
+      "step": 15740
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.379615978117014e-06,
+      "loss": 1.88,
+      "step": 15750
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.361380654262155e-06,
+      "loss": 1.9082,
+      "step": 15760
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.343212005351339e-06,
+      "loss": 1.9052,
+      "step": 15770
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.325110084894619e-06,
+      "loss": 1.8896,
+      "step": 15780
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.307074946205512e-06,
+      "loss": 1.9033,
+      "step": 15790
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.289106642400865e-06,
+      "loss": 1.9454,
+      "step": 15800
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.271205226400663e-06,
+      "loss": 2.0155,
+      "step": 15810
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.253370750927916e-06,
+      "loss": 2.0,
+      "step": 15820
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2356032685084597e-06,
+      "loss": 1.8817,
+      "step": 15830
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.2179028314708446e-06,
+      "loss": 1.9762,
+      "step": 15840
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.200269491946141e-06,
+      "loss": 1.9036,
+      "step": 15850
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.1827033018678194e-06,
+      "loss": 1.8031,
+      "step": 15860
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.1652043129715692e-06,
+      "loss": 1.9264,
+      "step": 15870
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.147772576795179e-06,
+      "loss": 1.9577,
+      "step": 15880
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.130408144678339e-06,
+      "loss": 1.8868,
+      "step": 15890
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.1131110677625437e-06,
+      "loss": 1.8024,
+      "step": 15900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.0958813969908896e-06,
+      "loss": 1.8309,
+      "step": 15910
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.0787191831079716e-06,
+      "loss": 1.9402,
+      "step": 15920
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.061624476659693e-06,
+      "loss": 1.9716,
+      "step": 15930
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.044597327993153e-06,
+      "loss": 1.8549,
+      "step": 15940
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.0276377872564615e-06,
+      "loss": 1.9476,
+      "step": 15950
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.0107459043986228e-06,
+      "loss": 2.0257,
+      "step": 15960
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.993921729169376e-06,
+      "loss": 1.832,
+      "step": 15970
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.977165311119042e-06,
+      "loss": 1.9528,
+      "step": 15980
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9604766995983957e-06,
+      "loss": 1.8408,
+      "step": 15990
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9438559437584906e-06,
+      "loss": 1.8888,
+      "step": 16000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.927303092550553e-06,
+      "loss": 1.8368,
+      "step": 16010
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9108181947258e-06,
+      "loss": 1.9077,
+      "step": 16020
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.89440129883533e-06,
+      "loss": 1.8759,
+      "step": 16030
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8780524532299447e-06,
+      "loss": 1.8338,
+      "step": 16040
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8617717060600449e-06,
+      "loss": 2.0187,
+      "step": 16050
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.845559105275449e-06,
+      "loss": 1.8527,
+      "step": 16060
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8294146986252924e-06,
+      "loss": 1.8925,
+      "step": 16070
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8133385336578435e-06,
+      "loss": 1.8209,
+      "step": 16080
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7973306577204058e-06,
+      "loss": 1.8469,
+      "step": 16090
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.781391117959147e-06,
+      "loss": 1.7746,
+      "step": 16100
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7655199613189788e-06,
+      "loss": 1.8485,
+      "step": 16110
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7497172345434021e-06,
+      "loss": 1.8458,
+      "step": 16120
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.733982984174398e-06,
+      "loss": 1.8836,
+      "step": 16130
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7183172565522488e-06,
+      "loss": 1.7653,
+      "step": 16140
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.7027200978154439e-06,
+      "loss": 1.8755,
+      "step": 16150
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.687191553900508e-06,
+      "loss": 1.9445,
+      "step": 16160
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6732745678534722e-06,
+      "loss": 2.0124,
+      "step": 16170
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6578765179313727e-06,
+      "loss": 1.8397,
+      "step": 16180
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6425472149038361e-06,
+      "loss": 1.8247,
+      "step": 16190
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6272867039184986e-06,
+      "loss": 1.8082,
+      "step": 16200
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.612095029920413e-06,
+      "loss": 1.9342,
+      "step": 16210
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5969722376518726e-06,
+      "loss": 1.8988,
+      "step": 16220
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5819183716523245e-06,
+      "loss": 1.9012,
+      "step": 16230
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5669334762581977e-06,
+      "loss": 1.7213,
+      "step": 16240
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5520175956027977e-06,
+      "loss": 1.9705,
+      "step": 16250
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5371707736161644e-06,
+      "loss": 1.9695,
