diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,17016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9998763851485586,
+  "eval_steps": 500,
+  "global_step": 28312,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7058823529411767e-07,
+      "loss": 4.9244,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.411764705882353e-07,
+      "loss": 4.5069,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4117647058823531e-06,
+      "loss": 3.6782,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8823529411764707e-06,
+      "loss": 2.9642,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 2.5767,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8235294117647062e-06,
+      "loss": 2.401,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2941176470588236e-06,
+      "loss": 2.3275,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7647058823529414e-06,
+      "loss": 2.2191,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.235294117647059e-06,
+      "loss": 2.0439,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 1.7901,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.176470588235295e-06,
+      "loss": 1.4897,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6470588235294125e-06,
+      "loss": 1.251,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.11764705882353e-06,
+      "loss": 1.1331,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.588235294117647e-06,
+      "loss": 1.1103,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.058823529411766e-06,
+      "loss": 1.1048,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.529411764705883e-06,
+      "loss": 1.0945,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.0806,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.470588235294118e-06,
+      "loss": 1.0726,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.941176470588237e-06,
+      "loss": 1.0658,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 1.0578,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.882352941176472e-06,
+      "loss": 1.0535,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.035294117647059e-05,
+      "loss": 1.0402,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0823529411764706e-05,
+      "loss": 1.047,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1294117647058825e-05,
+      "loss": 1.0332,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 1.0275,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.223529411764706e-05,
+      "loss": 1.0338,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2705882352941177e-05,
+      "loss": 1.0286,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3176470588235294e-05,
+      "loss": 1.0248,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3647058823529413e-05,
+      "loss": 1.0303,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4117647058823532e-05,
+      "loss": 1.0269,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4588235294117647e-05,
+      "loss": 1.019,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5058823529411765e-05,
+      "loss": 1.0283,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5529411764705882e-05,
+      "loss": 1.0178,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.0279,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.647058823529412e-05,
+      "loss": 1.0125,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6941176470588237e-05,
+      "loss": 1.0162,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7411764705882353e-05,
+      "loss": 1.0113,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7882352941176474e-05,
+      "loss": 1.0219,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8352941176470587e-05,
+      "loss": 1.0163,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8823529411764708e-05,
+      "loss": 1.0142,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9294117647058825e-05,
+      "loss": 1.0063,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9764705882352945e-05,
+      "loss": 0.9995,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.023529411764706e-05,
+      "loss": 1.0093,
+      "step": 430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.070588235294118e-05,
+      "loss": 1.0074,
+      "step": 440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1176470588235296e-05,
+      "loss": 1.003,
+      "step": 450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1647058823529413e-05,
+      "loss": 1.0029,
+      "step": 460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.2117647058823533e-05,
+      "loss": 1.0042,
+      "step": 470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.258823529411765e-05,
+      "loss": 1.0121,
+      "step": 480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.3058823529411763e-05,
+      "loss": 1.0019,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.3529411764705884e-05,
+      "loss": 1.0064,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.4e-05,
+      "loss": 1.0074,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.447058823529412e-05,
+      "loss": 1.0118,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.4941176470588238e-05,
+      "loss": 0.9997,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.5411764705882355e-05,
+      "loss": 1.0053,
+      "step": 540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.5882352941176475e-05,
+      "loss": 0.9937,
+      "step": 550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.635294117647059e-05,
+      "loss": 1.0067,
+      "step": 560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.6823529411764706e-05,
+      "loss": 1.0023,
+      "step": 570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7294117647058826e-05,
+      "loss": 0.9994,
+      "step": 580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7764705882352943e-05,
+      "loss": 0.9861,
+      "step": 590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.8235294117647063e-05,
+      "loss": 0.999,
+      "step": 600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.870588235294118e-05,
+      "loss": 0.9987,
+      "step": 610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9176470588235294e-05,
+      "loss": 0.9934,
+      "step": 620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9647058823529414e-05,
+      "loss": 0.9884,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.011764705882353e-05,
+      "loss": 0.9996,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.0588235294117644e-05,
+      "loss": 0.9902,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.1058823529411765e-05,
+      "loss": 0.9965,
+      "step": 660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.1529411764705885e-05,
+      "loss": 0.9936,
+      "step": 670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.9821,
+      "step": 680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.247058823529412e-05,
+      "loss": 0.9889,
+      "step": 690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.294117647058824e-05,
+      "loss": 0.9944,
+      "step": 700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.341176470588236e-05,
+      "loss": 0.9821,
+      "step": 710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.388235294117647e-05,
+      "loss": 0.9928,
+      "step": 720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.4352941176470587e-05,
+      "loss": 0.991,
+      "step": 730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.482352941176471e-05,
+      "loss": 0.9895,
+      "step": 740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.529411764705883e-05,
+      "loss": 0.9899,
+      "step": 750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.576470588235295e-05,
+      "loss": 0.99,
+      "step": 760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.623529411764706e-05,
+      "loss": 0.9951,
+      "step": 770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6705882352941175e-05,
+      "loss": 0.9871,
+      "step": 780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7176470588235295e-05,
+      "loss": 0.9781,
+      "step": 790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7647058823529415e-05,
+      "loss": 0.9838,
+      "step": 800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.811764705882353e-05,
+      "loss": 0.9724,
+      "step": 810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.858823529411765e-05,
+      "loss": 0.9764,
+      "step": 820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.905882352941177e-05,
+      "loss": 0.981,
+      "step": 830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.952941176470589e-05,
+      "loss": 0.9841,
+      "step": 840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4e-05,
+      "loss": 0.9871,
+      "step": 850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.999998691314254e-05,
+      "loss": 0.9905,
+      "step": 860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.999994765258726e-05,
+      "loss": 0.9767,
+      "step": 870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.999988221838556e-05,
+      "loss": 0.9712,
+      "step": 880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9999790610623065e-05,
+      "loss": 0.9699,
+      "step": 890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9999672829419655e-05,
+      "loss": 0.9781,
+      "step": 900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.999952887492948e-05,
+      "loss": 0.9844,
+      "step": 910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9999358747340924e-05,
+      "loss": 0.9808,
+      "step": 920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999916244687663e-05,
+      "loss": 0.979,
+      "step": 930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.99989399737935e-05,
+      "loss": 0.9738,
+      "step": 940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9998691328382675e-05,
+      "loss": 0.9724,
+      "step": 950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999841651096955e-05,
+      "loss": 0.9729,
+      "step": 960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999811552191379e-05,
+      "loss": 0.9831,
+      "step": 970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999778836160929e-05,
+      "loss": 0.9809,
+      "step": 980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999743503048419e-05,
+      "loss": 0.9688,
+      "step": 990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999705552900088e-05,
+      "loss": 0.9832,
+      "step": 1000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9996649857656045e-05,
+      "loss": 0.97,
+      "step": 1010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999621801698055e-05,
+      "loss": 0.9693,
+      "step": 1020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999576000753955e-05,
+      "loss": 0.9661,
+      "step": 1030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999527582993243e-05,
+      "loss": 0.9646,
+      "step": 1040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.999476548479283e-05,
+      "loss": 0.9751,
+      "step": 1050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9994228972788636e-05,
+      "loss": 0.9712,
+      "step": 1060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.999366629462197e-05,
+      "loss": 0.9702,
+      "step": 1070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9993077451029193e-05,
+      "loss": 0.9658,
+      "step": 1080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9992462442780927e-05,
+      "loss": 0.9715,
+      "step": 1090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.999182127068202e-05,
+      "loss": 0.9743,
+      "step": 1100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.999115393557157e-05,
+      "loss": 0.9724,
+      "step": 1110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.99904604383229e-05,
+      "loss": 0.9667,
+      "step": 1120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.998974077984359e-05,
+      "loss": 0.9678,
+      "step": 1130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9988994961075434e-05,
+      "loss": 0.967,
+      "step": 1140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.998822298299448e-05,
+      "loss": 0.9712,
+      "step": 1150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.998742484661101e-05,
+      "loss": 0.9719,
+      "step": 1160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.998660055296953e-05,
+      "loss": 0.96,
+      "step": 1170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.998575010314878e-05,
+      "loss": 0.9663,
+      "step": 1180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9984873498261734e-05,
+      "loss": 0.9636,
+      "step": 1190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.998397073945559e-05,
+      "loss": 0.963,
+      "step": 1200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.998304182791178e-05,
+      "loss": 0.9634,
+      "step": 1210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.998208676484596e-05,
+      "loss": 0.9656,
+      "step": 1220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.998110555150799e-05,
+      "loss": 0.961,
+      "step": 1230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9980098189181985e-05,
+      "loss": 0.9524,
+      "step": 1240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.997906467918627e-05,
+      "loss": 0.9728,
+      "step": 1250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.997800502287336e-05,
+      "loss": 0.9667,
+      "step": 1260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.997691922163004e-05,
+      "loss": 0.9617,
+      "step": 1270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.997580727687727e-05,
+      "loss": 0.9552,
+      "step": 1280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.997466919007024e-05,
+      "loss": 0.9621,
+      "step": 1290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.997350496269834e-05,
+      "loss": 0.958,
+      "step": 1300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.997231459628519e-05,
+      "loss": 0.971,
+      "step": 1310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9971098092388596e-05,
+      "loss": 0.9472,
+      "step": 1320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.9969855452600585e-05,
+      "loss": 0.964,
+      "step": 1330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.996858667854738e-05,
+      "loss": 0.9638,
+      "step": 1340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.99672917718894e-05,
+      "loss": 0.9605,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.996597073432129e-05,
+      "loss": 0.9614,
+      "step": 1360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.996462356757185e-05,
+      "loss": 0.9517,
+      "step": 1370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.996325027340411e-05,
+      "loss": 0.9506,
+      "step": 1380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9961850853615285e-05,
+      "loss": 0.9477,
+      "step": 1390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.996042531003677e-05,
+      "loss": 0.9546,
+      "step": 1400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.995897364453415e-05,
+      "loss": 0.944,
+      "step": 1410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9957495859007206e-05,
+      "loss": 0.9562,
+      "step": 1420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9955991955389885e-05,
+      "loss": 0.955,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.995446193565033e-05,
+      "loss": 0.9622,
+      "step": 1440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9952905801790866e-05,
+      "loss": 0.9558,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.995132355584797e-05,
+      "loss": 0.9489,
+      "step": 1460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.994971519989231e-05,
+      "loss": 0.9579,
+      "step": 1470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9948080736028714e-05,
+      "loss": 0.9577,
+      "step": 1480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.994642016639619e-05,
+      "loss": 0.9568,
+      "step": 1490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9944733493167885e-05,
+      "loss": 0.9444,
+      "step": 1500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.994302071855114e-05,
+      "loss": 0.9514,
+      "step": 1510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.994128184478745e-05,
+      "loss": 0.9394,
+      "step": 1520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.993951687415242e-05,
+      "loss": 0.9464,
+      "step": 1530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.993772580895587e-05,
+      "loss": 0.9436,
+      "step": 1540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.993590865154173e-05,
+      "loss": 0.9454,
+      "step": 1550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9934065404288086e-05,
+      "loss": 0.95,
+      "step": 1560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9932196069607175e-05,
+      "loss": 0.9569,
+      "step": 1570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9930300649945375e-05,
+      "loss": 0.9461,
+      "step": 1580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.992837914778318e-05,
+      "loss": 0.9453,
+      "step": 1590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.992643156563524e-05,
+      "loss": 0.9464,
+      "step": 1600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.992445790605032e-05,
+      "loss": 0.942,
+      "step": 1610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.992245817161134e-05,
+      "loss": 0.9395,
+      "step": 1620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.992043236493531e-05,
+      "loss": 0.9497,
+      "step": 1630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.991838048867338e-05,
+      "loss": 0.9503,
+      "step": 1640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9916302545510795e-05,
+      "loss": 0.9431,
+      "step": 1650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.991419853816694e-05,
+      "loss": 0.9549,
+      "step": 1660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9912068469395315e-05,
+      "loss": 0.9464,
+      "step": 1670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.990991234198349e-05,
+      "loss": 0.94,
+      "step": 1680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9907730158753164e-05,
+      "loss": 0.9347,
+      "step": 1690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9905521922560134e-05,
+      "loss": 0.9414,
+      "step": 1700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9903287636294286e-05,
+      "loss": 0.9365,
+      "step": 1710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9901027302879585e-05,
+      "loss": 0.9406,
+      "step": 1720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.989874092527411e-05,
+      "loss": 0.9446,
+      "step": 1730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.989642850647002e-05,
+      "loss": 0.9456,
+      "step": 1740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.989409004949352e-05,
+      "loss": 0.9483,
+      "step": 1750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.989172555740494e-05,
+      "loss": 0.9379,
+      "step": 1760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.988933503329864e-05,
+      "loss": 0.943,
+      "step": 1770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.988691848030307e-05,
+      "loss": 0.938,
+      "step": 1780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9884475901580736e-05,
+      "loss": 0.9429,
+      "step": 1790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.988200730032821e-05,
+      "loss": 0.9352,
+      "step": 1800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.987951267977612e-05,
+      "loss": 0.9414,
+      "step": 1810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9876992043189135e-05,
+      "loss": 0.9367,
+      "step": 1820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.987444539386597e-05,
+      "loss": 0.9427,
+      "step": 1830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.98718727351394e-05,
+      "loss": 0.9416,
+      "step": 1840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.986927407037622e-05,
+      "loss": 0.9362,
+      "step": 1850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.986664940297726e-05,
+      "loss": 0.9452,
+      "step": 1860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.98639987363774e-05,
+      "loss": 0.9435,
+      "step": 1870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.986132207404551e-05,
+      "loss": 0.9354,
+      "step": 1880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9858619419484526e-05,
+      "loss": 0.9401,
+      "step": 1890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.985589077623135e-05,
+      "loss": 0.9341,
+      "step": 1900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.9853136147856924e-05,
+      "loss": 0.9342,
+      "step": 1910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.98503555379662e-05,
+      "loss": 0.9357,
+      "step": 1920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.984754895019812e-05,
+      "loss": 0.9393,
+      "step": 1930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.984471638822562e-05,
+      "loss": 0.9401,
+      "step": 1940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.984185785575562e-05,
+      "loss": 0.9344,
+      "step": 1950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.983897335652907e-05,
+      "loss": 0.9319,
+      "step": 1960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.9836062894320864e-05,
+      "loss": 0.9449,
+      "step": 1970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.983312647293987e-05,
+      "loss": 0.9364,
+      "step": 1980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.983016409622895e-05,
+      "loss": 0.9347,
+      "step": 1990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.982717576806492e-05,
+      "loss": 0.9319,
+      "step": 2000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.9824161492358565e-05,
+      "loss": 0.9361,
+      "step": 2010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.982112127305463e-05,
+      "loss": 0.9315,
+      "step": 2020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.9818055114131795e-05,
+      "loss": 0.9288,
+      "step": 2030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.981496301960271e-05,
+      "loss": 0.9358,
+      "step": 2040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.981184499351395e-05,
+      "loss": 0.9358,
+      "step": 2050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9808701039946026e-05,
+      "loss": 0.9335,
+      "step": 2060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.98055311630134e-05,
+      "loss": 0.941,
+      "step": 2070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.980233536686442e-05,
+      "loss": 0.9366,
+      "step": 2080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9799113655681407e-05,
+      "loss": 0.9355,
+      "step": 2090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.979586603368055e-05,
+      "loss": 0.9345,
+      "step": 2100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.979259250511197e-05,
+      "loss": 0.9362,
+      "step": 2110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.978929307425969e-05,
+      "loss": 0.9266,
+      "step": 2120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.978596774544163e-05,
+      "loss": 0.9343,
+      "step": 2130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.978261652300959e-05,
+      "loss": 0.9275,
+      "step": 2140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.977923941134928e-05,
+      "loss": 0.9312,
+      "step": 2150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9775836414880266e-05,
+      "loss": 0.9326,
+      "step": 2160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9772407538056014e-05,
+      "loss": 0.9425,
+      "step": 2170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.976895278536383e-05,
+      "loss": 0.935,
+      "step": 2180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.976547216132492e-05,
+      "loss": 0.9308,
+      "step": 2190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.976196567049431e-05,
+      "loss": 0.9319,
+      "step": 2200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.97584333174609e-05,
+      "loss": 0.933,
+      "step": 2210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.975487510684743e-05,
+      "loss": 0.9285,
+      "step": 2220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.975129104331048e-05,
+      "loss": 0.9304,
+      "step": 2230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.974768113154046e-05,
+      "loss": 0.9349,
+      "step": 2240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.974404537626162e-05,
+      "loss": 0.9303,
+      "step": 2250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.9740383782232e-05,
+      "loss": 0.9273,
+      "step": 2260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.97366963542435e-05,
+      "loss": 0.9319,
+      "step": 2270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.9732983097121795e-05,
+      "loss": 0.9261,
+      "step": 2280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.972924401572637e-05,
+      "loss": 0.9354,
+      "step": 2290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.972547911495051e-05,
+      "loss": 0.9292,
+      "step": 2300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.972168839972128e-05,
+      "loss": 0.9302,
+      "step": 2310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.9717871874999545e-05,
+      "loss": 0.9232,
+      "step": 2320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.971402954577993e-05,
+      "loss": 0.9251,
+      "step": 2330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.971016141709083e-05,
+      "loss": 0.9447,
+      "step": 2340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.970626749399443e-05,
+      "loss": 0.9549,
+      "step": 2350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.9702347781586645e-05,
+      "loss": 0.9613,
+      "step": 2360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.9698402284997135e-05,
+      "loss": 1.0125,
+      "step": 2370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.9694431009389315e-05,
+      "loss": 0.9856,
+      "step": 2380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.969043395996035e-05,
+      "loss": 0.9733,
+      "step": 2390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.968641114194111e-05,
+      "loss": 0.9635,
+      "step": 2400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.9682362560596214e-05,
+      "loss": 0.9609,
+      "step": 2410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.967828822122397e-05,
+      "loss": 0.9423,
+      "step": 2420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.967418812915641e-05,
+      "loss": 0.9416,
+      "step": 2430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.967006228975927e-05,
+      "loss": 0.9448,
+      "step": 2440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.966591070843197e-05,
+      "loss": 0.942,
+      "step": 2450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.9661733390607635e-05,
+      "loss": 0.9356,
+      "step": 2460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.965753034175305e-05,
+      "loss": 0.9379,
+      "step": 2470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.96533015673687e-05,
+      "loss": 0.9402,
+      "step": 2480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.96490470729887e-05,
+      "loss": 0.937,
+      "step": 2490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.964476686418087e-05,
+      "loss": 0.9311,
+      "step": 2500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.9640460946546645e-05,
+      "loss": 0.9384,
+      "step": 2510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.9636129325721123e-05,
+      "loss": 0.9325,
+      "step": 2520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.963177200737303e-05,
+      "loss": 0.9311,
+      "step": 2530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.9627388997204726e-05,
+      "loss": 0.9279,
+      "step": 2540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.9622980300952204e-05,
+      "loss": 0.9336,
+      "step": 2550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.961854592438504e-05,
+      "loss": 0.9325,
+      "step": 2560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.961408587330647e-05,
+      "loss": 0.9294,
+      "step": 2570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.960960015355327e-05,
+      "loss": 0.9353,
+      "step": 2580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.960508877099586e-05,
+      "loss": 0.9248,
+      "step": 2590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.960055173153821e-05,
+      "loss": 0.9253,
+      "step": 2600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.959598904111788e-05,
+      "loss": 0.9241,
+      "step": 2610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.9591400705706e-05,
+      "loss": 0.9208,
+      "step": 2620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.9586786731307254e-05,
+      "loss": 0.9267,
+      "step": 2630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.9582147123959894e-05,
+      "loss": 0.9344,
+      "step": 2640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.9577481889735707e-05,
+      "loss": 0.9254,
+      "step": 2650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.957279103474001e-05,
+      "loss": 0.929,
+      "step": 2660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.9568074565111665e-05,
+      "loss": 0.9263,
+      "step": 2670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.956333248702304e-05,
+      "loss": 0.9292,
+      "step": 2680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.955856480668004e-05,
+      "loss": 0.9239,
+      "step": 2690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.9553771530322044e-05,
+      "loss": 0.93,
+      "step": 2700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.954895266422195e-05,
+      "loss": 0.9225,
+      "step": 2710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.9544108214686154e-05,
+      "loss": 0.9265,
+      "step": 2720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.9539238188054494e-05,
+      "loss": 0.9242,
+      "step": 2730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.953434259070032e-05,
+      "loss": 0.9193,
+      "step": 2740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.9529421429030426e-05,
+      "loss": 0.9261,
+      "step": 2750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.952447470948506e-05,
+      "loss": 0.925,
+      "step": 2760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9519502438537944e-05,
+      "loss": 0.9155,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.95145046226962e-05,
+      "loss": 0.9255,
+      "step": 2780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.95094812685004e-05,
+      "loss": 0.9201,
+      "step": 2790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.950443238252453e-05,
+      "loss": 0.9211,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9499357971376014e-05,
+      "loss": 0.9248,
+      "step": 2810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.949425804169565e-05,
+      "loss": 0.9207,
+      "step": 2820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.948913260015764e-05,
+      "loss": 0.9201,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9483981653469586e-05,
+      "loss": 0.9185,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.947880520837245e-05,
+      "loss": 0.9272,
+      "step": 2850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9473603271640575e-05,
+      "loss": 0.9164,
+      "step": 2860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9468375850081664e-05,
+      "loss": 0.9204,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.946312295053676e-05,
+      "loss": 0.9218,
+      "step": 2880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9457844579880274e-05,
+      "loss": 0.9187,
+      "step": 2890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9452540745019916e-05,
+      "loss": 0.9192,
+      "step": 2900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9447211452896754e-05,
+      "loss": 0.9207,
+      "step": 2910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.944185671048514e-05,
+      "loss": 0.913,
+      "step": 2920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.943647652479277e-05,
+      "loss": 0.9088,
+      "step": 2930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9431070902860605e-05,
+      "loss": 0.908,
+      "step": 2940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.94256398517629e-05,
+      "loss": 0.9183,
+      "step": 2950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.942018337860721e-05,
+      "loss": 0.9119,
+      "step": 2960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.941470149053433e-05,
+      "loss": 0.9215,
+      "step": 2970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9409194194718325e-05,
+      "loss": 0.9189,
+      "step": 2980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9403661498366536e-05,
+      "loss": 0.9221,
+      "step": 2990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.93981034087195e-05,
+      "loss": 0.9125,
+      "step": 3000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9392519933051025e-05,
+      "loss": 0.9196,
+      "step": 3010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.938691107866812e-05,
+      "loss": 0.9212,
+      "step": 3020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.938127685291101e-05,
+      "loss": 0.918,
+      "step": 3030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9375617263153136e-05,
+      "loss": 0.9136,
+      "step": 3040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9369932316801115e-05,
+      "loss": 0.9161,
+      "step": 3050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.936422202129475e-05,
+      "loss": 0.9088,
+      "step": 3060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.935848638410704e-05,
+      "loss": 0.9151,
+      "step": 3070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9352725412744114e-05,
+      "loss": 0.9188,
+      "step": 3080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9346939114745284e-05,
+      "loss": 0.9169,
+      "step": 3090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.934112749768299e-05,
+      "loss": 0.9156,
+      "step": 3100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9335290569162815e-05,
+      "loss": 0.9141,
+      "step": 3110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9329428336823464e-05,
+      "loss": 0.9148,
+      "step": 3120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9323540808336755e-05,
+      "loss": 0.9114,
+      "step": 3130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.931762799140762e-05,
+      "loss": 0.9081,
+      "step": 3140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.931168989377407e-05,
+      "loss": 0.9148,
+      "step": 3150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.930572652320721e-05,
+      "loss": 0.9127,
+      "step": 3160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.929973788751122e-05,
+      "loss": 0.9137,
+      "step": 3170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.929372399452334e-05,
+      "loss": 0.921,
+      "step": 3180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.928768485211387e-05,
+      "loss": 0.907,
+      "step": 3190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.928162046818615e-05,
+      "loss": 0.9095,
+      "step": 3200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.927553085067654e-05,
+      "loss": 0.9105,
+      "step": 3210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.926941600755445e-05,
+      "loss": 0.9149,
+      "step": 3220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.926327594682229e-05,
+      "loss": 0.913,
+      "step": 3230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.925711067651546e-05,
+      "loss": 0.9057,
+      "step": 3240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9250920204702366e-05,
+      "loss": 0.905,
+      "step": 3250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.924470453948439e-05,
+      "loss": 0.9102,
+      "step": 3260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.923846368899588e-05,
+      "loss": 0.9165,
+      "step": 3270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.923219766140415e-05,
+      "loss": 0.9106,
+      "step": 3280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.922590646490946e-05,
+      "loss": 0.9157,
+      "step": 3290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.9219590107745016e-05,
+      "loss": 0.9071,
+      "step": 3300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.921324859817695e-05,
+      "loss": 0.9113,
+      "step": 3310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.920688194450429e-05,
+      "loss": 0.91,
+      "step": 3320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.920049015505898e-05,
+      "loss": 0.9212,
+      "step": 3330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.919407323820588e-05,
+      "loss": 0.9073,
+      "step": 3340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.918763120234272e-05,
+      "loss": 0.9135,
+      "step": 3350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.918116405590009e-05,
+      "loss": 0.9148,
+      "step": 3360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.9174671807341454e-05,
+      "loss": 0.9134,
+      "step": 3370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.916815446516312e-05,
+      "loss": 0.9051,
+      "step": 3380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.9161612037894256e-05,
+      "loss": 0.9033,
+      "step": 3390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.915504453409683e-05,
+      "loss": 0.9111,
+      "step": 3400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.914845196236564e-05,
+      "loss": 0.9113,
+      "step": 3410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.91418343313283e-05,
+      "loss": 0.9065,
+      "step": 3420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.913519164964521e-05,
+      "loss": 0.9063,
+      "step": 3430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.912852392600954e-05,
+      "loss": 0.9047,
+      "step": 3440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.912183116914725e-05,
+      "loss": 0.9116,
+      "step": 3450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.911511338781707e-05,
+      "loss": 0.9057,
+      "step": 3460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.9108370590810445e-05,
+      "loss": 0.9045,
+      "step": 3470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.910160278695158e-05,
+      "loss": 0.9155,
+      "step": 3480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.909480998509742e-05,
+      "loss": 0.9151,
+      "step": 3490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.908799219413759e-05,
+      "loss": 0.9137,
+      "step": 3500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.9081149422994445e-05,
+      "loss": 0.907,
+      "step": 3510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.907428168062302e-05,
+      "loss": 0.9205,
+      "step": 3520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.906738897601103e-05,
+      "loss": 0.9105,
+      "step": 3530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.906047131817886e-05,
+      "loss": 0.9154,
+      "step": 3540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.905352871617955e-05,
+      "loss": 0.904,
+      "step": 3550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.904656117909878e-05,
+      "loss": 0.9005,
+      "step": 3560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.903956871605488e-05,
+      "loss": 0.9097,
+      "step": 3570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.903255133619877e-05,
+      "loss": 0.9048,
+      "step": 3580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.902550904871401e-05,
+      "loss": 0.9028,
+      "step": 3590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.9018441862816726e-05,
+      "loss": 0.9088,
+      "step": 3600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.901134978775566e-05,
+      "loss": 0.9128,
+      "step": 3610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.90042328328121e-05,
+      "loss": 0.94,
+      "step": 3620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.89970910072999e-05,
+      "loss": 0.9483,
+      "step": 3630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.898992432056547e-05,
+      "loss": 0.9345,
+      "step": 3640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.898273278198776e-05,
+      "loss": 0.9188,
+      "step": 3650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.897551640097822e-05,
+      "loss": 0.9143,
+      "step": 3660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.8968275186980825e-05,
+      "loss": 0.9203,
+      "step": 3670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.8961009149472055e-05,