+      "step": 16260
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5223930540249536e-06,
+      "loss": 1.8941,
+      "step": 16270
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5076844803522922e-06,
+      "loss": 2.025,
+      "step": 16280
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4930450959176723e-06,
+      "loss": 1.9247,
+      "step": 16290
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.478474943836794e-06,
+      "loss": 1.9061,
+      "step": 16300
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4639740670214758e-06,
+      "loss": 1.9744,
+      "step": 16310
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4495425081794883e-06,
+      "loss": 1.8611,
+      "step": 16320
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.435180309814463e-06,
+      "loss": 1.9599,
+      "step": 16330
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4208875142257356e-06,
+      "loss": 1.9711,
+      "step": 16340
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4066641635082561e-06,
+      "loss": 1.9774,
+      "step": 16350
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3925102995524298e-06,
+      "loss": 1.8972,
+      "step": 16360
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3784259640440279e-06,
+      "loss": 1.8353,
+      "step": 16370
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3644111984640245e-06,
+      "loss": 1.8143,
+      "step": 16380
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3504660440885248e-06,
+      "loss": 1.8271,
+      "step": 16390
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3365905419885888e-06,
+      "loss": 1.9049,
+      "step": 16400
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3227847330301608e-06,
+      "loss": 1.8046,
+      "step": 16410
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3090486578739098e-06,
+      "loss": 1.8512,
+      "step": 16420
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.295382356975139e-06,
+      "loss": 1.9666,
+      "step": 16430
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.281785870583635e-06,
+      "loss": 1.963,
+      "step": 16440
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2682592387435881e-06,
+      "loss": 1.9229,
+      "step": 16450
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2548025012934367e-06,
+      "loss": 1.8743,
+      "step": 16460
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2414156978657833e-06,
+      "loss": 1.899,
+      "step": 16470
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2280988678872452e-06,
+      "loss": 1.8706,
+      "step": 16480
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2148520505783718e-06,
+      "loss": 1.8438,
+      "step": 16490
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2016752849534963e-06,
+      "loss": 1.9913,
+      "step": 16500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1885686098206478e-06,
+      "loss": 2.068,
+      "step": 16510
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1755320637814205e-06,
+      "loss": 1.8634,
+      "step": 16520
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.162565685230868e-06,
+      "loss": 1.8784,
+      "step": 16530
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1496695123573847e-06,
+      "loss": 1.9008,
+      "step": 16540
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1368435831426021e-06,
+      "loss": 1.9433,
+      "step": 16550
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.124087935361262e-06,
+      "loss": 2.0302,
+      "step": 16560
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.1114026065811218e-06,
+      "loss": 1.8227,
+      "step": 16570
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0987876341628317e-06,
+      "loss": 1.9816,
+      "step": 16580
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0862430552598363e-06,
+      "loss": 1.9166,
+      "step": 16590
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0737689068182455e-06,
+      "loss": 1.9659,
+      "step": 16600
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0613652255767548e-06,
+      "loss": 2.019,
+      "step": 16610
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.049032048066506e-06,
+      "loss": 1.9461,
+      "step": 16620
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0367694106110049e-06,
+      "loss": 1.9227,
+      "step": 16630
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0245773493259946e-06,
+      "loss": 1.6706,
+      "step": 16640
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.012455900119369e-06,
+      "loss": 1.9249,
+      "step": 16650
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.000405098691043e-06,
+      "loss": 1.8078,
+      "step": 16660
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.884249805328732e-07,
+      "loss": 1.9171,
+      "step": 16670
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.76515580928533e-07,
+      "loss": 1.8064,
+      "step": 16680
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.646769349534207e-07,
+      "loss": 1.7982,
+      "step": 16690
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.52909077474548e-07,
+      "loss": 1.9003,
+      "step": 16700
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.412120431504495e-07,
+      "loss": 1.9546,
+      "step": 16710
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.295858664310569e-07,
+      "loss": 1.9828,
+      "step": 16720
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.180305815576301e-07,
+      "loss": 1.8911,
+      "step": 16730
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.065462225626292e-07,
+      "loss": 1.9123,
+      "step": 16740
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.951328232696288e-07,
+      "loss": 1.8117,
+      "step": 16750
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.837904172932093e-07,
+      "loss": 1.7962,
+      "step": 16760
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.725190380388687e-07,
+      "loss": 1.9551,
+      "step": 16770
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.613187187029137e-07,