+      "loss": 0.9199,
+      "step": 3680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.895371829796087e-05,
+      "loss": 0.9076,
+      "step": 3690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.89464026419887e-05,
+      "loss": 0.9149,
+      "step": 3700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.8939062191129445e-05,
+      "loss": 0.9114,
+      "step": 3710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.893169695498944e-05,
+      "loss": 0.902,
+      "step": 3720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.892430694320747e-05,
+      "loss": 0.9026,
+      "step": 3730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.891689216545474e-05,
+      "loss": 0.9055,
+      "step": 3740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.890945263143486e-05,
+      "loss": 0.905,
+      "step": 3750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.890198835088384e-05,
+      "loss": 0.9146,
+      "step": 3760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.889449933357009e-05,
+      "loss": 0.908,
+      "step": 3770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.888698558929436e-05,
+      "loss": 0.9011,
+      "step": 3780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.887944712788979e-05,
+      "loss": 0.9,
+      "step": 3790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.887188395922187e-05,
+      "loss": 0.9093,
+      "step": 3800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.886429609318839e-05,
+      "loss": 0.9,
+      "step": 3810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.88566835397195e-05,
+      "loss": 0.9048,
+      "step": 3820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.884904630877763e-05,
+      "loss": 0.906,
+      "step": 3830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.8841384410357505e-05,
+      "loss": 0.9115,
+      "step": 3840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.8833697854486165e-05,
+      "loss": 0.9083,
+      "step": 3850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.882598665122288e-05,
+      "loss": 0.9096,
+      "step": 3860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.8818250810659195e-05,
+      "loss": 0.9031,
+      "step": 3870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.881049034291891e-05,
+      "loss": 0.8998,
+      "step": 3880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.880270525815801e-05,
+      "loss": 0.8949,
+      "step": 3890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.879489556656474e-05,
+      "loss": 0.9016,
+      "step": 3900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.8787061278359536e-05,
+      "loss": 0.9036,
+      "step": 3910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.877920240379502e-05,
+      "loss": 0.8993,
+      "step": 3920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.877131895315597e-05,
+      "loss": 0.9081,
+      "step": 3930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.8763410936759374e-05,
+      "loss": 0.9063,
+      "step": 3940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.8755478364954315e-05,
+      "loss": 0.9037,
+      "step": 3950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.874752124812205e-05,
+      "loss": 0.8991,
+      "step": 3960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.8739539596675945e-05,
+      "loss": 0.907,
+      "step": 3970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.8731533421061464e-05,
+      "loss": 0.9055,
+      "step": 3980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.872350273175619e-05,
+      "loss": 0.8946,
+      "step": 3990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.8715447539269754e-05,
+      "loss": 0.8988,
+      "step": 4000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.870736785414389e-05,
+      "loss": 0.8999,
+      "step": 4010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.869926368695235e-05,
+      "loss": 0.9024,
+      "step": 4020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.869113504830095e-05,
+      "loss": 0.8982,
+      "step": 4030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.868298194882752e-05,
+      "loss": 0.8997,
+      "step": 4040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.867480439920191e-05,
+      "loss": 0.8995,
+      "step": 4050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.866660241012596e-05,
+      "loss": 0.901,
+      "step": 4060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8658375992333495e-05,
+      "loss": 0.8933,
+      "step": 4070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8650125156590316e-05,
+      "loss": 0.8971,
+      "step": 4080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.864184991369417e-05,
+      "loss": 0.9019,
+      "step": 4090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8633550274474744e-05,
+      "loss": 0.8938,
+      "step": 4100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.862522624979367e-05,
+      "loss": 0.8998,
+      "step": 4110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.861687785054447e-05,
+      "loss": 0.9028,
+      "step": 4120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8608505087652584e-05,
+      "loss": 0.8935,
+      "step": 4130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.860010797207532e-05,
+      "loss": 0.8973,
+      "step": 4140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.859168651480186e-05,
+      "loss": 0.8981,
+      "step": 4150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.858324072685326e-05,
+      "loss": 0.899,
+      "step": 4160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.857477061928239e-05,
+      "loss": 0.9,
+      "step": 4170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.856627620317396e-05,
+      "loss": 0.9012,
+      "step": 4180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.855775748964449e-05,
+      "loss": 0.8935,
+      "step": 4190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.8549214489842316e-05,
+      "loss": 0.8997,
+      "step": 4200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.8540647214947516e-05,
+      "loss": 0.9008,
+      "step": 4210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.853205567617197e-05,
+      "loss": 0.9024,
+      "step": 4220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.852343988475931e-05,
+      "loss": 0.9004,
+      "step": 4230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.851479985198489e-05,
+      "loss": 0.8942,
+      "step": 4240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.85061355891558e-05,
+      "loss": 0.8981,
+      "step": 4250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.849744710761084e-05,
+      "loss": 0.9069,
+      "step": 4260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.8488734418720494e-05,
+      "loss": 0.8969,
+      "step": 4270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.847999753388694e-05,
+      "loss": 0.8929,
+      "step": 4280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.847123646454402e-05,
+      "loss": 0.9022,
+      "step": 4290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.846245122215722e-05,
+      "loss": 0.8959,
+      "step": 4300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.845364181822365e-05,
+      "loss": 0.8916,
+      "step": 4310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8444808264272054e-05,
+      "loss": 0.8907,
+      "step": 4320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.843595057186279e-05,
+      "loss": 0.8874,
+      "step": 4330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.842706875258777e-05,
+      "loss": 0.8962,
+      "step": 4340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.841816281807054e-05,
+      "loss": 0.8932,
+      "step": 4350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8409232779966145e-05,
+      "loss": 0.8923,
+      "step": 4360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8400278649961197e-05,
+      "loss": 0.8946,
+      "step": 4370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.839130043977385e-05,
+      "loss": 0.9001,
+      "step": 4380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8382298161153745e-05,
+      "loss": 0.8927,
+      "step": 4390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.837327182588205e-05,
+      "loss": 0.8923,
+      "step": 4400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8364221445771395e-05,
+      "loss": 0.9049,
+      "step": 4410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.835514703266589e-05,
+      "loss": 0.8904,
+      "step": 4420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8346048598441076e-05,
+      "loss": 0.8916,
+      "step": 4430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.833692615500396e-05,
+      "loss": 0.8882,
+      "step": 4440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.832777971429295e-05,
+      "loss": 0.8914,
+      "step": 4450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.831860928827785e-05,
+      "loss": 0.888,
+      "step": 4460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.830941488895988e-05,
+      "loss": 0.8958,
+      "step": 4470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8300196528371616e-05,
+      "loss": 0.8957,
+      "step": 4480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.829095421857699e-05,
+      "loss": 0.8938,
+      "step": 4490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8281687971671285e-05,
+      "loss": 0.8947,
+      "step": 4500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.827239779978111e-05,
+      "loss": 0.894,
+      "step": 4510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8263083715064363e-05,
+      "loss": 0.8921,
+      "step": 4520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8253745729710284e-05,
+      "loss": 0.8947,
+      "step": 4530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.824438385593934e-05,
+      "loss": 0.8978,
+      "step": 4540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8234998106003283e-05,
+      "loss": 0.8894,
+      "step": 4550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8225588492185105e-05,
+      "loss": 0.896,
+      "step": 4560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.821615502679906e-05,
+      "loss": 0.8856,
+      "step": 4570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8206697722190563e-05,
+      "loss": 0.8892,
+      "step": 4580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.819721659073626e-05,
+      "loss": 0.8863,
+      "step": 4590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.818771164484398e-05,
+      "loss": 0.8932,
+      "step": 4600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.81781828969527e-05,
+      "loss": 0.8907,
+      "step": 4610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.816863035953257e-05,
+      "loss": 0.8863,
+      "step": 4620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8159054045084846e-05,
+      "loss": 0.8903,
+      "step": 4630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.814945396614192e-05,
+      "loss": 0.8884,
+      "step": 4640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.813983013526728e-05,
+      "loss": 0.9007,
+      "step": 4650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.813018256505549e-05,
+      "loss": 0.885,
+      "step": 4660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8120511268132195e-05,
+      "loss": 0.8925,
+      "step": 4670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.811081625715408e-05,
+      "loss": 0.8887,
+      "step": 4680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8101097544808866e-05,
+      "loss": 0.8858,
+      "step": 4690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.80913551438153e-05,
+      "loss": 0.8851,
+      "step": 4700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.808158906692311e-05,
+      "loss": 0.8947,
+      "step": 4710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8071799326913045e-05,
+      "loss": 0.8966,
+      "step": 4720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8061985936596776e-05,
+      "loss": 0.8866,
+      "step": 4730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.805214890881695e-05,
+      "loss": 0.8919,
+      "step": 4740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.804228825644716e-05,
+      "loss": 0.8876,
+      "step": 4750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8032403992391885e-05,
+      "loss": 0.8922,
+      "step": 4760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.802249612958653e-05,
+      "loss": 0.8934,
+      "step": 4770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.801256468099736e-05,
+      "loss": 0.8935,
+      "step": 4780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.800260965962154e-05,
+      "loss": 0.8863,
+      "step": 4790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.799263107848705e-05,
+      "loss": 0.8887,
+      "step": 4800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7982628950652725e-05,
+      "loss": 0.8858,
+      "step": 4810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.79726032892082e-05,
+      "loss": 0.8957,
+      "step": 4820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7962554107273926e-05,
+      "loss": 0.885,
+      "step": 4830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.795248141800112e-05,
+      "loss": 0.8843,
+      "step": 4840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.794238523457175e-05,
+      "loss": 0.8899,
+      "step": 4850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.793226557019857e-05,
+      "loss": 0.9039,
+      "step": 4860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7922122438125035e-05,
+      "loss": 0.9641,
+      "step": 4870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.791195585162531e-05,
+      "loss": 0.9534,
+      "step": 4880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7901765824004264e-05,
+      "loss": 0.9882,
+      "step": 4890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7891552368597456e-05,
+      "loss": 1.0478,
+      "step": 4900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.788131549877107e-05,
+      "loss": 0.9491,
+      "step": 4910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.787105522792195e-05,
+      "loss": 0.9363,
+      "step": 4920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7860771569477587e-05,
+      "loss": 0.9231,
+      "step": 4930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7850464536896036e-05,
+      "loss": 0.9272,
+      "step": 4940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.784013414366598e-05,
+      "loss": 0.9205,
+      "step": 4950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7829780403306654e-05,
+      "loss": 0.9177,
+      "step": 4960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.781940332936784e-05,
+      "loss": 0.9212,
+      "step": 4970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.780900293542988e-05,
+      "loss": 0.9116,
+      "step": 4980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.779857923510361e-05,
+      "loss": 0.9004,
+      "step": 4990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.778813224203038e-05,
+      "loss": 0.906,
+      "step": 5000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7777661969882034e-05,
+      "loss": 0.9011,
+      "step": 5010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.776716843236086e-05,
+      "loss": 0.9084,
+      "step": 5020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.775665164319959e-05,
+      "loss": 0.9045,
+      "step": 5030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.774611161616141e-05,
+      "loss": 0.9059,
+      "step": 5040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.77355483650399e-05,
+      "loss": 0.9022,
+      "step": 5050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.772496190365903e-05,
+      "loss": 0.9047,
+      "step": 5060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7714352245873154e-05,
+      "loss": 0.9057,
+      "step": 5070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7703719405566986e-05,
+      "loss": 0.8948,
+      "step": 5080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7693063396655574e-05,
+      "loss": 0.8981,
+      "step": 5090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.768238423308427e-05,
+      "loss": 0.8995,
+      "step": 5100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.767168192882876e-05,
+      "loss": 0.8987,
+      "step": 5110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.766095649789498e-05,
+      "loss": 0.899,
+      "step": 5120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.765020795431917e-05,
+      "loss": 0.8897,
+      "step": 5130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.763943631216778e-05,
+      "loss": 0.8925,
+      "step": 5140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7628641585537505e-05,
+      "loss": 0.8906,
+      "step": 5150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.761782378855525e-05,
+      "loss": 0.892,
+      "step": 5160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.760698293537813e-05,
+      "loss": 0.8906,
+      "step": 5170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.759611904019338e-05,
+      "loss": 0.8962,
+      "step": 5180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.758523211721846e-05,
+      "loss": 0.8891,
+      "step": 5190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.757432218070091e-05,
+      "loss": 0.8903,
+      "step": 5200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7563389244918415e-05,
+      "loss": 0.8921,
+      "step": 5210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.755243332417875e-05,
+      "loss": 0.8881,
+      "step": 5220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7541454432819776e-05,
+      "loss": 0.8935,
+      "step": 5230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.753045258520941e-05,
+      "loss": 0.8919,
+      "step": 5240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.751942779574561e-05,
+      "loss": 0.8902,
+      "step": 5250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.750838007885636e-05,
+      "loss": 0.8911,
+      "step": 5260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.749730944899966e-05,
+      "loss": 0.8917,
+      "step": 5270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.7486215920663476e-05,
+      "loss": 0.8857,
+      "step": 5280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.747509950836575e-05,
+      "loss": 0.8885,
+      "step": 5290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.746396022665438e-05,
+      "loss": 0.8928,
+      "step": 5300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.745279809010718e-05,
+      "loss": 0.893,
+      "step": 5310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7441613113331876e-05,
+      "loss": 0.8953,
+      "step": 5320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.74304053109661e-05,
+      "loss": 0.8903,
+      "step": 5330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.741917469767731e-05,
+      "loss": 0.8934,
+      "step": 5340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.74079212881629e-05,
+      "loss": 0.8851,
+      "step": 5350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7396645097150014e-05,
+      "loss": 0.8973,
+      "step": 5360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7385346139395645e-05,
+      "loss": 0.8952,
+      "step": 5370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.737402442968658e-05,
+      "loss": 0.8808,
+      "step": 5380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.736267998283939e-05,
+      "loss": 0.886,
+      "step": 5390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.735131281370037e-05,
+      "loss": 0.8864,
+      "step": 5400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7339922937145596e-05,
+      "loss": 0.8817,
+      "step": 5410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.732851036808082e-05,
+      "loss": 0.8875,
+      "step": 5420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.731707512144151e-05,
+      "loss": 0.8716,
+      "step": 5430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.7305617212192816e-05,
+      "loss": 0.8835,
+      "step": 5440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.729413665532955e-05,
+      "loss": 0.8775,
+      "step": 5450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.728263346587612e-05,
+      "loss": 0.8899,
+      "step": 5460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.727110765888663e-05,
+      "loss": 0.889,
+      "step": 5470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7259559249444704e-05,
+      "loss": 0.8818,
+      "step": 5480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.724798825266359e-05,
+      "loss": 0.884,
+      "step": 5490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.72363946836861e-05,
+      "loss": 0.8796,
+      "step": 5500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.722477855768456e-05,
+      "loss": 0.893,
+      "step": 5510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.721313988986083e-05,
+      "loss": 0.8833,
+      "step": 5520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.720147869544628e-05,
+      "loss": 0.8789,
+      "step": 5530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.718979498970172e-05,
+      "loss": 0.8888,
+      "step": 5540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7178088787917484e-05,
+      "loss": 0.8824,
+      "step": 5550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.7166360105413285e-05,
+      "loss": 0.8788,
+      "step": 5560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.71546089575383e-05,
+      "loss": 0.8857,
+      "step": 5570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.714283535967108e-05,
+      "loss": 0.8796,
+      "step": 5580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.713103932721957e-05,
+      "loss": 0.8757,
+      "step": 5590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7119220875621065e-05,
+      "loss": 0.8838,
+      "step": 5600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.71073800203422e-05,
+      "loss": 0.8872,
+      "step": 5610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.709551677687895e-05,
+      "loss": 0.8847,
+      "step": 5620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.708363116075656e-05,
+      "loss": 0.8829,
+      "step": 5630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.707172318752956e-05,
+      "loss": 0.8821,
+      "step": 5640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.705979287278175e-05,
+      "loss": 0.8831,
+      "step": 5650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.704784023212618e-05,
+      "loss": 0.8776,
+      "step": 5660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.703586528120507e-05,
+      "loss": 0.884,
+      "step": 5670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.70238680356899e-05,
+      "loss": 0.8811,
+      "step": 5680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.7011848511281266e-05,
+      "loss": 0.8814,
+      "step": 5690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.699980672370896e-05,
+      "loss": 0.8762,
+      "step": 5700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.69877426887319e-05,
+      "loss": 0.8859,
+      "step": 5710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.697565642213812e-05,
+      "loss": 0.8864,
+      "step": 5720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.6963547939744737e-05,
+      "loss": 0.8829,
+      "step": 5730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.695141725739795e-05,
+      "loss": 0.8848,
+      "step": 5740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6939264390973016e-05,
+      "loss": 0.8864,
+      "step": 5750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.692708935637421e-05,
+      "loss": 0.8772,
+      "step": 5760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.691489216953483e-05,
+      "loss": 0.8742,
+      "step": 5770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.690267284641716e-05,
+      "loss": 0.881,
+      "step": 5780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.689043140301246e-05,
+      "loss": 0.8787,
+      "step": 5790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.687816785534092e-05,
+      "loss": 0.8845,
+      "step": 5800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.686588221945168e-05,
+      "loss": 0.877,
+      "step": 5810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.685357451142278e-05,
+      "loss": 0.8824,
+      "step": 5820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6841244747361126e-05,
+      "loss": 0.8767,
+      "step": 5830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.682889294340252e-05,
+      "loss": 0.8736,
+      "step": 5840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6816519115711586e-05,
+      "loss": 0.8808,
+      "step": 5850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.680412328048178e-05,
+      "loss": 0.8791,
+      "step": 5860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.679170545393534e-05,
+      "loss": 0.8793,
+      "step": 5870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6779265652323315e-05,
+      "loss": 0.8797,
+      "step": 5880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.67668038919255e-05,
+      "loss": 0.8802,
+      "step": 5890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6754320189050404e-05,
+      "loss": 0.8805,
+      "step": 5900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.674181456003528e-05,
+      "loss": 0.88,
+      "step": 5910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.672928702124607e-05,
+      "loss": 0.8706,
+      "step": 5920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.6716737589077386e-05,
+      "loss": 0.8814,
+      "step": 5930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.670416627995249e-05,
+      "loss": 0.8861,
+      "step": 5940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.669157311032326e-05,
+      "loss": 0.8821,
+      "step": 5950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.667895809667022e-05,
+      "loss": 0.884,
+      "step": 5960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.666632125550245e-05,
+      "loss": 0.8849,
+      "step": 5970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.66536626033576e-05,
+      "loss": 0.8863,
+      "step": 5980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.664098215680187e-05,
+      "loss": 0.8699,
+      "step": 5990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.662827993242998e-05,
+      "loss": 0.871,
+      "step": 6000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.661555594686515e-05,
+      "loss": 0.878,
+      "step": 6010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.660281021675908e-05,
+      "loss": 0.8794,
+      "step": 6020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.659004275879193e-05,
+      "loss": 0.8727,
+      "step": 6030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.657725358967228e-05,
+      "loss": 0.8734,
+      "step": 6040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.6564442726137134e-05,
+      "loss": 0.8767,
+      "step": 6050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.655161018495189e-05,
+      "loss": 0.8723,
+      "step": 6060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.6538755982910314e-05,
+      "loss": 0.8725,
+      "step": 6070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.6525880136834514e-05,
+      "loss": 0.8683,
+      "step": 6080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.651298266357493e-05,
+      "loss": 0.8709,
+      "step": 6090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.65000635800103e-05,
+      "loss": 0.8735,
+      "step": 6100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.648712290304764e-05,
+      "loss": 0.8882,
+      "step": 6110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.647416064962224e-05,
+      "loss": 0.8879,
+      "step": 6120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.64611768366976e-05,
+      "loss": 0.8852,
+      "step": 6130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.644817148126546e-05,
+      "loss": 0.8906,
+      "step": 6140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.643514460034575e-05,
+      "loss": 0.906,
+      "step": 6150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.642209621098655e-05,
+      "loss": 0.9239,
+      "step": 6160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.640902633026411e-05,
+      "loss": 0.9071,
+      "step": 6170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.63959349752828e-05,
+      "loss": 0.9016,
+      "step": 6180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.638282216317508e-05,
+      "loss": 0.896,
+      "step": 6190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.63696879111015e-05,
+      "loss": 0.8891,
+      "step": 6200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.635653223625067e-05,
+      "loss": 0.8903,
+      "step": 6210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.634335515583924e-05,
+      "loss": 0.8887,
+      "step": 6220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.633015668711187e-05,
+      "loss": 0.8891,
+      "step": 6230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.63169368473412e-05,
+      "loss": 0.8896,
+      "step": 6240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.630369565382785e-05,
+      "loss": 0.8906,
+      "step": 6250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.629043312390037e-05,
+      "loss": 0.8869,
+      "step": 6260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.627714927491527e-05,
+      "loss": 0.8818,
+      "step": 6270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.6263844124256905e-05,
+      "loss": 0.8876,
+      "step": 6280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.6250517689337545e-05,
+      "loss": 0.886,
+      "step": 6290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.6237169987597314e-05,
+      "loss": 0.8833,
+      "step": 6300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.622380103650415e-05,
+      "loss": 0.883,
+      "step": 6310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.6210410853553816e-05,
+      "loss": 0.8848,
+      "step": 6320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.6196999456269845e-05,
+      "loss": 0.8759,
+      "step": 6330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.6183566862203555e-05,
+      "loss": 0.8765,
+      "step": 6340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.617011308893398e-05,
+      "loss": 0.888,
+      "step": 6350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.6156638154067876e-05,
+      "loss": 0.8796,
+      "step": 6360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.614314207523971e-05,
+      "loss": 0.8835,
+      "step": 6370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.61296248701116e-05,
+      "loss": 0.8799,
+      "step": 6380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.611608655637333e-05,
+      "loss": 0.8861,
+      "step": 6390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.610252715174229e-05,
+      "loss": 0.8689,
+      "step": 6400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.608894667396347e-05,
+      "loss": 0.8771,
+      "step": 6410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.607534514080946e-05,
+      "loss": 0.8784,
+      "step": 6420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.60617225700804e-05,
+      "loss": 0.8831,
+      "step": 6430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.604807897960394e-05,
+      "loss": 0.882,
+      "step": 6440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.603441438723526e-05,
+      "loss": 0.8744,
+      "step": 6450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.602072881085701e-05,
+      "loss": 0.8706,
+      "step": 6460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.6007022268379316e-05,
+      "loss": 0.8786,
+      "step": 6470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.599329477773973e-05,
+      "loss": 0.8805,
+      "step": 6480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.597954635690324e-05,
+      "loss": 0.877,
+      "step": 6490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.5965777023862176e-05,
+      "loss": 0.878,
+      "step": 6500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.5951986796636295e-05,
+      "loss": 0.8697,
+      "step": 6510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.5938175693272655e-05,
+      "loss": 0.8735,
+      "step": 6520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.5924343731845664e-05,
+      "loss": 0.8823,
+      "step": 6530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.5910490930457006e-05,
+      "loss": 0.8761,
+      "step": 6540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.589661730723563e-05,
+      "loss": 0.8735,
+      "step": 6550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.5882722880337776e-05,
+      "loss": 0.8684,
+      "step": 6560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.5868807667946876e-05,
+      "loss": 0.8788,
+      "step": 6570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.585487168827355e-05,
+      "loss": 0.8685,
+      "step": 6580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.584091495955563e-05,
+      "loss": 0.8677,
+      "step": 6590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.58269375000581e-05,
+      "loss": 0.8762,
+      "step": 6600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.581293932807304e-05,
+      "loss": 0.8684,
+      "step": 6610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.579892046191967e-05,
+      "loss": 0.8732,
+      "step": 6620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.578488091994428e-05,
+      "loss": 0.8729,
+      "step": 6630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.5770820720520216e-05,
+      "loss": 0.8734,
+      "step": 6640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.575673988204786e-05,
+      "loss": 0.8739,
+      "step": 6650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.57426384229546e-05,
+      "loss": 0.88,
+      "step": 6660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.572851636169484e-05,
+      "loss": 0.8732,
+      "step": 6670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.5714373716749886e-05,
+      "loss": 0.8738,
+      "step": 6680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.5700210506628026e-05,
+      "loss": 0.8734,
+      "step": 6690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.5686026749864466e-05,
+      "loss": 0.8692,
+      "step": 6700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.567182246502127e-05,
+      "loss": 0.8713,
+      "step": 6710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.56575976706874e-05,
+      "loss": 0.8797,
+      "step": 6720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.5643352385478616e-05,
+      "loss": 0.8732,
+      "step": 6730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.562908662803754e-05,
+      "loss": 0.8665,
+      "step": 6740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.5614800417033565e-05,
+      "loss": 0.8634,
+      "step": 6750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.560049377116284e-05,
+      "loss": 0.872,
+      "step": 6760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.558616670914828e-05,
+      "loss": 0.8687,
+      "step": 6770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.55718192497395e-05,
+      "loss": 0.8679,
+      "step": 6780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.555745141171282e-05,
+      "loss": 0.8648,
+      "step": 6790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.554306321387122e-05,
+      "loss": 0.87,
+      "step": 6800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.552865467504432e-05,
+      "loss": 0.8689,
+      "step": 6810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.55142258140884e-05,
+      "loss": 0.8704,
+      "step": 6820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.549977664988628e-05,
+      "loss": 0.8718,
+      "step": 6830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.548530720134738e-05,
+      "loss": 0.8733,
+      "step": 6840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.547081748740766e-05,
+      "loss": 0.8699,
+      "step": 6850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.5456307527029606e-05,
+      "loss": 0.8721,
+      "step": 6860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.54417773392022e-05,
+      "loss": 0.8687,
+      "step": 6870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.542722694294088e-05,
+      "loss": 0.8551,
+      "step": 6880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.541265635728755e-05,
+      "loss": 0.8632,
+      "step": 6890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.539806560131053e-05,
+      "loss": 0.8739,
+      "step": 6900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.538345469410453e-05,
+      "loss": 0.8674,
+      "step": 6910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.536882365479063e-05,
+      "loss": 0.8617,
+      "step": 6920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.535417250251627e-05,
+      "loss": 0.8618,
+      "step": 6930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.5339501256455216e-05,
+      "loss": 0.8731,
+      "step": 6940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.53248099358075e-05,
+      "loss": 0.8584,
+      "step": 6950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.531009855979945e-05,
+      "loss": 0.8692,
+      "step": 6960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.529536714768363e-05,
+      "loss": 0.8698,
+      "step": 6970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.528061571873883e-05,
+      "loss": 0.8577,
+      "step": 6980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.526584429227005e-05,
+      "loss": 0.8652,
+      "step": 6990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.525105288760843e-05,
+      "loss": 0.8676,
+      "step": 7000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.5236241524111264e-05,
+      "loss": 0.8682,
+      "step": 7010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.522141022116199e-05,
+      "loss": 0.8664,
+      "step": 7020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.520655899817012e-05,
+      "loss": 0.8714,
+      "step": 7030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.519168787457122e-05,
+      "loss": 0.8672,
+      "step": 7040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.5176796869826943e-05,
+      "loss": 0.8731,
+      "step": 7050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.5161886003424915e-05,
+      "loss": 0.8679,
+      "step": 7060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.5146955294878784e-05,
+      "loss": 0.8692,
+      "step": 7070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.5132004763728144e-05,
+      "loss": 0.8751,
+      "step": 7080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.511703442953856e-05,
+      "loss": 0.8656,
+      "step": 7090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.510204431190149e-05,
+      "loss": 0.8742,
+      "step": 7100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.508703443043427e-05,
+      "loss": 0.87,
+      "step": 7110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.5072004804780135e-05,
+      "loss": 0.8608,
+      "step": 7120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.505695545460814e-05,
+      "loss": 0.8727,
+      "step": 7130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.504188639961315e-05,
+      "loss": 0.8731,
+      "step": 7140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.502679765951583e-05,
+      "loss": 0.871,
+      "step": 7150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.501168925406259e-05,
+      "loss": 0.8688,
+      "step": 7160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.49965612030256e-05,
+      "loss": 0.8682,
+      "step": 7170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.4981413526202704e-05,
+      "loss": 0.8718,
+      "step": 7180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.4966246243417465e-05,
+      "loss": 0.8663,
+      "step": 7190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.495105937451908e-05,
+      "loss": 0.861,
+      "step": 7200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.4935852939382396e-05,
+      "loss": 0.8612,
+      "step": 7210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.492062695790786e-05,
+      "loss": 0.8704,
+      "step": 7220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.4905381450021484e-05,
+      "loss": 0.8608,
+      "step": 7230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.489011643567486e-05,
+      "loss": 0.8659,
+      "step": 7240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.4874831934845095e-05,
+      "loss": 0.8692,
+      "step": 7250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.485952796753479e-05,
+      "loss": 0.8565,
+      "step": 7260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.484420455377202e-05,
+      "loss": 0.8642,
+      "step": 7270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.482886171361034e-05,
+      "loss": 0.8631,
+      "step": 7280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.481349946712869e-05,
+      "loss": 0.8671,
+      "step": 7290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4798117834431436e-05,
+      "loss": 0.8676,
+      "step": 7300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.478271683564829e-05,
+      "loss": 0.8689,
+      "step": 7310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4767296490934324e-05,
+      "loss": 0.869,
+      "step": 7320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.475185682046992e-05,
+      "loss": 0.8651,
+      "step": 7330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4736397844460766e-05,
+      "loss": 0.8637,
+      "step": 7340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4720919583137786e-05,
+      "loss": 0.8707,
+      "step": 7350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.470542205675717e-05,
+      "loss": 0.8607,
+      "step": 7360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4689905285600314e-05,
+      "loss": 0.8707,
+      "step": 7370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.467436928997379e-05,
+      "loss": 0.8762,
+      "step": 7380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.465881409020933e-05,
+      "loss": 0.8809,
+      "step": 7390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4643239706663813e-05,
+      "loss": 0.8796,
+      "step": 7400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.46276461597192e-05,
+      "loss": 0.8746,
+      "step": 7410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4612033469782555e-05,
+      "loss": 0.8796,
+      "step": 7420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.4596401657285975e-05,
+      "loss": 0.8775,
+      "step": 7430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.458075074268659e-05,
+      "loss": 0.871,
+      "step": 7440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.456508074646654e-05,
+      "loss": 0.8694,
+      "step": 7450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.4549391689132914e-05,
+      "loss": 0.8706,
+      "step": 7460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.453368359121775e-05,
+      "loss": 0.8639,
+      "step": 7470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.4517956473278037e-05,
+      "loss": 0.8688,
+      "step": 7480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.4502210355895615e-05,
+      "loss": 0.8723,
+      "step": 7490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.44864452596772e-05,
+      "loss": 0.8668,
+      "step": 7500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.4470661205254354e-05,
+      "loss": 0.8661,
+      "step": 7510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.445485821328345e-05,
+      "loss": 0.8662,
+      "step": 7520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.443903630444562e-05,
+      "loss": 0.8715,
+      "step": 7530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.4423195499446796e-05,
+      "loss": 0.8683,
+      "step": 7540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.440733581901759e-05,
+      "loss": 0.868,
+      "step": 7550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.439145728391335e-05,
+      "loss": 0.8663,
+      "step": 7560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.437555991491409e-05,
+      "loss": 0.8629,
+      "step": 7570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.435964373282447e-05,
+      "loss": 0.8564,
+      "step": 7580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.434370875847377e-05,
+      "loss": 0.8574,
+      "step": 7590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.432775501271586e-05,
+      "loss": 0.8704,
+      "step": 7600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.431178251642918e-05,
+      "loss": 0.8659,
+      "step": 7610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.429579129051672e-05,
+      "loss": 0.865,
+      "step": 7620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.4279781355905956e-05,
+      "loss": 0.8623,
+      "step": 7630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.4263752733548866e-05,
+      "loss": 0.8667,
+      "step": 7640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.424770544442187e-05,
+      "loss": 0.8646,
+      "step": 7650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.423163950952585e-05,
+      "loss": 0.8646,
+      "step": 7660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.421555494988605e-05,
+      "loss": 0.8655,
+      "step": 7670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.4199451786552094e-05,
+      "loss": 0.8669,
+      "step": 7680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.418333004059798e-05,
+      "loss": 0.8621,
+      "step": 7690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.4167189733122e-05,
+      "loss": 0.8618,
+      "step": 7700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.415103088524675e-05,
+      "loss": 0.8609,
+      "step": 7710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.413485351811908e-05,
+      "loss": 0.8595,
+      "step": 7720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.411865765291008e-05,
+      "loss": 0.8593,
+      "step": 7730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.410244331081503e-05,
+      "loss": 0.8648,
+      "step": 7740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.4086210513053446e-05,
+      "loss": 0.8623,
+      "step": 7750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.4069959280868934e-05,
+      "loss": 0.8567,
+      "step": 7760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.405368963552925e-05,
+      "loss": 0.8696,
+      "step": 7770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.403740159832625e-05,
+      "loss": 0.8615,
+      "step": 7780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.4021095190575865e-05,
+      "loss": 0.8553,
+      "step": 7790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.400477043361805e-05,
+      "loss": 0.8628,
+      "step": 7800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.3988427348816775e-05,
+      "loss": 0.8558,
+      "step": 7810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.397206595756001e-05,
+      "loss": 0.8637,
+      "step": 7820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.395568628125968e-05,
+      "loss": 0.8641,
+      "step": 7830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.393928834135163e-05,
+      "loss": 0.8641,
+      "step": 7840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.39228721592956e-05,
+      "loss": 0.8638,
+      "step": 7850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.390643775657523e-05,
+      "loss": 0.8666,
+      "step": 7860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.388998515469798e-05,
+      "loss": 0.8678,
+      "step": 7870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.387351437519513e-05,
+      "loss": 0.8633,
+      "step": 7880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.385702543962176e-05,
+      "loss": 0.857,
+      "step": 7890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.384051836955672e-05,
+      "loss": 0.8629,
+      "step": 7900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.382399318660255e-05,
+      "loss": 0.8648,
+      "step": 7910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.3807449912385535e-05,
+      "loss": 0.8627,
+      "step": 7920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.379088856855562e-05,
+      "loss": 0.8617,
+      "step": 7930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.377430917678641e-05,
+      "loss": 0.855,
+      "step": 7940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.37577117587751e-05,
+      "loss": 0.8582,
+      "step": 7950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.374109633624251e-05,
+      "loss": 0.861,
+      "step": 7960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.3724462930933e-05,
+      "loss": 0.8596,
+      "step": 7970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.370781156461447e-05,
+      "loss": 0.8588,
+      "step": 7980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.369114225907833e-05,
+      "loss": 0.857,
+      "step": 7990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.3674455036139455e-05,
+      "loss": 0.8584,
+      "step": 8000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.365774991763618e-05,
+      "loss": 0.861,
+      "step": 8010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.364102692543026e-05,
+      "loss": 0.8559,
+      "step": 8020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.362428608140682e-05,
+      "loss": 0.8608,
+      "step": 8030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.3607527407474395e-05,
+      "loss": 0.8532,
+      "step": 8040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.3590750925564794e-05,
+      "loss": 0.8544,
+      "step": 8050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.357395665763317e-05,
+      "loss": 0.8605,
+      "step": 8060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.355714462565795e-05,
+      "loss": 0.8473,
+      "step": 8070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.354031485164078e-05,
+      "loss": 0.8579,
+      "step": 8080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.352346735760656e-05,
+      "loss": 0.8528,
+      "step": 8090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.3506602165603365e-05,
+      "loss": 0.8618,
+      "step": 8100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.348971929770243e-05,
+      "loss": 0.8677,
+      "step": 8110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.347281877599813e-05,
+      "loss": 0.8551,
+      "step": 8120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.345590062260792e-05,
+      "loss": 0.8561,
+      "step": 8130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.3438964859672364e-05,
+      "loss": 0.8535,
+      "step": 8140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.342201150935504e-05,
+      "loss": 0.8597,
+      "step": 8150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.340504059384256e-05,
+      "loss": 0.8583,
+      "step": 8160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.338805213534453e-05,
+      "loss": 0.8661,
+      "step": 8170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.337104615609349e-05,
+      "loss": 0.863,
+      "step": 8180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.335402267834492e-05,
+      "loss": 0.8563,
+      "step": 8190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.3336981724377214e-05,
+      "loss": 0.8523,
+      "step": 8200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.331992331649163e-05,
+      "loss": 0.8642,
+      "step": 8210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.3302847477012246e-05,
+      "loss": 0.858,
+      "step": 8220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.328575422828598e-05,
+      "loss": 0.8653,
+      "step": 8230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.326864359268251e-05,
+      "loss": 0.8537,
+      "step": 8240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.325151559259431e-05,
+      "loss": 0.8556,
+      "step": 8250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.323437025043653e-05,
+      "loss": 0.8584,
+      "step": 8260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.321720758864703e-05,
+      "loss": 0.8562,
+      "step": 8270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.320002762968635e-05,
+      "loss": 0.8572,
+      "step": 8280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.318283039603765e-05,
+      "loss": 0.8592,
+      "step": 8290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.316561591020671e-05,
+      "loss": 0.8488,
+      "step": 8300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.314838419472189e-05,
+      "loss": 0.8493,
+      "step": 8310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.3131135272134076e-05,
+      "loss": 0.8506,
+      "step": 8320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.31138691650167e-05,
+      "loss": 0.8557,
+      "step": 8330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.309658589596565e-05,
+      "loss": 0.8631,
+      "step": 8340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.3079285487599326e-05,
+      "loss": 0.8636,
+      "step": 8350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.306196796255849e-05,
+      "loss": 0.8544,
+      "step": 8360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.304463334350637e-05,
+      "loss": 0.8587,
+      "step": 8370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.302728165312852e-05,
+      "loss": 0.8607,
+      "step": 8380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.300991291413285e-05,
+      "loss": 0.8569,
+      "step": 8390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.299252714924958e-05,
+      "loss": 0.8508,
+      "step": 8400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.297512438123122e-05,
+      "loss": 0.8474,
+      "step": 8410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.295770463285252e-05,
+      "loss": 0.8497,
+      "step": 8420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.2940267926910455e-05,
+      "loss": 0.859,
+      "step": 8430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.29228142862242e-05,
+      "loss": 0.8549,
+      "step": 8440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.290534373363507e-05,
+      "loss": 0.8596,
+      "step": 8450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.288785629200655e-05,
+      "loss": 0.8524,
+      "step": 8460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.28703519842242e-05,
+      "loss": 0.8589,
+      "step": 8470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.285283083319565e-05,
+      "loss": 0.8562,
+      "step": 8480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.283529286185058e-05,
+      "loss": 0.8591,
+      "step": 8490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.2817738093140685e-05,
+      "loss": 0.8594,
+      "step": 8500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.280016655003964e-05,
+      "loss": 0.8512,
+      "step": 8510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.2782578255543084e-05,
+      "loss": 0.8542,
+      "step": 8520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.2764973232668555e-05,
+      "loss": 0.8552,
+      "step": 8530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.27473515044555e-05,
+      "loss": 0.8498,
+      "step": 8540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.272971309396522e-05,
+      "loss": 0.8549,
+      "step": 8550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.271205802428086e-05,
+      "loss": 0.8541,
+      "step": 8560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.269438631850735e-05,
+      "loss": 0.8507,
+      "step": 8570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.26766979997714e-05,
+      "loss": 0.86,
+      "step": 8580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.265899309122147e-05,
+      "loss": 0.8541,
+      "step": 8590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2641271616027705e-05,
+      "loss": 0.8491,
+      "step": 8600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.262353359738196e-05,
+      "loss": 0.8591,
+      "step": 8610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.260577905849772e-05,
+      "loss": 0.8558,
+      "step": 8620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.258800802261011e-05,
+      "loss": 0.8472,
+      "step": 8630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.257022051297581e-05,
+      "loss": 0.8519,
+      "step": 8640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2552416552873096e-05,
+      "loss": 0.8617,
+      "step": 8650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.253459616560175e-05,
+      "loss": 0.8605,
+      "step": 8660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2516759374483066e-05,
+      "loss": 0.8592,
+      "step": 8670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.249890620285979e-05,
+      "loss": 0.861,
+      "step": 8680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2481036674096116e-05,
+      "loss": 0.8666,
+      "step": 8690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.246315081157765e-05,
+      "loss": 0.8559,
+      "step": 8700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.244524863871135e-05,
+      "loss": 0.8587,
+      "step": 8710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.242733017892555e-05,
+      "loss": 0.8553,
+      "step": 8720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2409395455669866e-05,
+      "loss": 0.8505,
+      "step": 8730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.239144449241523e-05,
+      "loss": 0.8571,
+      "step": 8740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.23734773126538e-05,
+      "loss": 0.8586,
+      "step": 8750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2355493939898976e-05,
+      "loss": 0.87,
+      "step": 8760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.233749439768534e-05,
+      "loss": 0.8534,
+      "step": 8770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.231947870956864e-05,
+      "loss": 0.8524,
+      "step": 8780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2301446899125746e-05,
+      "loss": 0.8516,
+      "step": 8790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2283398989954623e-05,
+      "loss": 0.8598,
+      "step": 8800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.226533500567433e-05,
+      "loss": 0.8614,
+      "step": 8810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.224725496992493e-05,
+      "loss": 0.8446,
+      "step": 8820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2229158906367515e-05,
+      "loss": 0.8587,
+      "step": 8830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.221104683868415e-05,
+      "loss": 0.8602,
+      "step": 8840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.219291879057783e-05,
+      "loss": 0.8543,
+      "step": 8850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2174774785772487e-05,
+      "loss": 0.8479,
+      "step": 8860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2156614848012905e-05,
+      "loss": 0.8525,
+      "step": 8870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2138439001064745e-05,
+      "loss": 0.852,
+      "step": 8880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2120247268714473e-05,
+      "loss": 0.8557,
+      "step": 8890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.210203967476936e-05,
+      "loss": 0.8597,
+      "step": 8900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2083816243057415e-05,
+      "loss": 0.8491,
+      "step": 8910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.206557699742739e-05,
+      "loss": 0.855,
+      "step": 8920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.204732196174872e-05,
+      "loss": 0.8586,
+      "step": 8930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2029051159911513e-05,
+      "loss": 0.8527,
+      "step": 8940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.201076461582651e-05,
+      "loss": 0.8521,
+      "step": 8950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.199246235342504e-05,
+      "loss": 0.8504,
+      "step": 8960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.197414439665902e-05,
+      "loss": 0.8455,
+      "step": 8970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1955810769500915e-05,
+      "loss": 0.8501,
+      "step": 8980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.193746149594365e-05,
+      "loss": 0.8529,
+      "step": 8990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1919096600000674e-05,
+      "loss": 0.8473,
+      "step": 9000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.190071610570587e-05,
+      "loss": 0.8482,
+      "step": 9010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.188232003711353e-05,
+      "loss": 0.8533,
+      "step": 9020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1863908418298304e-05,
+      "loss": 0.8582,
+      "step": 9030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.184548127335524e-05,
+      "loss": 0.8495,
+      "step": 9040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1827038626399665e-05,
+      "loss": 0.8522,
+      "step": 9050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.180858050156722e-05,
+      "loss": 0.8487,
+      "step": 9060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1790106923013784e-05,
+      "loss": 0.8523,
+      "step": 9070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.177161791491546e-05,
+      "loss": 0.8602,
+      "step": 9080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.175311350146856e-05,
+      "loss": 0.8475,
+      "step": 9090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.173459370688954e-05,
+      "loss": 0.8514,
+      "step": 9100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1716058555414985e-05,
+      "loss": 0.8554,
+      "step": 9110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.16975080713016e-05,
+      "loss": 0.8494,
+      "step": 9120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.167894227882613e-05,
+      "loss": 0.8458,
+      "step": 9130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.166036120228536e-05,
+      "loss": 0.8438,
+      "step": 9140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1641764865996084e-05,
+      "loss": 0.8474,
+      "step": 9150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.162315329429506e-05,
+      "loss": 0.8478,
+      "step": 9160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1604526511538996e-05,
+      "loss": 0.8474,
+      "step": 9170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1585884542104476e-05,
+      "loss": 0.8415,
+      "step": 9180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1567227410388004e-05,
+      "loss": 0.8496,
+      "step": 9190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.154855514080589e-05,
+      "loss": 0.8499,
+      "step": 9200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.152986775779426e-05,
+      "loss": 0.8456,
+      "step": 9210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.151116528580904e-05,
+      "loss": 0.8463,
+      "step": 9220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.149244774932588e-05,
+      "loss": 0.8463,
+      "step": 9230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1473715172840154e-05,
+      "loss": 0.8533,
+      "step": 9240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.145496758086693e-05,
+      "loss": 0.8517,
+      "step": 9250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.143620499794089e-05,
+      "loss": 0.8505,
+      "step": 9260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.141742744861638e-05,
+      "loss": 0.8543,
+      "step": 9270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.139863495746731e-05,
+      "loss": 0.8473,
+      "step": 9280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.137982754908713e-05,
+      "loss": 0.8493,
+      "step": 9290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.136100524808884e-05,
+      "loss": 0.853,
+      "step": 9300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1342168079104914e-05,
+      "loss": 0.8481,
+      "step": 9310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.132331606678728e-05,
+      "loss": 0.8475,
+      "step": 9320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.130444923580732e-05,
+      "loss": 0.8443,
+      "step": 9330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.128556761085576e-05,
+      "loss": 0.8536,
+      "step": 9340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.126667121664272e-05,
+      "loss": 0.862,
+      "step": 9350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1247760077897656e-05,
+      "loss": 0.846,
+      "step": 9360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.122883421936929e-05,
+      "loss": 0.8382,
+      "step": 9370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1209893665825625e-05,
+      "loss": 0.8503,
+      "step": 9380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.119093844205391e-05,
+      "loss": 0.85,
+      "step": 9390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.117196857286055e-05,
+      "loss": 0.845,
+      "step": 9400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1152984083071165e-05,
+      "loss": 0.8503,
+      "step": 9410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.1133984997530475e-05,
+      "loss": 0.8424,
+      "step": 9420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.1114971341102306e-05,
+      "loss": 0.855,
+      "step": 9430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.109594313866958e-05,
+      "loss": 0.853,
+      "step": 9440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.107690041513422e-05,
+      "loss": 0.8463,
+      "step": 9450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.1057843195417166e-05,
+      "loss": 0.8502,
+      "step": 9460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.103877150445834e-05,
+      "loss": 0.8467,
+      "step": 9470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.1019685367216575e-05,
+      "loss": 0.8472,
+      "step": 9480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.100058480866964e-05,
+      "loss": 0.8547,
+      "step": 9490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.098146985381416e-05,
+      "loss": 0.8431,
+      "step": 9500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.096234052766561e-05,
+      "loss": 0.8443,
+      "step": 9510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0943196855258255e-05,
+      "loss": 0.8513,
+      "step": 9520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0924038861645155e-05,
+      "loss": 0.8471,
+      "step": 9530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0904866571898096e-05,
+      "loss": 0.8477,
+      "step": 9540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.088568001110759e-05,
+      "loss": 0.8484,
+      "step": 9550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.0866479204382805e-05,
+      "loss": 0.8471,
+      "step": 9560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.0847264176851575e-05,
+      "loss": 0.8402,
+      "step": 9570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.0828034953660324e-05,
+      "loss": 0.8387,
+      "step": 9580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.080879155997406e-05,
+      "loss": 0.8419,
+      "step": 9590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.078953402097635e-05,
+      "loss": 0.8425,
+      "step": 9600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.077026236186925e-05,
+      "loss": 0.8519,
+      "step": 9610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.07509766078733e-05,
+      "loss": 0.8449,
+      "step": 9620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.073167678422752e-05,
+      "loss": 0.8422,
+      "step": 9630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.0712362916189274e-05,
+      "loss": 0.847,
+      "step": 9640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.069303502903437e-05,
+      "loss": 0.8407,
+      "step": 9650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.0673693148056946e-05,
+      "loss": 0.8537,
+      "step": 9660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.065433729856943e-05,
+      "loss": 0.8499,
+      "step": 9670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.063496750590255e-05,
+      "loss": 0.847,
+      "step": 9680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.061558379540527e-05,
+      "loss": 0.8511,
+      "step": 9690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.059618619244479e-05,
+      "loss": 0.8594,
+      "step": 9700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0576774722406466e-05,
+      "loss": 0.8452,
+      "step": 9710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.055734941069383e-05,
+      "loss": 0.8435,
+      "step": 9720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.053791028272849e-05,
+      "loss": 0.8431,
+      "step": 9730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0518457363950154e-05,
+      "loss": 0.8477,
+      "step": 9740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0498990679816602e-05,
+      "loss": 0.8436,
+      "step": 9750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0479510255803587e-05,
+      "loss": 0.8414,
+      "step": 9760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.046001611740487e-05,
+      "loss": 0.8352,
+      "step": 9770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.044050829013215e-05,
+      "loss": 0.8465,
+      "step": 9780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0420986799515037e-05,
+      "loss": 0.8419,
+      "step": 9790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0401451671101036e-05,
+      "loss": 0.8477,
+      "step": 9800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.038190293045548e-05,
+      "loss": 0.8451,
+      "step": 9810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0362340603161552e-05,
+      "loss": 0.8537,
+      "step": 9820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.0342764714820166e-05,
+      "loss": 0.8401,
+      "step": 9830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0323175291050014e-05,
+      "loss": 0.8411,
+      "step": 9840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.03035723574875e-05,
+      "loss": 0.8516,
+      "step": 9850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.02839559397867e-05,
+      "loss": 0.8453,
+      "step": 9860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0264326063619346e-05,
+      "loss": 0.8418,
+      "step": 9870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.024468275467477e-05,
+      "loss": 0.8473,
+      "step": 9880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0225026038659892e-05,
+      "loss": 0.8435,
+      "step": 9890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0205355941299187e-05,
+      "loss": 0.8538,
+      "step": 9900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0185672488334607e-05,
+      "loss": 0.8559,
+      "step": 9910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0165975705525628e-05,
+      "loss": 0.8616,
+      "step": 9920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.014626561864914e-05,
+      "loss": 0.8619,
+      "step": 9930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0126542253499463e-05,
+      "loss": 0.8567,
+      "step": 9940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.010680563588827e-05,
+      "loss": 0.8562,
+      "step": 9950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0087055791644594e-05,
+      "loss": 0.8564,
+      "step": 9960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0067292746614776e-05,
+      "loss": 0.8443,
+      "step": 9970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.0047516526662437e-05,
+      "loss": 0.8505,
+      "step": 9980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.0027727157668428e-05,
+      "loss": 0.8452,
+      "step": 9990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.000792466553082e-05,
+      "loss": 0.8473,
+      "step": 10000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9988109076164843e-05,
+      "loss": 0.8509,
+      "step": 10010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9968280415502886e-05,
+      "loss": 0.8457,
+      "step": 10020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9948438709494424e-05,
+      "loss": 0.8476,
+      "step": 10030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9928583984106026e-05,
+      "loss": 0.85,
+      "step": 10040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.990871626532128e-05,
+      "loss": 0.8502,
+      "step": 10050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.988883557914079e-05,
+      "loss": 0.8528,
+      "step": 10060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.986894195158213e-05,
+      "loss": 0.8436,
+      "step": 10070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.98490354086798e-05,
+      "loss": 0.838,
+      "step": 10080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9829115976485213e-05,
+      "loss": 0.85,
+      "step": 10090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9809183681066646e-05,
+      "loss": 0.848,
+      "step": 10100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9789238548509202e-05,
+      "loss": 0.8383,
+      "step": 10110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9769280604914805e-05,
+      "loss": 0.8451,
+      "step": 10120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9749309876402122e-05,
+      "loss": 0.8435,
+      "step": 10130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9729326389106564e-05,
+      "loss": 0.8475,
+      "step": 10140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9709330169180238e-05,
+      "loss": 0.8423,
+      "step": 10150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9689321242791906e-05,
+      "loss": 0.846,
+      "step": 10160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9669299636126973e-05,
+      "loss": 0.845,
+      "step": 10170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.964926537538742e-05,
+      "loss": 0.843,
+      "step": 10180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9629218486791807e-05,
+      "loss": 0.847,
+      "step": 10190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9609158996575207e-05,
+      "loss": 0.846,
+      "step": 10200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9589086930989196e-05,
+      "loss": 0.8472,
+      "step": 10210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.956900231630179e-05,
+      "loss": 0.8405,
+      "step": 10220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9548905178797442e-05,
+      "loss": 0.8372,
+      "step": 10230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.952879554477699e-05,
+      "loss": 0.844,
+      "step": 10240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9508673440557625e-05,
+      "loss": 0.8434,
+      "step": 10250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.948853889247286e-05,
+      "loss": 0.8393,
+      "step": 10260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9468391926872487e-05,
+      "loss": 0.8438,
+      "step": 10270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9448232570122557e-05,
+      "loss": 0.8412,
+      "step": 10280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.942806084860533e-05,
+      "loss": 0.8415,
+      "step": 10290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9407876788719253e-05,
+      "loss": 0.8446,
+      "step": 10300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9387680416878915e-05,
+      "loss": 0.8449,
+      "step": 10310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.936747175951502e-05,
+      "loss": 0.8492,
+      "step": 10320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.934725084307435e-05,
+      "loss": 0.8442,
+      "step": 10330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9327017694019737e-05,
+      "loss": 0.843,
+      "step": 10340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.930677233883e-05,
+      "loss": 0.848,
+      "step": 10350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9286514803999963e-05,
+      "loss": 0.8436,
+      "step": 10360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.926624511604036e-05,
+      "loss": 0.8414,
+      "step": 10370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.924596330147785e-05,
+      "loss": 0.8418,
+      "step": 10380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9225669386854955e-05,
+      "loss": 0.8421,
+      "step": 10390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9205363398730032e-05,
+      "loss": 0.8387,
+      "step": 10400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.918504536367723e-05,
+      "loss": 0.8409,
+      "step": 10410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.9164715308286474e-05,
+      "loss": 0.8445,
+      "step": 10420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.914437325916343e-05,
+      "loss": 0.8413,
+      "step": 10430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.912401924292944e-05,
+      "loss": 0.8371,
+      "step": 10440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.9103653286221508e-05,
+      "loss": 0.8427,
+      "step": 10450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.908327541569228e-05,
+      "loss": 0.838,
+      "step": 10460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.9062885658009978e-05,
+      "loss": 0.8401,
+      "step": 10470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.9042484039858397e-05,
+      "loss": 0.8333,
+      "step": 10480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.9022070587936837e-05,
+      "loss": 0.8424,
+      "step": 10490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.9001645328960092e-05,
+      "loss": 0.8416,
+      "step": 10500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.898120828965841e-05,
+      "loss": 0.8489,
+      "step": 10510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8960759496777444e-05,
+      "loss": 0.8442,
+      "step": 10520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8940298977078254e-05,
+      "loss": 0.8352,
+      "step": 10530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8919826757337218e-05,
+      "loss": 0.8413,
+      "step": 10540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.889934286434604e-05,
+      "loss": 0.8369,
+      "step": 10550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.88788473249117e-05,
+      "loss": 0.8431,
+      "step": 10560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.885834016585642e-05,
+      "loss": 0.8444,
+      "step": 10570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.883782141401763e-05,
+      "loss": 0.8423,
+      "step": 10580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8817291096247918e-05,
+      "loss": 0.837,
+      "step": 10590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8796749239415027e-05,
+      "loss": 0.8352,
+      "step": 10600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8776195870401787e-05,
+      "loss": 0.8348,
+      "step": 10610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8755631016106095e-05,
+      "loss": 0.8442,
+      "step": 10620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8735054703440903e-05,
+      "loss": 0.8427,
+      "step": 10630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.871446695933411e-05,
+      "loss": 0.8401,
+      "step": 10640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.869386781072863e-05,
+      "loss": 0.8367,
+      "step": 10650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.867325728458225e-05,
+      "loss": 0.8501,
+      "step": 10660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.86526354078677e-05,
+      "loss": 0.8445,
+      "step": 10670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.863200220757251e-05,
+      "loss": 0.8357,
+      "step": 10680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8611357710699065e-05,
+      "loss": 0.8394,
+      "step": 10690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.859070194426452e-05,
+      "loss": 0.8384,
+      "step": 10700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.85700349353008e-05,
+      "loss": 0.83,
+      "step": 10710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.85493567108545e-05,
+      "loss": 0.8343,
+      "step": 10720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.852866729798693e-05,
+      "loss": 0.8314,
+      "step": 10730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8507966723774033e-05,
+      "loss": 0.8405,
+      "step": 10740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.848725501530635e-05,
+      "loss": 0.843,
+      "step": 10750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8466532199689e-05,
+      "loss": 0.8309,
+      "step": 10760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8445798304041635e-05,
+      "loss": 0.8388,
+      "step": 10770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8425053355498408e-05,
+      "loss": 0.8385,
+      "step": 10780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8404297381207944e-05,
+      "loss": 0.8487,
+      "step": 10790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8383530408333285e-05,