+      "loss": 2.0146,
+      "step": 16780
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.501894922723686e-07,
+      "loss": 2.031,
+      "step": 16790
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.39131391524875e-07,
+      "loss": 1.9982,
+      "step": 16800
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.281444490285923e-07,
+      "loss": 1.9703,
+      "step": 16810
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.172286971421167e-07,
+      "loss": 1.9458,
+      "step": 16820
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.063841680143597e-07,
+      "loss": 1.9845,
+      "step": 16830
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.956108935844836e-07,
+      "loss": 1.96,
+      "step": 16840
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.849089055817771e-07,
+      "loss": 1.9771,
+      "step": 16850
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.74278235525594e-07,
+      "loss": 1.9081,
+      "step": 16860
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.637189147252255e-07,
+      "loss": 1.9258,
+      "step": 16870
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.532309742798422e-07,
+      "loss": 2.0137,
+      "step": 16880
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.428144450783742e-07,
+      "loss": 1.9445,
+      "step": 16890
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.324693577994424e-07,
+      "loss": 1.7126,
+      "step": 16900
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.221957429112469e-07,
+      "loss": 1.9293,
+      "step": 16910
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.119936306714981e-07,
+      "loss": 1.9187,
+      "step": 16920
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.018630511273106e-07,
+      "loss": 1.8097,
+      "step": 16930
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.918040341151316e-07,
+      "loss": 1.873,
+      "step": 16940
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.81816609260627e-07,
+      "loss": 1.8917,
+      "step": 16950
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.719008059786286e-07,
+      "loss": 1.9634,
+      "step": 16960
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.620566534730172e-07,
+      "loss": 1.953,
+      "step": 16970
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.522841807366564e-07,
+      "loss": 1.9529,
+      "step": 16980
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.425834165512901e-07,
+      "loss": 2.0853,
+      "step": 16990
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.329543894874779e-07,
+      "loss": 2.014,
+      "step": 17000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.233971279044904e-07,
+      "loss": 1.9084,
+      "step": 17010
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.139116599502448e-07,
+      "loss": 2.0178,
+      "step": 17020
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.044980135612055e-07,
+      "loss": 2.0759,
+      "step": 17030
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.951562164623142e-07,
+      "loss": 1.843,
+      "step": 17040
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.858862961668988e-07,
+      "loss": 1.9735,
+      "step": 17050
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.766882799765977e-07,
+      "loss": 1.9039,
+      "step": 17060
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.675621949812804e-07,
+      "loss": 1.9123,
+      "step": 17070
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.585080680589661e-07,
+      "loss": 1.962,
+      "step": 17080
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.49525925875738e-07,
+      "loss": 1.9358,
+      "step": 17090
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.406157948856793e-07,
+      "loss": 2.023,
+      "step": 17100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.317777013307795e-07,
+      "loss": 2.1128,
+      "step": 17110
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.230116712408695e-07,
+      "loss": 1.8367,
+      "step": 17120
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.143177304335389e-07,
+      "loss": 1.9262,
+      "step": 17130
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.056959045140614e-07,
+      "loss": 1.9201,
+      "step": 17140
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.971462188753106e-07,
+      "loss": 1.9317,
+      "step": 17150
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.886686986977112e-07,
+      "loss": 1.9448,
+      "step": 17160
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.802633689491271e-07,
+      "loss": 1.9531,
+      "step": 17170
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.719302543848225e-07,
+      "loss": 1.8816,
+      "step": 17180
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.6366937954736846e-07,
+      "loss": 2.0139,
+      "step": 17190
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.5548076876658073e-07,
+      "loss": 1.9213,
+      "step": 17200
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.473644461594345e-07,
+      "loss": 1.8365,
+      "step": 17210
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.3932043563001404e-07,
+      "loss": 1.9953,
+      "step": 17220
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.313487608694239e-07,
+      "loss": 1.9761,
+      "step": 17230
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.2344944535573086e-07,
+      "loss": 1.832,
+      "step": 17240
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.1562251235388594e-07,
+      "loss": 1.843,
+      "step": 17250
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.0786798491566634e-07,
+      "loss": 1.8319,
+      "step": 17260
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.001858858795893e-07,
+      "loss": 1.9392,
+      "step": 17270
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.925762378708758e-07,
+      "loss": 1.8896,
+      "step": 17280
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.8503906330134563e-07,
+      "loss": 1.8744,
+      "step": 17290
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.775743843693863e-07,