+      "loss": 0.8389,
+      "step": 10800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8362752464051873e-05,
+      "loss": 0.836,
+      "step": 10810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8341963575555513e-05,
+      "loss": 0.8313,
+      "step": 10820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8321163770050312e-05,
+      "loss": 0.8407,
+      "step": 10830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8300353074756698e-05,
+      "loss": 0.8298,
+      "step": 10840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8279531516909314e-05,
+      "loss": 0.8368,
+      "step": 10850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8258699123757048e-05,
+      "loss": 0.8335,
+      "step": 10860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8237855922562956e-05,
+      "loss": 0.8334,
+      "step": 10870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8217001940604234e-05,
+      "loss": 0.8394,
+      "step": 10880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.819613720517219e-05,
+      "loss": 0.836,
+      "step": 10890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8175261743572207e-05,
+      "loss": 0.8328,
+      "step": 10900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8154375583123706e-05,
+      "loss": 0.8367,
+      "step": 10910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8133478751160104e-05,
+      "loss": 0.8341,
+      "step": 10920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.811257127502879e-05,
+      "loss": 0.8346,
+      "step": 10930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8091653182091083e-05,
+      "loss": 0.8342,
+      "step": 10940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8070724499722186e-05,
+      "loss": 0.8395,
+      "step": 10950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.804978525531117e-05,
+      "loss": 0.8364,
+      "step": 10960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8028835476260928e-05,
+      "loss": 0.8338,
+      "step": 10970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8007875189988136e-05,
+      "loss": 0.8367,
+      "step": 10980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7986904423923218e-05,
+      "loss": 0.8436,
+      "step": 10990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7965923205510324e-05,
+      "loss": 0.8386,
+      "step": 11000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.794493156220727e-05,
+      "loss": 0.8293,
+      "step": 11010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7923929521485525e-05,
+      "loss": 0.8347,
+      "step": 11020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.790291711083015e-05,
+      "loss": 0.8416,
+      "step": 11030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7881894357739803e-05,
+      "loss": 0.8335,
+      "step": 11040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7860861289726647e-05,
+      "loss": 0.8368,
+      "step": 11050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.783981793431637e-05,
+      "loss": 0.8385,
+      "step": 11060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7818764319048102e-05,
+      "loss": 0.8316,
+      "step": 11070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7797700471474415e-05,
+      "loss": 0.839,
+      "step": 11080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7776626419161263e-05,
+      "loss": 0.8286,
+      "step": 11090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7755542189687957e-05,
+      "loss": 0.8357,
+      "step": 11100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.773444781064713e-05,
+      "loss": 0.8369,
+      "step": 11110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7713343309644693e-05,
+      "loss": 0.8322,
+      "step": 11120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7692228714299814e-05,
+      "loss": 0.8313,
+      "step": 11130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.767110405224485e-05,
+      "loss": 0.8384,
+      "step": 11140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.764996935112535e-05,
+      "loss": 0.8403,
+      "step": 11150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7628824638600004e-05,
+      "loss": 0.8576,
+      "step": 11160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7607669942340588e-05,
+      "loss": 0.8731,
+      "step": 11170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7586505290031946e-05,
+      "loss": 0.8715,
+      "step": 11180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.756533070937197e-05,
+      "loss": 0.8995,
+      "step": 11190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7544146228071524e-05,
+      "loss": 0.8774,
+      "step": 11200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.752295187385444e-05,
+      "loss": 0.8696,
+      "step": 11210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.7501747674457462e-05,
+      "loss": 0.8675,
+      "step": 11220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.748053365763023e-05,
+      "loss": 0.8541,
+      "step": 11230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.745930985113522e-05,
+      "loss": 0.8546,
+      "step": 11240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.743807628274773e-05,
+      "loss": 0.8482,
+      "step": 11250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7416832980255823e-05,
+      "loss": 0.8488,
+      "step": 11260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7395579971460313e-05,
+      "loss": 0.8427,
+      "step": 11270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7374317284174705e-05,
+      "loss": 0.8478,
+      "step": 11280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7353044946225172e-05,
+      "loss": 0.8459,
+      "step": 11290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.733176298545053e-05,
+      "loss": 0.8439,
+      "step": 11300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.731047142970216e-05,
+      "loss": 0.8401,
+      "step": 11310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.728917030684404e-05,
+      "loss": 0.8487,
+      "step": 11320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.726785964475262e-05,
+      "loss": 0.848,
+      "step": 11330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7246539471316885e-05,
+      "loss": 0.8441,
+      "step": 11340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.722520981443823e-05,
+      "loss": 0.8377,
+      "step": 11350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.720387070203047e-05,
+      "loss": 0.8417,
+      "step": 11360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7182522162019808e-05,
+      "loss": 0.8355,
+      "step": 11370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.716116422234476e-05,
+      "loss": 0.8411,
+      "step": 11380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.7139796910956165e-05,
+      "loss": 0.8437,
+      "step": 11390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7118420255817122e-05,
+      "loss": 0.8393,
+      "step": 11400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7097034284902952e-05,
+      "loss": 0.8429,
+      "step": 11410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7075639026201167e-05,
+      "loss": 0.84,
+      "step": 11420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.705423450771144e-05,
+      "loss": 0.8401,
+      "step": 11430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.703282075744556e-05,
+      "loss": 0.8366,
+      "step": 11440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.7011397803427395e-05,
+      "loss": 0.8404,
+      "step": 11450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6989965673692857e-05,
+      "loss": 0.8454,
+      "step": 11460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6968524396289874e-05,
+      "loss": 0.8451,
+      "step": 11470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.694707399927834e-05,
+      "loss": 0.8412,
+      "step": 11480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.692561451073008e-05,
+      "loss": 0.8341,
+      "step": 11490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6904145958728826e-05,
+      "loss": 0.8362,
+      "step": 11500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6882668371370156e-05,
+      "loss": 0.8362,
+      "step": 11510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6861181776761496e-05,
+      "loss": 0.8395,
+      "step": 11520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.6839686203022032e-05,
+      "loss": 0.8384,
+      "step": 11530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6818181678282725e-05,
+      "loss": 0.8412,
+      "step": 11540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.679666823068624e-05,
+      "loss": 0.8325,
+      "step": 11550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.677514588838692e-05,
+      "loss": 0.8353,
+      "step": 11560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6753614679550738e-05,
+      "loss": 0.8426,
+      "step": 11570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6732074632355286e-05,
+      "loss": 0.8359,
+      "step": 11580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6710525774989718e-05,
+      "loss": 0.8344,
+      "step": 11590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6688968135654712e-05,
+      "loss": 0.8332,
+      "step": 11600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6667401742562452e-05,
+      "loss": 0.8356,
+      "step": 11610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.664582662393656e-05,
+      "loss": 0.8238,
+      "step": 11620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6624242808012098e-05,
+      "loss": 0.8394,
+      "step": 11630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6602650323035484e-05,
+      "loss": 0.8319,
+      "step": 11640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6581049197264503e-05,
+      "loss": 0.8348,
+      "step": 11650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.6559439458968233e-05,
+      "loss": 0.8315,
+      "step": 11660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.653782113642705e-05,
+      "loss": 0.8328,
+      "step": 11670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6516194257932522e-05,
+      "loss": 0.8345,
+      "step": 11680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6494558851787445e-05,
+      "loss": 0.8422,
+      "step": 11690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6472914946305768e-05,
+      "loss": 0.8417,
+      "step": 11700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6451262569812567e-05,
+      "loss": 0.8416,
+      "step": 11710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.642960175064399e-05,
+      "loss": 0.8346,
+      "step": 11720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6407932517147244e-05,
+      "loss": 0.8358,
+      "step": 11730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6386254897680546e-05,
+      "loss": 0.8367,
+      "step": 11740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6364568920613094e-05,
+      "loss": 0.8325,
+      "step": 11750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6342874614325004e-05,
+      "loss": 0.8356,
+      "step": 11760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.632117200720732e-05,
+      "loss": 0.8281,
+      "step": 11770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.629946112766192e-05,
+      "loss": 0.8346,
+      "step": 11780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.6277742004101536e-05,
+      "loss": 0.8351,
+      "step": 11790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.625601466494967e-05,
+      "loss": 0.8401,
+      "step": 11800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.623427913864058e-05,
+      "loss": 0.8367,
+      "step": 11810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.621253545361924e-05,
+      "loss": 0.8282,
+      "step": 11820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6190783638341302e-05,
+      "loss": 0.8396,
+      "step": 11830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6169023721273047e-05,
+      "loss": 0.8293,
+      "step": 11840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6147255730891384e-05,
+      "loss": 0.836,
+      "step": 11850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.612547969568377e-05,
+      "loss": 0.8323,
+      "step": 11860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6103695644148182e-05,
+      "loss": 0.8256,
+      "step": 11870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6081903604793098e-05,
+      "loss": 0.8355,
+      "step": 11880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6060103606137458e-05,
+      "loss": 0.8313,
+      "step": 11890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6038295676710607e-05,
+      "loss": 0.8319,
+      "step": 11900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.6016479845052266e-05,
+      "loss": 0.8282,
+      "step": 11910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5994656139712504e-05,
+      "loss": 0.8308,
+      "step": 11920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.59728245892517e-05,
+      "loss": 0.8287,
+      "step": 11930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.595098522224049e-05,
+      "loss": 0.8318,
+      "step": 11940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5929138067259735e-05,
+      "loss": 0.8346,
+      "step": 11950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.5907283152900508e-05,
+      "loss": 0.8366,
+      "step": 11960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.588542050776401e-05,
+      "loss": 0.8334,
+      "step": 11970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5863550160461587e-05,
+      "loss": 0.8316,
+      "step": 11980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.584167213961464e-05,
+      "loss": 0.8317,
+      "step": 11990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5819786473854627e-05,
+      "loss": 0.8327,
+      "step": 12000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5797893191823e-05,
+      "loss": 0.8366,
+      "step": 12010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5775992322171207e-05,
+      "loss": 0.8319,
+      "step": 12020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.575408389356058e-05,
+      "loss": 0.8367,
+      "step": 12030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.573216793466238e-05,
+      "loss": 0.8278,
+      "step": 12040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.57102444741577e-05,
+      "loss": 0.8318,
+      "step": 12050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5688313540737473e-05,
+      "loss": 0.83,
+      "step": 12060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5666375163102388e-05,
+      "loss": 0.8265,
+      "step": 12070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5644429369962894e-05,
+      "loss": 0.8296,
+      "step": 12080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5622476190039135e-05,
+      "loss": 0.8326,
+      "step": 12090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.560051565206092e-05,
+      "loss": 0.8325,
+      "step": 12100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5578547784767702e-05,
+      "loss": 0.8348,
+      "step": 12110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.555657261690851e-05,
+      "loss": 0.8244,
+      "step": 12120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5534590177241936e-05,
+      "loss": 0.8249,
+      "step": 12130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5512600494536077e-05,
+      "loss": 0.8305,
+      "step": 12140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5490603597568528e-05,
+      "loss": 0.8331,
+      "step": 12150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.546859951512631e-05,
+      "loss": 0.8354,
+      "step": 12160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5446588276005857e-05,
+      "loss": 0.8248,
+      "step": 12170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.542456990901295e-05,
+      "loss": 0.8329,
+      "step": 12180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5402544442962724e-05,
+      "loss": 0.8352,
+      "step": 12190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5380511906679587e-05,
+      "loss": 0.8318,
+      "step": 12200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5358472328997214e-05,
+      "loss": 0.8321,
+      "step": 12210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.533642573875848e-05,
+      "loss": 0.8292,
+      "step": 12220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.531437216481544e-05,
+      "loss": 0.8294,
+      "step": 12230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5292311636029293e-05,
+      "loss": 0.8348,
+      "step": 12240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.527024418127035e-05,
+      "loss": 0.8381,
+      "step": 12250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5248169829417957e-05,
+      "loss": 0.8278,
+      "step": 12260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.522608860936051e-05,
+      "loss": 0.8273,
+      "step": 12270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5204000549995397e-05,
+      "loss": 0.829,
+      "step": 12280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5181905680228935e-05,
+      "loss": 0.8233,
+      "step": 12290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5159804028976367e-05,
+      "loss": 0.8286,
+      "step": 12300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.513769562516181e-05,
+      "loss": 0.8304,
+      "step": 12310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5115580497718223e-05,
+      "loss": 0.8371,
+      "step": 12320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5093458675587344e-05,
+      "loss": 0.8297,
+      "step": 12330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.50713301877197e-05,
+      "loss": 0.8313,
+      "step": 12340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5049195063074522e-05,
+      "loss": 0.8318,
+      "step": 12350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5027053330619735e-05,
+      "loss": 0.8312,
+      "step": 12360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5004905019331903e-05,
+      "loss": 0.8293,
+      "step": 12370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.498275015819621e-05,
+      "loss": 0.8235,
+      "step": 12380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.49605887762064e-05,
+      "loss": 0.8282,
+      "step": 12390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4938420902364774e-05,
+      "loss": 0.8285,
+      "step": 12400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4916246565682097e-05,
+      "loss": 0.8278,
+      "step": 12410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4894065795177607e-05,
+      "loss": 0.8361,
+      "step": 12420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4871878619878973e-05,
+      "loss": 0.833,
+      "step": 12430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.484968506882223e-05,
+      "loss": 0.8351,
+      "step": 12440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4827485171051756e-05,
+      "loss": 0.8288,
+      "step": 12450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.480527895562025e-05,
+      "loss": 0.8269,
+      "step": 12460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.478306645158867e-05,
+      "loss": 0.824,
+      "step": 12470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4760847688026198e-05,
+      "loss": 0.8285,
+      "step": 12480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.473862269401021e-05,
+      "loss": 0.8401,
+      "step": 12490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4716391498626244e-05,
+      "loss": 0.8317,
+      "step": 12500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4694154130967945e-05,
+      "loss": 0.8345,
+      "step": 12510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4671910620137053e-05,
+      "loss": 0.8313,
+      "step": 12520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4649660995243312e-05,
+      "loss": 0.8277,
+      "step": 12530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.46274052854045e-05,
+      "loss": 0.8225,
+      "step": 12540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4605143519746352e-05,
+      "loss": 0.8263,
+      "step": 12550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4582875727402516e-05,
+      "loss": 0.8237,
+      "step": 12560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4560601937514537e-05,
+      "loss": 0.8312,
+      "step": 12570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4538322179231804e-05,
+      "loss": 0.8357,
+      "step": 12580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.451603648171153e-05,
+      "loss": 0.8255,
+      "step": 12590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4493744874118668e-05,
+      "loss": 0.8312,
+      "step": 12600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4471447385625945e-05,
+      "loss": 0.8232,
+      "step": 12610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4449144045413763e-05,
+      "loss": 0.8255,
+      "step": 12620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4426834882670182e-05,
+      "loss": 0.8213,
+      "step": 12630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4404519926590888e-05,
+      "loss": 0.8305,
+      "step": 12640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4382199206379146e-05,
+      "loss": 0.8276,
+      "step": 12650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4359872751245763e-05,
+      "loss": 0.8259,
+      "step": 12660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4337540590409054e-05,
+      "loss": 0.8283,
+      "step": 12670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4315202753094794e-05,
+      "loss": 0.8322,
+      "step": 12680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.42928592685362e-05,
+      "loss": 0.8296,
+      "step": 12690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4270510165973865e-05,
+      "loss": 0.8211,
+      "step": 12700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.424815547465575e-05,
+      "loss": 0.8254,
+      "step": 12710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4225795223837114e-05,
+      "loss": 0.8312,
+      "step": 12720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.42034294427805e-05,
+      "loss": 0.8276,
+      "step": 12730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4181058160755682e-05,
+      "loss": 0.8308,
+      "step": 12740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.415868140703965e-05,
+      "loss": 0.82,
+      "step": 12750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4136299210916532e-05,
+      "loss": 0.8272,
+      "step": 12760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4113911601677592e-05,
+      "loss": 0.8316,
+      "step": 12770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4091518608621183e-05,
+      "loss": 0.8289,
+      "step": 12780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4069120261052682e-05,
+      "loss": 0.8267,
+      "step": 12790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4046716588284492e-05,
+      "loss": 0.824,
+      "step": 12800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4024307619635984e-05,
+      "loss": 0.821,
+      "step": 12810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.400189338443345e-05,
+      "loss": 0.8322,
+      "step": 12820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3979473912010094e-05,
+      "loss": 0.8339,
+      "step": 12830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3957049231705946e-05,
+      "loss": 0.8256,
+      "step": 12840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3934619372867866e-05,
+      "loss": 0.8233,
+      "step": 12850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3912184364849492e-05,
+      "loss": 0.8185,
+      "step": 12860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.388974423701121e-05,
+      "loss": 0.8257,
+      "step": 12870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3867299018720084e-05,
+      "loss": 0.8285,
+      "step": 12880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3844848739349846e-05,
+      "loss": 0.8218,
+      "step": 12890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3822393428280873e-05,
+      "loss": 0.8207,
+      "step": 12900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3799933114900096e-05,
+      "loss": 0.8256,
+      "step": 12910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3777467828601016e-05,
+      "loss": 0.8304,
+      "step": 12920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3754997598783637e-05,
+      "loss": 0.8314,
+      "step": 12930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.373252245485441e-05,
+      "loss": 0.8228,
+      "step": 12940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3710042426226254e-05,
+      "loss": 0.8339,
+      "step": 12950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3687557542318447e-05,
+      "loss": 0.8282,
+      "step": 12960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.366506783255665e-05,
+      "loss": 0.8255,
+      "step": 12970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3642573326372825e-05,
+      "loss": 0.8301,
+      "step": 12980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3620074053205202e-05,
+      "loss": 0.8217,
+      "step": 12990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3597570042498262e-05,
+      "loss": 0.8214,
+      "step": 13000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.357506132370269e-05,
+      "loss": 0.8173,
+      "step": 13010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3552547926275313e-05,
+      "loss": 0.8229,
+      "step": 13020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3530029879679104e-05,
+      "loss": 0.8241,
+      "step": 13030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3507507213383108e-05,
+      "loss": 0.8256,
+      "step": 13040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3484979956862413e-05,
+      "loss": 0.8285,
+      "step": 13050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3462448139598123e-05,
+      "loss": 0.8183,
+      "step": 13060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3439911791077303e-05,
+      "loss": 0.8159,
+      "step": 13070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3417370940792944e-05,
+      "loss": 0.8265,
+      "step": 13080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3394825618243954e-05,
+      "loss": 0.8214,
+      "step": 13090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.337227585293506e-05,
+      "loss": 0.8272,
+      "step": 13100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3349721674376826e-05,
+      "loss": 0.822,
+      "step": 13110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.332716311208558e-05,
+      "loss": 0.8174,
+      "step": 13120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3304600195583394e-05,
+      "loss": 0.8198,
+      "step": 13130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3282032954398032e-05,
+      "loss": 0.831,
+      "step": 13140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3259461418062923e-05,
+      "loss": 0.8243,
+      "step": 13150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3236885616117112e-05,
+      "loss": 0.8253,
+      "step": 13160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3214305578105237e-05,
+      "loss": 0.8201,
+      "step": 13170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.319172133357746e-05,
+      "loss": 0.8184,
+      "step": 13180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3169132912089468e-05,
+      "loss": 0.8208,
+      "step": 13190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3146540343202408e-05,
+      "loss": 0.8271,
+      "step": 13200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3123943656482845e-05,
+      "loss": 0.83,
+      "step": 13210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.310134288150275e-05,
+      "loss": 0.819,
+      "step": 13220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3078738047839425e-05,
+      "loss": 0.8219,
+      "step": 13230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3056129185075503e-05,
+      "loss": 0.8263,
+      "step": 13240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3033516322798875e-05,
+      "loss": 0.821,
+      "step": 13250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3010899490602673e-05,
+      "loss": 0.8181,
+      "step": 13260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2988278718085223e-05,
+      "loss": 0.8123,
+      "step": 13270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.296565403485001e-05,
+      "loss": 0.8258,
+      "step": 13280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2943025470505625e-05,
+      "loss": 0.8197,
+      "step": 13290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2920393054665757e-05,
+      "loss": 0.8257,
+      "step": 13300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2897756816949128e-05,
+      "loss": 0.822,
+      "step": 13310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2875116786979454e-05,
+      "loss": 0.8286,
+      "step": 13320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2852472994385416e-05,
+      "loss": 0.8185,
+      "step": 13330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.282982546880063e-05,
+      "loss": 0.8149,
+      "step": 13340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.280717423986359e-05,
+      "loss": 0.8205,
+      "step": 13350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2784519337217637e-05,
+      "loss": 0.8162,
+      "step": 13360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2761860790510907e-05,
+      "loss": 0.8217,
+      "step": 13370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.273919862939633e-05,
+      "loss": 0.8192,
+      "step": 13380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2716532883531545e-05,
+      "loss": 0.8173,
+      "step": 13390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2693863582578905e-05,
+      "loss": 0.8156,
+      "step": 13400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2671190756205384e-05,
+      "loss": 0.8208,
+      "step": 13410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2648514434082593e-05,
+      "loss": 0.82,
+      "step": 13420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.262583464588671e-05,
+      "loss": 0.8197,
+      "step": 13430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.260315142129846e-05,
+      "loss": 0.8213,
+      "step": 13440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2580464790003046e-05,
+      "loss": 0.8208,
+      "step": 13450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.255777478169014e-05,
+      "loss": 0.8168,
+      "step": 13460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2535081426053834e-05,
+      "loss": 0.825,
+      "step": 13470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2512384752792605e-05,
+      "loss": 0.8188,
+      "step": 13480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.248968479160925e-05,
+      "loss": 0.8199,
+      "step": 13490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.24669815722109e-05,
+      "loss": 0.8231,
+      "step": 13500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.244427512430893e-05,
+      "loss": 0.8239,
+      "step": 13510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2421565477618938e-05,
+      "loss": 0.8165,
+      "step": 13520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2398852661860725e-05,
+      "loss": 0.8146,
+      "step": 13530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2376136706758222e-05,
+      "loss": 0.8142,
+      "step": 13540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2353417642039483e-05,
+      "loss": 0.8221,
+      "step": 13550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2330695497436618e-05,
+      "loss": 0.8191,
+      "step": 13560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2307970302685775e-05,
+      "loss": 0.8216,
+      "step": 13570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2285242087527092e-05,
+      "loss": 0.8183,
+      "step": 13580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2262510881704662e-05,
+      "loss": 0.8228,
+      "step": 13590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2239776714966492e-05,
+      "loss": 0.8149,
+      "step": 13600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.221703961706446e-05,
+      "loss": 0.8148,
+      "step": 13610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2194299617754274e-05,
+      "loss": 0.8123,
+      "step": 13620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.217155674679546e-05,
+      "loss": 0.814,
+      "step": 13630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2148811033951283e-05,
+      "loss": 0.8214,
+      "step": 13640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2126062508988736e-05,
+      "loss": 0.8275,
+      "step": 13650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.210331120167848e-05,
+      "loss": 0.8457,
+      "step": 13660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2080557141794836e-05,
+      "loss": 0.8277,
+      "step": 13670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2057800359115716e-05,
+      "loss": 0.8376,
+      "step": 13680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2035040883422595e-05,
+      "loss": 0.9098,
+      "step": 13690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2012278744500482e-05,
+      "loss": 0.8815,
+      "step": 13700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1989513972137852e-05,
+      "loss": 0.8618,
+      "step": 13710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1966746596126643e-05,
+      "loss": 0.8511,
+      "step": 13720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1943976646262198e-05,
+      "loss": 0.847,
+      "step": 13730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1921204152343233e-05,
+      "loss": 0.8395,
+      "step": 13740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1898429144171763e-05,
+      "loss": 0.838,
+      "step": 13750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.187565165155314e-05,
+      "loss": 0.832,
+      "step": 13760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.185287170429593e-05,
+      "loss": 0.8328,
+      "step": 13770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1830089332211936e-05,
+      "loss": 0.8363,
+      "step": 13780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.180730456511611e-05,
+      "loss": 0.8291,
+      "step": 13790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1784517432826563e-05,
+      "loss": 0.8242,
+      "step": 13800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1761727965164488e-05,
+      "loss": 0.8305,
+      "step": 13810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1738936191954134e-05,
+      "loss": 0.8369,
+      "step": 13820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1716142143022772e-05,
+      "loss": 0.8263,
+      "step": 13830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1693345848200647e-05,
+      "loss": 0.8374,
+      "step": 13840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1670547337320948e-05,
+      "loss": 0.8276,
+      "step": 13850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1647746640219762e-05,
+      "loss": 0.8313,
+      "step": 13860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.162494378673603e-05,
+      "loss": 0.8276,
+      "step": 13870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1602138806711524e-05,
+      "loss": 0.8181,
+      "step": 13880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1579331729990795e-05,
+      "loss": 0.8233,
+      "step": 13890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.155652258642115e-05,
+      "loss": 0.8214,
+      "step": 13900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1533711405852578e-05,
+      "loss": 0.8231,
+      "step": 13910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.151089821813775e-05,
+      "loss": 0.8134,
+      "step": 13920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1488083053131956e-05,
+      "loss": 0.8192,
+      "step": 13930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1465265940693084e-05,
+      "loss": 0.8182,
+      "step": 13940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1442446910681557e-05,
+      "loss": 0.8205,
+      "step": 13950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.141962599296032e-05,
+      "loss": 0.8181,
+      "step": 13960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1396803217394777e-05,
+      "loss": 0.8177,
+      "step": 13970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.137397861385278e-05,
+      "loss": 0.8257,
+      "step": 13980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.135115221220455e-05,
+      "loss": 0.8191,
+      "step": 13990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1328324042322678e-05,
+      "loss": 0.8218,
+      "step": 14000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1305494134082068e-05,
+      "loss": 0.814,
+      "step": 14010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1282662517359885e-05,
+      "loss": 0.8173,
+      "step": 14020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1259829222035554e-05,
+      "loss": 0.8182,
+      "step": 14030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.123699427799067e-05,
+      "loss": 0.8199,
+      "step": 14040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.121415771510902e-05,
+      "loss": 0.8215,
+      "step": 14050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.119131956327646e-05,
+      "loss": 0.8149,
+      "step": 14060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1168479852380973e-05,
+      "loss": 0.8138,
+      "step": 14070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.114563861231256e-05,
+      "loss": 0.8089,
+      "step": 14080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.112279587296322e-05,
+      "loss": 0.8176,
+      "step": 14090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1099951664226927e-05,
+      "loss": 0.8194,
+      "step": 14100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1077106015999566e-05,
+      "loss": 0.8233,
+      "step": 14110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1054258958178914e-05,
+      "loss": 0.823,
+      "step": 14120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1031410520664597e-05,
+      "loss": 0.8173,
+      "step": 14130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1008560733358027e-05,
+      "loss": 0.8159,
+      "step": 14140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0985709626162404e-05,
+      "loss": 0.8173,
+      "step": 14150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0962857228982636e-05,
+      "loss": 0.8057,
+      "step": 14160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0940003571725346e-05,
+      "loss": 0.8023,
+      "step": 14170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0917148684298773e-05,
+      "loss": 0.7996,
+      "step": 14180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.089429259661279e-05,
+      "loss": 0.7989,
+      "step": 14190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0871435338578833e-05,
+      "loss": 0.8005,
+      "step": 14200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.084857694010987e-05,
+      "loss": 0.8064,
+      "step": 14210
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0825717431120362e-05,
+      "loss": 0.8033,
+      "step": 14220
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0802856841526217e-05,
+      "loss": 0.8011,
+      "step": 14230
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0779995201244773e-05,
+      "loss": 0.7976,
+      "step": 14240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0757132540194722e-05,
+      "loss": 0.7957,
+      "step": 14250
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0734268888296105e-05,
+      "loss": 0.7972,
+      "step": 14260
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.071140427547026e-05,
+      "loss": 0.8002,
+      "step": 14270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.068853873163979e-05,
+      "loss": 0.8021,
+      "step": 14280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0665672286728484e-05,
+      "loss": 0.7892,
+      "step": 14290
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.064280497066135e-05,
+      "loss": 0.799,
+      "step": 14300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0619936813364506e-05,
+      "loss": 0.7956,
+      "step": 14310
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0597067844765202e-05,
+      "loss": 0.7962,
+      "step": 14320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0574198094791713e-05,
+      "loss": 0.8008,
+      "step": 14330
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0551327593373357e-05,
+      "loss": 0.7955,
+      "step": 14340
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0528456370440445e-05,
+      "loss": 0.799,
+      "step": 14350
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.0505584455924214e-05,
+      "loss": 0.7979,
+      "step": 14360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0482711879756808e-05,
+      "loss": 0.7982,
+      "step": 14370
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0459838671871247e-05,
+      "loss": 0.8023,
+      "step": 14380
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0436964862201365e-05,
+      "loss": 0.7957,
+      "step": 14390
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.04140904806818e-05,
+      "loss": 0.7962,
+      "step": 14400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0391215557247933e-05,
+      "loss": 0.7968,
+      "step": 14410
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.036834012183583e-05,
+      "loss": 0.7918,
+      "step": 14420
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0345464204382262e-05,
+      "loss": 0.7985,
+      "step": 14430
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.032258783482462e-05,
+      "loss": 0.7925,
+      "step": 14440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0299711043100867e-05,
+      "loss": 0.7933,
+      "step": 14450
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0276833859149553e-05,
+      "loss": 0.7961,
+      "step": 14460
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.025395631290971e-05,
+      "loss": 0.7971,
+      "step": 14470
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.023107843432086e-05,
+      "loss": 0.7971,
+      "step": 14480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0208200253322957e-05,
+      "loss": 0.7985,
+      "step": 14490
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0185321799856345e-05,
+      "loss": 0.7994,
+      "step": 14500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0162443103861746e-05,
+      "loss": 0.7939,
+      "step": 14510
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0139564195280164e-05,
+      "loss": 0.7873,
+      "step": 14520
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0116685104052908e-05,
+      "loss": 0.7951,
+      "step": 14530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.009380586012152e-05,
+      "loss": 0.7948,
+      "step": 14540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.007092649342775e-05,
+      "loss": 0.7902,
+      "step": 14550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.0048047033913475e-05,
+      "loss": 0.7946,
+      "step": 14560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.002516751152074e-05,
+      "loss": 0.7934,
+      "step": 14570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.000228795619164e-05,
+      "loss": 0.7947,
+      "step": 14580
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9979408397868325e-05,
+      "loss": 0.7976,
+      "step": 14590
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9956528866492944e-05,
+      "loss": 0.7934,
+      "step": 14600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9933649392007616e-05,
+      "loss": 0.7982,
+      "step": 14610
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.991077000435438e-05,
+      "loss": 0.7887,
+      "step": 14620
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.988789073347517e-05,
+      "loss": 0.7873,
+      "step": 14630
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.986501160931176e-05,
+      "loss": 0.7951,
+      "step": 14640
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.984213266180574e-05,
+      "loss": 0.7933,
+      "step": 14650
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.981925392089845e-05,
+      "loss": 0.7912,
+      "step": 14660
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.979637541653097e-05,
+      "loss": 0.7907,
+      "step": 14670
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9773497178644085e-05,
+      "loss": 0.791,
+      "step": 14680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9750619237178216e-05,
+      "loss": 0.7937,
+      "step": 14690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9727741622073393e-05,
+      "loss": 0.7979,
+      "step": 14700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9704864363269232e-05,
+      "loss": 0.7989,
+      "step": 14710
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9681987490704875e-05,
+      "loss": 0.7939,
+      "step": 14720
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9659111034318952e-05,
+      "loss": 0.7917,
+      "step": 14730
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9636235024049562e-05,
+      "loss": 0.7988,
+      "step": 14740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.961335948983421e-05,
+      "loss": 0.794,
+      "step": 14750
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.959048446160978e-05,
+      "loss": 0.7931,
+      "step": 14760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9567609969312497e-05,
+      "loss": 0.7941,
+      "step": 14770
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9544736042877886e-05,
+      "loss": 0.7937,
+      "step": 14780
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9521862712240728e-05,
+      "loss": 0.7994,
+      "step": 14790
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.949899000733503e-05,
+      "loss": 0.7856,
+      "step": 14800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.947611795809396e-05,
+      "loss": 0.7982,
+      "step": 14810
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.945324659444985e-05,
+      "loss": 0.7924,
+      "step": 14820
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9430375946334134e-05,
+      "loss": 0.7956,
+      "step": 14830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9407506043677294e-05,
+      "loss": 0.7907,
+      "step": 14840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.938463691640885e-05,
+      "loss": 0.7919,
+      "step": 14850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9361768594457305e-05,
+      "loss": 0.7909,
+      "step": 14860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.93389011077501e-05,
+      "loss": 0.7928,
+      "step": 14870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9316034486213604e-05,
+      "loss": 0.7941,
+      "step": 14880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9293168759773015e-05,
+      "loss": 0.7874,
+      "step": 14890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9270303958352398e-05,
+      "loss": 0.7867,
+      "step": 14900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9247440111874588e-05,
+      "loss": 0.7871,
+      "step": 14910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.922457725026118e-05,
+      "loss": 0.7918,
+      "step": 14920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.920171540343247e-05,
+      "loss": 0.7964,
+      "step": 14930
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9178854601307433e-05,
+      "loss": 0.7904,
+      "step": 14940
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.915599487380369e-05,
+      "loss": 0.7914,
+      "step": 14950
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.913313625083741e-05,
+      "loss": 0.7967,
+      "step": 14960
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9110278762323366e-05,
+      "loss": 0.8,
+      "step": 14970
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9087422438174822e-05,
+      "loss": 0.7965,
+      "step": 14980
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9064567308303527e-05,
+      "loss": 0.7912,
+      "step": 14990
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9041713402619662e-05,
+      "loss": 0.7936,
+      "step": 15000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.901886075103181e-05,
+      "loss": 0.791,
+      "step": 15010
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.899600938344691e-05,
+      "loss": 0.7937,
+      "step": 15020
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8973159329770226e-05,
+      "loss": 0.797,
+      "step": 15030
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8950310619905282e-05,
+      "loss": 0.7926,
+      "step": 15040
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8927463283753872e-05,
+      "loss": 0.7952,
+      "step": 15050
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.890461735121597e-05,
+      "loss": 0.7869,
+      "step": 15060
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.888177285218973e-05,
+      "loss": 0.7847,
+      "step": 15070
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.885892981657142e-05,
+      "loss": 0.7882,
+      "step": 15080
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8836088274255395e-05,
+      "loss": 0.7913,
+      "step": 15090
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8813248255134055e-05,
+      "loss": 0.7918,
+      "step": 15100
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8790409789097815e-05,
+      "loss": 0.7948,
+      "step": 15110
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8767572906035036e-05,
+      "loss": 0.7938,
+      "step": 15120
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8744737635832036e-05,
+      "loss": 0.7946,
+      "step": 15130
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8721904008372994e-05,
+      "loss": 0.7965,
+      "step": 15140
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8699072053539962e-05,
+      "loss": 0.7981,
+      "step": 15150
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8676241801212786e-05,
+      "loss": 0.8019,
+      "step": 15160
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.86534132812691e-05,
+      "loss": 0.7994,
+      "step": 15170
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.863058652358426e-05,
+      "loss": 0.7903,
+      "step": 15180
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8607761558031325e-05,
+      "loss": 0.7847,
+      "step": 15190
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8584938414480983e-05,
+      "loss": 0.7934,
+      "step": 15200
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8562117122801576e-05,
+      "loss": 0.8028,
+      "step": 15210
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.853929771285899e-05,
+      "loss": 0.7888,
+      "step": 15220
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8516480214516674e-05,
+      "loss": 0.7946,
+      "step": 15230
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8493664657635554e-05,
+      "loss": 0.7962,
+      "step": 15240
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8470851072074026e-05,
+      "loss": 0.7919,
+      "step": 15250
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8448039487687907e-05,
+      "loss": 0.7944,
+      "step": 15260
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8425229934330386e-05,
+      "loss": 0.7953,
+      "step": 15270
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8402422441852005e-05,
+      "loss": 0.7964,
+      "step": 15280
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8379617040100602e-05,
+      "loss": 0.7941,
+      "step": 15290
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8356813758921282e-05,
+      "loss": 0.7925,
+      "step": 15300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8334012628156378e-05,
+      "loss": 0.7921,
+      "step": 15310
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.83112136776454e-05,
+      "loss": 0.7871,
+      "step": 15320
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8288416937225015e-05,
+      "loss": 0.7935,
+      "step": 15330
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.826562243672899e-05,
+      "loss": 0.798,
+      "step": 15340
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8242830205988157e-05,
+      "loss": 0.7912,
+      "step": 15350
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8220040274830385e-05,
+      "loss": 0.7991,
+      "step": 15360
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819725267308054e-05,
+      "loss": 0.7856,
+      "step": 15370
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8174467430560423e-05,
+      "loss": 0.8025,
+      "step": 15380
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8151684577088762e-05,
+      "loss": 0.8004,
+      "step": 15390
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.812890414248115e-05,
+      "loss": 0.7877,
+      "step": 15400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8106126156550016e-05,
+      "loss": 0.7864,
+      "step": 15410
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8083350649104597e-05,
+      "loss": 0.7816,
+      "step": 15420
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8060577649950856e-05,
+      "loss": 0.7897,
+      "step": 15430
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.80378071888915e-05,
+      "loss": 0.7888,
+      "step": 15440
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.801503929572591e-05,
+      "loss": 0.7932,
+      "step": 15450
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7992274000250098e-05,
+      "loss": 0.7967,
+      "step": 15460
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7969511332256688e-05,
+      "loss": 0.7956,
+      "step": 15470
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7946751321534857e-05,
+      "loss": 0.7927,
+      "step": 15480
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7923993997870312e-05,
+      "loss": 0.7871,
+      "step": 15490
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7901239391045226e-05,
+      "loss": 0.7951,
+      "step": 15500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7878487530838234e-05,
+      "loss": 0.7905,
+      "step": 15510
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7855738447024372e-05,
+      "loss": 0.7808,
+      "step": 15520
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.783299216937504e-05,
+      "loss": 0.7829,
+      "step": 15530
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.781024872765797e-05,
+      "loss": 0.7944,
+      "step": 15540
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.778750815163718e-05,
+      "loss": 0.7938,
+      "step": 15550
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7764770471072936e-05,
+      "loss": 0.7893,
+      "step": 15560
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7742035715721725e-05,
+      "loss": 0.7819,
+      "step": 15570
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.771930391533618e-05,
+      "loss": 0.7923,
+      "step": 15580
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7696575099665096e-05,
+      "loss": 0.7904,
+      "step": 15590
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7673849298453347e-05,
+      "loss": 0.7904,
+      "step": 15600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7651126541441866e-05,
+      "loss": 0.7931,
+      "step": 15610
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.76284068583676e-05,
+      "loss": 0.7949,
+      "step": 15620
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7605690278963473e-05,
+      "loss": 0.7868,
+      "step": 15630
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7582976832958355e-05,
+      "loss": 0.7861,
+      "step": 15640
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7560266550077015e-05,
+      "loss": 0.7889,
+      "step": 15650
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7537559460040054e-05,
+      "loss": 0.7905,
+      "step": 15660
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7514855592563932e-05,
+      "loss": 0.7918,
+      "step": 15670
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7492154977360875e-05,
+      "loss": 0.7877,
+      "step": 15680
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.746945764413885e-05,
+      "loss": 0.7881,
+      "step": 15690
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.744676362260154e-05,
+      "loss": 0.7921,
+      "step": 15700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7424072942448286e-05,
+      "loss": 0.7937,
+      "step": 15710
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7401385633374065e-05,
+      "loss": 0.7783,
+      "step": 15720
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7378701725069418e-05,
+      "loss": 0.7909,
+      "step": 15730
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7356021247220462e-05,
+      "loss": 0.7827,
+      "step": 15740
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7333344229508814e-05,
+      "loss": 0.7882,
+      "step": 15750
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7310670701611564e-05,
+      "loss": 0.7846,
+      "step": 15760
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7288000693201235e-05,
+      "loss": 0.7869,
+      "step": 15770
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.726533423394574e-05,
+      "loss": 0.7937,
+      "step": 15780
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.724267135350836e-05,
+      "loss": 0.7895,
+      "step": 15790
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7220012081547688e-05,
+      "loss": 0.7871,
+      "step": 15800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.719735644771757e-05,
+      "loss": 0.7858,
+      "step": 15810
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7174704481667123e-05,
+      "loss": 0.7838,
+      "step": 15820
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.715205621304065e-05,
+      "loss": 0.7846,
+      "step": 15830
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.712941167147762e-05,
+      "loss": 0.7829,
+      "step": 15840
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7106770886612616e-05,
+      "loss": 0.7894,
+      "step": 15850
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7084133888075317e-05,
+      "loss": 0.7889,
+      "step": 15860
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7061500705490438e-05,
+      "loss": 0.7868,
+      "step": 15870
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7038871368477707e-05,
+      "loss": 0.7835,
+      "step": 15880
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7016245906651806e-05,
+      "loss": 0.7901,
+      "step": 15890
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6993624349622365e-05,
+      "loss": 0.7892,
+      "step": 15900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6971006726993878e-05,
+      "loss": 0.7878,
+      "step": 15910
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.694839306836572e-05,
+      "loss": 0.7886,
+      "step": 15920
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.692578340333205e-05,
+      "loss": 0.7834,
+      "step": 15930
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.690317776148183e-05,
+      "loss": 0.7907,
+      "step": 15940
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6880576172398733e-05,
+      "loss": 0.7897,
+      "step": 15950
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6857978665661137e-05,
+      "loss": 0.7907,
+      "step": 15960
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6835385270842075e-05,
+      "loss": 0.7827,
+      "step": 15970
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6812796017509203e-05,
+      "loss": 0.7916,
+      "step": 15980
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6790210935224752e-05,
+      "loss": 0.7851,
+      "step": 15990
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.676763005354551e-05,
+      "loss": 0.7883,
+      "step": 16000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6745053402022736e-05,
+      "loss": 0.7907,
+      "step": 16010
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6722481010202182e-05,
+      "loss": 0.7848,
+      "step": 16020
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6699912907624018e-05,
+      "loss": 0.7879,
+      "step": 16030
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.667734912382279e-05,
+      "loss": 0.7878,
+      "step": 16040
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.665478968832741e-05,
+      "loss": 0.7869,
+      "step": 16050
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.663223463066108e-05,
+      "loss": 0.7861,
+      "step": 16060
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6609683980341288e-05,
+      "loss": 0.7858,
+      "step": 16070
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6587137766879748e-05,
+      "loss": 0.7858,
+      "step": 16080
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6564596019782373e-05,
+      "loss": 0.795,
+      "step": 16090
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6542058768549216e-05,
+      "loss": 0.7886,
+      "step": 16100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6519526042674466e-05,
+      "loss": 0.7834,
+      "step": 16110
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.649699787164637e-05,
+      "loss": 0.7832,
+      "step": 16120
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6474474284947232e-05,
+      "loss": 0.7863,
+      "step": 16130
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6451955312053346e-05,
+      "loss": 0.7867,
+      "step": 16140
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.642944098243497e-05,
+      "loss": 0.7882,
+      "step": 16150
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.640693132555629e-05,
+      "loss": 0.7855,
+      "step": 16160
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.638442637087537e-05,
+      "loss": 0.7836,
+      "step": 16170
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6361926147844137e-05,
+      "loss": 0.7834,
+      "step": 16180
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6339430685908287e-05,
+      "loss": 0.7902,
+      "step": 16190
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6316940014507325e-05,
+      "loss": 0.7874,
+      "step": 16200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.629445416307447e-05,
+      "loss": 0.7877,
+      "step": 16210
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6271973161036636e-05,
+      "loss": 0.7844,
+      "step": 16220
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.624949703781439e-05,
+      "loss": 0.7867,
+      "step": 16230
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.622702582282191e-05,
+      "loss": 0.7911,
+      "step": 16240
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6204559545466963e-05,
+      "loss": 0.7912,
+      "step": 16250
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6182098235150847e-05,
+      "loss": 0.7861,
+      "step": 16260
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6159641921268347e-05,
+      "loss": 0.786,
+      "step": 16270
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.613719063320772e-05,
+      "loss": 0.7801,
+      "step": 16280
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.611474440035066e-05,
+      "loss": 0.7904,
+      "step": 16290
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6092303252072224e-05,
+      "loss": 0.781,
+      "step": 16300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6069867217740824e-05,
+      "loss": 0.791,
+      "step": 16310
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.604743632671818e-05,
+      "loss": 0.7813,
+      "step": 16320
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6025010608359277e-05,
+      "loss": 0.7837,
+      "step": 16330
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6002590092012343e-05,
+      "loss": 0.7894,
+      "step": 16340
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.598017480701877e-05,
+      "loss": 0.7827,
+      "step": 16350
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.595776478271313e-05,
+      "loss": 0.7891,
+      "step": 16360
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.593536004842311e-05,
+      "loss": 0.7884,
+      "step": 16370
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5912960633469455e-05,
+      "loss": 0.7927,
+      "step": 16380
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5890566567165967e-05,
+      "loss": 0.7881,
+      "step": 16390
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5868177878819436e-05,
+      "loss": 0.7984,
+      "step": 16400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5845794597729636e-05,
+      "loss": 0.8095,
+      "step": 16410
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5823416753189224e-05,
+      "loss": 0.7979,
+      "step": 16420
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5801044374483776e-05,
+      "loss": 0.7975,
+      "step": 16430
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5778677490891706e-05,
+      "loss": 0.7938,
+      "step": 16440
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.575631613168424e-05,
+      "loss": 0.7905,
+      "step": 16450
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5733960326125363e-05,
+      "loss": 0.7916,
+      "step": 16460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.57116101034718e-05,
+      "loss": 0.7958,
+      "step": 16470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5689265492972975e-05,
+      "loss": 0.7927,
+      "step": 16480
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5666926523870964e-05,
+      "loss": 0.7879,
+      "step": 16490
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5644593225400445e-05,
+      "loss": 0.7854,
+      "step": 16500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.562226562678869e-05,
+      "loss": 0.7952,
+      "step": 16510
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.559994375725551e-05,
+      "loss": 0.7925,
+      "step": 16520
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5577627646013223e-05,
+      "loss": 0.7793,
+      "step": 16530
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.55553173222666e-05,
+      "loss": 0.7753,
+      "step": 16540
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5533012815212848e-05,
+      "loss": 0.7837,
+      "step": 16550
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5510714154041553e-05,
+      "loss": 0.7857,
+      "step": 16560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5488421367934666e-05,
+      "loss": 0.7915,
+      "step": 16570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5466134486066425e-05,
+      "loss": 0.7869,
+      "step": 16580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5443853537603356e-05,
+      "loss": 0.789,
+      "step": 16590
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5421578551704222e-05,
+      "loss": 0.7842,
+      "step": 16600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.539930955751998e-05,
+      "loss": 0.7865,
+      "step": 16610
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.537704658419375e-05,
+      "loss": 0.7831,
+      "step": 16620
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5354789660860758e-05,
+      "loss": 0.7936,
+      "step": 16630
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5332538816648327e-05,
+      "loss": 0.7854,
+      "step": 16640
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5310294080675828e-05,
+      "loss": 0.7841,
+      "step": 16650
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.528805548205462e-05,
+      "loss": 0.7887,
+      "step": 16660
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5265823049888037e-05,
+      "loss": 0.7822,
+      "step": 16670
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5243596813271349e-05,
+      "loss": 0.7826,
+      "step": 16680
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5221376801291719e-05,
+      "loss": 0.7818,
+      "step": 16690
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5199163043028158e-05,
+      "loss": 0.7868,
+      "step": 16700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5176955567551495e-05,
+      "loss": 0.788,
+      "step": 16710
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.515475440392434e-05,
+      "loss": 0.7819,
+      "step": 16720
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5132559581201031e-05,
+      "loss": 0.7826,
+      "step": 16730
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5110371128427623e-05,
+      "loss": 0.7797,
+      "step": 16740
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5088189074641826e-05,
+      "loss": 0.7831,
+      "step": 16750
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5066013448872981e-05,
+      "loss": 0.7848,
+      "step": 16760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5043844280142005e-05,
+      "loss": 0.777,
+      "step": 16770
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.502168159746138e-05,
+      "loss": 0.786,
+      "step": 16780
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.499952542983509e-05,
+      "loss": 0.7905,
+      "step": 16790
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4977375806258599e-05,
+      "loss": 0.7827,
+      "step": 16800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4955232755718795e-05,
+      "loss": 0.7797,
+      "step": 16810
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4933096307193986e-05,
+      "loss": 0.7855,
+      "step": 16820
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4910966489653814e-05,
+      "loss": 0.7775,
+      "step": 16830
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4888843332059267e-05,
+      "loss": 0.7747,
+      "step": 16840
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4866726863362595e-05,
+      "loss": 0.7908,
+      "step": 16850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4844617112507317e-05,
+      "loss": 0.7838,
+      "step": 16860
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.482251410842814e-05,
+      "loss": 0.7815,
+      "step": 16870
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4800417880050955e-05,
+      "loss": 0.777,
+      "step": 16880
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4778328456292776e-05,
+      "loss": 0.7814,
+      "step": 16890
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.475624586606172e-05,
+      "loss": 0.7774,
+      "step": 16900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.4734170138256958e-05,
+      "loss": 0.776,
+      "step": 16910
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4712101301768681e-05,
+      "loss": 0.7724,
+      "step": 16920
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.469003938547806e-05,
+      "loss": 0.7892,
+      "step": 16930
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4667984418257211e-05,
+      "loss": 0.7826,
+      "step": 16940
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4645936428969165e-05,
+      "loss": 0.7769,
+      "step": 16950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4623895446467789e-05,
+      "loss": 0.7726,
+      "step": 16960
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4601861499597818e-05,
+      "loss": 0.7739,
+      "step": 16970
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4579834617194762e-05,
+      "loss": 0.7842,
+      "step": 16980
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4557814828084884e-05,
+      "loss": 0.78,
+      "step": 16990
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4535802161085175e-05,
+      "loss": 0.7809,
+      "step": 17000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4513796645003293e-05,
+      "loss": 0.7807,
+      "step": 17010
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4491798308637544e-05,
+      "loss": 0.7799,
+      "step": 17020
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4469807180776849e-05,
+      "loss": 0.7815,
+      "step": 17030
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4447823290200664e-05,
+      "loss": 0.7719,
+      "step": 17040
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.4425846665679001e-05,
+      "loss": 0.7802,
+      "step": 17050
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4403877335972359e-05,
+      "loss": 0.7795,
+      "step": 17060
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4381915329831682e-05,
+      "loss": 0.7861,
+      "step": 17070
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4359960675998337e-05,
+      "loss": 0.7819,
+      "step": 17080
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4338013403204067e-05,
+      "loss": 0.7751,
+      "step": 17090
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4316073540170953e-05,
+      "loss": 0.788,
+      "step": 17100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4294141115611393e-05,
+      "loss": 0.7822,
+      "step": 17110
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4272216158228017e-05,
+      "loss": 0.7749,
+      "step": 17120
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4250298696713716e-05,
+      "loss": 0.7748,
+      "step": 17130
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4228388759751557e-05,
+      "loss": 0.7839,
+      "step": 17140
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4206486376014765e-05,
+      "loss": 0.7822,
+      "step": 17150
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4184591574166677e-05,
+      "loss": 0.787,
+      "step": 17160
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4162704382860707e-05,
+      "loss": 0.7797,
+      "step": 17170
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4140824830740317e-05,
+      "loss": 0.7831,
+      "step": 17180
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.411895294643895e-05,
+      "loss": 0.7822,
+      "step": 17190
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.4097088758580037e-05,
+      "loss": 0.7835,
+      "step": 17200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4075232295776932e-05,
+      "loss": 0.7766,
+      "step": 17210
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4053383586632871e-05,
+      "loss": 0.7819,
+      "step": 17220
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.4031542659740955e-05,
+      "loss": 0.7788,
+      "step": 17230
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.400970954368409e-05,
+      "loss": 0.7751,
+      "step": 17240
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3987884267034961e-05,
+      "loss": 0.7889,
+      "step": 17250
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.396606685835601e-05,
+      "loss": 0.7736,
+      "step": 17260
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3944257346199347e-05,
+      "loss": 0.7837,
+      "step": 17270
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3922455759106783e-05,
+      "loss": 0.7826,
+      "step": 17280
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3900662125609738e-05,
+      "loss": 0.7753,
+      "step": 17290
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3878876474229239e-05,
+      "loss": 0.7858,
+      "step": 17300
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3857098833475849e-05,
+      "loss": 0.7784,
+      "step": 17310
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.383532923184966e-05,
+      "loss": 0.7766,
+      "step": 17320
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3813567697840237e-05,
+      "loss": 0.7798,
+      "step": 17330
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.37918142599266e-05,
+      "loss": 0.7743,
+      "step": 17340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3770068946577147e-05,
+      "loss": 0.7714,
+      "step": 17350
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3748331786249665e-05,
+      "loss": 0.7757,
+      "step": 17360
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3726602807391267e-05,
+      "loss": 0.7759,
+      "step": 17370
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3704882038438361e-05,
+      "loss": 0.7728,
+      "step": 17380
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3683169507816603e-05,
+      "loss": 0.7767,
+      "step": 17390
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3661465243940875e-05,
+      "loss": 0.7778,
+      "step": 17400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3639769275215238e-05,
+      "loss": 0.7761,
+      "step": 17410
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3618081630032895e-05,
+      "loss": 0.7779,
+      "step": 17420
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3596402336776164e-05,
+      "loss": 0.7819,
+      "step": 17430
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3574731423816419e-05,
+      "loss": 0.7729,
+      "step": 17440
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3553068919514076e-05,
+      "loss": 0.7782,
+      "step": 17450
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.353141485221855e-05,
+      "loss": 0.7793,
+      "step": 17460
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3509769250268208e-05,
+      "loss": 0.779,
+      "step": 17470
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.348813214199034e-05,
+      "loss": 0.7814,
+      "step": 17480
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3466503555701126e-05,
+      "loss": 0.7801,
+      "step": 17490
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3444883519705583e-05,
+      "loss": 0.7762,
+      "step": 17500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3423272062297543e-05,
+      "loss": 0.7824,
+      "step": 17510
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3401669211759615e-05,
+      "loss": 0.7857,
+      "step": 17520
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3380074996363148e-05,
+      "loss": 0.7791,
+      "step": 17530
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3358489444368173e-05,
+      "loss": 0.7753,
+      "step": 17540
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3336912584023395e-05,
+      "loss": 0.776,
+      "step": 17550
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3315344443566152e-05,
+      "loss": 0.7805,
+      "step": 17560
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.329378505122236e-05,
+      "loss": 0.779,
+      "step": 17570
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.327223443520648e-05,
+      "loss": 0.7764,
+      "step": 17580
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3250692623721504e-05,
+      "loss": 0.7822,
+      "step": 17590
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322915964495889e-05,
+      "loss": 0.7789,
+      "step": 17600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3207635527098543e-05,
+      "loss": 0.7817,
+      "step": 17610
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3186120298308767e-05,
+      "loss": 0.7808,
+      "step": 17620
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.316461398674624e-05,
+      "loss": 0.7765,
+      "step": 17630
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3143116620555963e-05,
+      "loss": 0.7795,
+      "step": 17640
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3121628227871227e-05,
+      "loss": 0.7727,
+      "step": 17650
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3100148836813591e-05,
+      "loss": 0.7779,
+      "step": 17660
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3078678475492823e-05,
+      "loss": 0.7786,
+      "step": 17670
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3057217172006887e-05,
+      "loss": 0.7838,
+      "step": 17680
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3035764954441876e-05,
+      "loss": 0.778,
+      "step": 17690
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.301432185087201e-05,
+      "loss": 0.7826,
+      "step": 17700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2992887889359565e-05,
+      "loss": 0.7774,
+      "step": 17710
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2971463097954874e-05,
+      "loss": 0.7824,
+      "step": 17720
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.295004750469624e-05,
+      "loss": 0.7718,
+      "step": 17730
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2928641137609952e-05,
+      "loss": 0.776,
+      "step": 17740
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2907244024710215e-05,
+      "loss": 0.783,
+      "step": 17750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2885856193999128e-05,
+      "loss": 0.788,
+      "step": 17760
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.286447767346664e-05,
+      "loss": 0.7777,
+      "step": 17770
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2843108491090517e-05,
+      "loss": 0.7823,
+      "step": 17780
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2821748674836303e-05,
+      "loss": 0.7774,
+      "step": 17790
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2800398252657292e-05,
+      "loss": 0.781,
+      "step": 17800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.277905725249446e-05,
+      "loss": 0.7709,
+      "step": 17810
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.275772570227648e-05,
+      "loss": 0.774,
+      "step": 17820
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.273640362991965e-05,
+      "loss": 0.7757,
+      "step": 17830