+      "loss": 1.7374,
+      "step": 17300
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.701822230598562e-07,
+      "loss": 1.8881,
+      "step": 17310
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6286260114405666e-07,
+      "loss": 1.9391,
+      "step": 17320
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5561554017962686e-07,
+      "loss": 1.7762,
+      "step": 17330
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.484410615105127e-07,
+      "loss": 1.7973,
+      "step": 17340
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.4133918626688687e-07,
+      "loss": 1.8061,
+      "step": 17350
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.350095917681956e-07,
+      "loss": 1.8806,
+      "step": 17360
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.2804572048074357e-07,
+      "loss": 1.8364,
+      "step": 17370
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.211545126868526e-07,
+      "loss": 1.879,
+      "step": 17380
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.143359886824127e-07,
+      "loss": 1.9531,
+      "step": 17390
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.075901685492405e-07,
+      "loss": 1.889,
+      "step": 17400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.0091707215503804e-07,
+      "loss": 1.9469,
+      "step": 17410
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.9431671915330917e-07,
+      "loss": 1.7712,
+      "step": 17420
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.8778912898332343e-07,
+      "loss": 1.9142,
+      "step": 17430
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.813343208700442e-07,
+      "loss": 1.9511,
+      "step": 17440
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.749523138240867e-07,
+      "loss": 1.9044,
+      "step": 17450
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.6864312664164616e-07,
+      "loss": 1.8736,
+      "step": 17460
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.624067779044559e-07,
+      "loss": 2.0145,
+      "step": 17470
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.5624328597971534e-07,
+      "loss": 2.0157,
+      "step": 17480
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.5015266902005953e-07,
+      "loss": 1.9218,
+      "step": 17490
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.441349449634867e-07,
+      "loss": 1.942,
+      "step": 17500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3819013153331404e-07,
+      "loss": 2.103,
+      "step": 17510
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3231824623812237e-07,
+      "loss": 1.8327,
+      "step": 17520
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2651930637170581e-07,
+      "loss": 1.9656,
+      "step": 17530
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.207933290130193e-07,
+      "loss": 2.0078,
+      "step": 17540
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1514033102613408e-07,
+      "loss": 1.6987,
+      "step": 17550
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0956032906017664e-07,
+      "loss": 1.797,
+      "step": 17560
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0405333954929273e-07,
+      "loss": 1.8454,
+      "step": 17570
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9861937871259173e-07,
+      "loss": 1.8936,
+      "step": 17580
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9325846255409674e-07,
+      "loss": 1.9581,
+      "step": 17590
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8797060686270574e-07,
+      "loss": 2.033,
+      "step": 17600
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8275582721213324e-07,
+      "loss": 1.9129,
+      "step": 17610
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.776141389608743e-07,
+      "loss": 1.9413,
+      "step": 17620
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7254555725216282e-07,
+      "loss": 1.8104,
+      "step": 17630
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6755009701391045e-07,
+      "loss": 1.8079,
+      "step": 17640
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6262777295867892e-07,
+      "loss": 1.8694,
+      "step": 17650
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5777859958362174e-07,
+      "loss": 1.87,
+      "step": 17660
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5300259117046467e-07,
+      "loss": 1.9871,
+      "step": 17670
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4829976178543648e-07,
+      "loss": 1.8625,
+      "step": 17680
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4367012527924383e-07,
+      "loss": 1.8871,
+      "step": 17690
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3911369528702978e-07,
+      "loss": 1.9717,
+      "step": 17700
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3463048522832643e-07,
+      "loss": 2.0892,
+      "step": 17710
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3022050830701903e-07,
+      "loss": 1.8019,
+      "step": 17720
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.2588377751131253e-07,
+      "loss": 1.876,
+      "step": 17730
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.2162030561368442e-07,
+      "loss": 1.9194,
+      "step": 17740
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1743010517085428e-07,
+      "loss": 1.8412,
+      "step": 17750
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1331318852373929e-07,
+      "loss": 1.9027,
+      "step": 17760
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0926956779742648e-07,
+      "loss": 1.9806,
+      "step": 17770
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0529925490112835e-07,
+      "loss": 1.8933,
+      "step": 17780
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.0140226152815791e-07,
+      "loss": 1.8299,
+      "step": 17790
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.757859915588141e-08,
+      "loss": 2.0548,
+      "step": 17800
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.382827904569902e-08,
+      "loss": 2.0427,
+      "step": 17810
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.015131224300032e-08,
+      "loss": 2.0755,
+      "step": 17820
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.654770957713665e-08,
+      "loss": 1.9084,
+      "step": 17830
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.30174816613849e-08,
+      "loss": 1.8893,
+      "step": 17840
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.956063889292542e-08,
+      "loss": 2.0192,
+      "step": 17850
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.617719145280311e-08,
+      "loss": 1.8799,
+      "step": 17860
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.286714930589966e-08,
+      "loss": 2.0046,
+      "step": 17870
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.963052220090582e-08,
+      "loss": 1.7649,
+      "step": 17880
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.646731967028808e-08,
+      "loss": 1.8506,
+      "step": 17890
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.337755103026643e-08,
+      "loss": 1.8589,
+      "step": 17900
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.036122538078393e-08,
+      "loss": 1.881,
+      "step": 17910
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.7418351605484364e-08,
+      "loss": 1.7037,
+      "step": 17920
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.454893837167352e-08,
+      "loss": 1.8876,
+      "step": 17930
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.1752994130307984e-08,
+      "loss": 1.7534,
+      "step": 17940
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.9030527115964676e-08,
+      "loss": 1.7854,
+      "step": 17950
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.638154534681582e-08,
+      "loss": 1.9643,
+      "step": 17960
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.380605662460402e-08,
+      "loss": 1.9423,
+      "step": 17970
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.1304068534625536e-08,
+      "loss": 1.8708,
+      "step": 17980
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.8875588445694276e-08,
+      "loss": 1.9405,
+      "step": 17990
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.652062351014174e-08,
+      "loss": 1.9404,
+      "step": 18000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.423918066377263e-08,
+      "loss": 2.1083,
+      "step": 18010
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.2031266625864865e-08,
+      "loss": 1.901,
+      "step": 18020
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.9896887899133474e-08,
+      "loss": 1.9372,
+      "step": 18030
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.783605076972229e-08,
+      "loss": 1.9413,
+      "step": 18040
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.584876130717895e-08,
+      "loss": 1.9087,
+      "step": 18050
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.393502536444103e-08,
+      "loss": 1.9429,
+      "step": 18060
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.209484857781663e-08,
+      "loss": 1.9209,
+      "step": 18070
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0328236366967678e-08,
+      "loss": 1.9739,
+      "step": 18080
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8635193934893325e-08,
+      "loss": 1.8776,
+      "step": 18090
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.701572626792436e-08,
+      "loss": 1.9401,
+      "step": 18100
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5469838135692693e-08,
+      "loss": 1.8213,
+      "step": 18110
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3997534091123033e-08,
+      "loss": 1.9722,
+      "step": 18120
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2598818470424545e-08,
+      "loss": 1.8914,
+      "step": 18130
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1273695393079763e-08,
+      "loss": 2.1437,
+      "step": 18140
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.002216876181683e-08,
+      "loss": 2.1365,
+      "step": 18150
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.84424226261782e-09,
+      "loss": 1.8102,
+      "step": 18160
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.739919364696535e-09,
+      "loss": 1.8124,
+      "step": 18170
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.709203320484636e-09,
+      "loss": 1.9223,
+      "step": 18180
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.752097165637182e-09,
+      "loss": 1.8989,
+      "step": 18190
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.868603719007658e-09,
+      "loss": 2.0224,
+      "step": 18200
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.0587255826507465e-09,
+      "loss": 1.9957,
+      "step": 18210
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.3224651418028996e-09,
+      "loss": 1.9629,
+      "step": 18220
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.6598245648934426e-09,
+      "loss": 1.8888,
+      "step": 18230
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0708058035195932e-09,
+      "loss": 1.9362,
+      "step": 18240
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5554105924492357e-09,
+      "loss": 1.863,
+      "step": 18250
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1136404496236986e-09,
+      "loss": 1.8412,
+      "step": 18260
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.454966761299975e-10,
+      "loss": 1.8972,
+      "step": 18270
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.509803562258164e-10,
+      "loss": 1.7978,
+      "step": 18280
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.300923573145264e-10,
+      "loss": 1.8905,
+      "step": 18290
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.283332995351334e-11,
+      "loss": 1.8056,
+      "step": 18300
+    },
+    {
+      "epoch": 3.0,
+      "step": 18306,
+      "total_flos": 9.176450281611264e+17,
+      "train_loss": 1.9485092873158096,
+      "train_runtime": 27427.4704,
+      "train_samples_per_second": 5.34,
+      "train_steps_per_second": 0.667
+    }
+  ],
+  "max_steps": 18306,
+  "num_train_epochs": 3,
+  "total_flos": 9.176450281611264e+17,
+  "trial_name": null,
+  "trial_params": null
+}