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2715091063327854e-05,
+      "loss": 0.7785,
+      "step": 17840
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2693788030392553e-05,
+      "loss": 0.7691,
+      "step": 17850
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2672494558992719e-05,
+      "loss": 0.7802,
+      "step": 17860
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2651210676994821e-05,
+      "loss": 0.7814,
+      "step": 17870
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.262993641225276e-05,
+      "loss": 0.7712,
+      "step": 17880
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2608671792607866e-05,
+      "loss": 0.7732,
+      "step": 17890
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2587416845888844e-05,
+      "loss": 0.7754,
+      "step": 17900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2566171599911748e-05,
+      "loss": 0.7761,
+      "step": 17910
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2544936082479917e-05,
+      "loss": 0.7728,
+      "step": 17920
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.252371032138398e-05,
+      "loss": 0.7741,
+      "step": 17930
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2502494344401783e-05,
+      "loss": 0.7775,
+      "step": 17940
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2481288179298378e-05,
+      "loss": 0.7684,
+      "step": 17950
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2460091853825962e-05,
+      "loss": 0.7746,
+      "step": 17960
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2438905395723867e-05,
+      "loss": 0.7686,
+      "step": 17970
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.241772883271851e-05,
+      "loss": 0.7773,
+      "step": 17980
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2396562192523356e-05,
+      "loss": 0.769,
+      "step": 17990
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2375405502838889e-05,
+      "loss": 0.7747,
+      "step": 18000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2354258791352562e-05,
+      "loss": 0.7804,
+      "step": 18010
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.233312208573878e-05,
+      "loss": 0.774,
+      "step": 18020
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2311995413658853e-05,
+      "loss": 0.7788,
+      "step": 18030
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2290878802760942e-05,
+      "loss": 0.775,
+      "step": 18040
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2269772280680058e-05,
+      "loss": 0.7803,
+      "step": 18050
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2248675875038015e-05,
+      "loss": 0.7619,
+      "step": 18060
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2227589613443367e-05,
+      "loss": 0.7816,
+      "step": 18070
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2206513523491413e-05,
+      "loss": 0.7729,
+      "step": 18080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2185447632764124e-05,
+      "loss": 0.7765,
+      "step": 18090
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2164391968830138e-05,
+      "loss": 0.7675,
+      "step": 18100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2143346559244705e-05,
+      "loss": 0.7817,
+      "step": 18110
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.212231143154964e-05,
+      "loss": 0.7745,
+      "step": 18120
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2101286613273322e-05,
+      "loss": 0.7725,
+      "step": 18130
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2080272131930628e-05,
+      "loss": 0.7739,
+      "step": 18140
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2059268015022916e-05,
+      "loss": 0.7725,
+      "step": 18150
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2038274290037971e-05,
+      "loss": 0.763,
+      "step": 18160
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2017290984449979e-05,
+      "loss": 0.7796,
+      "step": 18170
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.19963181257195e-05,
+      "loss": 0.7732,
+      "step": 18180
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1975355741293412e-05,
+      "loss": 0.7786,
+      "step": 18190
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1954403858604885e-05,
+      "loss": 0.7692,
+      "step": 18200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.193346250507335e-05,
+      "loss": 0.773,
+      "step": 18210
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.191253170810446e-05,
+      "loss": 0.7784,
+      "step": 18220
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1891611495090051e-05,
+      "loss": 0.7644,
+      "step": 18230
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1870701893408105e-05,
+      "loss": 0.778,
+      "step": 18240
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1849802930422726e-05,
+      "loss": 0.771,
+      "step": 18250
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1828914633484087e-05,
+      "loss": 0.7774,
+      "step": 18260
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1808037029928397e-05,
+      "loss": 0.77,
+      "step": 18270
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1787170147077884e-05,
+      "loss": 0.769,
+      "step": 18280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1766314012240745e-05,
+      "loss": 0.7719,
+      "step": 18290
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1745468652711094e-05,
+      "loss": 0.7754,
+      "step": 18300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1724634095768965e-05,
+      "loss": 0.7764,
+      "step": 18310
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1703810368680244e-05,
+      "loss": 0.7751,
+      "step": 18320
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1682997498696645e-05,
+      "loss": 0.78,
+      "step": 18330
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1662195513055679e-05,
+      "loss": 0.7641,
+      "step": 18340
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1641404438980604e-05,
+      "loss": 0.7689,
+      "step": 18350
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1620624303680393e-05,
+      "loss": 0.7726,
+      "step": 18360
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1599855134349731e-05,
+      "loss": 0.7788,
+      "step": 18370
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1579096958168914e-05,
+      "loss": 0.7699,
+      "step": 18380
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1558349802303895e-05,
+      "loss": 0.7722,
+      "step": 18390
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1537613693906158e-05,
+      "loss": 0.7759,
+      "step": 18400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.151688866011278e-05,
+      "loss": 0.7749,
+      "step": 18410
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1496174728046281e-05,
+      "loss": 0.7713,
+      "step": 18420
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1475471924814717e-05,
+      "loss": 0.7802,
+      "step": 18430
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1454780277511537e-05,
+      "loss": 0.7752,
+      "step": 18440
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1434099813215616e-05,
+      "loss": 0.7708,
+      "step": 18450
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1413430558991169e-05,
+      "loss": 0.7765,
+      "step": 18460
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1392772541887772e-05,
+      "loss": 0.769,
+      "step": 18470
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1372125788940257e-05,
+      "loss": 0.7721,
+      "step": 18480
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1351490327168765e-05,
+      "loss": 0.7872,
+      "step": 18490
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1330866183578596e-05,
+      "loss": 0.7731,
+      "step": 18500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1310253385160296e-05,
+      "loss": 0.7751,
+      "step": 18510
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1289651958889527e-05,
+      "loss": 0.7734,
+      "step": 18520
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1269061931727094e-05,
+      "loss": 0.7737,
+      "step": 18530
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1248483330618858e-05,
+      "loss": 0.779,
+      "step": 18540
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.122791618249576e-05,
+      "loss": 0.7709,
+      "step": 18550
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1207360514273714e-05,
+      "loss": 0.7696,
+      "step": 18560
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1186816352853644e-05,
+      "loss": 0.7815,
+      "step": 18570
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1166283725121398e-05,
+      "loss": 0.7693,
+      "step": 18580
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.114576265794773e-05,
+      "loss": 0.7704,
+      "step": 18590
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1125253178188263e-05,
+      "loss": 0.7757,
+      "step": 18600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1104755312683476e-05,
+      "loss": 0.7717,
+      "step": 18610
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1084269088258614e-05,
+      "loss": 0.7692,
+      "step": 18620
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1063794531723729e-05,
+      "loss": 0.7643,
+      "step": 18630
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1043331669873566e-05,
+      "loss": 0.7724,
+      "step": 18640
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1022880529487581e-05,
+      "loss": 0.7716,
+      "step": 18650
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1002441137329898e-05,
+      "loss": 0.7751,
+      "step": 18660
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0982013520149248e-05,
+      "loss": 0.7748,
+      "step": 18670
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0961597704678974e-05,
+      "loss": 0.7717,
+      "step": 18680
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.094119371763696e-05,
+      "loss": 0.7685,
+      "step": 18690
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0920801585725597e-05,
+      "loss": 0.7703,
+      "step": 18700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0900421335631802e-05,
+      "loss": 0.7715,
+      "step": 18710
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0880052994026906e-05,
+      "loss": 0.762,
+      "step": 18720
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0859696587566657e-05,
+      "loss": 0.7681,
+      "step": 18730
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0839352142891214e-05,
+      "loss": 0.771,
+      "step": 18740
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0819019686625046e-05,
+      "loss": 0.7667,
+      "step": 18750
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0798699245376959e-05,
+      "loss": 0.7657,
+      "step": 18760
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0778390845740017e-05,
+      "loss": 0.7642,
+      "step": 18770
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0758094514291541e-05,
+      "loss": 0.7644,
+      "step": 18780
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0737810277593046e-05,
+      "loss": 0.7765,
+      "step": 18790
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0717538162190227e-05,
+      "loss": 0.7701,
+      "step": 18800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0697278194612899e-05,
+      "loss": 0.767,
+      "step": 18810
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0677030401375012e-05,
+      "loss": 0.7637,
+      "step": 18820
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.065679480897455e-05,
+      "loss": 0.7754,
+      "step": 18830
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0636571443893555e-05,
+      "loss": 0.7745,
+      "step": 18840
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0616360332598044e-05,
+      "loss": 0.7705,
+      "step": 18850
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0596161501538027e-05,
+      "loss": 0.7735,
+      "step": 18860
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0575974977147419e-05,
+      "loss": 0.7734,
+      "step": 18870
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.055580078584403e-05,
+      "loss": 0.7665,
+      "step": 18880
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0535638954029536e-05,
+      "loss": 0.7688,
+      "step": 18890
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0515489508089453e-05,
+      "loss": 0.7694,
+      "step": 18900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0495352474393057e-05,
+      "loss": 0.7779,
+      "step": 18910
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0475227879293413e-05,
+      "loss": 0.7756,
+      "step": 18920
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0455115749127276e-05,
+      "loss": 0.7666,
+      "step": 18930
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0435016110215122e-05,
+      "loss": 0.7694,
+      "step": 18940
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0414928988861051e-05,
+      "loss": 0.7677,
+      "step": 18950
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0394854411352791e-05,
+      "loss": 0.7609,
+      "step": 18960
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0374792403961652e-05,
+      "loss": 0.7709,
+      "step": 18970
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.035474299294251e-05,
+      "loss": 0.7764,
+      "step": 18980
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.033470620453373e-05,
+      "loss": 0.7632,
+      "step": 18990
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0314682064957182e-05,
+      "loss": 0.7778,
+      "step": 19000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0294670600418164e-05,
+      "loss": 0.7681,
+      "step": 19010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0274671837105403e-05,
+      "loss": 0.7676,
+      "step": 19020
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.0254685801190993e-05,
+      "loss": 0.7673,
+      "step": 19030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.023471251883037e-05,
+      "loss": 0.7622,
+      "step": 19040
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0214752016162281e-05,
+      "loss": 0.7703,
+      "step": 19050
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0194804319308762e-05,
+      "loss": 0.7665,
+      "step": 19060
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.017486945437507e-05,
+      "loss": 0.7621,
+      "step": 19070
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0154947447449686e-05,
+      "loss": 0.7698,
+      "step": 19080
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0135038324604248e-05,
+      "loss": 0.7778,
+      "step": 19090
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.011514211189356e-05,
+      "loss": 0.7713,
+      "step": 19100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0095258835355482e-05,
+      "loss": 0.767,
+      "step": 19110
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0075388521010998e-05,
+      "loss": 0.7692,
+      "step": 19120
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0055531194864092e-05,
+      "loss": 0.7625,
+      "step": 19130
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0035686882901778e-05,
+      "loss": 0.7691,
+      "step": 19140
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.0015855611094007e-05,
+      "loss": 0.7696,
+      "step": 19150
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.996037405393702e-06,
+      "loss": 0.7806,
+      "step": 19160
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.976232291736645e-06,
+      "loss": 0.7691,
+      "step": 19170
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.956440296041531e-06,
+      "loss": 0.7702,
+      "step": 19180
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.936661444209832e-06,
+      "loss": 0.7712,
+      "step": 19190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.916895762125873e-06,
+      "loss": 0.7694,
+      "step": 19200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.897143275656706e-06,
+      "loss": 0.7669,
+      "step": 19210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.877404010652143e-06,
+      "loss": 0.7648,
+      "step": 19220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.857677992944663e-06,
+      "loss": 0.7595,
+      "step": 19230
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.837965248349439e-06,
+      "loss": 0.7695,
+      "step": 19240
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.818265802664245e-06,
+      "loss": 0.76,
+      "step": 19250
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.798579681669484e-06,
+      "loss": 0.7691,
+      "step": 19260
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.778906911128078e-06,
+      "loss": 0.7775,
+      "step": 19270
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.75924751678552e-06,
+      "loss": 0.7691,
+      "step": 19280
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.739601524369765e-06,
+      "loss": 0.7622,
+      "step": 19290
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.719968959591257e-06,
+      "loss": 0.76,
+      "step": 19300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.70034984814284e-06,
+      "loss": 0.759,
+      "step": 19310
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.68074421569978e-06,
+      "loss": 0.7698,
+      "step": 19320
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.661152087919682e-06,
+      "loss": 0.7735,
+      "step": 19330
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.641573490442484e-06,
+      "loss": 0.7696,
+      "step": 19340
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.62200844889041e-06,
+      "loss": 0.7689,
+      "step": 19350
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.602456988867965e-06,
+      "loss": 0.7662,
+      "step": 19360
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.582919135961852e-06,
+      "loss": 0.7684,
+      "step": 19370
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.563394915740996e-06,
+      "loss": 0.7652,
+      "step": 19380
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.54388435375645e-06,
+      "loss": 0.769,
+      "step": 19390
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.524387475541423e-06,
+      "loss": 0.7684,
+      "step": 19400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.504904306611196e-06,
+      "loss": 0.7692,
+      "step": 19410
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.485434872463104e-06,
+      "loss": 0.7589,
+      "step": 19420
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.465979198576535e-06,
+      "loss": 0.7709,
+      "step": 19430
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.446537310412845e-06,
+      "loss": 0.7696,
+      "step": 19440
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.427109233415348e-06,
+      "loss": 0.7601,
+      "step": 19450
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.407694993009304e-06,
+      "loss": 0.7661,
+      "step": 19460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.388294614601843e-06,
+      "loss": 0.7671,
+      "step": 19470
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.368908123581974e-06,
+      "loss": 0.7629,
+      "step": 19480
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.349535545320516e-06,
+      "loss": 0.7619,
+      "step": 19490
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.330176905170077e-06,
+      "loss": 0.7654,
+      "step": 19500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.31083222846505e-06,
+      "loss": 0.7666,
+      "step": 19510
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.291501540521524e-06,
+      "loss": 0.7641,
+      "step": 19520
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.272184866637307e-06,
+      "loss": 0.7625,
+      "step": 19530
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.252882232091847e-06,
+      "loss": 0.7728,
+      "step": 19540
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.23359366214622e-06,
+      "loss": 0.7611,
+      "step": 19550
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.21431918204312e-06,
+      "loss": 0.7709,
+      "step": 19560
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.195058817006772e-06,
+      "loss": 0.7546,
+      "step": 19570
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.175812592242939e-06,
+      "loss": 0.7669,
+      "step": 19580
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.156580532938895e-06,
+      "loss": 0.7693,
+      "step": 19590
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.137362664263343e-06,
+      "loss": 0.7698,
+      "step": 19600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.118159011366452e-06,
+      "loss": 0.7722,
+      "step": 19610
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.098969599379754e-06,
+      "loss": 0.7673,
+      "step": 19620
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.079794453416174e-06,
+      "loss": 0.7636,
+      "step": 19630
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.060633598569942e-06,
+      "loss": 0.7657,
+      "step": 19640
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.041487059916596e-06,
+      "loss": 0.7651,
+      "step": 19650
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.022354862512935e-06,
+      "loss": 0.7616,
+      "step": 19660
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.003237031397002e-06,
+      "loss": 0.7649,
+      "step": 19670
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.984133591588022e-06,
+      "loss": 0.7676,
+      "step": 19680
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.965044568086405e-06,
+      "loss": 0.7708,
+      "step": 19690
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.94596998587367e-06,
+      "loss": 0.7642,
+      "step": 19700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.926909869912463e-06,
+      "loss": 0.7674,
+      "step": 19710
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.907864245146485e-06,
+      "loss": 0.7593,
+      "step": 19720
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.888833136500468e-06,
+      "loss": 0.7691,
+      "step": 19730
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.869816568880144e-06,
+      "loss": 0.7614,
+      "step": 19740
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.850814567172245e-06,
+      "loss": 0.7645,
+      "step": 19750
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.831827156244403e-06,
+      "loss": 0.7625,
+      "step": 19760
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.812854360945185e-06,
+      "loss": 0.7614,
+      "step": 19770
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.793896206104002e-06,
+      "loss": 0.7694,
+      "step": 19780
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.77495271653114e-06,
+      "loss": 0.7626,
+      "step": 19790
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.756023917017662e-06,
+      "loss": 0.7673,
+      "step": 19800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.737109832335419e-06,
+      "loss": 0.763,
+      "step": 19810
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.718210487237e-06,
+      "loss": 0.7682,
+      "step": 19820
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.69932590645572e-06,
+      "loss": 0.7612,
+      "step": 19830
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.680456114705546e-06,
+      "loss": 0.7552,
+      "step": 19840
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.661601136681124e-06,
+      "loss": 0.7688,
+      "step": 19850
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.642760997057675e-06,
+      "loss": 0.7632,
+      "step": 19860
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.623935720491046e-06,
+      "loss": 0.7629,
+      "step": 19870
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.605125331617578e-06,
+      "loss": 0.7661,
+      "step": 19880
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.586329855054184e-06,
+      "loss": 0.7627,
+      "step": 19890
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.567549315398216e-06,
+      "loss": 0.763,
+      "step": 19900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.548783737227518e-06,
+      "loss": 0.7632,
+      "step": 19910
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.530033145100318e-06,
+      "loss": 0.7639,
+      "step": 19920
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.511297563555263e-06,
+      "loss": 0.7601,
+      "step": 19930
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.492577017111327e-06,
+      "loss": 0.7529,
+      "step": 19940
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.473871530267846e-06,
+      "loss": 0.7673,
+      "step": 19950
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.455181127504391e-06,
+      "loss": 0.7625,
+      "step": 19960
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.43650583328085e-06,
+      "loss": 0.764,
+      "step": 19970
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.417845672037301e-06,
+      "loss": 0.7681,
+      "step": 19980
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.399200668194045e-06,
+      "loss": 0.7608,
+      "step": 19990
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.380570846151517e-06,
+      "loss": 0.7686,
+      "step": 20000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.361956230290313e-06,
+      "loss": 0.7665,
+      "step": 20010
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.343356844971105e-06,
+      "loss": 0.7683,
+      "step": 20020
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.324772714534662e-06,
+      "loss": 0.7626,
+      "step": 20030
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.306203863301743e-06,
+      "loss": 0.7639,
+      "step": 20040
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.28765031557316e-06,
+      "loss": 0.7715,
+      "step": 20050
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.269112095629662e-06,
+      "loss": 0.7636,
+      "step": 20060
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.250589227731967e-06,
+      "loss": 0.7688,
+      "step": 20070
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.232081736120676e-06,
+      "loss": 0.7656,
+      "step": 20080
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.213589645016291e-06,
+      "loss": 0.7715,
+      "step": 20090
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.195112978619145e-06,
+      "loss": 0.7648,
+      "step": 20100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.176651761109381e-06,
+      "loss": 0.7658,
+      "step": 20110
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.15820601664693e-06,
+      "loss": 0.7699,
+      "step": 20120
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.13977576937149e-06,
+      "loss": 0.7604,
+      "step": 20130
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.121361043402442e-06,
+      "loss": 0.7697,
+      "step": 20140
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.102961862838899e-06,
+      "loss": 0.7677,
+      "step": 20150
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.084578251759583e-06,
+      "loss": 0.7661,
+      "step": 20160
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.066210234222882e-06,
+      "loss": 0.7636,
+      "step": 20170
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.047857834266755e-06,
+      "loss": 0.7559,
+      "step": 20180
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.029521075908713e-06,
+      "loss": 0.7571,
+      "step": 20190
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.011199983145827e-06,
+      "loss": 0.7604,
+      "step": 20200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.992894579954644e-06,
+      "loss": 0.7662,
+      "step": 20210
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.974604890291175e-06,
+      "loss": 0.7659,
+      "step": 20220
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.956330938090892e-06,
+      "loss": 0.7634,
+      "step": 20230
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.938072747268644e-06,
+      "loss": 0.7661,
+      "step": 20240
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.919830341718673e-06,
+      "loss": 0.7678,
+      "step": 20250
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.901603745314552e-06,
+      "loss": 0.7629,
+      "step": 20260
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.88339298190916e-06,
+      "loss": 0.763,
+      "step": 20270
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.865198075334682e-06,
+      "loss": 0.7637,
+      "step": 20280
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.84701904940251e-06,
+      "loss": 0.758,
+      "step": 20290
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.828855927903298e-06,
+      "loss": 0.7609,
+      "step": 20300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.810708734606854e-06,
+      "loss": 0.7663,
+      "step": 20310
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.792577493262143e-06,
+      "loss": 0.7644,
+      "step": 20320
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.774462227597278e-06,
+      "loss": 0.7542,
+      "step": 20330
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.756362961319442e-06,
+      "loss": 0.7616,
+      "step": 20340
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.738279718114878e-06,
+      "loss": 0.7644,
+      "step": 20350
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.720212521648885e-06,
+      "loss": 0.7574,
+      "step": 20360
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.70216139556573e-06,
+      "loss": 0.7637,
+      "step": 20370
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.68412636348868e-06,
+      "loss": 0.7657,
+      "step": 20380
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.666107449019909e-06,
+      "loss": 0.7583,
+      "step": 20390
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.648104675740527e-06,
+      "loss": 0.7612,
+      "step": 20400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.630118067210498e-06,
+      "loss": 0.7544,
+      "step": 20410
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.612147646968646e-06,
+      "loss": 0.7626,
+      "step": 20420
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.594193438532596e-06,
+      "loss": 0.7672,
+      "step": 20430
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.576255465398772e-06,
+      "loss": 0.756,
+      "step": 20440
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.558333751042335e-06,
+      "loss": 0.7611,
+      "step": 20450
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.540428318917192e-06,
+      "loss": 0.7595,
+      "step": 20460
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.522539192455909e-06,
+      "loss": 0.7578,
+      "step": 20470
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.504666395069749e-06,
+      "loss": 0.7634,
+      "step": 20480
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.486809950148575e-06,
+      "loss": 0.7602,
+      "step": 20490
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.468969881060868e-06,
+      "loss": 0.7594,
+      "step": 20500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.451146211153659e-06,
+      "loss": 0.7609,
+      "step": 20510
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.43333896375255e-06,
+      "loss": 0.7702,
+      "step": 20520
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.415548162161617e-06,
+      "loss": 0.7633,
+      "step": 20530
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.39777382966344e-06,
+      "loss": 0.7584,
+      "step": 20540
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.38001598951902e-06,
+      "loss": 0.7572,
+      "step": 20550
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.362274664967814e-06,
+      "loss": 0.7625,
+      "step": 20560
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.3445498792276095e-06,
+      "loss": 0.7584,
+      "step": 20570
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.326841655494605e-06,
+      "loss": 0.7545,
+      "step": 20580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.309150016943282e-06,
+      "loss": 0.7573,
+      "step": 20590
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.2914749867264545e-06,
+      "loss": 0.7566,
+      "step": 20600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.273816587975167e-06,
+      "loss": 0.7626,
+      "step": 20610
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.256174843798727e-06,
+      "loss": 0.7629,
+      "step": 20620
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.238549777284618e-06,
+      "loss": 0.7531,
+      "step": 20630
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.220941411498539e-06,
+      "loss": 0.7574,
+      "step": 20640
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.203349769484278e-06,
+      "loss": 0.7552,
+      "step": 20650
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.185774874263785e-06,
+      "loss": 0.7556,
+      "step": 20660
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.16821674883706e-06,
+      "loss": 0.755,
+      "step": 20670
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.1506754161821915e-06,
+      "loss": 0.7573,
+      "step": 20680
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.1331508992552504e-06,
+      "loss": 0.7631,
+      "step": 20690
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.1156432209903405e-06,
+      "loss": 0.7516,
+      "step": 20700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.098152404299496e-06,
+      "loss": 0.7592,
+      "step": 20710
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.080678472072719e-06,
+      "loss": 0.7655,
+      "step": 20720
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.06322144717787e-06,
+      "loss": 0.7659,
+      "step": 20730
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.045781352460728e-06,
+      "loss": 0.7642,
+      "step": 20740
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.028358210744881e-06,
+      "loss": 0.7553,
+      "step": 20750
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.010952044831765e-06,
+      "loss": 0.7536,
+      "step": 20760
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.993562877500562e-06,
+      "loss": 0.7636,
+      "step": 20770
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.976190731508243e-06,
+      "loss": 0.7579,
+      "step": 20780
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.958835629589482e-06,
+      "loss": 0.7663,
+      "step": 20790
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.941497594456654e-06,
+      "loss": 0.7531,
+      "step": 20800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.924176648799789e-06,
+      "loss": 0.761,
+      "step": 20810
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.906872815286578e-06,
+      "loss": 0.755,
+      "step": 20820
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.889586116562288e-06,
+      "loss": 0.7614,
+      "step": 20830
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.872316575249785e-06,
+      "loss": 0.754,
+      "step": 20840
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.855064213949461e-06,
+      "loss": 0.7567,
+      "step": 20850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.837829055239249e-06,
+      "loss": 0.762,
+      "step": 20860
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.820611121674548e-06,
+      "loss": 0.7569,
+      "step": 20870
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.80341043578822e-06,
+      "loss": 0.7598,
+      "step": 20880
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.786227020090554e-06,
+      "loss": 0.762,
+      "step": 20890
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.769060897069255e-06,
+      "loss": 0.7578,
+      "step": 20900
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.75191208918937e-06,
+      "loss": 0.7582,
+      "step": 20910
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.734780618893308e-06,
+      "loss": 0.7628,
+      "step": 20920
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.7176665086007735e-06,
+      "loss": 0.7613,
+      "step": 20930
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.700569780708766e-06,
+      "loss": 0.7644,
+      "step": 20940
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.683490457591528e-06,
+      "loss": 0.7628,
+      "step": 20950
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.666428561600515e-06,
+      "loss": 0.7623,
+      "step": 20960
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.649384115064405e-06,
+      "loss": 0.7609,
+      "step": 20970
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.632357140289012e-06,
+      "loss": 0.7645,
+      "step": 20980
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.6153476595572894e-06,
+      "loss": 0.7595,
+      "step": 20990
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.598355695129317e-06,
+      "loss": 0.7635,
+      "step": 21000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.581381269242222e-06,
+      "loss": 0.7535,
+      "step": 21010
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.564424404110206e-06,
+      "loss": 0.7519,
+      "step": 21020
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.547485121924473e-06,
+      "loss": 0.7651,
+      "step": 21030
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.530563444853211e-06,
+      "loss": 0.7534,
+      "step": 21040
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.5136593950415895e-06,
+      "loss": 0.7519,
+      "step": 21050
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.4967729946116885e-06,
+      "loss": 0.7578,
+      "step": 21060
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.479904265662509e-06,
+      "loss": 0.7548,
+      "step": 21070
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.463053230269911e-06,
+      "loss": 0.7487,
+      "step": 21080
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.4462199104866e-06,
+      "loss": 0.7577,
+      "step": 21090
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.42940432834211e-06,
+      "loss": 0.7521,
+      "step": 21100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.412606505842751e-06,
+      "loss": 0.7612,
+      "step": 21110
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.395826464971586e-06,
+      "loss": 0.7577,
+      "step": 21120
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.379064227688427e-06,
+      "loss": 0.7522,
+      "step": 21130
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.362319815929765e-06,
+      "loss": 0.758,
+      "step": 21140
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.345593251608784e-06,
+      "loss": 0.759,
+      "step": 21150
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.328884556615289e-06,
+      "loss": 0.7535,
+      "step": 21160
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.31219375281572e-06,
+      "loss": 0.7531,
+      "step": 21170
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.295520862053093e-06,
+      "loss": 0.753,
+      "step": 21180
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.278865906146978e-06,
+      "loss": 0.7554,
+      "step": 21190
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.262228906893475e-06,
+      "loss": 0.7584,
+      "step": 21200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.2456098860652e-06,
+      "loss": 0.7546,
+      "step": 21210
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.2290088654112144e-06,
+      "loss": 0.7531,
+      "step": 21220
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.212425866657051e-06,
+      "loss": 0.7481,
+      "step": 21230
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.19586091150463e-06,
+      "loss": 0.7578,
+      "step": 21240
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.179314021632286e-06,
+      "loss": 0.7568,
+      "step": 21250
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.162785218694693e-06,
+      "loss": 0.7618,
+      "step": 21260
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.146274524322857e-06,
+      "loss": 0.7497,
+      "step": 21270
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.1297819601240835e-06,
+      "loss": 0.7547,
+      "step": 21280
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.113307547681971e-06,
+      "loss": 0.7524,
+      "step": 21290
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.09685130855633e-06,
+      "loss": 0.7544,
+      "step": 21300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.080413264283225e-06,
+      "loss": 0.7559,
+      "step": 21310
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.063993436374873e-06,
+      "loss": 0.7596,
+      "step": 21320
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.0475918463196895e-06,
+      "loss": 0.755,
+      "step": 21330
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.031208515582176e-06,
+      "loss": 0.7565,
+      "step": 21340
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.014843465602984e-06,
+      "loss": 0.7614,
+      "step": 21350
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.998496717798807e-06,
+      "loss": 0.7482,
+      "step": 21360
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.982168293562416e-06,
+      "loss": 0.7538,
+      "step": 21370
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.96585821426257e-06,
+      "loss": 0.7568,
+      "step": 21380
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.949566501244053e-06,
+      "loss": 0.7567,
+      "step": 21390
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.933293175827586e-06,
+      "loss": 0.7613,
+      "step": 21400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.917038259309853e-06,
+      "loss": 0.7526,
+      "step": 21410
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.900801772963409e-06,
+      "loss": 0.755,
+      "step": 21420
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.884583738036733e-06,
+      "loss": 0.7552,
+      "step": 21430
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.8683841757541205e-06,
+      "loss": 0.7619,
+      "step": 21440
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.852203107315721e-06,
+      "loss": 0.759,
+      "step": 21450
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.836040553897455e-06,
+      "loss": 0.7568,
+      "step": 21460
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.8198965366510415e-06,
+      "loss": 0.7597,
+      "step": 21470
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.8037710767039106e-06,
+      "loss": 0.7506,
+      "step": 21480
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.787664195159239e-06,
+      "loss": 0.7603,
+      "step": 21490
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.771575913095853e-06,
+      "loss": 0.7516,
+      "step": 21500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.75550625156827e-06,
+      "loss": 0.753,
+      "step": 21510
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.739455231606621e-06,
+      "loss": 0.7581,
+      "step": 21520
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.723422874216656e-06,
+      "loss": 0.7536,
+      "step": 21530
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.707409200379681e-06,
+      "loss": 0.7588,
+      "step": 21540
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.691414231052577e-06,
+      "loss": 0.7596,
+      "step": 21550
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.67543798716772e-06,
+      "loss": 0.7553,
+      "step": 21560
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.659480489633e-06,
+      "loss": 0.759,
+      "step": 21570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.643541759331756e-06,
+      "loss": 0.7557,
+      "step": 21580
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.627621817122793e-06,
+      "loss": 0.7551,
+      "step": 21590
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.611720683840296e-06,
+      "loss": 0.7595,
+      "step": 21600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.595838380293865e-06,
+      "loss": 0.7495,
+      "step": 21610
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.579974927268434e-06,
+      "loss": 0.7483,
+      "step": 21620
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.56413034552429e-06,
+      "loss": 0.7447,
+      "step": 21630
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.548304655797003e-06,
+      "loss": 0.7487,
+      "step": 21640
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.532497878797429e-06,
+      "loss": 0.7528,
+      "step": 21650
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.516710035211665e-06,
+      "loss": 0.7585,
+      "step": 21660
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.500941145701048e-06,
+      "loss": 0.7533,
+      "step": 21670
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.485191230902089e-06,
+      "loss": 0.7564,
+      "step": 21680
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.469460311426489e-06,
+      "loss": 0.7536,
+      "step": 21690
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.453748407861066e-06,
+      "loss": 0.7473,
+      "step": 21700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.438055540767772e-06,
+      "loss": 0.7541,
+      "step": 21710
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.422381730683639e-06,
+      "loss": 0.7522,
+      "step": 21720
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.406726998120751e-06,
+      "loss": 0.7535,
+      "step": 21730
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.391091363566241e-06,
+      "loss": 0.7538,
+      "step": 21740
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.375474847482241e-06,
+      "loss": 0.7602,
+      "step": 21750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.359877470305852e-06,
+      "loss": 0.7511,
+      "step": 21760
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.344299252449156e-06,
+      "loss": 0.7533,
+      "step": 21770
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.328740214299129e-06,
+      "loss": 0.7595,
+      "step": 21780
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.3132003762176755e-06,
+      "loss": 0.7527,
+      "step": 21790
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.297679758541554e-06,
+      "loss": 0.7514,
+      "step": 21800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.28217838158237e-06,
+      "loss": 0.7525,
+      "step": 21810
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.2666962656265674e-06,
+      "loss": 0.7539,
+      "step": 21820
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.251233430935358e-06,
+      "loss": 0.7488,
+      "step": 21830
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.235789897744743e-06,
+      "loss": 0.759,
+      "step": 21840
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.220365686265452e-06,
+      "loss": 0.7564,
+      "step": 21850
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.204960816682922e-06,
+      "loss": 0.7494,
+      "step": 21860
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.1895753091572995e-06,
+      "loss": 0.7569,
+      "step": 21870
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.174209183823373e-06,
+      "loss": 0.7549,
+      "step": 21880
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.1588624607905634e-06,
+      "loss": 0.7524,
+      "step": 21890
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.143535160142923e-06,
+      "loss": 0.7531,
+      "step": 21900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.128227301939059e-06,
+      "loss": 0.7551,
+      "step": 21910
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.1129389062121615e-06,
+      "loss": 0.755,
+      "step": 21920
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.097669992969918e-06,
+      "loss": 0.754,
+      "step": 21930
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.082420582194556e-06,
+      "loss": 0.7501,
+      "step": 21940
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.067190693842752e-06,
+      "loss": 0.7556,
+      "step": 21950
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.051980347845644e-06,
+      "loss": 0.7575,
+      "step": 21960
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.0367895641087925e-06,
+      "loss": 0.7593,
+      "step": 21970
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.0216183625121685e-06,
+      "loss": 0.7473,
+      "step": 21980
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.006466762910096e-06,
+      "loss": 0.7568,
+      "step": 21990
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.99133478513127e-06,
+      "loss": 0.7501,
+      "step": 22000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.9762224489786805e-06,
+      "loss": 0.7537,
+      "step": 22010
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.961129774229645e-06,
+      "loss": 0.7477,
+      "step": 22020
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.94605678063571e-06,
+      "loss": 0.7489,
+      "step": 22030
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.931003487922703e-06,
+      "loss": 0.7607,
+      "step": 22040
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.915969915790644e-06,
+      "loss": 0.7579,
+      "step": 22050
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.900956083913764e-06,
+      "loss": 0.7555,
+      "step": 22060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.885962011940439e-06,
+      "loss": 0.7547,
+      "step": 22070
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.870987719493212e-06,
+      "loss": 0.7514,
+      "step": 22080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.85603322616871e-06,
+      "loss": 0.7536,
+      "step": 22090
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.841098551537686e-06,
+      "loss": 0.7564,
+      "step": 22100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.826183715144912e-06,
+      "loss": 0.7546,
+      "step": 22110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.8112887365092385e-06,
+      "loss": 0.7494,
+      "step": 22120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7964136351235024e-06,
+      "loss": 0.7507,
+      "step": 22130
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.781558430454544e-06,
+      "loss": 0.7543,
+      "step": 22140
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.7667231419431505e-06,
+      "loss": 0.7511,
+      "step": 22150
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.75190778900406e-06,
+      "loss": 0.7466,
+      "step": 22160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.737112391025906e-06,
+      "loss": 0.7514,
+      "step": 22170
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.722336967371233e-06,
+      "loss": 0.747,
+      "step": 22180
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7075815373764e-06,
+      "loss": 0.7498,
+      "step": 22190
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.692846120351646e-06,
+      "loss": 0.7574,
+      "step": 22200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.678130735580995e-06,
+      "loss": 0.7525,
+      "step": 22210
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.663435402322265e-06,
+      "loss": 0.744,
+      "step": 22220
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.648760139807025e-06,
+      "loss": 0.747,
+      "step": 22230
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.634104967240587e-06,
+      "loss": 0.7524,
+      "step": 22240
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.619469903801965e-06,
+      "loss": 0.7509,
+      "step": 22250
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.604854968643857e-06,
+      "loss": 0.7456,
+      "step": 22260
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.590260180892614e-06,
+      "loss": 0.7488,
+      "step": 22270
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.575685559648238e-06,
+      "loss": 0.7551,
+      "step": 22280
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.561131123984315e-06,
+      "loss": 0.7491,
+      "step": 22290
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.546596892948043e-06,
+      "loss": 0.7442,
+      "step": 22300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.5320828855601495e-06,
+      "loss": 0.7534,
+      "step": 22310
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.51758912081492e-06,
+      "loss": 0.7554,
+      "step": 22320
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.503115617680134e-06,
+      "loss": 0.7552,
+      "step": 22330
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.488662395097056e-06,
+      "loss": 0.7464,
+      "step": 22340
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4742294719804095e-06,
+      "loss": 0.7507,
+      "step": 22350
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4598168672183655e-06,
+      "loss": 0.7532,
+      "step": 22360
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.445424599672481e-06,
+      "loss": 0.7538,
+      "step": 22370
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.431052688177724e-06,
+      "loss": 0.7575,
+      "step": 22380
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.416701151542402e-06,
+      "loss": 0.7544,
+      "step": 22390
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4023700085481735e-06,
+      "loss": 0.7555,
+      "step": 22400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.3880592779499965e-06,
+      "loss": 0.7504,
+      "step": 22410
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.373768978476123e-06,
+      "loss": 0.7582,
+      "step": 22420
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.359499128828055e-06,
+      "loss": 0.7472,
+      "step": 22430
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.345249747680553e-06,
+      "loss": 0.7517,
+      "step": 22440
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.331020853681571e-06,
+      "loss": 0.7523,
+      "step": 22450
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.316812465452267e-06,
+      "loss": 0.75,
+      "step": 22460
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.302624601586946e-06,
+      "loss": 0.7565,
+      "step": 22470
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.288457280653077e-06,
+      "loss": 0.7487,
+      "step": 22480
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.274310521191225e-06,
+      "loss": 0.7505,
+      "step": 22490
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.260184341715042e-06,
+      "loss": 0.7546,
+      "step": 22500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.246078760711274e-06,
+      "loss": 0.7602,
+      "step": 22510
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.231993796639686e-06,
+      "loss": 0.7505,
+      "step": 22520
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.217929467933064e-06,
+      "loss": 0.7526,
+      "step": 22530
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.203885792997207e-06,
+      "loss": 0.7463,
+      "step": 22540
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.1898627902108615e-06,
+      "loss": 0.7544,
+      "step": 22550
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.175860477925739e-06,
+      "loss": 0.7487,
+      "step": 22560
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.1618788744664654e-06,
+      "loss": 0.7534,
+      "step": 22570
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.147917998130555e-06,
+      "loss": 0.752,
+      "step": 22580
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.133977867188423e-06,
+      "loss": 0.7478,
+      "step": 22590
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.120058499883308e-06,
+      "loss": 0.7511,
+      "step": 22600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.106159914431298e-06,
+      "loss": 0.7515,
+      "step": 22610
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.092282129021268e-06,
+      "loss": 0.752,
+      "step": 22620
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.078425161814874e-06,
+      "loss": 0.7467,
+      "step": 22630
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.06458903094654e-06,
+      "loss": 0.7492,
+      "step": 22640
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.050773754523407e-06,
+      "loss": 0.7421,
+      "step": 22650
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.03697935062533e-06,
+      "loss": 0.7538,
+      "step": 22660
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.023205837304855e-06,
+      "loss": 0.7488,
+      "step": 22670
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.009453232587175e-06,
+      "loss": 0.7499,
+      "step": 22680
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.995721554470135e-06,
+      "loss": 0.7542,
+      "step": 22690
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.982010820924178e-06,
+      "loss": 0.7567,
+      "step": 22700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.968321049892356e-06,
+      "loss": 0.7429,
+      "step": 22710
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.954652259290272e-06,
+      "loss": 0.752,
+      "step": 22720
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.941004467006075e-06,
+      "loss": 0.7506,
+      "step": 22730
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.927377690900436e-06,
+      "loss": 0.7478,
+      "step": 22740
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.913771948806529e-06,
+      "loss": 0.7505,
+      "step": 22750
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.900187258529986e-06,
+      "loss": 0.7535,
+      "step": 22760
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.886623637848908e-06,
+      "loss": 0.7531,
+      "step": 22770
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8730811045138e-06,
+      "loss": 0.7439,
+      "step": 22780
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8595596762476e-06,
+      "loss": 0.7466,
+      "step": 22790
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.846059370745585e-06,
+      "loss": 0.745,
+      "step": 22800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.832580205675431e-06,
+      "loss": 0.7564,
+      "step": 22810
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.819122198677119e-06,
+      "loss": 0.7542,
+      "step": 22820
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.805685367362957e-06,
+      "loss": 0.7505,
+      "step": 22830
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.792269729317528e-06,
+      "loss": 0.7534,
+      "step": 22840
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.7788753020976975e-06,
+      "loss": 0.7504,
+      "step": 22850
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.7655021032325477e-06,
+      "loss": 0.7537,
+      "step": 22860
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.7521501502234105e-06,
+      "loss": 0.7492,
+      "step": 22870
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.7388194605437765e-06,
+      "loss": 0.755,
+      "step": 22880
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.725510051639345e-06,
+      "loss": 0.7457,
+      "step": 22890
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.712221940927938e-06,
+      "loss": 0.7531,
+      "step": 22900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.698955145799528e-06,
+      "loss": 0.7467,
+      "step": 22910
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6857096836161676e-06,
+      "loss": 0.7501,
+      "step": 22920
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6724855717120187e-06,
+      "loss": 0.744,
+      "step": 22930
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6592828273932756e-06,
+      "loss": 0.7464,
+      "step": 22940
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.646101467938199e-06,
+      "loss": 0.7469,
+      "step": 22950
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.632941510597021e-06,
+      "loss": 0.7519,
+      "step": 22960
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.61980297259201e-06,
+      "loss": 0.7477,
+      "step": 22970
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6066858711173723e-06,
+      "loss": 0.7529,
+      "step": 22980
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.593590223339283e-06,
+      "loss": 0.7442,
+      "step": 22990
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.5805160463958145e-06,
+      "loss": 0.7517,
+      "step": 23000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.567463357396972e-06,
+      "loss": 0.7467,
+      "step": 23010
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.5544321734246133e-06,
+      "loss": 0.7475,
+      "step": 23020
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.541422511532466e-06,
+      "loss": 0.7469,
+      "step": 23030
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.5284343887460846e-06,
+      "loss": 0.752,
+      "step": 23040
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.5154678220628503e-06,
+      "loss": 0.7487,
+      "step": 23050
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.502522828451913e-06,
+      "loss": 0.752,
+      "step": 23060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.4895994248542107e-06,
+      "loss": 0.7544,
+      "step": 23070
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.4766976281824106e-06,
+      "loss": 0.7512,
+      "step": 23080
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.4638174553209146e-06,
+      "loss": 0.7582,
+      "step": 23090
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.4509589231258223e-06,
+      "loss": 0.7454,
+      "step": 23100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.438122048424908e-06,
+      "loss": 0.7476,
+      "step": 23110
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.425306848017602e-06,
+      "loss": 0.7484,
+      "step": 23120
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.4125133386749856e-06,
+      "loss": 0.7507,
+      "step": 23130
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3997415371397313e-06,
+      "loss": 0.7465,
+      "step": 23140
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3869914601261235e-06,
+      "loss": 0.7483,
+      "step": 23150
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3742631243199964e-06,
+      "loss": 0.744,
+      "step": 23160
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3615565463787523e-06,
+      "loss": 0.743,
+      "step": 23170
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3488717429313034e-06,
+      "loss": 0.744,
+      "step": 23180
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.336208730578072e-06,
+      "loss": 0.7441,
+      "step": 23190
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.323567525890954e-06,
+      "loss": 0.7523,
+      "step": 23200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3109481454133262e-06,
+      "loss": 0.7553,
+      "step": 23210
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2983506056599834e-06,
+      "loss": 0.7473,
+      "step": 23220
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.285774923117151e-06,
+      "loss": 0.7498,
+      "step": 23230
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.273221114242442e-06,
+      "loss": 0.753,
+      "step": 23240
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2606891954648524e-06,
+      "loss": 0.7515,
+      "step": 23250
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.248179183184723e-06,
+      "loss": 0.7489,
+      "step": 23260
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2356910937737273e-06,
+      "loss": 0.7497,
+      "step": 23270
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.223224943574845e-06,
+      "loss": 0.7461,
+      "step": 23280
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.21078074890236e-06,
+      "loss": 0.7511,
+      "step": 23290
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.1983585260418026e-06,
+      "loss": 0.7473,
+      "step": 23300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.1859582912499663e-06,
+      "loss": 0.742,
+      "step": 23310
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.173580060754857e-06,
+      "loss": 0.7606,
+      "step": 23320
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.1612238507556925e-06,
+      "loss": 0.7481,
+      "step": 23330
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.1488896774228682e-06,
+      "loss": 0.7473,
+      "step": 23340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.136577556897933e-06,
+      "loss": 0.753,
+      "step": 23350
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.124287505293595e-06,
+      "loss": 0.7524,
+      "step": 23360
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.112019538693665e-06,
+      "loss": 0.7407,
+      "step": 23370
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.0997736731530504e-06,
+      "loss": 0.7406,
+      "step": 23380
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.087549924697748e-06,
+      "loss": 0.7502,
+      "step": 23390
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.0753483093247993e-06,
+      "loss": 0.7511,
+      "step": 23400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.0631688430022886e-06,
+      "loss": 0.7491,
+      "step": 23410
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.051011541669309e-06,
+      "loss": 0.7446,
+      "step": 23420
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.038876421235939e-06,
+      "loss": 0.7499,
+      "step": 23430
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.0267634975832517e-06,
+      "loss": 0.7489,
+      "step": 23440
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.014672786563244e-06,
+      "loss": 0.7445,
+      "step": 23450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.0026043039988707e-06,
+      "loss": 0.7436,
+      "step": 23460
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9905580656839728e-06,
+      "loss": 0.748,
+      "step": 23470
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9785340873832956e-06,
+      "loss": 0.7476,
+      "step": 23480
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.966532384832441e-06,
+      "loss": 0.7489,
+      "step": 23490
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.954552973737874e-06,
+      "loss": 0.7518,
+      "step": 23500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.942595869776874e-06,
+      "loss": 0.7474,
+      "step": 23510
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9306610885975394e-06,
+      "loss": 0.7465,
+      "step": 23520
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9187486458187385e-06,
+      "loss": 0.7451,
+      "step": 23530
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9068585570301256e-06,
+      "loss": 0.7474,
+      "step": 23540
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.894990837792082e-06,
+      "loss": 0.7497,
+      "step": 23550
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.8831455036357338e-06,
+      "loss": 0.7509,
+      "step": 23560
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.8713225700628844e-06,
+      "loss": 0.7473,
+      "step": 23570
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.859522052546051e-06,
+      "loss": 0.7461,
+      "step": 23580
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.8477439665283956e-06,
+      "loss": 0.7453,
+      "step": 23590
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.83598832742374e-06,
+      "loss": 0.7396,
+      "step": 23600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.8242551506165108e-06,
+      "loss": 0.7526,
+      "step": 23610
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.8125444514617606e-06,
+      "loss": 0.7451,
+      "step": 23620
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.8008562452851042e-06,
+      "loss": 0.7437,
+      "step": 23630
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.789190547382743e-06,
+      "loss": 0.7421,
+      "step": 23640
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.7775473730213942e-06,
+      "loss": 0.7453,
+      "step": 23650
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.7659267374383247e-06,
+      "loss": 0.7381,
+      "step": 23660
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.7543286558412874e-06,
+      "loss": 0.7502,
+      "step": 23670
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.7427531434085365e-06,
+      "loss": 0.7403,
+      "step": 23680
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.7312002152887673e-06,
+      "loss": 0.7411,
+      "step": 23690
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.7196698866011437e-06,
+      "loss": 0.7453,
+      "step": 23700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.708162172435238e-06,
+      "loss": 0.7438,
+      "step": 23710
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.6966770878510317e-06,
+      "loss": 0.7451,
+      "step": 23720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.6852146478788886e-06,
+      "loss": 0.7396,
+      "step": 23730
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.673774867519545e-06,
+      "loss": 0.745,
+      "step": 23740
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.662357761744072e-06,
+      "loss": 0.7442,
+      "step": 23750
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.650963345493882e-06,
+      "loss": 0.7468,
+      "step": 23760
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.6395916336806784e-06,
+      "loss": 0.752,
+      "step": 23770
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.6282426411864625e-06,
+      "loss": 0.7467,
+      "step": 23780
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.616916382863499e-06,
+      "loss": 0.7407,
+      "step": 23790
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.605612873534298e-06,
+      "loss": 0.7452,
+      "step": 23800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.594332127991599e-06,
+      "loss": 0.7442,
+      "step": 23810
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5830741609983577e-06,
+      "loss": 0.7419,
+      "step": 23820
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5718389872877114e-06,
+      "loss": 0.7486,
+      "step": 23830
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5606266215629783e-06,
+      "loss": 0.7488,
+      "step": 23840
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5494370784976095e-06,
+      "loss": 0.7454,
+      "step": 23850
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5382703727352144e-06,
+      "loss": 0.7504,
+      "step": 23860
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5271265188894957e-06,
+      "loss": 0.7443,
+      "step": 23870
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5160055315442542e-06,
+      "loss": 0.749,
+      "step": 23880
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5049074252533667e-06,
+      "loss": 0.7426,
+      "step": 23890
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.49383221454077e-06,
+      "loss": 0.7457,
+      "step": 23900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.482779913900433e-06,
+      "loss": 0.7461,
+      "step": 23910
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4717505377963467e-06,
+      "loss": 0.7436,
+      "step": 23920
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.460744100662491e-06,
+      "loss": 0.7469,
+      "step": 23930
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4497606169028474e-06,
+      "loss": 0.7501,
+      "step": 23940
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4388001008913296e-06,
+      "loss": 0.7466,
+      "step": 23950
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.427862566971817e-06,
+      "loss": 0.7483,
+      "step": 23960
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.416948029458097e-06,
+      "loss": 0.7462,
+      "step": 23970
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.40605650263388e-06,
+      "loss": 0.7428,
+      "step": 23980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.3951880007527417e-06,
+      "loss": 0.7382,
+      "step": 23990
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.3843425380381446e-06,
+      "loss": 0.7405,
+      "step": 24000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.373520128683382e-06,
+      "loss": 0.7453,
+      "step": 24010
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.3627207868515956e-06,
+      "loss": 0.745,
+      "step": 24020
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.3519445266757267e-06,
+      "loss": 0.7444,
+      "step": 24030
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.3411913622585127e-06,
+      "loss": 0.7512,
+      "step": 24040
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.3304613076724623e-06,
+      "loss": 0.7424,
+      "step": 24050
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.319754376959853e-06,
+      "loss": 0.7418,
+      "step": 24060
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.3090705841326844e-06,
+      "loss": 0.7506,
+      "step": 24070
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.2984099431726927e-06,
+      "loss": 0.7404,
+      "step": 24080
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.2877724680312996e-06,
+      "loss": 0.7451,
+      "step": 24090
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.277158172629621e-06,
+      "loss": 0.7411,
+      "step": 24100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.2665670708584364e-06,
+      "loss": 0.7466,
+      "step": 24110
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.2559991765781606e-06,
+      "loss": 0.7421,
+      "step": 24120
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.2454545036188556e-06,
+      "loss": 0.744,
+      "step": 24130
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.2349330657801804e-06,
+      "loss": 0.742,
+      "step": 24140
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.224434876831387e-06,
+      "loss": 0.7427,
+      "step": 24150
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.2139599505113153e-06,
+      "loss": 0.7437,
+      "step": 24160
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.203508300528341e-06,
+      "loss": 0.7487,
+      "step": 24170
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1930799405604007e-06,
+      "loss": 0.7491,
+      "step": 24180
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.182674884254934e-06,
+      "loss": 0.7418,
+      "step": 24190
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1722931452288877e-06,
+      "loss": 0.735,
+      "step": 24200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1619347370687004e-06,
+      "loss": 0.7446,
+      "step": 24210
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.151599673330269e-06,
+      "loss": 0.7482,
+      "step": 24220
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.14128796753895e-06,
+      "loss": 0.7475,
+      "step": 24230
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1309996331895235e-06,
+      "loss": 0.7507,
+      "step": 24240
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.120734683746184e-06,
+      "loss": 0.7439,
+      "step": 24250
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1104931326425194e-06,
+      "loss": 0.7481,
+      "step": 24260
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1002749932815146e-06,
+      "loss": 0.7429,
+      "step": 24270
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.0900802790354914e-06,
+      "loss": 0.7458,
+      "step": 24280
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.0799090032461346e-06,
+      "loss": 0.752,
+      "step": 24290
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.069761179224441e-06,
+      "loss": 0.7465,
+      "step": 24300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.05963682025073e-06,
+      "loss": 0.7504,
+      "step": 24310
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.0495359395746005e-06,
+      "loss": 0.7482,
+      "step": 24320
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.039458550414941e-06,
+      "loss": 0.7421,
+      "step": 24330
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.029404665959871e-06,
+      "loss": 0.7472,
+      "step": 24340
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.0193742993667807e-06,
+      "loss": 0.7486,
+      "step": 24350
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.009367463762255e-06,
+      "loss": 0.7429,
+      "step": 24360
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.999384172242109e-06,
+      "loss": 0.7438,
+      "step": 24370
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9894244378713233e-06,
+      "loss": 0.7448,
+      "step": 24380
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9794882736840714e-06,
+      "loss": 0.7365,
+      "step": 24390
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9695756926836586e-06,
+      "loss": 0.7396,
+      "step": 24400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.959686707842554e-06,
+      "loss": 0.7328,
+      "step": 24410
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.949821332102311e-06,
+      "loss": 0.7504,
+      "step": 24420
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9399795783736185e-06,
+      "loss": 0.742,
+      "step": 24430
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.930161459536235e-06,
+      "loss": 0.7398,
+      "step": 24440
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9203669884389998e-06,
+      "loss": 0.7429,
+      "step": 24450
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.910596177899786e-06,
+      "loss": 0.7433,
+      "step": 24460
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9008490407055258e-06,
+      "loss": 0.7433,
+      "step": 24470
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8911255896121528e-06,
+      "loss": 0.7407,
+      "step": 24480
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.881425837344608e-06,
+      "loss": 0.7443,
+      "step": 24490
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.87174979659682e-06,
+      "loss": 0.7432,
+      "step": 24500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.862097480031686e-06,
+      "loss": 0.7397,
+      "step": 24510
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8524689002810547e-06,
+      "loss": 0.747,
+      "step": 24520
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8428640699457135e-06,
+      "loss": 0.7472,
+      "step": 24530
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8332830015953606e-06,
+      "loss": 0.7493,
+      "step": 24540
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.823725707768611e-06,
+      "loss": 0.7451,
+      "step": 24550
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8141922009729573e-06,
+      "loss": 0.7415,
+      "step": 24560
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8046824936847617e-06,
+      "loss": 0.7415,
+      "step": 24570
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7951965983492403e-06,
+      "loss": 0.7422,
+      "step": 24580
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7857345273804538e-06,
+      "loss": 0.7478,
+      "step": 24590
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.776296293161277e-06,
+      "loss": 0.7417,
+      "step": 24600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7668819080433962e-06,
+      "loss": 0.7413,
+      "step": 24610
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7574913843472763e-06,
+      "loss": 0.746,
+      "step": 24620
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7481247343621688e-06,
+      "loss": 0.7476,
+      "step": 24630
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.738781970346073e-06,
+      "loss": 0.7466,
+      "step": 24640
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7294631045257283e-06,
+      "loss": 0.7496,
+      "step": 24650
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7201681490966016e-06,
+      "loss": 0.7462,
+      "step": 24660
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7108971162228716e-06,
+      "loss": 0.7426,
+      "step": 24670
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.701650018037404e-06,
+      "loss": 0.7436,
+      "step": 24680
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6924268666417498e-06,
+      "loss": 0.7473,
+      "step": 24690
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.683227674106107e-06,
+      "loss": 0.7454,
+      "step": 24700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6740524524693413e-06,
+      "loss": 0.7382,
+      "step": 24710
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6649012137389164e-06,
+      "loss": 0.7424,
+      "step": 24720
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6557739698909436e-06,
+      "loss": 0.739,
+      "step": 24730
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6466707328701059e-06,
+      "loss": 0.7401,
+      "step": 24740
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6375915145896871e-06,
+      "loss": 0.7431,
+      "step": 24750
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6285363269315247e-06,
+      "loss": 0.7389,
+      "step": 24760
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6195051817460217e-06,
+      "loss": 0.7445,
+      "step": 24770
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6104980908521017e-06,
+      "loss": 0.744,
+      "step": 24780
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6015150660372225e-06,
+      "loss": 0.7454,
+      "step": 24790
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.592556119057338e-06,
+      "loss": 0.7467,
+      "step": 24800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5836212616368918e-06,
+      "loss": 0.751,
+      "step": 24810
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.574710505468804e-06,
+      "loss": 0.7426,
+      "step": 24820
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5658238622144595e-06,
+      "loss": 0.7419,
+      "step": 24830
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5569613435036756e-06,
+      "loss": 0.7385,
+      "step": 24840
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5481229609347103e-06,
+      "loss": 0.7417,
+      "step": 24850
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5393087260742222e-06,
+      "loss": 0.7505,
+      "step": 24860
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5305186504572823e-06,
+      "loss": 0.7433,
+      "step": 24870
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5217527455873326e-06,
+      "loss": 0.7432,
+      "step": 24880
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5130110229361884e-06,
+      "loss": 0.7461,
+      "step": 24890
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.504293493944018e-06,
+      "loss": 0.7399,
+      "step": 24900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.495600170019329e-06,
+      "loss": 0.744,
+      "step": 24910
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.4869310625389433e-06,
+      "loss": 0.744,
+      "step": 24920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.47828618284801e-06,
+      "loss": 0.7471,
+      "step": 24930
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.4696655422599525e-06,
+      "loss": 0.7407,
+      "step": 24940
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.461069152056478e-06,
+      "loss": 0.7421,
+      "step": 24950
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.4524970234875667e-06,
+      "loss": 0.7397,
+      "step": 24960
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.4439491677714323e-06,
+      "loss": 0.7438,
+      "step": 24970
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.435425596094544e-06,
+      "loss": 0.7448,
+      "step": 24980
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4269263196115657e-06,
+      "loss": 0.7468,
+      "step": 24990
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4184513494453888e-06,
+      "loss": 0.7461,
+      "step": 25000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.410000696687084e-06,
+      "loss": 0.7434,
+      "step": 25010
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4015743723958975e-06,
+      "loss": 0.7494,
+      "step": 25020
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3931723875992376e-06,
+      "loss": 0.7451,
+      "step": 25030
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3847947532926687e-06,
+      "loss": 0.7415,
+      "step": 25040
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3764414804398763e-06,
+      "loss": 0.7453,
+      "step": 25050
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3681125799726736e-06,
+      "loss": 0.7397,
+      "step": 25060
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3598080627909682e-06,
+      "loss": 0.7449,
+      "step": 25070
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.351527939762769e-06,
+      "loss": 0.7433,
+      "step": 25080
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3432722217241501e-06,
+      "loss": 0.7437,
+      "step": 25090
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3350409194792625e-06,
+      "loss": 0.7451,
+      "step": 25100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.326834043800278e-06,
+      "loss": 0.7513,
+      "step": 25110
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.31865160542743e-06,
+      "loss": 0.7445,
+      "step": 25120
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.310493615068953e-06,
+      "loss": 0.7422,
+      "step": 25130
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3023600834010974e-06,
+      "loss": 0.7436,
+      "step": 25140
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2942510210680935e-06,
+      "loss": 0.738,
+      "step": 25150
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2861664386821638e-06,
+      "loss": 0.7458,
+      "step": 25160
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2781063468234823e-06,
+      "loss": 0.7391,
+      "step": 25170
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.270070756040176e-06,
+      "loss": 0.7378,
+      "step": 25180
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.262059676848304e-06,
+      "loss": 0.738,
+      "step": 25190
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2540731197318578e-06,
+      "loss": 0.7463,
+      "step": 25200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2461110951427258e-06,
+      "loss": 0.7358,
+      "step": 25210
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2381736135007038e-06,
+      "loss": 0.7372,
+      "step": 25220
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2302606851934518e-06,
+      "loss": 0.7424,
+      "step": 25230
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2223723205765125e-06,
+      "loss": 0.7378,
+      "step": 25240
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.214508529973275e-06,
+      "loss": 0.7418,
+      "step": 25250
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2066693236749671e-06,
+      "loss": 0.7413,
+      "step": 25260
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.1988547119406468e-06,
+      "loss": 0.7428,
+      "step": 25270
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1910647049971913e-06,
+      "loss": 0.7375,
+      "step": 25280
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1832993130392612e-06,
+      "loss": 0.7435,
+      "step": 25290
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.175558546229325e-06,
+      "loss": 0.7422,
+      "step": 25300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1678424146976064e-06,
+      "loss": 0.7391,
+      "step": 25310
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1601509285421009e-06,
+      "loss": 0.7358,
+      "step": 25320
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1524840978285456e-06,
+      "loss": 0.7418,
+      "step": 25330
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1448419325904102e-06,
+      "loss": 0.7392,
+      "step": 25340
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1372244428288904e-06,
+      "loss": 0.7473,
+      "step": 25350
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1296316385128847e-06,
+      "loss": 0.7468,
+      "step": 25360
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1220635295789872e-06,
+      "loss": 0.7424,
+      "step": 25370
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1145201259314797e-06,
+      "loss": 0.7498,
+      "step": 25380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.1070014374422988e-06,
+      "loss": 0.7421,
+      "step": 25390
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.099507473951056e-06,
+      "loss": 0.7492,
+      "step": 25400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.0920382452649814e-06,
+      "loss": 0.746,
+      "step": 25410
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0845937611589564e-06,
+      "loss": 0.7377,
+      "step": 25420
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0771740313754674e-06,
+      "loss": 0.7419,
+      "step": 25430
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0697790656246121e-06,
+      "loss": 0.7442,
+      "step": 25440
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0624088735840755e-06,
+      "loss": 0.7429,
+      "step": 25450
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.055063464899122e-06,
+      "loss": 0.737,
+      "step": 25460
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0477428491825848e-06,
+      "loss": 0.7399,
+      "step": 25470
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0404470360148533e-06,
+      "loss": 0.7374,
+      "step": 25480
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0331760349438435e-06,
+      "loss": 0.7455,
+      "step": 25490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0259298554850195e-06,
+      "loss": 0.7437,
+      "step": 25500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0187085071213465e-06,
+      "loss": 0.7442,
+      "step": 25510
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0115119993033096e-06,
+      "loss": 0.744,
+      "step": 25520
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.0043403414488662e-06,
+      "loss": 0.7413,
+      "step": 25530
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.971935429434709e-07,
+      "loss": 0.7419,
+      "step": 25540
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.90071613140029e-07,
+      "loss": 0.7504,
+      "step": 25550
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.82974561358918e-07,
+      "loss": 0.7532,
+      "step": 25560
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.7590239688794e-07,
+      "loss": 0.7443,
+      "step": 25570
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.688551289823422e-07,
+      "loss": 0.7488,
+      "step": 25580
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.618327668647765e-07,
+      "loss": 0.7493,
+      "step": 25590
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.548353197253135e-07,
+      "loss": 0.7431,
+      "step": 25600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.478627967214104e-07,
+      "loss": 0.7412,
+      "step": 25610
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.409152069779104e-07,
+      "loss": 0.7416,
+      "step": 25620
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.339925595870225e-07,
+      "loss": 0.743,
+      "step": 25630
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.270948636083221e-07,
+      "loss": 0.7482,
+      "step": 25640
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.202221280687196e-07,
+      "loss": 0.7479,
+      "step": 25650
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.133743619624669e-07,
+      "loss": 0.74,
+      "step": 25660
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.065515742511421e-07,
+      "loss": 0.7431,
+      "step": 25670
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.997537738636275e-07,
+      "loss": 0.7466,
+      "step": 25680
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.929809696961044e-07,
+      "loss": 0.7428,
+      "step": 25690
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.862331706120497e-07,
+      "loss": 0.7492,
+      "step": 25700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.795103854422127e-07,
+      "loss": 0.7444,
+      "step": 25710
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.728126229846001e-07,
+      "loss": 0.7422,
+      "step": 25720
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.66139892004485e-07,
+      "loss": 0.7455,
+      "step": 25730
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.594922012343709e-07,
+      "loss": 0.7393,
+      "step": 25740
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.528695593739988e-07,
+      "loss": 0.7382,
+      "step": 25750
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.462719750903225e-07,
+      "loss": 0.7391,
+      "step": 25760
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.396994570175132e-07,
+      "loss": 0.7367,
+      "step": 25770
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.331520137569238e-07,
+      "loss": 0.7451,
+      "step": 25780
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.266296538771046e-07,
+      "loss": 0.7361,
+      "step": 25790
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.201323859137722e-07,
+      "loss": 0.7364,
+      "step": 25800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.136602183698094e-07,
+      "loss": 0.7475,
+      "step": 25810
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.072131597152499e-07,
+      "loss": 0.7416,
+      "step": 25820
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.007912183872712e-07,
+      "loss": 0.7433,
+      "step": 25830
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.943944027901684e-07,
+      "loss": 0.7406,
+      "step": 25840
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.880227212953717e-07,
+      "loss": 0.7435,
+      "step": 25850
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.81676182241402e-07,
+      "loss": 0.7407,
+      "step": 25860
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.753547939338912e-07,
+      "loss": 0.7368,
+      "step": 25870
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.690585646455439e-07,
+      "loss": 0.7374,
+      "step": 25880
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.627875026161514e-07,
+      "loss": 0.7376,
+      "step": 25890
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.565416160525596e-07,
+      "loss": 0.7456,
+      "step": 25900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.503209131286727e-07,
+      "loss": 0.7358,
+      "step": 25910
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.441254019854316e-07,
+      "loss": 0.7419,
+      "step": 25920
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.379550907308219e-07,
+      "loss": 0.7456,
+      "step": 25930
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.318099874398355e-07,
+      "loss": 0.7345,
+      "step": 25940
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.256901001544836e-07,
+      "loss": 0.737,
+      "step": 25950
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.195954368837732e-07,
+      "loss": 0.7425,
+      "step": 25960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.135260056037086e-07,
+      "loss": 0.733,
+      "step": 25970
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.074818142572604e-07,
+      "loss": 0.7378,
+      "step": 25980
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.014628707543836e-07,
+      "loss": 0.7504,
+      "step": 25990
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.954691829719773e-07,
+      "loss": 0.7406,
+      "step": 26000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.895007587538982e-07,
+      "loss": 0.7431,
+      "step": 26010
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.835576059109406e-07,
+      "loss": 0.7386,
+      "step": 26020
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.776397322208205e-07,
+      "loss": 0.7413,
+      "step": 26030
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.717471454281721e-07,
+      "loss": 0.7458,
+      "step": 26040
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.658798532445465e-07,
+      "loss": 0.7445,
+      "step": 26050
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.600378633483795e-07,
+      "loss": 0.736,
+      "step": 26060
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.542211833850043e-07,
+      "loss": 0.7434,
+      "step": 26070
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.484298209666229e-07,
+      "loss": 0.7394,
+      "step": 26080
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.426637836723126e-07,
+      "loss": 0.738,
+      "step": 26090
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.369230790480041e-07,
+      "loss": 0.735,
+      "step": 26100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.312077146064743e-07,
+      "loss": 0.7378,
+      "step": 26110
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.255176978273358e-07,
+      "loss": 0.7453,
+      "step": 26120
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.198530361570387e-07,
+      "loss": 0.7322,
+      "step": 26130
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.142137370088397e-07,
+      "loss": 0.7367,
+      "step": 26140
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.085998077628152e-07,
+      "loss": 0.7477,
+      "step": 26150
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.030112557658264e-07,
+      "loss": 0.738,
+      "step": 26160
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.974480883315425e-07,
+      "loss": 0.7499,
+      "step": 26170
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.919103127403891e-07,
+      "loss": 0.745,
+      "step": 26180
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.863979362395844e-07,
+      "loss": 0.746,
+      "step": 26190
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.80910966043089e-07,
+      "loss": 0.7375,
+      "step": 26200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.754494093316276e-07,
+      "loss": 0.7428,
+      "step": 26210
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.700132732526586e-07,
+      "loss": 0.7325,
+      "step": 26220
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.646025649203801e-07,
+      "loss": 0.7417,
+      "step": 26230
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.592172914157057e-07,
+      "loss": 0.7385,
+      "step": 26240
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.538574597862689e-07,
+      "loss": 0.7347,
+      "step": 26250
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.48523077046399e-07,
+      "loss": 0.7477,
+      "step": 26260
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.432141501771316e-07,
+      "loss": 0.7356,
+      "step": 26270
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.379306861261824e-07,
+      "loss": 0.7392,
+      "step": 26280
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.326726918079472e-07,
+      "loss": 0.7339,
+      "step": 26290
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.27440174103484e-07,
+      "loss": 0.7443,
+      "step": 26300
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.222331398605174e-07,
+      "loss": 0.7421,
+      "step": 26310
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.170515958934185e-07,
+      "loss": 0.7405,
+      "step": 26320
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.118955489832012e-07,
+      "loss": 0.7418,
+      "step": 26330
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.067650058775076e-07,
+      "loss": 0.7438,
+      "step": 26340
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.016599732906091e-07,
+      "loss": 0.7483,
+      "step": 26350
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.965804579033861e-07,
+      "loss": 0.7471,
+      "step": 26360
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.915264663633301e-07,
+      "loss": 0.7394,
+      "step": 26370
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.864980052845281e-07,
+      "loss": 0.7347,
+      "step": 26380
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.814950812476559e-07,
+      "loss": 0.7383,
+      "step": 26390
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.765177007999677e-07,
+      "loss": 0.7392,
+      "step": 26400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.7156587045528834e-07,
+      "loss": 0.7341,
+      "step": 26410
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.666395966940096e-07,
+      "loss": 0.7451,
+      "step": 26420
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.617388859630767e-07,
+      "loss": 0.7451,
+      "step": 26430
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.5686374467597936e-07,
+      "loss": 0.7447,
+      "step": 26440
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.5201417921274524e-07,
+      "loss": 0.7386,
+      "step": 26450
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.47190195919931e-07,
+      "loss": 0.7451,
+      "step": 26460
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.4239180111061763e-07,
+      "loss": 0.7415,
+      "step": 26470
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.37619001064391e-07,
+      "loss": 0.7482,
+      "step": 26480
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.3287180202735033e-07,
+      "loss": 0.7386,
+      "step": 26490
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.2815021021208604e-07,
+      "loss": 0.7418,
+      "step": 26500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.234542317976753e-07,
+      "loss": 0.749,
+      "step": 26510
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.187838729296845e-07,
+      "loss": 0.7392,
+      "step": 26520
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.141391397201422e-07,
+      "loss": 0.7388,
+      "step": 26530
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0952003824754396e-07,
+      "loss": 0.7429,
+      "step": 26540
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0492657455684314e-07,
+      "loss": 0.7443,
+      "step": 26550
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.003587546594401e-07,
+      "loss": 0.7467,
+      "step": 26560
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.9581658453317294e-07,
+      "loss": 0.7502,
+      "step": 26570
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.913000701223202e-07,
+      "loss": 0.7382,
+      "step": 26580
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.8680921733757373e-07,
+      "loss": 0.7415,
+      "step": 26590
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.823440320560545e-07,
+      "loss": 0.7477,
+      "step": 26600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.7790452012128167e-07,
+      "loss": 0.7433,
+      "step": 26610
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.734906873431876e-07,
+      "loss": 0.7484,
+      "step": 26620
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.6910253949808517e-07,
+      "loss": 0.7384,
+      "step": 26630
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.6474008232868727e-07,
+      "loss": 0.7437,
+      "step": 26640
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.604033215440739e-07,
+      "loss": 0.7392,
+      "step": 26650
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5609226281970497e-07,
+      "loss": 0.739,
+      "step": 26660
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.51806911797401e-07,
+      "loss": 0.7391,
+      "step": 26670
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.475472740853403e-07,
+      "loss": 0.7352,
+      "step": 26680
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.433133552580503e-07,
+      "loss": 0.7362,
+      "step": 26690
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.391051608563989e-07,
+      "loss": 0.7474,
+      "step": 26700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.3492269638759176e-07,
+      "loss": 0.7354,
+      "step": 26710
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.307659673251595e-07,
+      "loss": 0.7399,
+      "step": 26720
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.266349791089529e-07,
+      "loss": 0.7369,
+      "step": 26730
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.225297371451408e-07,
+      "loss": 0.7404,
+      "step": 26740
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1845024680619007e-07,
+      "loss": 0.7451,
+      "step": 26750
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1439651343087683e-07,
+      "loss": 0.7424,
+      "step": 26760
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.103685423242597e-07,
+      "loss": 0.7387,
+      "step": 26770
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.0636633875769094e-07,
+      "loss": 0.7374,
+      "step": 26780
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.023899079687942e-07,
+      "loss": 0.7432,
+      "step": 26790
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9843925516147123e-07,
+      "loss": 0.7372,
+      "step": 26800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.945143855058796e-07,
+      "loss": 0.74,
+      "step": 26810
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.906153041384441e-07,
+      "loss": 0.7438,
+      "step": 26820
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.8674201616183617e-07,
+      "loss": 0.7433,
+      "step": 26830
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.8289452664497453e-07,
+      "loss": 0.7367,
+      "step": 26840
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.790728406230092e-07,
+      "loss": 0.7389,
+      "step": 26850
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.752769630973329e-07,
+      "loss": 0.7485,
+      "step": 26860
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.715068990355496e-07,
+      "loss": 0.7371,
+      "step": 26870
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6776265337149277e-07,
+      "loss": 0.7361,
+      "step": 26880
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.640442310052027e-07,
+      "loss": 0.74,
+      "step": 26890
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6035163680292464e-07,
+      "loss": 0.7406,
+      "step": 26900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.5668487559710184e-07,
+      "loss": 0.7445,
+      "step": 26910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.530439521863781e-07,
+      "loss": 0.7422,
+      "step": 26920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.494288713355708e-07,
+      "loss": 0.7398,
+      "step": 26930
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.458396377756955e-07,
+      "loss": 0.7338,
+      "step": 26940
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4227625620391936e-07,
+      "loss": 0.7503,
+      "step": 26950
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3873873128359424e-07,
+      "loss": 0.7438,
+      "step": 26960
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3522706764422808e-07,
+      "loss": 0.7441,
+      "step": 26970
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3174126988148692e-07,
+      "loss": 0.7454,
+      "step": 26980
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.2828134255718171e-07,
+      "loss": 0.744,
+      "step": 26990
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.2484729019927265e-07,
+      "loss": 0.7411,
+      "step": 27000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.2143911730185152e-07,
+      "loss": 0.7417,
+      "step": 27010
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.1805682832515053e-07,
+      "loss": 0.7407,
+      "step": 27020
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.147004276955178e-07,
+      "loss": 0.7449,
+      "step": 27030
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.1136991980543086e-07,
+      "loss": 0.7399,
+      "step": 27040
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.0806530901347655e-07,
+      "loss": 0.7404,
+      "step": 27050
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.047865996443532e-07,
+      "loss": 0.7438,
+      "step": 27060
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.0153379598885748e-07,
+      "loss": 0.7461,
+      "step": 27070
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9830690230389082e-07,
+      "loss": 0.7495,
+      "step": 27080
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9510592281243968e-07,
+      "loss": 0.7421,
+      "step": 27090
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9193086170358643e-07,
+      "loss": 0.7392,
+      "step": 27100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8878172313248067e-07,
+      "loss": 0.7402,
+      "step": 27110
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8565851122035904e-07,
+      "loss": 0.7453,
+      "step": 27120
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.825612300545232e-07,
+      "loss": 0.735,
+      "step": 27130
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7948988368834408e-07,
+      "loss": 0.7369,
+      "step": 27140
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7644447614124427e-07,
+      "loss": 0.7383,
+      "step": 27150
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7342501139870904e-07,
+      "loss": 0.748,
+      "step": 27160
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.704314934122686e-07,
+      "loss": 0.7387,
+      "step": 27170
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.674639260994937e-07,
+      "loss": 0.7493,
+      "step": 27180
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6452231334399993e-07,
+      "loss": 0.7348,
+      "step": 27190
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.616066589954346e-07,
+      "loss": 0.7424,
+      "step": 27200
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.587169668694699e-07,
+      "loss": 0.7419,
+      "step": 27210
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5585324074780972e-07,
+      "loss": 0.7416,
+      "step": 27220
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5301548437816726e-07,
+      "loss": 0.7357,
+      "step": 27230
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.502037014742763e-07,
+      "loss": 0.7384,
+      "step": 27240
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.474178957158756e-07,
+      "loss": 0.7415,
+      "step": 27250
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4465807074870885e-07,
+      "loss": 0.7404,
+      "step": 27260
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.419242301845225e-07,
+      "loss": 0.7369,
+      "step": 27270
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3921637760105022e-07,
+      "loss": 0.7335,
+      "step": 27280
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3653451654202622e-07,
+      "loss": 0.7362,
+      "step": 27290
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3387865051715854e-07,
+      "loss": 0.7372,
+      "step": 27300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.312487830021447e-07,
+      "loss": 0.7433,
+      "step": 27310
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.286449174386517e-07,
+      "loss": 0.7417,
+      "step": 27320
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2606705723432476e-07,
+      "loss": 0.7447,
+      "step": 27330
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2351520576276755e-07,
+      "loss": 0.7474,
+      "step": 27340
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.209893663635575e-07,
+      "loss": 0.7361,
+      "step": 27350
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1848954234222166e-07,
+      "loss": 0.7401,
+      "step": 27360
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1601573697024526e-07,
+      "loss": 0.742,
+      "step": 27370
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1356795348506088e-07,
+      "loss": 0.7335,
+      "step": 27380
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1114619509004831e-07,
+      "loss": 0.7407,
+      "step": 27390
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0875046495453012e-07,
+      "loss": 0.7432,
+      "step": 27400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0638076621376059e-07,
+      "loss": 0.7333,
+      "step": 27410
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0403710196893235e-07,
+      "loss": 0.7341,
+      "step": 27420
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0171947528716753e-07,
+      "loss": 0.7447,
+      "step": 27430
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.942788920150881e-08,
+      "loss": 0.7455,
+      "step": 27440
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.716234671092173e-08,
+      "loss": 0.7393,
+      "step": 27450
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.492285078029017e-08,
+      "loss": 0.7416,
+      "step": 27460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.270940434041198e-08,
+      "loss": 0.7408,
+      "step": 27470
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.052201028799224e-08,
+      "loss": 0.7361,
+      "step": 27480
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.836067148564109e-08,
+      "loss": 0.7428,
+      "step": 27490
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.622539076187153e-08,
+      "loss": 0.7449,
+      "step": 27500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.411617091109714e-08,
+      "loss": 0.7376,
+      "step": 27510
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.203301469362102e-08,
+      "loss": 0.7401,
+      "step": 27520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.997592483564465e-08,
+      "loss": 0.7405,
+      "step": 27530
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.794490402924793e-08,
+      "loss": 0.7414,
+      "step": 27540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.593995493240025e-08,
+      "loss": 0.7392,
+      "step": 27550
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.396108016894943e-08,
+      "loss": 0.7384,
+      "step": 27560
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.200828232862388e-08,
+      "loss": 0.749,
+      "step": 27570
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.008156396701716e-08,
+      "loss": 0.7342,
+      "step": 27580
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.818092760560113e-08,
+      "loss": 0.7367,
+      "step": 27590
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.630637573171061e-08,
+      "loss": 0.7325,
+      "step": 27600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.445791079854547e-08,
+      "loss": 0.746,
+      "step": 27610
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.263553522516396e-08,
+      "loss": 0.7448,
+      "step": 27620
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.083925139648727e-08,
+      "loss": 0.7466,
+      "step": 27630
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.906906166328164e-08,
+      "loss": 0.7364,
+      "step": 27640
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.732496834217172e-08,
+      "loss": 0.75,
+      "step": 27650
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.560697371562507e-08,
+      "loss": 0.741,
+      "step": 27660
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.391508003195878e-08,
+      "loss": 0.7501,
+      "step": 27670
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.224928950533059e-08,
+      "loss": 0.7428,
+      "step": 27680
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.0609604315736693e-08,
+      "loss": 0.7423,
+      "step": 27690
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.8996026609007265e-08,
+      "loss": 0.7412,
+      "step": 27700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.740855849681314e-08,
+      "loss": 0.7376,
+      "step": 27710
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5847202056645836e-08,
+      "loss": 0.7412,
+      "step": 27720
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.4311959331833075e-08,
+      "loss": 0.7412,
+      "step": 27730
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.280283233152327e-08,
+      "loss": 0.7395,
+      "step": 27740
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.131982303069437e-08,
+      "loss": 0.7357,
+      "step": 27750
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.98629333701317e-08,
+      "loss": 0.7429,
+      "step": 27760
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.8432165256454543e-08,
+      "loss": 0.7428,
+      "step": 27770
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.70275205620807e-08,
+      "loss": 0.7393,
+      "step": 27780
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.564900112525527e-08,
+      "loss": 0.7445,
+      "step": 27790
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.429660875002405e-08,
+      "loss": 0.7409,
+      "step": 27800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.297034520624243e-08,
+      "loss": 0.7491,
+      "step": 27810
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.1670212229575336e-08,
+      "loss": 0.7507,
+      "step": 27820
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.03962115214862e-08,
+      "loss": 0.7378,
+      "step": 27830
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.914834474924133e-08,
+      "loss": 0.7389,
+      "step": 27840
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.7926613545907753e-08,
+      "loss": 0.7383,
+      "step": 27850
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.6731019510348732e-08,
+      "loss": 0.7411,
+      "step": 27860
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5561564207217115e-08,
+      "loss": 0.7375,
+      "step": 27870
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.4418249166966447e-08,
+      "loss": 0.7404,
+      "step": 27880
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3301075885835413e-08,
+      "loss": 0.7414,
+      "step": 27890
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.221004582585007e-08,
+      "loss": 0.7382,
+      "step": 27900
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.11451604148305e-08,
+      "loss": 0.741,
+      "step": 27910
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.010642104637528e-08,
+      "loss": 0.7394,
+      "step": 27920
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.90938290798659e-08,
+      "loss": 0.733,
+      "step": 27930
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8107385840469006e-08,
+      "loss": 0.7422,
+      "step": 27940
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7147092619127504e-08,
+      "loss": 0.733,
+      "step": 27950
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.621295067256723e-08,
+      "loss": 0.7319,
+      "step": 27960
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.53049612232814e-08,
+      "loss": 0.7442,
+      "step": 27970
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.4423125459543941e-08,
+      "loss": 0.7435,
+      "step": 27980
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3567444535402818e-08,
+      "loss": 0.7445,
+      "step": 27990
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2737919570675606e-08,
+      "loss": 0.7415,
+      "step": 28000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1934551650947257e-08,
+      "loss": 0.74,
+      "step": 28010
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1157341827574552e-08,
+      "loss": 0.7343,
+      "step": 28020
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0406291117683875e-08,
+      "loss": 0.7388,
+      "step": 28030
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.68140050416011e-09,
+      "loss": 0.7415,
+      "step": 28040
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.982670935659966e-09,
+      "loss": 0.7397,
+      "step": 28050
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.310103326603092e-09,
+      "loss": 0.7381,
+      "step": 28060
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.663698557163201e-09,
+      "loss": 0.7388,
+      "step": 28070
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.043457473285831e-09,
+      "loss": 0.7467,
+      "step": 28080
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.449380886668355e-09,
+      "loss": 0.7417,
+      "step": 28090
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.881469574775533e-09,
+      "loss": 0.7335,
+      "step": 28100
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.339724280817304e-09,
+      "loss": 0.7385,
+      "step": 28110
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.82414571377543e-09,
+      "loss": 0.7382,
+      "step": 28120
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.334734548374631e-09,
+      "loss": 0.739,
+      "step": 28130
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.871491425102569e-09,
+      "loss": 0.7399,
+      "step": 28140
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.434416950198749e-09,
+      "loss": 0.7416,
+      "step": 28150
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.023511695658954e-09,
+      "loss": 0.7326,
+      "step": 28160
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6387761992241467e-09,
+      "loss": 0.7333,
+      "step": 28170
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.280210964393792e-09,
+      "loss": 0.738,
+      "step": 28180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.947816460419194e-09,
+      "loss": 0.7372,
+      "step": 28190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6415931222990567e-09,
+      "loss": 0.7342,
+      "step": 28200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3615413507839237e-09,
+      "loss": 0.7401,
+      "step": 28210
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1076615123717381e-09,
+      "loss": 0.7416,
+      "step": 28220
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.79953939314504e-10,
+      "loss": 0.7394,
+      "step": 28230
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.78418929607183e-10,
+      "loss": 0.7403,
+      "step": 28240
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.030567469965775e-10,
+      "loss": 0.7358,
+      "step": 28250
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.538676209746683e-10,
+      "loss": 0.7399,
+      "step": 28260
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.308517467874971e-10,
+      "loss": 0.7328,
+      "step": 28270
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3400928542184333e-10,
+      "loss": 0.7416,
+      "step": 28280
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.334036361410612e-11,
+      "loss": 0.7446,
+      "step": 28290
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.884507384808387e-11,
+      "loss": 0.7405,
+      "step": 28300
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.234743527537944e-13,
+      "loss": 0.7479,
+      "step": 28310
+    },
+    {
+      "epoch": 2.0,
+      "step": 28312,
+      "total_flos": 4.0304188158043765e+23,
+      "train_loss": 0.8292061477166914,
+      "train_runtime": 175162.7697,
+      "train_samples_per_second": 41.38,
+      "train_steps_per_second": 0.162
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 28312,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 3000,
+  "total_flos": 4.0304188158043765e+23,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}